ELECTRONIC COMMERCE:
CONTINUALLY ADAPTING TO THE MARKET

Information technology is altering the way businesses are conducted – organizations are rethinking their strategy, policy, market demographics, service quality, and innovation. The Internet trading increasingly pulls producers and consumers on-line providing opportunities for new products and services. Employees are able to add value through innovation, data mining and information handling. Information content, accuracy, format, ease of use, and timeliness are continually improved affecting interorganizational relationship, supply chain, demand assessment, pre- and postsale services, and integration. This panelist will describe how electronic commerce has altered the market place. Challenges and opportunities provided by this technology will be discussed.

Leong Kai Hin, Universiti Malaya
Chan K. Thim, Universiti Telekom

This paper begins with an introductory background of the financial crisis that led to exchange control measures. There were many reasons that such a large magnitude of economic uncertainty was created in this part of the world. This "contagion effect", so called the worst financial crisis of the century is far from over and if not contained, will result in more damaging and shocking effects on the economies of these nations; in particular, Malaysia.

Section 2 explains the exchange control mechanism in terms of exchange rate risks and short-term capital risks. These two factors were the major cause of the economic calamities that spread like wild fires across the Southeast Asian nations. This type of risk factors was rather difficult to evaluate and subject to broad classifications. The exchange rate mechanism in Malaysia is not unlike those of its neighbouring states and the usage of risk management is very limited, probably in only a few financial instruments.

In section 3, there emerged empirical evidences from macroeconomics data that project the working (effectiveness) of these exchange control measures, which formed the basis for policy determination. Although implemented in less than six months, positive signals of recovery have surfaced to support the selective exchange control measures.

With references to Singapore (& Brunei) and China, section 4 examines the effectiveness of the exchange control measures with comparison and contrast as to draw peculiar trends of such policies to be recommended in the future. Recommendations for policies must consider the risk impact on global finances in order to avoid the "financial flu" attacking again.

Finally, the conclusion summarized the consequences of exchange control measures that have affected global finances. The key to understand the exchange control measures is pertinent in order to manage risk successfully and to reduce aftershock effects of the financial crisis.

BANKS’ EXECUTIVE COMPENSATION AND RISK MANAGEMENT-A FUNCTIONAL OR
DYSFUNCTIONAL RELATIONSHIP?

The question of the possible adverse impact of the structure of executive remuneration packages on the risk management of banks, and hence the regulatory goals of stability and efficiency of the financial system is the subject of this paper. This topic has recently been reviewed by financial system regulators, such as the Bank of England (Davies, 1997), and gained recognition by formal legislative recognition in the USA. The Federal Deposit Insurance Corporation in passing legislation banning the linking of executive compensation to volume of lending in FDIC supervised institutions, has recognized the adverse effects of a management reward focus. Also there is a similar recognition evident in the introduction of new core principles of prudential supervision by the Bank of International Settlement (BIS, 1997).

The hypothesis that the structure of executive remuneration packages in the financial sector is linked to adverse risk behavior in financial institutions has its basis in a set of constructs regarding management behavior known as agency theory. An increase in risk profile of banks can increase shareholders returns in terms of capital gains. It can also increase the return from the exercise of executive options.

In Australia the practice of rewarding executives according to volume of loans generated fell out of favor after the Australian Financial System incurred a record level of bad and doubtful debts by 1991. Such executive remuneration contracts were rewritten to grant only executive options upon employment or as a bonus. Buyback clauses were inserted in the articles of association of Australian banks after the practice of granting an increasing number of executive options was introduced in the early 1990’s. These buyback clauses also increase the potential for increasing executive remuneration, although the original intention may have been to return surplus capital. Some Australian banks announce buybacks before opening of trade (CBA) - others after close of trade (NAB) during which times executive options can be exercised with notification the next day.

What is evident is that if the net effect of the restructuring of executive remuneration packages is to increase the price of bank stock at the expense of risk management, with eventual effects on stability and efficiency, then agency relationships are being distorted by the structuring of contractual obligations.

The argument that incorrectly designed executive remunerations contracts could increase the risk taking propensity of bank management is examined empirically in this paper by testing whether there was any significant similarity in the direction of movement between the degree of income received in the form of profit related bonuses or share options, to the firm’s overall risk profile, in particular credit risk.

Microeconomic indicators of the performance of Australian banks between 1973 and 1993 were used in this study. The ratios calculated were indicative of executive compensation and credit, interest rate, liquidity and leverage risk. Other indicators of changes in agency relationships were tested such as financing, investing and lending patterns, dividend yield, and the ratio of off-balance sheet assets to on-balance sheet assets.

Two crucial turning points are clearly observable in the history of regulation the Australian Financial System which facilitates an events study of this nature which attempts to test for significant differences. Accompanying the changes in regulation were major changes in the role of executive management in banks and hence in their agency relationships and remuneration. The concentration of assets in four major public listed banks also makes data collection and analysis transparent.

The purpose of both the parametric and non-parametric tests applied were to determine if there were significant differences pre and post changes in regulatory arrangements which also dramatically changed the ways executives were rewarded.

The results showed that credit risk deteriorated significantly while top executive remuneration increased significantly post the events selected as being turning points in the restructuring of agency relationships between management and shareholders. Also the results do not show dividend yield changing significantly, but they do show a dramatic increase in both the amount and type of executive remuneration.

The trend in Australia and elsewhere to buybacks and executive options would appear to refute any claims that shareholder rewards are the exclusive motivation for bank management behavior. The results of this study have implications for formal legalization of control parameters to executive compensation, particularly when executive options are used in direct conjunction to buyback programs.

This paper is divided into seven sections – the first two reviewing the literature and formulating hypotheses, the third explaining data collection, analysis and testing and the fourth listing and interpreting the results. The final sections interpret the implications of the results in terms of the hypothesis, analyze why some banks outperformed or underperformed and finally draw conclusion regarding appropriate structuring of executive remuneration in financial institutions in order to promote agency relationships.

THE EFFECTS OF BACKGROUND RISK ON OPTIMAL PORTFOLIOS

The aim of this paper is to develop an analytical framework for investment decision making in globalize international financial markets. To this end we will use some results coming from the area of risk theory. We will also show how this framework can be applied to the particular case of noisy markets.

Our attention will be focused on the dependence between the risky asset and the background risk, which will be viewed as noisy risk in our applications. Background risk might arise due to any reason like war, earthquakes, moral hazard, informational asymmetries, nonmarketable assets (human capital, irreplaceable commodities,...), cross border risk, market risk, political risk, inefficiency.

In our framework the background risk will be an additive risk rather than a multiplicative risk. In our model, we consider a risk-averted investor, a risky asset, a riskless asset, and an additive background risk. The investor is supposed to maximize the expected utility of a von Neumann Morgenstern utility function.

First of all, we define the dependence between the risky asset and the background risk in several ways :

• the background risk is stochastically decreasing or increasing in the risky asset

• the risky asset and the background risk are positively or negatively likelihood ratio dependent- the background risk is positively or negatively dependent on the risky asset in the sense of the first, the second or the third stochastic dominance criteria

We point out the effect of the dependence between the two sources of risk on the demand for risky asset. We find that it is not necessarily true that the individual invests only in the riskless asset when the risky asset is fair, where fair means that the expected return of the risky asset is equal to the risk-free rate. In fact, the demand for risky asset is positive when the background risk and the risky asset are negatively dependent. In this case high returns are likely to be accompanied by low values of the background risk. The background acts like a hedge-portfolio. When the risky asset is fair, the absolute value of the demand for risky asset is lower than the percentage of additional background risk per asset’s risk.

By studying the increase in the strength of the dependence, we give the ways to obtain diversified portfolio in both positive and negative dependence : the demand for risky asset is an increasing function of this strength when dependence goes from perfect positive dependence to perfect negative dependence.

And finally we link our framework with the Capital Asset Pricing Model in the special case of noisy markets. We show that the demand is shifted in order to obtain a diversified portfolio because the investor willingness to buy marketable risk is modified by the presence of the background risk. The systematic risk of any risky asset is given by the usual beta added with a corrective term.

OWNERSHIP STRUCTURE, RISK, AND PERFORMANCE:
AN EMPIRICAL INVESTIGATION IN TURKISH COMPANIES

The relationship between equity ownership structure and firm performance has become a key issue in understanding the effectiveness of alternative corporate governance mechanisms. In light of massive privatization efforts in former Eastern block countries as well as experiences of developed economies of USA, Japan and Western Europe, researchers face vast amount of data to test various corporate governance issues brought out by the theory. In this paper we examine whether concentration of ownership and ownership mix have any impact on the risk taking behavior and performance of Turkish nonfinancial companies listed on Istanbul Stock Exchange (ISE). With public offerings of equity through IPOs, direct foreign investment and a large public sector in the economy, the Turkish market offers a very rich combination of corporate governance schemes to be compared. Moreover, privatization of publicly owned companies is still being debated on the basis of the impact of ownership mix on performance. A related issue surfaces with respect to the method of privatization. The merits of public offering of equity which leads to a more diffuse ownership versus private placement through block sales that results in a concentrated ownership is another controversy to be resolved. Hence we address ownership structure and ownership mix issues in the Turkish market in order to shed some light on this debate.

Our sample of companies includes 145 nonfinancial corporations listed on ISE as of the end of 1995. We excluded banks, leasing companies, investment companies, holding companies and insurance firms. Data on ownership structure and financial statements are obtained from ISE (1997). Market value data are provided in ISE (1996) and monthly bulletins of ISE.

We define ownership structure alternatively as the percentage share ownership of largest three stockholders and percentage owned by minority shareholders. Ownership mix refers to the type of majority shareholder(s). Hence we identify them as foreign owners, government and a conglomerate. In our empirical models, ownership mix variables are taken as dummy variables. We also employ control variables to account for differences in firm size and leverage. The results indicate that firms with concentrated ownership command a higher earnings multiple. Firms with government ownership have lower P/E ratios. Accounting measures of performance are not related with ownership structure and mix. In terms of risk taking, our findings reveal that highly concentrated and less diffuse companies have higher risk as suggested by larger standard deviation of monthly stock returns. Government owned firms in our sample display higher risk although they are larger on the average.

Findings with respect to ownership structure and mix are mostly similar to those reported in the literature for other countries. However, accounting measures of performance are a notable exception. There, we do not find any relationship between them and ownership measures and the signs of leverage and firm size as control variables are contrary to expectations. We attribute this anomaly to the distortion of balance sheets due to persistent inflation experienced in the country.

Another surprise result is the lack of any effect of foreign ownership and affiliation to a conglomerate. To an observer of Turkish corporate governance mechanisms, conglomerates and joint ventures should have certain advantages over other companies. Yet failure to disclose any such effects is an issue to be investigated further. One plausible explanation is the nature of our sample. As we include only the largest publicly traded companies in our sample, we suspect that large domestic companies with no conglomerate affiliation are not handicapped in terms of access to capital markets or procurement of projects, government incentives, qualified labor etc.

PRICE, VOLUME AND VOLATILITY SPILLOVER AMONG
NEW YORK, TOKYO AND LONDON STOCK MARKETS

This study objects to disentangle two possible interpretations: informational link hypothesis and contagion hypothesis by examining the effect of trading volume on inter-market dependence on stock returns. First, if correlation between international stock returns is caused by international contagion of liquidity traders’ sentiments or by resolution of heterogeneous interpretations of foreign news, such correlation will be positively influenced by foreign trading volume. Second, if international return interdependence is associated with the information containing in stock price changes in one market to another market, these interdependence are likely to be positively influenced by foreign price volatility but not by foreign trading volume. The use of trading volume enables us to assess the two possible channels of international transmission of international stock return and volatility by examining the causal relationship among the correlation of international stock returns, trading volume, and volatility.

In this paper, we examine the dynamic relationship among the U.S., Japan and U.K. daily stock market return volatility and trading volume using a multivariate generalized autoregressive conditional heteroskedastic (GARCH) model. The vultivariate GARCH model uses information from the history of more than one market. According to Conrad, Gultekin, and Kaul (1991), multivariate models provide more precise estimates of the parameters because they utilize information in entire variance-covariance matrix of the errors.

Data

The data set comprises daily market price index and trading volume series for the three largest stock exchanges: New York, Tokyo and London. For the New York Stock Exchange, we use the NYSE composite index. The data cover the period of January 2, 1970 - December 30, 1995, and consist of 6,568 observations for each series. For the Tokyo Stock Exchange, we use the Nikkei 225 index, which is taken from the PACAP database of the University of Rhode Island. The data cover the period of January 4, 1975 - December 30, 1995, and consist of 5,696 observations. For London, we use the FT-SE 100 index. The index covers the period of January 4, 1984 – December 30, 1995, and consists of 3,023 observations for each variable.

Method

We employ a three-step procedures developed by Gallant, Rossi and Tauchen (1992) to adjust for seasonal regularities. Then, we use these adjusted data to test for the robustness of the dynamic relations. In step one, we regress the original stock return series on dummy variables for day-of-the-week (one for each day from Tuesday through Friday), dummy variables for pre-holiday, dummy variables for turn-of-the year, and dummy variables for turn-of-the-month.

The following multivariate GARCH model is posited for the joint processes governing the daily rates of return for the Japan, U.S. and U.K. markets:

or

where the returns vector is denoted by r’_t . The residual vector is given by , with its corresponding conditional covariance matrix . e_t is represented by a column vector of forecast errors of the best linear predictor of r_t conditional on past information, denoted by , and including the P lagged values of r_t and trading volume. Where vech(.) denotes the column-stack operator of the lower portion of a symmetric matrix, is an vector of innovation, a isparameter vector, and b and c are matrices of constant parameters.

Empirical Evidences and concluding remarks:

Empirical results are consistent with the hypothesis that the cause of international transmission of stock returns and volatility is transmission of information from one stock market to another. Before the crash, the correlation between international stock returns might be caused by international contagion of liquidity traders’ sentiments or by resolution of heterogeneous interpretations of foreign news. With the development of communication technology, the more efficient stock markets become, the less contagion effect among national stock markets.

VOLATILITY ARBITRAGE IN FIXED INCOME MARKETS

Gunter Meissner, Hawaii Pacific University

In the fixed income markets, various derivatives can be replicated by other derivatives. This bears arbitrage opportunities. The article investigates three types of volatility-arbitrage in the fixed income markets: Cap-floor parity arbitrage, cap-floor forward volatility arbitrage, and bond option – swaption arbitrage.

1. Cap-Floor Parity Arbitrage

A standard popular type of arbitrage in the fixed income market is cap-floor parity arbitrage: Buying a cap and selling a floor with the same strike is equivalent to entering into a swap one period forward, where the strike is paid and the Libor rate is received.

If the floor volatility is higher than the cap volatility, as it is not unusual in the cap-floor market, an investor can buy the cap and sell the floor at the mid-market strike (zero cost) and enter into a long forward swap at the mid-market swap rate. To achieve this type of cap-floor parity arbitrage, only slight differences in the cap-floor implied volatility are necessary.

2. Cap-Floor Forward Volatility Arbitrage

Caps and floors are usually quoted as flat yearly implied volatilities. However, the volatility curve is usually inverted. So although we can derive the total value of the cap by pricing each caplet with a constant volatility, the first caplets actually have a higher volatility than the later ones.

The specific forward volatility, which is applied to each single caplet, is implicitly contained in the flat volatility curve and can be extracted by iteration. The paper suggests two methods, both based on Newton-Raphson, to generate the forward volatility.

If the flat volatility curve has a certain structure, especially being too inverted, arbitrage opportunities especially for long dated caps exist. Arbitrage opportunities can be found quite often in trading practice, as e.g. in August 1996 in the Japanese cap-floor market.

Cap-floor forward volatility arbitrage means that the trader buys the cap or floor at a lower price than the intrinsic value. This means that the cap has a positive theta and gamma. Therefore the trader will daily make money on the theta and while hedging also on the gamma, thus realizing a risk-less profit over time.

3. Bond option – Swaption arbitrage                                                                                                                              A payers-swaption, which gives the holder the right to pay a fixed rate and to receive a floating rate, can be viewed as a  put on a fixed rate bond with the strike equal to the par value of the floater. Also, a receivers-swaption, which gives the  holder the right to receive a fixed rate and to pay a floating rate, is equivalent to a call on a bond with the strike equal to the par value of the floater.

For the equations payer = put and receivers = call to hold, it is also necessary that the forward price of the bond option is equal to the value of the fixed side of the swaption (including a payment of the principal at swap maturity). Furthermore, the floating side of the swaption has to have a constant principal amount and the underlying swap has to start at option maturity T.

A difference between swaption and bond options is though, that in a swaption the geometric Brownian motion is modeled with the fair forward swap rate. Thus, the fair forward swap rate is log-normally distributed and prices are normally distributed. In a bond option, the geometric Brownian motion is modeled with the fair forward bond price, thus the forward bond price is log-normally distributed and rates are normally distributed. Consequently, the market prices bond options with bond price volatility and swaptions with rate or yield volatility. The elasticity based conversion from yield volatility to price volatility bears arbitrage opportunities due to different discounting used in trading practice.

For a cash-settled swaption, the expected swap cash flows are discounted with the mid market forward swap rate. For a swap-settled swaption, the expected swap cash flows are discounted with the forward discount factors, resulting from the swap curve. For a bond option, the discounting of the future cash flows is done with the constant yield of the bond.

Due to the different discounting, arbitrage between swaptions and bond option exists. The arbitrage is more likely, the steeper the yield curve and the higher the levels of volatilities and interest rates. Traders should check volatility arbitrage opportunities, which occur more often in non-mature markets.

ECONOMIC UNCERTAINTY AND CREDIT CRUNCH:
EVIDENCE FROM THE TURKISH BANKING SYSTEM

The most important role of financial institutions is the transfer of funds collected from surplus units as "deposits" to deficit units in the form of "direct loans." In recent years, the Turkish banks are being criticized by entrepreneurs and by the business community for staying away from a bank's true line of business and for decreasing the amount of loans extended. In response to this criticism, banks claim that the level of economic uncertainty makes it impossible to extend credit even to highly creditworthy firms.

The goal of this paper is twofold. First, it aims to document the changes in the amount and type of loans distributed by the Turkish banking system over the time period of 1986 to 1998, using monthly loan data. Second, the paper aims to see whether there is a relationship between the changes in the amount of loans distributed by banks and the level of economic uncertainty. This analysis is conducted in a multivariate regression framework after controlling for other factors that might affect the loan supply by banks and the loan demand by borrowers.

The balance sheet data for Turkish banks are obtained from the web site publications of the Central Bank of Turkey. It covers the period from 1986 to 1998. The inflation information is obtained from the International Monetary Funds’ International Financial Statistics Database.

In this paper, overall economic uncertainty is proxied by uncertainty of annual expected inflation, following Berument and Guner (1997). Berument and Guner (1997) show that interest rates increase as inflation rate or the variability of inflation increases.

Also, the conditional variability of inflation is estimated. Engle (1982) introduces the Autoregressive Conditional Heteroscedasticity (ARCH) model that allows the forecast variance to vary systematically over time. Bollerslev (1982) extends the ARCH model and models the conditional variance of the variable as a function of its own lagged values as well as the lagged values of squared residuals from the ARCH model. Following these developments, the conditional variance of inflation is estimated using a Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model.

REVIEW OF LITERATURE

Previous studies that address the issue of credit crunch generally define credit crunch as a significant reduction in the supply of credit available to commercial borrowers. In order to decide that an economy is indeed experiencing an episode of credit crunch, at least two distinct time periods are compared against each other to test the existence of a secular decline in the amount of total credit extended by the banking system.

Research conducted in the late 1980s and the 1990s has focused on the impact of introducing Risk-Based Capital Standards on the lending behavior of American banks. The main discussion in this literature is about distinguishing between two completely different types of reductions in the amount of credit extended by the banking system. One possibility is that there is a weak loan growth rate as a result of the normal procyclical pattern of both loan demand and the creditworthiness of borrowers before the economy goes into a recession. The other possibility is that there is a downshift in credit supply as a result of new and increased requirements on bank capital that are now tied to the amount of risk the bank takes on in the asset portfolio. Since it is rather difficult to determine whether the observed slow credit growth is a demand or supply phenomenon, the evidence regarding the existence of a credit crunch remains controversial today.

CONTRIBUTION

The contribution this paper makes comes from the question it tackles concerning the relationship between economic uncertainty and the amount of bank credit available in a given economy. Previous research has examined the economic costs of inflation volatility on real growth and investment but not on bank credit. In this sense, by measuring economic uncertainty as the volatility of inflation, this paper provides an added dimension to the discussion regarding the impact of inflation on the workings of an economic system.

INVESTOR SENTIMENT AND CLOSED-END FUND PUZZLE IN AN EMERGING MARKET

The shares of closed-end funds generally sell at a price lower than the net asset value (NAV) of the underlying asset portfolio in the U.S. markets. Since these funds invest in publicly traded securities, like stocks and bonds, and the U.S. markets are efficient, this discount, named as closed-end fund puzzle, is surprising. Lee, Shleifer and Thaler (1991) identify four components of this puzzle that are well documented in the U.S. for closed-end mutual funds. First, closed-end funds start out at a premium of almost 10 percent. Second, although they start at a premium, they move to an average discount of over 10 percent within 120 days from the beginning of trading (Weiss, 1989). Third, discounts on closed-end funds are subject to wide fluctuations over time. Last, when closed-end funds are terminated through either liquidation or an open-ending, share prices rise and discounts shrink.

Although the closed-end puzzle is well documented in the developed markets, there is no study that examines the behavior of the closed-end funds in the emerging markets. The Istanbul Stock Exchange (ISE) is one of the emerging markets which has been attracting the attention of international fund managers. The empirical studies examining the ISE found that the market is not semi-strong form efficient and the evidence on the weak-form efficiency of the ISE is inconclusive (Aydogan and Muradoglu, 1998; Balaban and Kunter, 1996). Furthermore, the ISE has been operating since 1986, investment in stocks is a new alternative and principles of portfolio management are not used well by Turkish investors (Yüce, Önder and Mugan, 1998). If portfolio managers are using modern portfolio management techniques, then investors should be willing to pay higher price for the portfolio that is already formed. Hence, zero or positive premium is expected in the closed-end funds traded on the ISE.

This paper has three purposes. First, the relationship between NAV and stock prices of eleven mutual funds in the ISE, an emerging market, is investigated for the period between July 1995 and June 1998. Second, the behaviour of the discount or premium on closed-end funds is examined over time. Third, the validity of investor sentiment hypotheses for mutual funds in an emerging market is tested for the ISE. The investor sentiment hypothesis makes three predictions about closed-end fund discounts. First, discounts on funds should be correlated with each other. Second, new funds should start when old funds are trading at a premium. Third, changes in the discounts on the closed-end funds should be correlated with the returns on portfolio of stocks that are not related with the funds.

The data used in the analyses come from two different sources. The portfolio holdings and NAVs of closed-end funds are taken from the weekly bulletins of the ISE. The closing prices of the closed-end funds and the ISE-100 composite index on Fridays are obtained from the Datastream.

The results show that the Turkish closed-end funds are sold at a discount as well. The average value-weighted discount is 12.88 percent which is slightly higher than the reported discount for the U.S. closed-end funds. These discounts range from -12.38 percent to 41.97 percent. For four of the closed end funds, a premium is found. Furthermore, these discounts fluctuate from week to week. In addition, the fluctuations on the average discount changes are more than the fluctuation in the return on the ISE-composite index over the period analyzed.

In order to test the investor sentiment hypothesis, ten decile portfolios are formed based on the market capitalization of stocks traded on the ISE-National Market. Stocks are assigned to a portfolio based on their market capitalization at the end of each year. Hence, the first portfolio consists of 10 percent of all stocks that have the smallest equity value in the ISE-National market and the tenth portfolio consists of all stocks that have the highest equity value in the ISE-National market. Continuously compounded weekly returns on the ISE-composite index are used as the market return.

The results show that discounts on funds are highly correlated. The average pairwise correlation of monthly discounts of individual funds is 0.3322. Out of 110 correlations, only eleven of them are insignificant and twelve of them are negative. The examination of the average discount on closed-end funds before and after the introduction of the funds shows that except two funds that were started during the sample period, all of the other funds are started when the existing funds are selling at a premium. Furthermore, all except two funds enjoyed, on average, a premium for the first four weeks of the fund introduction.

The comparison of discounts on ten size portfolios and market returns shows that there is a negative relationship between the changes in discount and the return on all portfolios. When the discounts decline, stocks do well. The coefficient on the change in the value weighted discount (VWD) is lowest for the lowest value portfolio and it is highest for the largest size portfolio. However, the coefficients do not increase monotonically with portfolio size.

The results of this study suggest that mutual funds in emerging markets trade at a discount as well and the size of the discount is similar to what was observed for mutual funds in the U.S. Investor sentiment hypothesis can explain the discounts observed for the Turkish mutual funds. These findings imply that the modern portfolio management does not seem to have much value in emerging markets or Turkish fund managers do not use modern portfolio techniques as well.

WHAT DRIVES EXCHANGE RATES? :
THE CASE OF THE YEN/DOLLAR RATE

The purpose of this paper is to examine the mechanism of dynamic multilateral interactions between the yen/dollar exchange rate and macroeconomic variables under the flexible exchange regime. We found the dynamic behavior of the yen/U.S. dollar exchange rate is almost self-generating in the short run. In the long run, however, the predictions by the monetary approach to exchange rate determination appear to be valid. Our finding is also consistent with the portfolio-balance approach to exchange rate determination: While the value of the U.S. dollar depreciates when there is a positive shock in the Japanese trade balance, the value of the U.S. dollar appreciates when there is a positive shock in the U.S. trade balance. It implies that the macroeconomic variables identified by the theories of exchange rate determination do matter in the long run for the yen/dollar exchange rate during the flexible exchange period, but not in the short run. It also implies that the policy coordination between the two countries to stabilize the exchange rate will have an efficacy in the long run, but will have a limited or no efficacy in the short run.

AN EMPIRICAL STUDY OF THE THAI BAHT

This paper seeks to find the long-run relationship between Thai Baht and the currencies of its five major trading partners and to provide an empirical explanation for the collapse of the Thai Baht in 1997. Han (1998) presents a model which implies that the real exchange rates of major trading partners should have a long-run relationship (or co-movement) in order to maintain either a stabilized trade balance or price level. The relationship of the Thai Baht and the currencies of its five major trading partners: Japan, the United States, Germany, Netherlands, and Singapore is examined. A vector autoregression (VAR) is used to estimate a system of interrelated real exchange rates and to analyze the dynamic impact of random disturbances on the system of variables. It is found that, between 1981Q1 and 1998 Q1, when the real exchange rate of the Japanese Yen (against the US Dollar) increases (the Japanese Yen depreciates), the real exchange rate of the Thai Baht (against the US Dollar) decreases. This could cause a trade deficit for Thailand. The cointegrating relationship between the above five real exchange rates is further investigated using canonical cointegrating regression (CCR) and Johansen's LR test. Results from both methods conclude that the real exchange rates (against the US Dollar) of Thailand, Germany, Japan, Netherlands, and Singapore are cointegrated.

Han (1998) presents a general equilibrium model to capture the relations between the choice of currency basket weights and the balance of trade for an economy. It is shown in the model that in order to keep trade balance of the home country unchanged, an appropriate choice of currency basket weights has to be made. The choice of currency basket weights then implies a long-run relationship between the currency of the home country and its trading partners. For example, when the currency of partner A against the numeraire depreciates by 1%, the home country currency against the numeraire should also
depreciate, according to the currency basket weights of partner A, by to maintain the competitiveness. The total impact from all currencies in the basket on the value of the home currency should add up to one. This implication can be used to test the overall competitiveness of the home country when there are changes in the real exchange rates of its trading partners. It can also be used to analyze the effect of exchange rate movement of the home country on its balance of trade. It is well known that the inevitable collapse of the Thai Baht in 1997 was following months of trade deficits and a deteriorating current account. The model developed by Han (1998) provides an empirical implication to examine the property of the Thai Baht that started the financial crisis in Asia and later expended to Latin America and Russia.

Two econometric methodologies are used to examine the long-run property of the Thai Baht. The non-structural approach of vector autoregression (VAR) is first applied to estimate the response of the home country currency (Thai Baht) when the values of its trading partners' currencies depreciate or appreciate. The VAR results are further examined to assess the cause of trade balance surplus or deficit in the home country. The impulse response analysis and variance decomposition are also performed to show the impact of an external shock to the long-term path of Thai Baht. The AIC criterion is used to determine the lag periods. It is found that the log level of the Thai Baht mainly depends on its own past values, which indicates the relative inflexibility of the Thai Baht responding to changes in the currency values of its trading partners. The results also show that log levels of the five real exchange rates are close to I(1) process. It is also found that when the real exchange for the Japanese Yen against the US Dollar increases (the Japanese Yen depreciates), the real exchange rate for the Thai Baht against the US Dollar decreases (the Thai Baht appreciates). This would make Thai products relatively non-competitive, could cause a trade deficit with Japan, and then causes an increase in the deficit of its current account. The impulse response functions and variance decomposition are calculated using the ordering: log DM, log Guilder, log Yen, log the Singapore Dollar, and log Baht. It is observed that an increase in the log level of the Deutsche Mark and the Singapore Dollar is associated with an increase in the log level of the Thai Baht for the first two to three quarters, but this increase is reverse afterwards. Within about 12 quarters, the shock has almost disappeared. When there is a positive shock to the log levels of the Japanese Yen, there is a contemporaneous decrease in the log level of the Thai Baht. This effect reaches zero within about three-quarters, but reappears again at the fourth quarter and persists over the next eight quarters. When there is a positive shock to the log level of the Dutch Guilder, there is no immediate effect on the Thai Baht for about two quarters, and then a positive effect on the Thai Baht re-emerges at the third quarter and persists for the next nine quarters. It means that the Thai Baht has about six months lag to catch up with Guilder's depreciation. From the analysis of variance decomposition, it is found 72.5% of the variation in the Thai Baht, considering one period (quarter) horizon, comes from its past fluctuation, followed by the changes in the Singapore Dollar, the Deutsche Mark, the Japanese Yen, and the Dutch Guilder. Among the major trading partners, the Thai Baht is most sensitive to the change in the Singapore Dollar.

The long-run relationship between real exchange rates of trade-related countries is also investigated using the concept of cointegration. If a country pegs its currency to a basket, there should be a long-run relationship between the log levels of the real exchange rates (all measured against the numeraire) of countries involved. It means that the percentage change in the real exchange rates among trading partners should be cointegrated. The augmented Dickey-Fuller test is applied to test the nonstationarity of the log levels of real exchange rates. Johansen's LR test and the canonical cointegrating regression proposed by Park (1992) are used to test and to estimate the cointegrating vector.

After normalizing the cointegrating vector ()', it is implied in the model of Han (1998) that, if the trade balance of the home country is stabilized, the summation of the elements in the cointegrating vector ought to be zero. That is, the summation of the impact of all the currencies in the basket is negative one. If the summation of the elements in the cointegrating vector is negative, the percentage change in the home country's real exchange rate is less than the uniform percentage change in its trading partners' real exchange rates. It indicates that the home country currency is relatively inflexible, even comparing to the currencies chosen in its currency basket. The inflexibility will affect the home country's competitiveness in exports market. It may also cause international speculative attacks on the home currency.

THE DOLLAR, THE EURO, AND THE YEN AS INTERNATIONAL RESERVE
AND INVESTMENT CURRENCIES

Since EMU represents a currency area with a GDP level and a world market share comparable to the United States, it is widely expected that the euro becomes an important international currency. This paper suggests simple methods how to quantify the effects EMU may exert on the role of the dollar and the yen as the other major international currencies. Estimates presented indicate that indeed the euro may indeed lead to a significant decline in the market share of the dollar as an official reserve currency of central banks and as an investment currency of in private portfolios.

These effects hinge on the assumption that the euro achieves a reputation similar to the deutschmark before EMU started. The projected shifts can be expected to materialize only in the medium term so that no immediate abrupt changes in the demand for dollars will occur.

HOSTILE TAKEOVERS, BREACH OF TRUST AND THE CORPORATE HEDGING DECISION

This study examines the "breach of trust" problem associated with hostile takeovers in the context of the corporate hedging decision. A takeover threat exposes managers to the risk that their relationship-specific rents are appropriated by the raider. As a consequence, it reduces their incentives to engage in implicit contracting with company owners or to invest in firm-specific human capital. This has profound implications for the design of the corporate hedging program. Managers overemphasize the importance of short-term exposures and systematically undervalue the benefits of flexibility-based operative hedging using real options. It is shown that charter-based antitakeover measures may resolve this contracting problem in conjunction with golden parachutes. Statutory control of risk management and takeovers may, under certain conditions, achieve the same result.

The paper employs an incomplete contracting approach to examine how the threat of a hostile takeover affects management’s hedging decisions. The firm may reduce its exposure to financial risk with financial as well as operative means. The hold-up problem associated with the takeover threat reduces the incentives to carry out the effort-intensive acquisition of operational flexibility and, as a result, curtails the company’s ability to manage so-called operating exposures. In addition, operative hedging overemphasizes the benefits of diversification relative to operational flexibility. The takeover literature has advanced two explanations for the short-termism of management behavior. This study argues along the lines of Schnitzer (1995,1997) that the threat of rent appropriation forces managers to give short-term payoffs a greater role. It stands in contrast to Stein (1988) who argues that the takeover threat leads management to behave myopically and engage in short-term profit-boosting if the likelihood of an ownership change is negatively related to short-term performance. While both approaches yield the same behavioral outcome in the general takeover case, it can be demonstrated that the myopia hypothesis has very different implications for corporate hedging as it would rather support the excessive use of operative hedging in order to minimize short-term earnings volatility.

This study develops the missing link between the takeover and the hedging literature. Following Froot/Stein/Scharfstein (1993), we rationalize risk management on the basis of a capital market imperfections argument and add the corporate governance dimension to the literature analyzing the choice between financial and operative hedging instruments as well as real-option- vs. diversification-based operative hedging, in particular Chowdry/Howe (1996), Hommel (1999), Kogut/Kulatilaka (1994) and Mello/Parsons/Triantis (1995). The paper ties in directly with the ongoing governance reform debate in Europe and recent regulatory initiatives in Germany aimed at fixing minimum standards for carrying out takeovers and the supervisory board’s role in monitoring risk management. A critical evaluation of these regulatory changes concludes the paper.

DENATIONALIZATION OF GERMAN CORPORAT GOVERANCE – THE CASE OF HOECHST

Stefan Eckert, Otto-Friedrich-University Bamberg, Germany

In the economic literature a convergence of national corporate governance systems is predicted or even diagnosed. From the national perspective this development implies a transcending of national corporate governance systems, in other words a denationalization of corporate governance. The objective of this study is to analyze the process of denationalization of corporate governance for German public corporations and to explore the implications that arise for the internationalization of these companies.

Therefore on the one side it shall be analyzed whether the process of denationalization of corporate governance can be characterized in the case of German corporations as a gradual development or a process marked by radical changes. Furthermore the underlying factors that influence this process and the events that triggered changes in the denationalization process shall be explored. And it shall be judged whether there has been a radical change concerning the denationalization of corporate governance in the case of German companies in the 1990s.

On the other side, the consequences for the internationalization of companies shall be considered. Presumable implications drawn from previous theoretical and empirical findings are summarized in two hypotheses:

1. The increasing denationalization of corporate governance of German corporations induces a reduction in the relevance of international diversification at the corporate level.
2. The increasing denationalization of the governance of German corporations leads to an increasing relevance of internationalizing as a tool for internalizing internationally segmented markets.

As research method a case study design was chosen. An especially interesting research case seems to be Hoechst, a German company which is mainly operating in the pharmaceutical and the chemical sector. Case data was gathered from documents like annual reports, published interviews with company executives etc.

First empirical results are:

1. The process of the denationalization of corporate governance at Hoechst can not be characterized as a continuous process. Rather, it can be split into phases, which differ considerably according to the intensity of denationalization.
2. The most important "critical events" for the denationalization of corporate governance at Hoechst are: a) the taking over of a 24.9% stake of Hoechst by Kuwait in 1982, b) Jürgen Dormann becoming CEO in 1994, and c) the planned merger between Hoechst and Rhône-Poulenc, which has been announced in 1998.
3. A reduction in the relevance of international diversification during the phase of intense denationalization can not be found. Whereas product diversification has been reduced heavily, regional diversification seems to have been expanded.
4. Weak empirical evidence points to an increase in the relevance of internalizing internationally segmented markets.

CHANGING EXCHANGE RATE PROPERTIES
DURING THE EMS PERIOD?

The performance of the European Monetary System is still debated. Regarding microeconomic characteristics it has been claimed that the realizations of higher moments of exchange rate distribution would indicate increasing riskiness for agents. Other critics have argued that any progress of the EMS is fictitious as it has not performed better than other currencies over the same period. We analyze theses propositions by searching for trends in distributional properties of EMS exchange rates and comparing them to an outside benchmark. We find properties indicating decreasing risk. Moreover, this decline seems to be faster than for the world benchmark.

MULTIVARIATE STATISTICAL MODELLING FOR THE
CLASSIFICATION OF THE SHARES TRADED
AT THE IMKB AS TO THEIR AVERAGE EARNINGS

This study aims at developing multivariate models to classify the shares traded at the IMKB according to their average earnings. The use of the model that is successful in this classification will make it possible for the shares traded at the IMKB to be ranked within objective criteria. Though the brokers seem to rank the shares traded at the IMKB by considering various indicators, the univariate characteristic of this approach -the use of only one independent variable- produces a restrictive effect on its serviceability. The facts that the univariate approach yields contradictory results and that it does not have the capacity to evaluate all the features of a company and the relations between these features, and that these models’ capacity to predict and rank is lower than that of the multivariate models make the use of multivariate models more attractive for this purpose. Though in some researches are made some rankings based on the scores obtained through the attachment of subjective weight to several variables, this approach is hardly scientific because of the subjectivity it displayed in the handling of the selection of variables, the attachment of weight to the variables and the rankings according to the scores obtained. Hence, this study pursues a scientific approach in which several criteria are made use of, and aims at ranking the shares traded at the IMKB within objective and varied criteria.

In this study the IMKB earnings average is taken as the basis for the ranking criterion, and an attempt is made to develop a model that distinguishes between the 30 shares that have the highest average earnings and 30 shares that have to lowest average earnings. The success of the Multiple Regression Model, Multiple Discriminant Analysis, Logit and Probit that are the multivariate statistical techniques used in ranking (predicting financial failure) depends on how distinct the difference between the two or more groups to be distinguished. The job of the analyst gets easy when the difference is like the oppositeness of black and white. It gets difficult when it has gray tones. For example, the difference between the shares above and the shares below the IMKB earnings average is more abstract than the difference between the companies that went bankrupt and that didn’t go bankrupt, and the gray area in this definition is larger than that of the bankruptcy. Therefore, the gray area varies according to the selected definition, and the success of the model increases or decreases depending on the size of the gray area. While the size of the gray area is a significant factor in the selection of the definition, there is another factor, namely the appropriateness of the definition for the purpose. For example, when the brokers develop models to predict the performance of shares, they are expected to base the definition of financial failure on a definition that is also used here, such as earning below the average.

In the second stage, it has been decided on which independent variables to be used in model development, and the values of these variables have been calculated one by one for each of the companies. In this study, it has been decided to make use of financial ratios rather than mere accounting data. The reason for this choice is to overcome the effects of inflation on a large scale and have control over important variables, such as the size of the company, difference in sector and difference in risk. It is known that it doesn’t suffice to predict financial failure using only financial ratios, also qualitative variables, such as company news, are added to the analysis to test whether the usage of qualitative variables together with the quantitative ones, such as financial ratios, are improving the model performance or not.

At the end of the first two stages, the values of the dependent and independent variables were obtained, and in the third stage, the data table ready for analysis was checked for the last time and the ultimate form of the table was made. In this stage, decision was made on whether there was any outlier between the measured values.

Fourthly, models have been obtained through the use of Logit, one of the alternative multivariate statistical techniques.

The following picture is viewed after the models have been evaluated:

• General correct prediction percentage of the models does not differ a lot, whether the examined period is of 3, 6, 9, or 12 months. This shows that the financial panorama of the firms does not go through much change throughout the year.

• The fact that models involving small number of financial ratios are obtained for each examined period, shows both the serviceability of this type of analyses and the high correlation between financial ratios.

• Though the performance of the models obtained is found to be statistically significant, the fact that the predictive power of these models can not be more than 75 percent displays the inadequacy of using only financial ratios in these models. It is expected that, besides financial ratios, the use of qualitative data as an explanatory variable, such as company news, will add to the predictive power of the models.

SECURITY EXCHANGE COMMISSION IN BANGLADESH:
IS IT OPTIMALLY SETUP?

Introduction

Bangladesh has continued to maintain steady stability in macroeconomic front. In 1995-96, there was marked improvement in foreign exchange reserve (+ 32%), reduction in current account deficit (- 10.26%), improvement in revenue collection
(+ 10.95%), increase in investment (+ 9%) and further containment of rate of inflation (1.3%). The macro-economic stability has been lauded by multilateral donors and development organizations. Given the above scenario about the Bangladesh general economy, the natural question arises: why did the market crash in 1996? In Hoque (1997) paper, it was showed that the Dhaka Stock Exchange exhibited chaos reflecting a breakaway movement. The breakaway movement did take place to the downside subsequently. Why did the market move to a breakaway point? The paper addresses this question by analyzing primary historical documents and data collected from Security Exchange Commission (SEC) of Bangladesh and Investment Corporation of Bangladesh.

The primary thesis of this investigation is that the SEC of Bangladesh did not perform its role before and during the crash. It was setup for failure to begin with. A comparative analysis is drawn between SEC of USA and SEC of Bangladesh. Further, the analysis was corroborated with facts that are pending legal investigation.

Constitution and Function of Bangladesh SEC

The Securities and Exchange Commission (SEC) was established on 8 June 1993 under the Securities and Exchange Commission Act, 1993 in order to protect the interest of investors in securities, develop and regulate the securities market and ensure proper issuance of securities and compliance of the relevant law. Consistent with overall policies, SEC acts as a central regulatory agency performing wide range of functions covering the entire capital market including the proper issue of capital the establishment of fair trading practices and the close supervision of issuers, market and intermediaries.

The Board of SEC is the policy-making and oversight body, while the implementation and day to day regulatory functions are taken care of by the full-time Chairman and members. Thus the work and strategies of the Commission are to protect investors, foster investors' confidence, promote a healthy, active and properly administered securities market, facilitate capital information and inhibit fraud in the public offering of securities.

Is It Optimally Set Up?

As can be observed from above that the SEC of Bangladesh follows U.S. SEC in name and overall organizational structure. However, this is an ill adapted version of the SEC of USA. The SEC needs to be an independent, nonpartisan, quasijudicial regulatory agency with responsibility for administering the government securities laws. The purpose of these laws is to protect investors in securities markets that operate fairly and to ensure that investors have access to disclosure of all material information concerning publicly traded securities. The Commission also regulates firms engaged in the purchase or sale of securities, people who provide investment advice, and investment companies.

Biases of Bangladesh SEC

i) Political Bias

Five Commissioners sit on the SEC, with one designated as Chairman by the President of Bangladesh. All Commission members are appointed by the President, with the advice and consent of the parliament, Ministry of Finance. However, it does not follow US in restricting members to no more than three from the same political party.

ii) Inefficiency Bias

The Commission employs lawyers, accountants, financial analysts and examiners, investigators, economists and other professionals. From popular accounts it is revealed that SEC does not have enough expertise to conduct its business. The Commission is headquartered in Dhaka, and does not have regional or district offices in other cities, especially in Chittagong. More importantly, it does not have rigorous principal divisions to carry out its objectives. The Division of Corporation Finance, Division of Market Regulation, Division of Market Regulation, Division of Investment Management, Division of Enforcement, Office of Compliance Inspections and Examinations divisions a la its US counterpart are missing in rigor and /or structure, and are main reasons for turning a manageable problem into a full blown crisis. These divisions are hallmark of any SEC organization. These divisions found its expression in formal structures of U.S. SEC over a period of 40 some years. The case in point is Division of Enforcement which was created under U.S. SEC in August 1972, thirty eight years after the creation of SEC. This was created to consolidate enforcement activities that previously had been handled by various operating divisions. Thus Bangladesh SEC is beset with organizational bottleneck in addition to shortage of well trained personnel. Existence of SEC has raised complacency rather than efficiency and fairness of security dealings in Bangladesh.

Evidence of Mismanagement of Crisis

The SEC instituted a four member Enquiry Committee to look into alleged fraudulent acts and insider trading preceding the crash of 1996. Both Dhaka and Chittagong Stock exchanges experienced a bullish run from July to mid November 1996. During this period market capitalization, turnover and share price index increased by 265%, 1000% and 260% respectively. This bullish run was not justified by movement in fundamentals. Rather it was largely motivated by market manipulation and exploitation of uniformed and misinformed investors. In this respect the SEC has failed to carry out its basic duties. The executive summary of the Enquiry Committee states:

"SEC, instead of taking direct action against the wrongdoers. such as suspension of trading, canceling license of brokers, etc., took measures of corrective nature which failed, that too belatedly did not produce any positive results. SEC should have been more aggressive in taking punitive actions against the manipulators. Strong action could stop the debacles at the earlier stage and at a lesser cost to the investors."

The report goes on to illustrate the cases of flagrant violations of security rules by individuals and institutions. The delivery versus payment (DVP) mechanism was used as one of the main vehicle of manipulation. It is a system of settlement that allows the buyer and sellers of contracts to settle transactions directly without routing through the clearinghouse of the exchanges. The usual practice of settlement had not been followed and exchanges did not stop the violations. The report also cited inordinate delays in clearing applications for new IPOs by the SEC authorities thereby failing to ease bullish pressure.

Accounts from various sources including the medium of press confirm the findings stated above. It is also mentioned in popular as well financial press the SEC continues to suffer from paralysis of actions. The clearance for IPOs continues to be delayed. The disclosure requirement in a quarterly and annual basis for the existing companies are not strictly followed. Further, there is an enthusiasm for over-regulating appears to be prevalent among members of SEC. Until last year SEC used to fix the price of IPOs, a task that should never be under its jurisdiction. In a way the existence of SEC has created a false sense of security and complacency in the world of investment in Bangladesh. The uneasiness in governing SEC is reflected in the fact that SEC has seen three Chairmen in five years.

Given this complacency and inefficiency biases it is not surprising that the market moved away from fundamentals to a grossly overvalued situation warranting a crash.

Conclusion

Unexploited profit opportunities arise when the securities are mispriced. This happens when the prices move away from fundamentals. In this paper we posit that the movement away from fundamentals in Bangladesh Stock Markets essentially took place because of regulatory bias rather than irrational exuberance. Evidence of such bias can be traced to the inner working of the SEC. Because of its lack of knowledgeable personnel and rigorous structure, the SEC could not take appropriate actions in a timely manner to contain this movement. The result was a brutal adjustment precipitated by the market resulting in a crash to pre 1994 level. If short selling were allowed this would have been a great opportunity for profit and which in turn would have allowed a market correction. However, the crash made the ill informed and ordinary investors hanging with virtually valueless paper.

One of the lessons learned from the crash is that the SEC should be streamlined and made more visible and accountable. The process towards that will be to structure its inner mechanisms in accordance with the SEC of the U.S.

CHAOTIC SYSTEMS ANALYSIS OF MAJOR
STOCK MARKET INDICES

This paper considers deterministic chaos as a new test of weak-form market efficiency. Using the embedding theorem and the information dimension, it is demonstrated that the daily return of the stock price indices expressed in US dollars of different countries are either random or high-dimensional deterministic. Although 4,000 observations per market are used in the study, this figure is low relative to what is required in order to be able to further test the model. However, it does support the idea that thousands of data points are not enough in order to find regularity in the data, thus supporting the weak-form efficiency of financial markets in general and stock markets in particular. Since a low-order deterministic chaos is not found in the data, the weak-form market efficiency hypothesis is substantiated. The behavior of these stock indices cannot be predicted based solely on past price behavior.

The data set used in this study consists of daily stock market price indices for Canada, Europe 14, Europe Excluding the UK, the World Excluding USA, France, Germany, Japan, the UK, and USA. Except for the World Excluding USA stock price index, which includes data from January 1, 1982 to October 24, 1995, all of the daily stock price indices cover January 1, 1982 to September 5, 1997. Thus, all of the daily stock price indices consist of about 4,000 data points. For the purposes of this paper, we analyze the relative daily changes in the stock price indices expressed in percentages. These percentages are calculated from returns converted to U.S. Dollars.

The weak-form market efficiency hypothesis states that future securities prices cannot be predicted from current and past price and market information. This hypothesis suggests that investors cannot reliably earn abnormal returns merely by looking at this universally available information. Weak-form market efficiency has long been the subject of empirical scrutiny. The most basic test of the weak-form efficiency hypothesis is auto correlation. This test fits the time series of excess returns to a linear regression model. Auto correlation studies strongly support the weak-form market efficiency theory. Some studies do show, however, some small correlation between successive daily returns. Many of the authors of auto correlation studies have also conducted runs tests. A run is defined as two or more consecutive positive or negative changes. For a given time-series, these tests compare the expected number of runs, which is based on a random distribution, to the actual number. Consistent with a small positive relationship between successive one-day returns, Fama (1965) finds fewer runs than expected. For longer intervals, however, the number of runs is consistent with the number expected for a random series.

It is clear that the preponderance of traditional measures support the weak-form market efficiency hypothesis. It is possible, however, that price changes are governed by a process which appears stochastic given traditional methodologies but is actually deterministic. In the last few decades, researchers in the physical and biological sciences have recognized that certain low-dimension, nonlinear, deterministic systems can exhibit stochastic-like behavior. Such systems are termed chaotic systems.

While demonstration of the absence of determinism in the time-series will support the weak-form market efficiency hypothesis, the presence of determinism does not necessarily contradict the hypothesis since chaotic systems are extremely sensitive to small perturbations and may elude predictions. Recognizing the existence of deterministic chaos in economic data is important from both theoretical and practical points of view. From the theoretical point of view, knowing that a system is chaotic may assist in constructing mathematical models which provide a deeper understanding of its underlying dynamics. From the practical point of view, such a model may facilitate process's control and, in some cases, short-term predictions. The high sensitivity of chaotic systems to small perturbations makes long-term predictions impossible. Nevertheless, in some cases, short-term predictions within estimable error margins are not beyond the realm of possibility.

The investigation of chaos initially assumes that a time-series represents a deterministic dynamic system, with an unknown number of degrees of freedom, that is dense on an attractor. One assumes various embedding dimensions and constructs the phase space portrait. Subsequently, one computes various measures of the attractor such as (fractal) dimensions and Lyapunov exponents as functions of the embedding space's dimension. When the data represents a chaotic system, the attractor's dimension will initially increase as the embedding space dimension increases but eventually reaches an asymptotic value. When the data represents a truly random system, the attractor's dimension will continue to increase with the embedding space dimension.

Motivated by the study of linear systems, a frequent starting point in analyzing time-series is the construction of the power spectrum, which is equivalent to the computation of the auto correlation. The power spectrum may assist in the discovery of periodic or quasi-periodic behavior. In linear systems, modes in the power spectrum correspond to generalized degrees of freedom of the system, and broad-band power spectra are generated by an infinite-dimensional system. This is not true, however, in nonlinear systems; some low-dimension chaotic systems may exhibit broad-band power spectra.

The power spectra of the daily returns of stock price indices are depicted as a function of frequency -- all of the daily returns have a broad-band power spectrum, which implies lack of periodicity in the data. Although this type of power spectrum is consistent with random behavior, it is also common in many chaotic systems. The study has also determined that all the daily returns of stock price indices are nearly Gaussian. Although the Gaussian probability distribution is common in many stochastic processes, it is also exhibited by some chaotic, deterministic systems. The average mutual information is evaluated for the daily series of the returns of the stock price indices. Then the information dimension for the Canadian stock price indices is analyzed, by first computing the log of the average distance between the reference points and their p-th nearest neighbor as a function of the log of the number of selected points, k. Small values of k are then discarded due to their susceptibility to noise. Once these values are removed, the data forms nearly a straight line. The information dimension is the negative inverse of the slope of the line. The information dimension as a function of the embedding space dimension is then computed for various other stock price indices. The results of the study indicate that the daily returns of all the stocks' price indices are either random or a result of a high-dimension, deterministic process.

AN INTERNATIONAL ECONOMIC ANALYSIS OF FOREIGN DIRECT INVESTMENT
AND INTERNATIONAL INDEBTEDNESS

Saziye Gazioglu, University of Aberdeen, England
W. David McCausland, University of Aberdeen, England

This paper develops the micro-foundations of foreign direct investment and integrates this with a macro level analysis. In doing so, it highlights the importance of profit repatriation in generating different effects of foreign direct investment on net international
debt, trade, and competitiveness in developed economies compared to less developed economies.

AN INTERNATIONAL INVESTIGATION OF THE INFLUENCE
OF GROWTH OPPORTUNITIES ON FIRM LIQUIDITY

Working capital management is gaining increasing attention as an important factor in corporate performance. Analysts commonly assess the efficiency of working capital management by comparing liquidity ratios to benchmarks for companies in the same industry and country. There is evidence that working capital requirements differ across industries. For a sample of 1,181 U.S. firms during 1960-79, Hawawini, Viallet, and Vora (1986) find a significant industry effect on investment in working capital, suggesting that working capital policies conform to industry benchmarks. However, with increasing global competition spurred by integration of world markets, benchmarks representing efficiency levels in individual countries may be inappropriate. Relatively efficient companies in countries with generally inefficient working capital management may face a competitive disadvantage against mediocre companies from countries with more developed working capital management practices. It is important to evaluate working capital management across countries.

There is scant empirical evidence on differences in working capital policies and their determinants. Researchers have only recently started addressing some of these issues and have focused on cash management. Opler, Pinkowitz, Stulz, and Williamson (1998) and Kim, Mauer, and Sherman (1998) examine the determinants of cash holdings of U.S. firms from 1971-94. Their results show that firms with stronger growth opportunities and riskier activities tend to hold more cash, while those with greater access to capital markets hold less cash.

This paper studies trends in liquidity measures of companies in five countries spread across four continents, and investigates relations between changes in liquidity and growth opportunities. The countries, selected on the basis of availability of data on sufficiently large numbers of companies throughout the study period, are Australia, Canada, Japan, United Kingdom (U.K.) and U.S.A.

The data indicate that both the levels of liquidity measures and their trends vary across countries. Further, different liquidity measures rank countries differently. We find that real economic growth has a positive impact on changes in the cash ratio, quick ratio, and net working capital ratio, and a negative influence on changes in the current assets turnover. These findings, suggesting that firms adjust liquidity to reflect growth opportunities, caution against static comparisons of liquidity across countries, or even for firms in the same country. Since liquidity may be a strategic weapon, evaluation of working capital management must take into account differential growth opportunities available to firms.

Selected References

1. Baskin, Jonathan, "Corporate Liquidity in Games of Monopoly Power," Review of Economics and Statistics, 1987, 69, 312-319.
2. Hawawini, Gabriel, Claude Viallet and Ashok Vora, "Industry Influence on Corporate Working Capital Decisions," Sloan Management Review, 1986, Summer 15-24.
3. Kim, C., D. Mauer and A. Sherman, "The Determinants of Corporate Liquidity: Theory and Evidence," Journal of Financial and Quantitative Analysis, 1998, Sept. 335-.
4. Opler, Tim, Lee Pinkowitz, Rene Stulz and Rohan Williamson, "The Determinants and Implications of Corporate Cash Holdings," Journal of Financial Economics, 1998, forthcoming.
5. Pinkowitz, Lee and Rohan Williamson, "Bank Power and Cash Holdings: Evidence from Japan," Manuscript, Ohio State University, 1999.

FINANCIAL ANALYSIS OF EXPONENTIALLY INCREASING
INVENTORY HOLDING AND ORDERING COSTS

Sadik Cokelez, California State University, Dominguez Hills

 

This paper analyzes the total of annual inventory holding costs and ordering costs when holding costs and fixed costs of ordering and receiving increase exponentially and concurrently. The traditional model assumes that the holding cost per unit is constant during the entire year as well as the fixed order cost; but in real life such costs vary with inflation almost on a monthly basis. Especially, in those countries with soaring inflation rates such costs may even vary on a weekly basis.

This study develops the new total cost equation for the total of annual inventory holding costs and ordering costs under exponentially increasing holding and ordering costs. Modified notations for the varying holding costs and ordering costs are substituted into the traditional formula after finding the averages of these exponentially varying costs over a year.

The traditional total cost formula is based on the assumption that both H(annual inventory holding cost per unit) and C(fixed cost of ordering and receiving per order) are constant; but in real life it is plausible to assume that such costs increase in an exponential manner where the multiplying component is a^t(for the holding cost) or b^t(for the ordering cost).

In summary this paper contributes to quantitative analysis of inventory management related costs;the traditional total cost equation for the inventory carrying and ordering costs is modified to make it more suitable for real life operating situations by incorporating the possibility of exponential increases in the holding costs as well as in the ordering costs. The simultaneous changes in these two components are analyzed and a new total cost equation is derived.

This new total cost equation has implications on the new modified economic order quantity (EOQ) that is one of the major determinants of inventory management policies; this study can be extended by computing the modified EOQ on the basis of this new total cost equation.

CONGLOMERATE DIVERSIFICATION, FIRM PERFORMANCE
AND CORPORATE GOVERANCE

Akin Sayrak, University of Texas, Austin
John Martin, Baylor University

In this paper, we examine the relationship between corporate governance and diversification by analyzing the difference in buy-and-hold returns of conglomerates and pseudo-conglomerates (portfolios of focused firms matched by two-digit SIC classification). We identify conglomerates using the unrelated diversification component of the Entropy Measure and document a secular decline in the unrelated diversification of all firms over the period 1978-95. On average, conglomerates perform on par with the pseudo conglomerates. However, we find that firms with independent boards and low corporate complexity tend to over-perform their pseudo-conglomerate counterparts, whereas firms with dependent boards and high corporate complexity tend to underperform. Furthermore, the impact of corporate complexity is more pronounced for firms with independent boards. Our analysis indicates that an active market for corporate control has a positive impact on conglomerate performance. This suggests that external corporate control brings discipline and acts to the benefit of the shareholders.

ACCOUNTING DISCLOSURES UNDER ISLAMIC BANKING

One of the main objectives of financial reporting is to produce reports that fulfill users requirement and the figures presented reflect the true picture of the financial position of the organizations. For years, standards issued by the various professional bodies have guided the accountants and auditors in discharging their duties in preparing, presenting and auditing financial statements. In the case of accounting disclosures, the reporting requirements are usually governed by the International Accounting Standards (IAS) issued by the International Accounting Standards Committee (IASC). Although these standards of reporting are on voluntarily basis, many developed and developing countries choose to adopt the standards issued by the said body. By adopting these standards, the financial information prepared and presented are more acceptable and reliable.

As at to-date, IASC has issued standards that adequately cover the normal accounting policies and procedures of business entities worldwide. On the contrary, these standards are mainly meant for business transactions with no religious restrictions. Islam, for example, prohibits its followers from dealing with interest. Therefore, any organization established on Islamic principles must conform to this law and Islamic banks are one of the organizations, which fall in this category. Presently, there are more than 150 Islamic financial institutions operating globally and serve the banking need of the Muslim communities. The operations of these institutions are different with those of conventional banks because they do not associate themselves with interest. All deposits and lending facilities are governed by the Islamic banking principles. However, no attempt has been made by the IASC to promulgate guidelines or standards to deal with this kind of institutions. In the absence of standards for Islamic financial institutions, each bank has its own style of reporting its financial affairs. The Accounting and Auditing Organization have initiated the step toward standardizing Islamic financial reporting for Islamic Financial Institutions (AAOIFI). AAOIFI has just completed its initial works in establishing standards in the area of Islamic accounting principles and practices of disclosure. These standards, however, are yet to be fully adopted by Islamic financial institutions.

Another distinctive feature of Islamic financial institutions is that they usually have a special body call Shariah Supervisory Board (SSB) who will issue a statement that their operations are in accordance with Islamic principles. At present, even the reporting technique and format used by the various SSBs of Islamic banks worldwide differ. Each SSB has its own style of reporting.

This paper will discuss the accounting disclosure practices among the various Islamic banks in the world and the efforts by the AAOIFI to promulgate accounting and auditing standards for these banks. The reporting style and contents of the various SSB reports will also be highlighted.

BANK INTEREST RATES AND ADAPTIVE EXPECTATION OF
"DIVIDEND YIELDS" IN ISLAMIC BANKING:
A TEST OF RATIONAL EXPECTATION

Why study: At the start of 1997, there were 166 Islamic financial institutions that are designed on the basis of profit-sharing contracts in deposit and loan activities while also using the more common method of mark-up and lease arrangements. As is well known, conventional banks accept deposits and make loans based on pre-fixed mandatory interest rates. Pre-fixing of interest rates, despite being an odd method compared with most transactions (example share purchase, venture capital, etc.), has been justified in that this lets banks assume the risk of loans/deposits going bad. But Islam considers such contracts stipulating giving/taking of pre-fixed interest rates as one-sided, and thus unfair.

The question that arises is how do economic agents form expectations about the likely yield an Islamic deposits or loans at a future date since the yields are not pre-specified. Some writers note that, in an economy with conventional and Islamic banks existing side by side as they do in all but three countries, Islamic banks appear to use the conventional bank interest rates to determine the yield, the "dividend rate", offered by the Islamic banks. But this idea as to whether this is in fact the case has not been directly tested. The aim of this paper is to present evidence on this practical question.

Known knowledge: There are several instances of similar behavior in the mainstream economics. Two examples are: expected future sales affect the inventory decisions of firms; the unobserved permanent income may actually determine the consumption behavior of individuals, though individual’s income may vary much more erratically. This kind of situations are modeled as partial adjustment behavior, whereby individuals can be conceived as using revealed information, such as the conventional bank interest rates, to form expectations about the currently unobservable future dividend rates of Islamic banks.

Method of Study: Thus, we use the powerful econometric method of partial lagged adjustment model to test this behavior. The model requires that the data be available on both series. That is, we observe the announced dividend rates over time and the pre-fixed interest rates on deposits, investment funds and loans. If indeed bank management decides on dividend rates at a future date by following the interest rates at current times, then there ought to be a reliable statistical relation between the two variables.

Findings: Findings relate to how banks and credit unions determine the dividend rates. In the cases of Islamic banks, we report a strong relation between conventional interest rates and dividend rates. The conventional bank interest rate variables for savings, and term deposits are strongly influencing current dividend rates of Islamic banks. We also find the past dividend rate established by the bank is also a strong factor in the determination of the current dividend rate, as it should be. The latter suggests that the banks adjusts the past dividend rates by reference to the conventional interest rates. However, the evidence from the credit unions is just opposite meaning that conventional interest rates offered by the credit unions do not affect the dividend rates offered by the credit unions. This is a puzzle on which further research is needed.

DETERMINANTS OF BANK PROFITABILITY:
FINDINGS ON ISLAMIC BANKS

Sudin Haron, The Northern University of Malaysia

Why study: Researchers have managed, over the last three decades, to identify factors that significantly influence bank profitability. But such studies have been all about conventional banks. Today, there are more than 150 Islamic banks operating in 35 Muslim and non-Muslim countries with fund mobilized totaling US$80 billion. Except Pakistan, Iran and Sudan, which have converted the conventional banking entirely to Islamic banking, Islamic banks in other countries operate side by side with conventional banks. Therefore, it is interesting to know whether factors, which have influences on profitability of conventional banks, have similar effects on Islamic banks to add new findings to the body of knowledge in banking literature.

Known knowledge: Factors that influence bank profitability are divided into internal and external. Internal factors are further classified into two broad categories: (1) financial statement variables relating to the decisions which directly involve the items in balance sheets and income statements, and (2) non-financial statement variables such as number of branches, location, size of branches and banks, technology, and efficiency. The external factors verified as impacting on profitability are regulation, competition, concentration, market share, ownership, money supply, interest rate and inflation.

Method of Study: The approach used in this study is similar to those used in conventional bank studies. While almost all-external and internal factors, which have been shown to affect profitability, are examined, some variables such as number of branches, location, concentration and ownership were excluded in this study. New variables such as profit sharing and mark-up ratios with borrowers and profit-sharing ratios with depositors are included in this study because these ratios are only applicable to Islamic banks.

Findings: Since there were conflicting findings from studies of conventional banking, thus, the results of this study also in some cases confirmed the findings of several researchers while contradicting findings of others. This study finds that all three sources of funds for Islamic banks are positively related with profitability. Similarly funds invested in mark-up activities are positively related to profitability compared to activities based on profit-sharing. The profit-sharing ratio between banks and the borrowers seems to be very favorable to the banks, whereas the profit-sharing ratio between the banks and the depositors indicates a mutual advantage. In terms of expense management, this study offers no peculiar findings. While interest rates, inflation and size significantly impact on the profits of conventional banks, similar results were found for Islamic banking. In the case of market share and money supply, these factors were found to have an adverse effect on profits and these results are opposite to the findings of earlier studies. This study also found competition has no effect on bank profitability.

The findings reported in this worldwide study of Islamic financial institutions affirms that the unique factors of this class of banks affect profitability, and that such affects are based on economic rationale. Since this form of banking is only about 25 years old, we find there is need for more competition to improve the variety of services as well as standardize the Shariah ethical standards for audit of Islamic banks across countries.

ISLAMIC BANKING AND STOCK MARKET:
THE CASE OF IRAN

The relationship among businesses as providers of capital is a basic variable that shapes accounting system in each country. In the British-American accounting models, Choi and Mueller (1997) have argued that majority of financing capital is through stock market while in the continental accounting model it is through banks. Today, banks have done majority of financing capital in Iran. The role of Tehran Stock Exchange for mobilization of funds is limited. Islam permits profit sharing while prohibits interest (riba). The central requirement of the Islamic banking system is the replacement of the rate of interest with the rate of return on real activities as a mechanism for allocating resources. Islamic banks are considered as financial institutions and not as a monetary institution or intermediaries. In other words, they act in the capacity of an investor and not as a lender for carrying on trade, investment or service with the objective of generating profits.

In the Iranian banking system, the difference between expected profit and predetermined profit is not clear. When banks provide financial facilities to customers, the expected share of the bank's profit is predetermined based on an economic income. While the difference between profit and riba is the element of uncertainty in the expected profit, the share of profit that banks pay to depositors is lower than the inflation rate. In the case of hyperinflation, which exists in Iran, people do not have an incentive to deposit or investment in the banking system. Consequently, there is a huge amount of money circulating outside the banking system in search of profitable investment opportunities. Of course, this is contrary to the goals of Islamic banking which is supposed to eliminate the distance between money market, commodities and services markets. Harei (1994) has stated that in the Islamic view there is not recognition of purchasing power for lender, while according to Sadr (1984) maintenance f the purchasing power equivalent to the rate of inflation is permitted. According to Mutahhari (1985) modern banking in general and Islamic banking in particular is a new subject. Traditional contracts such as Musharaka (joint- venture or profit sharing) and Mudaraba (trust funding) and other ancient contracts are not sufficient for today sophisticated financial system. According to Behshti (1992) the best way from Islamic view for mobilization of funds is the stock exchange rather than banks. Despite the mobilization of funds, Islamic laws by stock market is more practical than banks the share of stock market for financing of needed capital is very limited in Iran.

REFERENCES

Beheshti, S.M. (1992). Islamic Economic, Fajar Press, Tehran, Iran.
Choi, F. and Mueller GG, 1997. International Accounting, 2nd Ed., Prentice Hall.

SYNDICATED LOANS

The market for syndicated loans is one of the fastest growing sectors of the global money and capital markets. Over $1 trillion dollars of new credit lines were extended to borrowers in 1997 alone. While finance research has focused extensively on loan sales, very little work has been done on the larger syndication market. Syndicated loans are also of interest from a financial contracting perspective, since they constitute a hybrid of public and private debt. These loans represent private debt in that they are shorter in maturity, contain more restrictive covenants, and are more likely to be collateralized than public debt contracts. The representative syndicated loan is sizable ($242 million in our sample), however, and is sold to investors in an underwriting process like public debt contracts.

Our research is focused on identifying the set of factors that influence the "salability" of debt contracts such as syndicated loans. We hypothesize that the characteristics of the borrower, of the originating bank, and of the loan contract itself all will influence the extent to which a loan can be sold in the syndication process. By sorting out these influences empirically, we provide further evidence on the significance of information problems and mechanisms for resolving them in financial contracting.

In a syndicated loan, two or more banks or financial institutions agree jointly to make a loan to the borrower. One lender will act as the managing agent for the group, negotiating the loan agreement, then coordinating the documentation process, the loan closing, the funding of advances, and the administration of the loan. This "lead bank" also typically obtains waivers and amendments to loan documents and holds all pledged collateral on behalf of the syndicate members. In effect, the syndicate members delegate monitoring responsibilities to the lead or agent bank which earns a fee for its services.

This contractual setting involves agency costs in the form of adverse selection and moral hazard. Although the borrower’s financial statements are made available to the syndicate members, the lead bank may possess valuable idiosyncratic information about the borrower. It might attempt to exploit its advantage by offering only low quality loans, yielding the prospect of a "lemons market." Also, as the lead bank sells larger proportions of a loan, its incentive to monitor declines, resulting in a moral hazard problem. We hypothesize that these agency problems can be resolved by a combination of monitoring activities, contract design features, and reputation factors.

We specify a model which predicts the percentage of a particular loan that can be sold to outside investors. Our dependent variable ranges form zero (a non-syndicated loan) to one (complete sale of the entire loan). For the 3400 loans in our sample, the average proportion sold is 68 percent. We relate the proportions sold to measures of the scope of information problems about the borrower (proxied by credit rating information or the existence of a ticker symbol), to the maturity of the loan, to whether or not the loan is collateralized, to the size of the loan, and to the reputation of the lead bank. Reputation is measured by computing the volume of "repeat business" between the lead bank and various syndicate members. Since our dependent variable is discrete, we use the tobit estimation technique. Our sample includes 3400 loans originated by U.S. and global banks over the period 1988-1996.

Our empirical results reveal that as the character of the information about an individual borrower improves, a larger percentage of the loan can be sold to investors. These results are consistent with findings in the loan sales literature which indicate that selling banks hold larger proportions of riskier loans in their own portfolios, perhaps as a signal of debt quality. Lead banks with stronger reputations can syndicate larger portions of an individual loan, suggesting that reputation can serve to attenuate adverse selection and moral hazard problems in the syndication setting. We also find that commercial banks are more successful at syndication than investment banks, which are relatively new entrants into the syndication market.

Our results on the role of the loan’s characteristics in affecting its salability are interesting. An increase in maturity enhances the prospects for syndicating a loan. This may reflect the fact that lengthening maturity reduces duplicative monitoring costs. Alternatively, longer-term loans are less likely to involve the capture of rents by the originating bank at the loan renewal stage, a phenomenon emphasized by Rajan in loan contracting. Since syndicate agents cannot commit to sharing these rents upon renewal, longer-maturity loans can more readily be sold in a syndication setting. The presence of collateral discourages the capacity to syndicate. Syndicate members apparently treat the presence of collateral as a signal of higher borrower risk and consequent potential for financial distress. The presence of multiple creditors complicates the loan workout process in the event of such distress.

In an expanded version of our model, we examine whether certain characteristics of the lead bank also influence its syndication behavior. We find that banks which are capital constrained syndicate larger portions of their loans. However, liquidity constraints and overall loan portfolio quality are not factors which influence the scale of syndication activity.

The fundamental contribution of our paper is to provide further empirical support for the hypothesis that the character and quality of information concerning borrowers affects the salability of debt claims. Where information is less than fully transparent, debt contracts tend to be marketed to investors with specialized monitoring skills who rely on contractual characteristics and seller reputation to resolve information asymmetry and agency problems.

TO COVER OR NOT TO COVER:
EMPIRICAL FINDINGS OF A POTENTIAL FOREIGN FINANCING

The formal introduction of the new "euro" currency is just the latest unmistakable evidence of a rapidly changing and increasingly global business environment. The steady and inexorable integration of the world’s money and capital markets, the continued expansion of world trade, the dramatic improvement in telecommunications, the ongoing revolution in managing information systems, and now, the introduction of the euro, have all dramatically changed how firms do business in the post-Bretton Woods era of floating exchange rates.

In particular, the question of how to finance both short-term working capital transactions and longer-term capital projects has become increasingly important. To make optimal financing decisions in this new business environment, multinational corporations must clearly understand certain key relationships between interest rates, exchange rates, and the forward market for foreign exchange. Failing to understand these key relationships could lead in the short run to unnecessarily high financing costs as well as lower profitability. In the long run, such deficiencies could result in the failure of the firm itself.

To evaluate certain key theoretical relationships, and to determine the lowest effective financing costs, data on spot and forward exchange rates were obtained from the Wall Street Journal. In addition, data on money market and Eurocurrency interest rates were obtained from The Economist. These data included observations for thirty consecutive, workable trading days for two different time periods (i.e., January, February and March of 1997 and 1998). Data for five major hard currencies (i.e., the Japanese Yen, British Pound, French Franc, Swiss Franc, and German Mark, with respect to U. S. Dollar) were evaluated to determine the following:

1. The Efficiency of the Forward Market and Interest Rate Parity. In an efficient market, observed premiums/discounts in the forward market for foreign exchange should accurately reflect existing (n-period) interest rate differentials. If this were not true, arbitrageurs would find risk-free profit opportunities. Assuming that interest rate parity holds, a critical question is whether an uncovered foreign financing will result in equal, lower, or higher financing costs compared to domestic financing.

To test the validity of interest rate parity hypothesis, 30, 90 and 180-day premiums/discounts on the five major hard currencies were compared to contemporaneous, de-annualized interest differentials (i.e., 30, 90 and 180-day rates). Because of the inevitable ambiguity as to which interest rate to use as the predictor of currency movements, this study tested differentials in both (n-period) money market as well as eurocurrency rates.

2. The Forward Rate as a Forecaster of Future Spot Rates. With respect to financing decisions, the consistent overestimation or underestimation of future spot rates by forward rates has important implications for multinational corporations. Specifically, if forward rates are not unbiased predictors of future spot rates, uncovered foreign financing may result in either higher or lower costs compared to domestic financing.

Result 1

Allowing for a number of practical considerations including transactions cost and taxes, the results generally support the interest rate parity and efficient market hypotheses for all five currencies.

However, based on the fluctuations of the spot rates, it appears these factors are not sufficient to fully explain the observed violations of the other important theory, the International Fisher Effect Theory. This theory suggests that the n-period interest rate differential between two countries should accurately predict the n-period appreciation/depreciation of the exchange rate between two countries’ currencies.

The Implications of Interest Rate Parity for Financing Decisions:

For both time periods evaluated in this study, and for all five major currencies, interest rate parity holds. This notwithstanding, the reverse is true for the International Fisher Effect. During both time periods, the forward rate either overestimated or underestimated future spot rates. Consequently, multinationals with open financing positions would have either benefited from, or been adversely affected by, foreign financing as oppose to domestic financing.

Result 2

With respect to Japanese Yen, in 1997 all forward rates overestimated future spot rates. Similarly, in 1998, forward rates for the Japanese Yen overestimated future spot rates (with the exception of a limited number of 30-day rates).

In the case of British Pound, in 1997 two-thirds of the observations for 30-day forward rates resulted in underestimations of future spot rates. The 90-day and 180-day forward rates for the British Pound were mixed. In 1998, three-fourths of all forward rates were, ex post, underestimations of future spot rates.

During 1997, the 30-day, 90-day and 180-day forward rates for French Francs overestimated future spot rates for all but a few of the 30-day rates. In 1998, the results were mixed. However, for the 90-day and 180-day periods, the majority were underestimations.

With respect to the Swiss Franc, forward rates consistently overestimated future spot rates for both 1997 and 1998.

Finally, for the German Mark, in 1997 forward rates generally overestimated future spot rates except for a small number of 30-day rates. In 1998, a simple majority of the 30-day forward rates overestimated the future spot rates while a significant majority of the 90-day and 180-day forward rates underestimated future spot rates.

Based on the above, one can conclude that multinationals could have benefited (on average) from borrowing Japanese Yen, French Francs, Swiss Francs, and German Marks during the 1997 period of this study. However, in 1998, foreign financing would have been worthwhile (on average) only in Japanese Yen and Swiss Francs. Interestingly, these latter currencies are the only ones which are not euro-related.

BANK MARKET VALUE EFFECTS OF EXPOSURE TO
EMERGING MARKETS: A RECONSIDERATION

The paper was begun while Gerald Hanweck was Visiting Scholar, Division of Research and Statistics of the Federal Deposit Insurance Corporation. Only the views of the authors are represented here and do not necessarily reflect those of the FDIC or its staff.

Introduction

The causes of the devaluation by Mexico of the peso on December 20, 1994, following closely after the inauguration of NAFTA, the rapid depreciation of the Thai baht and Indonesian rupiah in the Fall of 1997, and the Russian ruble devaluation by over 60 percent in 2 weeks in August 1998, followed by or following massive infusions of capital to avert an international monetary crisis, serve as a reminder of the financial and political instability of many of the lesser developed countries and emerging markets. Although the 1994 Mexican crisis was contained, the Indonesian crisis has not been, only serving to emphasize the fragility of these nations' economies and markets. As a consequence, lenders and investors to Mexico, Russia, Indonesia or other emerging market nations, must rely upon the timeliness and willingness of international organizations, central bankers of major industrialized countries, and the world’s bankers to provide funding to avert crises and support investor
value – bank or nonbank.

The experience of the LDC debt crisis of the 1980s provides substantial evidence that few, if any, long term solutions are viable. The swap of bank loans for security debt under the Brady plan – so-called Brady bonds – is a solution applied to a special case. Bankers with large LDC debt exposures were closely monitored by federal bank regulators while working out these bad debts through formation of a special loan loss reserve for LDC debts. During the period beginning with the 1982 Mexican default to about 1990 with the institution of the Brady Plan, U.S. banks reserved large LDC losses, showed poor profitability relative to others in the industry, and were the subject of intense supervisory scrutiny and capital adequacy forbearance. From this experience, there may be little reason for banks in the US or globally to be willing to accept large emerging markets risk exposures again without substantial guarantees before hand. With the extensive growth of capital markets in many nations, the vehicle for investment has expanded as have the risks arising from the remaining fragility of emerging markets countries. In other words, the growth of these markets may compound the problem if several nations suffer economic set backs simultaneously as happened during the 1980s (e.g., Mexico 1982, Argentina 1982, Brazil 1983, Venezuela 1983, and South and East Asia in 1997 and 1998, etc.).

The devaluation in emerging markets economies since July 1997 to the recent devaluation of the Russian ruble in August 1998 have taken a considerable toll on commercial and investment banking companies' market valuations. The largest have suffered considerable losses in market value due simply to the Russian devaluation and financial system collapse, and the uncertainty of possible spillover to Latin American emerging markets (e.g., Brazil in particular). By the end of September, money center bank stock prices have fallen 37 percent from their highs in early July 1998 (see the S&P Money Center Bank Stock Index figure) and 10 to 30 percent for some banks in the 2 weeks following the suspension of the conversion of the ruble. Some banks (e.g., Bankers Trust, Citigroup, BankAmerica and J.P. Morgan) are trading at or below their previous 12-month lows several weeks after the Russian financial system collapse. Although, most U.S. banking companies have manageable direct exposures to Russia, they have considerable exposure to other banks that do have considerably greater direct exposure to Russia (e.g., German banks). These banks have experienced a considerable fall in trading operation volumes and revenue because of the global market turmoil. Furthermore, data on the debt of emerging markets countries showed an increase in the spread over comparable U.S. Treasury securities doubling and in some cases tripling after the Russians announced suspension on their foreign debt on August 17, 1998 (Russian debt spreads went from 300 bp 3 months before to over 10,000 bp after the announcement). This evidence clearly suggests a tight relationship between major banking company market values and the events in emerging markets today.

This study attempts to systematically analyze the relationships between changes in banking company market returns and changes in emerging market debt risk premiums using current period and historical data on bank country exposure and emerging markets debt from 1993 to 1998. This paper estimates the changes in bank market value resulting from banks' exposure to emerging markets investments and lending. The study employs individual bank data on exposure by major country on a quarterly basis from 1993 to 1998 collected by the Board of Governors of the Federal Reserve System. These data have never been used in this form in any studies of bank market value. These data, combined with information on market value of securitized emerging debt, Institutional Investor Country Credit Ratings and information from bank Reports of Income and Condition (Call Reports), defines the factors determining the market value of banks. The market value of emerging market debt and the extent of exposure allows a theoretical estimate of the impact this exposure would have on bank market values. The hypothesis that is of most concern is that bank credit exposures to emerging markets risks may be no greater than their exposures to domestic risks. An affirmative finding to this hypothesis means that the ability to trade debts, directly or through securitization, may be the most effective way of reducing the risk exposures that banks and other lenders face in lending to high risk borrowers regardless of origin, domestic or foreign.

The model is estimated using seemingly unrelated regression (SUR), quarterly over the period 1993 to 1998 for a panel of 20 bank holding companies with the largest emerging markets debt exposures. Since many of the factors determining bank market value are correlated, more efficient estimates will be made using the SUR methodology compared with simple pooling cross-section and time series. The authors have used this methodology in other studies and have found it considerably more reliable than the simple pooling approach (Goldberg, Hanweck and Sugrue, 1992). Preliminary results strongly suggest that the changes in market value of emerging markets debts has been significantly felt by banking companies, particularly since mid-1997 with the advent of the global emerging markets crises. One interpretation of the results on rising risk-premiums for banking companies is that securitization has added to the transparency of banking companies, thus providing a mechanism for investors to evaluate the impact of the emerging markets exposure of banks. This greater transparency is of value in that through securitization of emerging markets debts financial markets are more able to value banks' exposure and better value and trade the risks.

Gulser Meric, Rowan University
Mitchell Ratner, Rider University
Ricardo Leal, Federal University of Rio de Janeiro
Ilhan Meric, Rider University

INTRODUCTION

The emerging equity markets have received considerable attention in recent years because of their high returns and their low correlation with the developed equity markets [see: Ratner and Leal (1996), Aggarwal, Inclán, and Leal (1999), and Meric, Ratner, Leal, and Meric (1999)]. However, although there are a number of studies that investigate the co-movements of the developed equity markets, there are only a few studies that focus on the co-movements of the emerging equity markets. The objective of this paper is to study the co-movements of the four largest Latin American emerging equity markets and the U.S. equity market.

DATA AND METHODOLOGY

Monthly index returns are used in the study to avoid obscuring the co-movements of the five equity markets caused by possible speculative leads or lags in daily data. The data for the U.S. equity market are obtained from the S&P 500 index. The data for the four Latin American equity markets are obtained from the General Index (Argentina), the IBOVESPA Index (Brazil), the IGPA Index (Chile), and the IPC Index (Mexico). The index returns are computed as the natural log difference in the indices, ln (I_it/I_it-1). Dollar returns are computed from local currency. Equity indices and exchange rates for each country are obtained from the Economatica database.

To determine any changes in the patterns of the co-movements of the equity markets, the sample period (February 1984 - February 1995) is divided into three 44-month sub-periods. The pre-crash sample (Period I) is measured from February 1984 to September 1987. The post-crash periods are November 1987-June 1991 (Period II) and July 1991-February 1995 (Period III).

We use principal components analysis to study the co-movement patterns of the U.S. and the four Latin American equity markets during the February 1984-February 1995 period. Principal components analysis is applied to each sub-period separately to study the changes in the co-movement patterns of the markets. We use Box’s M test [see: Meric and Meric (1989, 1996, and 1997)] to study inter-temporal stability in the co-movements of the five equity markets. The test is applied to pairs of consecutive sub-periods to determine the inter-temporal stability of the variance/covariance matrix of index returns

FINDINGS

Correlation between the markets increased considerably from Period I to Period II. This implies a reduction in the portfolio diversification benefits within the five equity markets after the October 1987 crash. The correlation coefficients are also higher in Period III than in Period II. This implies that portfolio diversification benefits within the five equity markets continued to decrease after the crash.

The number of statistically significant principal components is three in Period I, two in period II, and only one in Period III. This implies that the co-movements of the five equity markets have become considerably closer during the February 1984-February 1995 period. This also implies that portfolio diversification benefits with these five equity markets have diminished substantially during this period.

The M test results indicate that the variance-covariance matrix of Period II is different from the variance-covariance matrix of Period I at about 8 percent level. However, the variance-covariance matrix of Period III is significantly different from the variance-covariance matrix of Period II at the one-percent level. These results imply that there were more significant changes in the co-movement patterns of the five equity markets from Period II to Period III than from Period I to Period II.

CONCLUSIONS

Low correlation among national equity markets is often presented as evidence in support of the benefits of international portfolio diversification. Our findings in this study indicate that correlation among the U.S., Argentine, Brazilian, Chilean, and Mexican equity markets have increased sharply during the February 1984-February 1995 period. This implies that international portfolio diversification benefits with these equity markets have diminished considerably during this period.

We have found significant changes in the co-movement patterns of the five equity markets during the February 1984-February 1995 period. Our findings imply that it may be difficult to make good ex ante portfolio diversification decisions with the ex post co-movement information from these markets.

COMPARING DAILY RETURN VARIANCE AND THE RELATIVE LIKELIHOOD
OF EXTREME OBSERVATIONS IN RETURNS BETWEEN THE NYSE and JSE

Frank A. Michello, Middle Tennessee State University

Introduction

This paper examines issues related to market microstructure between the New York Stock Exchange and the Johannesburg Stock Exchange. The paper tests the theoretical model developed by Easley and O’Hara (1991) which hypothesizes that daily return variance is higher with a market maker system but the relative likelihood of extreme observations is higher with a floor broker system. The theoretical model is developed as in Easley and O’Hara (1991) where the specialist conditions his expected value of the asset on the observed type of trade from the book and the information he infers from buy and sell orders. As a Bayesian who updates his beliefs, his posterior becomes his prior as he sets subsequent prices and since the relationship between quotes and prices are linear in beliefs, prices converge to their actual value. On the other hand, the floor broker cannot observe the behavior of traders since he does not have the book from which to infer information and hence can only condition his beliefs on whether an order is a buy or a sell order. The absence of the book in the floor broker market implies that there is a tradeoff for the loss in market informativeness in the form of reduced price volatility. These theoretical predictions imply that daily return variance will be higher with a market maker system but the relative likelihood of extreme observations is higher with a floor broker system as the inability to observe the form of order flow can lead to episodes where executing stop orders lead to cascades and extreme movements.

Literature

Issues related to microstructure on developed markets have received considerable attention from finance researchers. For instance, research on the bid-ask spread has provided important foundations, which have assessed the impact of various market regulations on spreads and have enabled researchers to make public policy prescriptions. This preliminary research has also provided the foundation for a new branch of market studies, which have yielded new insights into effects of information on market behavior. More recent studies of this nature include analyses of the impact of dividend policy on spreads [Conroy, Harris, and Benet, (1990)], earnings announcements on spreads [ Chiang and Venkatesh, (19988)], and optimal transparency in a dealer market with applications to foreign exchange [Lyons, (1996)]. In addition, the study by Berkart and Harvey (1997) has shown the important role market volatility plays in determining the cost of capital and in evaluating investment and asset allocation decisions by making it possible for investors to compare market volatilities in their international portfolio. This paper adds to this literature by providing the first known empirical results of the Easley and O’Hara (1991) information asymmetrical model by examining their theory on the NYSE and JSE markets that differ in their microstructure. The results will contribute to our understanding of the role of market features in determining daily return variance. The paper also examines the impact of tax, political, and economic structural changes in South Africa and investigates their impact on the daily return variance, skewness, and kurtosis.

Data and Methodology

The data used in the paper is daily closing prices from the NYSE and the JSE. NYSE data is from CRSP master files and JSE data is from Ntobi Technologies CC database in Johannesburg, South Africa. Both data set cover the period 1/1/1991 through February 29, 1996. 248 firms on the JSE are matched to 1333 firms on the NYSE by mean dollar trading volume that differ at most by 5% to form 248 matched pairs between the two exchanges.

The paper uses Kruskal-Wallis test, Sign test, Z test for proportions, and the Levene test for equality of variance to examine the testable hypotheses.

Results

The findings of the paper provide evidence in support of the Easley and O’Hara (1991) hypothesis that daily return variance is higher with a market maker system but the likelihood of extreme observations is higher with a floor broker system. The results show that average daily return variance is three times higher on the NYSE than on the JSE. The difference in average daily variance between the two exchanges is statistically significant. The results also show that average skewness and kurtosis are higher on the JSE than on the NYSE. These results imply that the relative likelihood of extreme observations is higher on the JSE than on the NYSE. However, the differences in average skewness and average kurtosis between the two exchanges are not statistically significant.

The results also show that the means of variance of daily return variance, skewness, and kurtosis differ between pre-event and post-event periods for each structural change. The means of variance for daily return variance, skewness, and kurtosis all decreased after tax, political and economic structural changes with the largest decrease associated with economic structural change. The differences in the means of variance are statistically significant at .01 level of significance.

THE EFFECT OF SPLIT ANNOUNCEMENTS
ON CANADIAN STOCKS

The finance theory predicts that stock splits have no effect on the market behavior around split announcements. Splits would only have the effect of reducing share prices according to the split factor. Yet, empirical evidence in the US markets concludes that splits tend to impact the share price beyond the theoretical expectation. To the authors' knowledge, no such evidence has been documented around the split announcement dates for Canadian stocks. This paper is an attempt to fill the gap in the Canadian literature. In particular, this paper examines the market behavior surrounding stock split announcements in the Canadian market for the 1978-1993 period. Using CAR methodology, findings show that the capital market's reaction to split announcements results in positive cumulative abnormal returns for Canadian stock splits. Among the many theories proposed to explain why these abnormal returns exist, the Canadian evidence offers strong support to the liquidity hypothesis and, to a lesser extent, the signaling hypothesis. Tests of individual split events, however, imply that the portfolio results were driven by few significant individual split events as shown from. Also, not all split events draw positive market reaction suggesting that the split event has no one unique motivation.

Next, means ratios of several proposed explanatory variables measures from post-announcement to pre-announcements are tested. The results show that in the Canadian market, number of transactions increases in the period following the split. The increase is, however, statistically insignificant. Trading volume increases and the absolute bid-ask spread decreases following the stock split announcement (and the change is statistically significant at the 5 percent level). These results show that split events are likely to enhance liquidity, thus supporting the trading range hypothesis. The same findings (i.e., increase in trading volume and decrease in bid-ask spread) may also be the result of split announcements signaling good information about the stock, thus inducing brokers to ease the required spread. However, the long-term increase in number of transactions associated with the reported changes in trading volume and bid-ask spread favor the liquidity hypothesis. On the other hand, the tests of firm-specific variables support the signaling hypothesis. In particular, the increase in future earnings per share (although statistically insignificant) and the growth in firm size both observed following the split suggest the presence of a possible signaling role for split announcements.

Using regression analysis, the relationship between CAR and the tested explanatory variables is generally weak. Most independent variables used in the regression model lend little support to what is leading to the abnormal return during the announcement event period. It may seem that this abnormal return is based upon future expectations by the market rather than on the past history of the firm. Nevertheless, the decrease in the absolute bid-ask spread variable and the increase in trading volume after the split announcement are significant enough, thus supporting the hypothesis of improving liquidity. Similarly, the absolute bid-ask spread seems to be especially relevant for its strong predictive power of the split factor.

There are many avenues open for future research. A study of the microstructure effects around the announcement date may give new light to whether the bid-ask spread is actually changing and in what manner. One limitation of this paper is the small sample size especially for the earnings and dividend changes test. The conclusions (based on the 1976-1995 period) would be strengthened with the use of a larger sample of stock split announcements over a longer period of time, and with the use of other possible set of explanatory variables.

 AGENCY PROBLEM AND AN AMPIRICAL INVESTIGATION OF
THE PROBLEM ON THE FINANCIAL DECISIONS
OF THE ISTANBUL STOCK EXCHANGE COMPANIES^*

Agency theory derives from the conflict of interests among corporate managers, outside stockholders, and bondholders. These conflicts lead to the agency problem and therefore effect the cost of financing alternatives. Conflict of interest between debt holders and stockholders causes the agency problem of debt, whereas the conflict of interest between managers and stockholders of the firm leads to the agency problem of equity. Agency problem has an influence on the degree of financial leverage, and due to these considerations, agency theory claims that there is an optimal capital structure and it can be obtained by a trade-off between agency cost of debt and equity.

Previous studies, which have attempted to examine the factors influencing financial decisions of Turkish companies have tested MM hypothesis, tax, and bankruptcy considerations on financial decisions of Turkish companies. However, agency consideration of capital structure theory has been ignored in these studies and effects of agency problem on the financial decisions of Turkish companies have remained unknown.

This study attempts to adopt the theoretical approach to test the effect of agency costs on the financial decision issue of Turkish companies and presents a survey that examines if the agency problem is an important issue on the financial decisions of the Istanbul Stock Exchange (ISE) companies. The principal hypothesis adopted for this study is: "agency problem is an important factor on the financial decisions of the ISE companies".

Findings of the previous studies in capital markets of developed countries have shown that there was a relationship between the capital structure and the agency related variables, which are managerial stockholding, free cash flow, growth opportunity, asset composition, age of the firm, and firm size. In this study, the relationship between the capital structure and the agency related variables was examined in order to test the principal hypothesis, which is tested in two stages. In the first stage multiple regression model was used to determine if there was a relationship between capital structure and the agency related variables. In the second stage two-way analysis of variance was used to determine if the agency problem was the main source of this relationship.

The research covers the period of 1990-1996 and the sample of fifty five ISE companies, whose stocks were traded in ISE in the period of 1990-1996. The data were collected from the ISE yearbooks of the companies for the years between 1990-1996. Also, the data gathered from the ISO’s computer files were used. The yearly statistical publications on Turkish economy of the State Institute of Statistical (SIS) were the other sources of data.

The averages of the values between the period of 1990-1996 were used to eliminate the negative effects of the extreme observations on the analysis. Using the average values, the models which were developed for testing of the hypothesis, were cross-sectionally analyzed. The outcomes of the analysis were all tested at the significance level of 0.10.

Because of readily available data, the study was limited for the period of 1990-1996, and the ISE companies. Because privately owned companies don’t have that same financing opportunities as publicly held companies, the outcomes of the study can only be generalized for ISE companies.

The findings show that ISE companies can be characterized by mean total debt ratio of 27%, mean short term debt to total debt ratio of 63%, low public ownership ratio (36%), and management is dominated by the owner-manager or his/her family. Also one other main characteristic of ISE companies is that ISE companies rely heavily on short term bank loans as the main source of debt.

The relationship between capital structure and the agency related variables was tested by the multiple regression model. Results of the test show that there is a significant relationship between the capital structure and the agency related variables, and 38 percent of the cross-sectional variation in capital structure can be explained by the agency related variables. Except managerial shareholding, the regression coefficient of the agency related variables are significant. The relationship between capital structure and agency related variables is dominated by free cash flow, asset composition and growth opportunity. The relationships between capital structure and firm size, asset composition, and age of the firm have the same direction with theoretical expectations, whereas the relationships between capital structure and free cash flow, and growth opportunity divert from the expectations. The results of the regression analysis indicate that there is a relationship between capital structure of the ISE companies and the agency related variables, yet, whether the agency problem is the main source of these relationship is still an issue which must be further tested. Two way analysis of variance model was used to test if the agency problem was the main source of the relationship between capital structure and the variables which we found to have the same direction with theoretically expected, namely firm size, asset structure, and age of the firm. The results of the test show that the relationships between capital structure and firm size, asset structure, and age of the firm are statistically significant but independent from the agency problem.

Results of the study indicate that even though there is a relationship between capital structure of ISE companies and managerial stockholdings, free cash flow, growth opportunity, company size, asset composition, and age of the firm; this is independent from the agency problem. Due to the fact that the high managerial control of the owners over the company, bank credits being the major source of debt, and the tax incentives having greatest impact on financial decisions, agency problem is not an important factor on capital structure of ISE companies.

Optimal Hedging Strategy
and Industry Structure

The evidence provided by recent research in motivations for risk management activities of non-financial firms documenting shareholder value maximizing role of these activities is inconclusive. Most of the rationales brought forth in favor of hedging revolve around explanations such as lowering expected taxes, financial distress costs and mitigating the underinvestment problem. Theories departing from shareholder value maximization by recognizing agency problems demonstrate motivation for hedging by the managers at the expense of shareholders. Most of the empirical studies testing these theories do not control for industry membership. The need for scrutiny at the industry level was recently recognized in Tufano (1996,1998) using gold mining industry.

In this paper we explore the possibility that there might be industry specific reasons for the existence, lack of and the nature of hedging activities. Especially, since studies that use samples across industries have thus far produced few consistent explanations for hedging activities there is merit in isolating companies in the same industry under the influence of similar cost and pricing pressures. In particular we argue that the competitive structure of the industry influences the optimal hedging strategy at the firm level and under certain conditions it could be the overwhelming factor.

First explicit acknowledgement of the industry competitive structure as a factor influencing hedging strategy was made by Froot et al. (1993). Far from being the main thrust of the article --the last of the seven conclusions-- the authors predict that optimal strategy for a given firm depends on both the nature of product market competition and on the hedging strategies adopted by its competitors. Ross (1996) goes a step further and states the drastic effect the competitive structure of a firm’s industry has on the optimal hedging policy. It claims that in a competitive industry firms should not hedge the prices of their variable inputs. Doing so, states Ross, is another way of speculating in favor of high input prices. These statements indicate the possibility that under certain circumstances full hedging of input price exposure might actually increase the riskiness of the firm and/or harm shareholders even without the need for explanations involving self-interested managers. Negative implications of certain hedging activities by firms have also appeared in popular press (Stillwater Mining’s hedging policy for metals has hurt stock and profit this year, WSJ C2 11/10/97).

In this paper we drive optimal hedging strategies for hedging input/output price risks under competitive and oligopolistic competitive structures using linear/non-linear and partial/full hedging strategies as a response to hedging strategies of other firms in the same industry. The possible combinations result in a rich set of outcomes. We construct several numerical examples that demonstrate more significant implications. Later in the paper we test the predictions of the model using industry level derivatives use data.

In the original MM (1958) world hedging is a superficial activity. The assumption relaxed under this setup is "perfectly competitive markets." Let’s first look at an example where full hedging of the input price risk becomes a risky proposition: Consider a firm (F) that implements a full (linear) hedging policy of it’s main input’s price. Assume further that there are other firms (P) in the same industry that chose to partially or not to hedge their market risk exposure to the same input. In the traditional sense F seems to be completely isolated from the market price fluctuations of the input. Most popular explanations for hedging would imply that ceteris paribus firm F has enhanced shareholder wealth by decreasing the probability of financial distress, or by increased debt capacity or by minimizing expected tax payments. At the industry level, however, one can imagine situations in which everything else does not necessarily remain the same. If the input price actually goes up in the time period of the hedge, firm P has two choices. It can leave the output price unchanged and therefore suffer declining profit margins, or it can keep the profit margin the same by increasing the output price accordingly which would result in declining market share. F is able to price its output at a level below which makes P barely viable. However, if the input price changes in the other direction the result will be opposite. In this case exactly because of the nature of its hedging activity F will experience increased probability of bankruptcy and lower debt capacity. P has made a bet on the direction of the input price changes and it is now able to price its product at such a low level that F cannot match. It is clear that under the title of full hedging F actually made a bet equally risky as the unhedged firm P. This scenario would fit mostly to an industry that is under competitive pressures and/or the elasticity of demand for the particular product is not inelastic. If the industry competitive structure is something akin to an oligopoly and/or the elasticity of demand is inelastic, the price of the product will not necessarily be depressed and both types of firms will still make a profit. However, the firm that is on the wrong side of the bet will have a lower profit margin and will face a long-term disadvantage that may lead to declining market share and hampered growth opportunities.

The extreme scenario depicted above would have slight variations when we introduce partial hedging and more drastic variations when we allow for non-linear hedging strategies. It is intuitive that some addition of non-linear hedging would mitigate the inherent bet of the fully hedged firm by eliminating the undesired portion of the hedge (locked-in input price when the market price went down). In this case the hedged firm can still participate in the prospect for declining input prices at the same time eliminating the risk of increased input prices.

As an example consider the airline industry. In 1997 the transportation index and airline industry stocks in particular experienced significant returns. At the same time, jet fuel prices --one of the key input commodities for this industry-- declined more than 20%. We can expect that at least some of the firms in this industry engage in fuel price hedging. (According to Ross (1996) they do). If we assume that every 5% decline in the price of jet fuel the operating margin of an airline by 2%, an unhedged airline can significantly reduce ticket prices without sacrificing profitability. However, a hedged airline could not match that reduction in the long run, thus lose market share to the unhedged airline. It is possible to conceive a situation under these conditions in which the unhedged airline experiences fairly stable cash flows whereas the hedged airline would experience lower cash flows, an outcome which it hoped to avoid by hedging the input price risk in the first place.

The examples above demonstrate situations where decreased cash flow volatility as the main goal of a risk management program without regard for the competitive structure of the industry might produce the exact opposite result and increase the cash flow volatility especially in the direction that is most undesired.

OPTION PRICING WITH FUTURES-STYLE MARGINING:
A NEURAL NETWORK APPROACH

THE ISSUER'S DECISION OF IPO MECHANISMS:
FIXED PRICE VERSUS AUCTION

Since early 70s, it has been well documented that initial public offerings(IPOs) have positive initial returns. Later on the focus of many studies was to explain the IPO underpricing phenomenon, the explanations include information asymmetry (Rock(1986) and Baron(1982)), the insurance hypothesis(see, e.g., Tinic(1988)), the reputation hypothesis(e.g., Beatty and Ritter(1986)), signalling(e.g., Allen and Faulhaber (1989)), market overreaction(Ritter, (1991)), and the list goes on. Since early 90s, a new interest on the relationship between IPO mechanisms and IPO discounts emerges. Welch(1992) proposes a cascades model which argues that offering proceeds depend on the ordering of information among investors, and if a few early investors believe that the offering is a bargain, all subsequent investors will act alike. Beveniste and Spint(1989) and Benveniste and Busaba(1997) argue that new issues would be underpriced to induce asymmetrically informed investors to reveal what they know to the underwriter. While these papers concentrated on the information gathering and share allocation mechanism of the underwriters in the IPO process, Stoughton and Zechner(1998), and Mello and Parsons(1998) study the optimal IPO allocation strategy between small holdings and controlling blocks. Although the resulting optimal methods were different, both suggest that it may be optimal to favor large shareholders because of the value created by external benefits of block control. Brennan and Franks(1997), on the other hand, point out that underpricing is needed to ensure oversubscription and reduce the controlling blocks.

Although a plethora of explanations have been advanced to explain why there is IPO underpricing, few touched the issue of how issuing firms decide their IPO mechanisms. Among the few that did look into this issue, Dunbar(1998) studies the choice between best-efforts and firm-commitment offering methods, and shows that issuers select the offering method that provides the greater probability of success. Applying Welch's(1992) approach, Benveniste and Busaba analyze bookbuilding versus fixed price methods and suggest that the discount required in bookbuilding may be lower than that in fixed price method, but the proceeds of the bookbuilding has higher ex ante variability. Fabozzi, Mroan, and Ma(1988) examine the offering of public utility debt and find that competitive bidding has lower cost than negotiated offering only when market uncertainty is low.

Before 1995, fixed price method was the only legitimate IPO underwriting mechanism in Taiwan. Changes in the underwriting regulations in 1995 allowed the issuer to choose among auction, fixed price, and bookbuilding for IPOs. However, since bookbuilding is restricted to new shares offers, it is not common to observe bookbuilding in IPOs where shares going public are usually old shares sold by existing owners. Therefore, fixed price and auction offers are the two major methods in IPOs in Taiwan. Since 1995, the number of auction offers in IPOs has almost rivaled that of fixed price offers. How do firms choose between the two? Is there a difference in IPO returns under different methods? This issue is obviously important to the issuing firms as well as investors in IPOs.

While dwelling in depth on the reasons for IPO discounts, the existing literature has largely ignored the decision of the firm's IPO mechanism. In this paper, we attempt to analyze the decision of the issuing firm's IPO mechanism, in particular, fixed-price versus auction offers, which to our knowledge has not been discussed thus far. Unlike previous models that concentrate either on the information asymmetry problem from underwriter's perspective(e.g., Benveniste and Spindt), or on ownership structure(e.g., Brennan and Franks) in IPOs, we attempt to incorporate both factors into the determination of IPO mechanisms from the view point of the issuing firm. We assume that the objectives of the issuing firm are to maximize the proceeds as well as to avoid loss of control in IPOs. The decision of IPO methods is proposed as a two-step process, in the first step the firm chooses a method that maximizes the proceeds, however, if the issuing firm is concerned about loss of control, this concern dominates the final choice of the IPO mechanism and the firm will choose the fixed price method which has less of the control problem. Using a simple valuation model we argue that the issuing firm will prefer auction to fixed price offering when the net effect of asymmetric information premium as well as risk premium required by investors on the value of the firm is positive, and when the issuing firm is less concerned about loss of control. Furthermore, the concern for loss of control dominates the final choice of IPO mechanisms. We suggest that firms having lower risk and higher information asymmetry, lower board ownership, and better performance tend to choose auction over fixed-price offering, while firms having higher risk and lower information asymmetry, more concentrated board ownership, and poorer performance would choose fixed-price offering. Using IPOs samples from 1995 to 1998 in Taiwan, the empirical results mostly support our hypotheses. In addition, we find no evidence that IPO discounts differ in the two methods.

PORTFOLIO MANAGERS' AND NOVICES’ FORECASTS OF RISK AND RETURN:
ARE THERE PREDICTABLE FORECAST ERRORS?

Recent empirical research on the predictability of asset prices is based on two controversial hypotheses, explaining market behaviour. The efficient markets hypothesis argues that, in frictionless markets, and with random information flow, prices reflect all available information. Investor forecasts of prices are then, rational in the sense that they do not contain a predictable error component. In contrast, the overreaction hypothesis argues that price movements are not only driven by the flow of new information but also by the overreaction of investors who violate Bayesian rules in updating their beliefs about company prospects. Investor forecasts of prices are then, expected to be adaptive rather than rational.

Although, both the efficient markets hypothesis and the overreaction hypothesis have important implications in terms of investors' forecasts of stock prices, literature on stock price forecasts is mainly concerned with the accuracy of such forecasts. Behavioral explanations of the inconclusive results of market efficiency tests are few and attempts to model investor behaviour are limited.

This study aims to investigate biases in predictions of stock price series, by conducting a controlled experiment. The purpose is to explore possible biases in return expectations and risk perceptions by using point and interval forecasts provided for different forecast horizons. Return expectations and risk perceptions of "experts" are first compared to those of "novices" at bull and bear markets. Next, comparisons within subject groups are made to compare the return predictions and risk perceptions across different forecast horizons.

The major contribution of this study is to investigate the two domains that were not captured by previous research on the overreaction hypothesis; utilization of actual portfolio managers as forecasters and the real-time, real-world assessment in the form of forecasting the prices of specific stocks traded at the stock exchange. The general claim of previous research that investors predict stock prices, by extrapolating from past trends, with proper hedging, is not substantiated for all subject groups, all forecast horizons, and all forecasting tasks. Differences are observed in return expectations and perceived risks due to the presence of contextual information, the trends in the stock market and the participants' level of expertise.

Investors are positive feedback traders when they are exposed to a time series without being supplied with real-time information. However, contextual information, and real-time forecasting behaviour is different. Subjects extrapolate bullish trends and expect price reversals in bearish trends. This study, shows that the expectation formation process can not be generalized to one that extrapolates the trends while hedging them at the same time. Different decision processes may be at work at different occasions. Such an immaculate optimism can be interpreted as a behavioral explanation to the higher volatilities and observed inefficiencies in emerging market. As is the case for many emerging markets, higher volatilities may be due to the speculative behaviour of investors. This does not mean to ignore or underestimate fundamental factors. Rather, one should also acknowledge that market behaviour could also be influenced by investor behaviour.

The possible implications of this study for finance are two. First, the behavioral assumption of the efficient markets hypothesis that expectations are rational should be treated with caution. Skepticism about the rational expectations hypothesis is hardly new. However, clearly a lot needs to be done to examine how investors form their beliefs, to explain empirical findings. Apparent anomalies can be due to methodology is but one explanation. Melding psychological and financial research is necessary for a better understanding of the market mechanism in general and financial markets in particular. Next, risk perceptions might differ across investors of different expertise, across bull versus bear markets, and across real world versus simulated environments. Variations in risk premia should not only be attributed to stocks' being more risky in terms of traditional risk measures or changes in risk aversion but also to differences in risk perceptions. This is especially true in evaluating the thinly traded emerging markets where economic aggregates and indicators are different from those in mature markets. Further research in this area is expected to validate the relevance of these findings. Studies combining the knowledge structures and cognitive theories with the actual behaviour of economic agents in financial settings will help financial theory be based on more realistic assumptions and thus practitioners to work with better models.

THE NEED FOR A SUPRA-MEGA REGULATOR FOR THE NEXT MILLENIUM – IS THERE EVIDENCE?

This paper puts forward an argument for a new international approach to the regulation of financial systems. The approach is based on both inductive and deductive reasoning. The Asian crisis, which became evident in 1997, with concomitant contagion effects for vulnerable emerging nations such as Russia and Brazil, provides a fertile database for the application of a systems theory approach to the analysis of financial crises. This approach was first developed by the OECD in response to crises during the eighties (OECD, 1991, 1992). Similarly a taxonomy of regulatory models developed by Currie (1992,1997, 1998) aids in our understanding of both the problem and solution to coordination of a uniform approach to solving the regulatory problems associated with globalization, innovation and deregulation.

In response to the recent series of financial crises the IMF has issued proposals for reform of Asian economies. Underlying these reforms is the complete recognition that the regulatory model of weak prudential and a mixture of strength in protective measures is not the optimal. This is shown by the concentration on increasing the strength of the enforcement mode, compliance audits and sanctions. These proposals include an emphasis on reconstructing international regulatory arrangements by,

• Introducing more effective surveillance of countries and greater transparency;
• Heightening regional surveillance by the IMF and the World Bank, financial sector reform including improved prudential regulation and supervision, with more effective structures for orderly workouts, including better bankruptcy laws at the national level. Sanctions to include widespread publicity of those who fail to make the grade, with "countries conditioning access to their markets by foreign banks on a strong home-country supervisory regime" (Hartcher, 1998, p.10);
• Strengthening international financial institutions, including the IMF .

This plan has been expanded upon further by Robert Rubin, the US Treasury Secretary to include the establishment of global standards along the lines of the "Core Principles for Effective Banking Supervision" already established by the Basle Committee, particularly centered to strengthen national bankruptcy laws, accountancy standards, disclosure, loan classification and overall corporate governance.

Policies suggested by the IMF have been criticized for a number of reasons. All these concerns can be summarized in the effect on confidence. Stiglitz (1997, p.2) maintains that the macroeconomic fundamentals of the East Asian countries are still very strong with low inflation, high savings and an impressive skill base, but, "Confidence has been adversely affected by concerns about the health of the financial system, and about the substance and perceptions of transparency and governance".

At the heart of this whole debate is the question of whether there is an optimum in the design of a regulatory model governing a financial system, and an optimum in designing the appropriate response to the evolution of the financial system, whether reregulation or further deregulation. This conclusion also suggests a new plan for supervision and monitoring on a regional basis for the IMF and the World Bank to follow - having a rating scale based on an objectively derived taxonomy of regulatory models as expounded in this paper, with a preconceived optimum which could remove claims of political interference in dictating a regulatory model. However for this to be monitored so that financial aid could be allocated on the basis of the rating presupposes the establishment of either a mega regulator or regional regulators would need to be established. This in turn raises other questions –

• Is a vital complementary role to set prudential measures comprising the enforcement mode, the requisite compliance audit types and sanctions and to benchmark optimum protective measures?
• Or is the future of the financial system in the next millennium assured by regional mega regulators operating in a coordinated consistent manner but adapting their prescription of prudential and protective measures to the stage of development of the underlying political and economic sub-systems?

The paper is hence divided into a five-part analysis. The first part analyses the structure of national and international regulatory models governing financial systems. The second part examines the role of international reparation banks, the Bank of International Settlements and the Asian Development Bank, in achieving the national and international regulatory goals of stability, safety and structure. The third section reviews major financial crises 1987-1998 - preconditions and characteristics, with particular reference to emerging nations. This leads to a discussion of a new solution to the achievement of national regulatory goals - mega regulators concentrating on strong prudential measures in an age of liberalization of protective measures.

CHAPTER 11, OPTIMAL CAPITAL STRUCTURE
AND THE DECISION TO DEFAULT

Erwan Morellec, EDHEC and HEC, FranceSince the pathbreaking papers by Black and Scholes (JPE, 1973) and Merton (JF, 1974) Contingent claims analysis has been widely applied to the pricing of corporate securities. However, standard models of the levered firm fail to explain observed credit spreads, capital structure decisions and the phenomenon called early default (Jones, Mason and Rosenfeld (JF, 1984)).

Several explanations have been proposed to account for these discrepancies. These are: taxes and costly financial distress (Leland (JF, 1994)), deviations from the APR (Franks and Torous (JF, 1989)), strategic debt service prior to formal bankruptcy proceedings (Anderson and Sundaresan (RFS, 1996) or Mella-Barral and Perraudin (JF, 1997)), dynamic capital structure strategy (Fisher, Heinkel and Zechner (JF, 1989) or Goldstein, Ju and Leland (1998)).

However none of these developments has paid attention to the bankruptcy law whereas it should be regarded as an integral aspect of a debt contract. In standard pricing models, it is assumed that default leads to immediate liquidation (Chapter 7 of the US Bankruptcy Code) whereas for most firms default leads to reorganization (Chapter 11, private workouts). In this paper, we present a contingent claims model of the levered firm analyzing the ex ante effects of possible Chapter 11 filings on the value of corporate debt, capital structure choices and the decision to default.

In the US bankruptcy law, an overwhelming majority of defaults are voluntary, i.e. declared by the manager. Moreover, the manager who has the choice of the chapter filing, generally selects Chapter 11. Under this bankruptcy procedure, she keeps control of the firm during an observation period, which length is fixed by the Court. Coupon payments on pre-bankruptcy debt are postponed. Empirical studies document that:

- the average observation period lies between 2 and 3 years (Franks and Torous (JF, 1989), Helwege (JF, 1998));

- most firms emerge from Ch.11 (Weiss (JFE, 1990), Morse and Shaw (JF, 1988));

- firms that elect to enter Ch.11 incur few real economic costs (Andrade and Kaplan (JF, 1998), Maksimovic and Phillips (JF, 1998)).

We therefore adopt the following model. Since the aim of the observation period is to distinguish firms that are economically sound from those that are economically unsound (Wruck (JFE, 1990), White (1994)), our model relies on the following specification:

- the firm operates its assets until the default threshold is reached;

- upon default, it files Chapter 11 and the Court grants it a period of observation that is d units of time long. Coupon payments are postponed;

- if the firm’s assets recover value within these d units of time, the firm resumes operations and creditors are repaid in full (including the time value of money);

- if the firm’s assets do not recover value, the firm is liquidated and creditors lose the coupon payments on the last default period.

We then derive, using the martingale approach, the ex ante values of corporate debt, equity and the firm. The default threshold is endogenously determined so as to maximize equity value. It is shown to be an increasing function of the observation period. An outline for this result is the following: when d=0 (Chapter 7 situation), the manager sets the default threshold as low as possible in order to minimize bankruptcy given limited liability, and as the anticipated d goes to infinity, the manager sets the default threshold as high as possible in order to postpone coupon payments without any risk of liquidation. Between these two extreme cases, we have Chapter 11 situation and the manager makes a trade-off. At this default threshold, equity value is positive even when the absolute priority rule is enforced. Chapter 11 therefore rationalizes early default as an equity value maximizing decision.

The closed-form expressions we obtain for the values of equity and debt also allow us to investigate the effects of Chapter 11 on capital structure choices. Optimal leverage is obtained when the coupon level maximizes firm value. We show it is strictly decreasing in d. Chapter 11 therefore accounts, at least partly, for the low debt levels observed in practice. Our simulations exhibit realistic leverage ratios (between 30% and 50%) for reasonable input parameter values. Finally, credit spreads are also analyzed. Those generated by our model are compatible with those empirically observed (Sarig and Warga (JF, 1989)).

INTERESR RATE LINKAGES WITHIN THE EMS:
AN EXAMINATION OF THE GERMAN DOMINANCE
HYPOTHESIS BEFORE AND AFTER  1992

Eleven European countries that form the European Monetary Union (EMU) recently introduced the Euro, a common currency unit that will gradually replace the domestic currencies of those member countries. The main purposes of having the Euro, as generally stated, are to reduce the cost of currency exchange and to eliminate the exchange rate uncertainty among EMU countries. On the other hand, having a common currency means surrendering the ability to use monetary policy tools to influence a country’s domestic economy. As members of the European Monetary System (EMS), EMU countries have experienced some degree of exchange rate stability in the last nineteen years. However, the monetary rules of the EMU are stricter than the rules that applied to EMS. During the EMS years, member countries retained some degree of freedom in domestic monetary policy. Furthermore, if the domestic economy spun out of control, it was possible for a country to suspend membership, or even to get out of the EMS. Unlike EMS, there is no turning back after joining the Euro, and failing to fulfill the domestic economic requirements carries significant penalties for the member countries.

With the apparent economic and governmental differences between EMU countries, the viability of Euro remains in question. While one can only wait and see the future of the Euro, the researcher can also make predictions by examining the past behavior in the EMS system. Behavior of the interest rates of the member countries of the EMS reveals the extent to which the fixed EMS exchange rates, with bands that allowed for partial adjustments, influenced domestic monetary policies of member countries.

Researchers have examined the issues surrounding common monetary policy arrangements in the EMS. Conditions for an equilibrium relationship among the domestic interest rates for the EMS countries have also been described. Through the years, much of the discussion has focused on Germany’s role as the primary policy maker, the so-called German dominance hypothesis. German dominance hypothesis states that Germany monetary authorities determine the level of domestic interest rate, and the remaining EMS countries follow the German lead in order to keep the bilateral exchange rates in line. Some authors, such as Karfakis and Moschos [1990] and Kirchgassner and Wolters [1993], find evidence in favor of the German dominance, while others, such as Katsimbris and Miller [1993], reject it. Our results based on multivariate cointegration tests support the existence of an equilibrium relationship among the short-term interest rates of the EMS countries, and also support the German dominance hypothesis between 1979 and 1992. However, it is highly doubtful that the relationships among EMS domestic interest rates, and the dominance of German monetary policy stayed the same after 1992. The unification of two German states, and the resulting changes in the economic conditions in Germany appear to have altered substantially the relationships among the EMS economies. In fact, the UK and Italy suspended their memberships in the EMS system during 1992, due to speculative attacks on their currencies. It is an interesting exercise to examine the relationships among the domestic interest rates for EMS countries after 1992. Results may allow us to see how well the system worked during troubled times in 1992, and to see if Germany still lead the system after 1992 in spite of the country’s domestic problems. Finding that the equilibrium among the interest rates has been preserved, and that the German dominance ended after 1992 (if it ever existed) will show that monetary policy for the EMS countries was a product of each individual member’s input. Such a result would suggest that the EMS and, by implementation the EMU, are healthier, even though it is much harder to govern and implement, if all member countries provide input to the common monetary policy decisions. On the other hand, evidence of German dominance, especially in the later years of the EMS system, would support the idea that German monetary authorities will determine the future of the economies of the EMU countries.

To determine whether or not the monetary policies of the EMS member countries were synchronous, we apply multivariate cointegration methods on the domestic interest rate series. By using Johansen’s cointegration method, we can identify equilibrium relationships among the system of exchange rate series. As mentioned above, we find equilibrium relationships (cointegration) among the interest rates of the EMS countries between 1979 and 1992, where the only common stochastic trend in the series is the stochastic component of the German interest rate series, supporting the German dominance hypothesis. By running a similar test for the post 1992 period, with the addition of interest rate series of the additional members to the EMS as they join the system, we will find out if Germany stayed at the center of EMS monetary policy decisions.

Currently we are in the process of completing our data set on the recent domestic short-term interest rates for the EMS countries. A presentable paper will be ready by mid- February.

HOW THE EUROPEAN ECONOMIC AND MONETARY UNION WILL CHANGE
GOVERNMENT DEBT MANAGEMENT

European Economic and Monetary Union (EMU) entered in the global financial market as a participant having unique fiscal and monetary policy. Its general economic and fiscal policies have been applied since Maastricht treaty was signed. From 1999 they will be supported by the single monetary policy of European Central Bank (ECB) and the national banks, adopting Euro (together forming Eurosystem). The success of the new established instruments and rules will change many of the present government financial principles, not only in the European countries but probably in the integrated economy as a whole. We expect one of the strongly affected area to be the government debt management.

The main purpose of this paper is to investigate some significant aspects in the process of changing government debt management of the European countries as they entered in the III stage of EMU on 1 January 1999. We study how the traditional debt policy is fitting in the EMU environment. Proving that the government will need to adapt it, we try to clarify what decisions can be expected.

The paper is organized in 3 parts.

The first part briefly reviews the fundamentals in the theory of public debt management: achievement of optimal maturity and cost minimization. We also extract those elements from EMU which are found to be relevant to the studied issue. The institutional structure and the role of both central bank and fiscal authority are highlighted as one of the main determinants for successful debt policy. The analysis shows that many traditional theoretical and practical approaches hardly will be applied in the EMU boundaries.

As implication in the second part we recognize two possible future opposite dimensions in the risk of government debt management. The rigorous fiscal discipline will increase the credibility of the European countries but it will also limit their flexibility. A deep euro denominated bond market expands the opportunity for the issuers but it will also intensify the competition. Probably the biggest impact will come from the Eurosystem. In its single monetary policy, price stability is announced as a primary goal. It has to support all general economic policies, including debt policy, based particularly on the stability of the interest rate. Our focus is on the impact of ECB debt issue; the applied wide range of assets, eligible for operations; the establishment of single market interest rate curve and the ECB institutional relations with European governments. Having in mind the main purposes of debt management, we prove that the new monetary policy picture leads to an extension of the risk from higher government debt costs.

The convergence in the debt management will be an inevitable consequence of EMU and an answer to the new risk exposure. In the third part we study what government decisions can be expected. We recognize a variance in the intensity of the above mentioned opposite effects in the risk of government debt management. It is sought in still relatively different fiscal positions of the European countries. Budgetary convergence probably will bring EMU participating countries to the target deficit level till the end of the century but debt ratio is stabilizing slower. Nevertheless, we expect the new risk position to be reached even if a complete consolidation is achieved. It is a result of specific fundamental EMU fiscal and monetary principles. Looking for the possible implications in the long term, we find that a concrete result can be the establishment of a new institutional structure for debt management.

The paper concludes that as far as EMU brings European governments closer to the position of the private issuers, we can expect their fiscal authorities to add in debt policy instruments and techniques from the private bonds management.

SECONDARY EQUITY OFFERINGS AND THE INFORMATION AND ELASTICITY HYPOTHESES:  FROM  EVIDENCE SENIOR SECURITIES

INTRODUCTION

It is a widely observed profile of empirical findings that the value of a firm's shares decline when a block of shares is offered for sale by way of secondary offerings. The decline in value is substantial, typically falling within the range of 2 to 3% of pre-announcement value in most previous studies. Although this profile is now widely known, there remains some controversy regarding the cause of such a decline. This study casts light on a long-standing controversy surrounding the causes of negative average market reaction for equity securities elicited by the announcements of secondary offerings. There has been debate in previous work as to whether this is caused by prior asymmetric information and the release of new information associated with the transaction, or securities being imperfect substitutes and the associated negatively sloped demand curve. Formally, the two primary competing hypotheses are the Information Release Hypotheses (IRH) and the Imperfect Substitute Hypothesis (ISH).

The logical underpinnings of the two hypotheses are quite clear, and distinct. However, despite substantial empirical work directed at testing these competing hypotheses, there is not an unambiguous conclusion as to which more completely describes the transaction and its implications. This paper provides stronger test of the hypotheses, and does so by examining both the equity and debt price changes associated with the announcement of secondary offerings of equity securities.

SAMPLE, DATA AMD METHODOLOGY

The Directory of Corporate Financing was used to identify a complete set of firms with secondary offerings over the interval 1970-1991. These firms were then screened for the availability of CRSP data on the equities, and traded debt with price data for those debt securities with sufficient volume to support the methodology. The debt data came from Trade Lines, and lack of debt trading data was the major factor in sample size attrition. The final empirical sample, for which data were available for both equity and debt securities, and which had unconfounded events, was 26 firms.

Standard event methodology was employed to calibrate equity security abnormal returns. The interval of estimation was event days –161 to –30 relative to the announcement. For debt instruments, the interval from –60 to –10 was used to establish mean returns for the calculation of mean adjusted returns in the immediate event window (–10, +10). The methods used by Hite and Owers (1983) and Handjinicoulia and Kalay (1984) were used to control for infrequent trading.

RESULTS

In accordance with prior studies we find average negative abnormal returns on the announcement of secondary issues. The two-day (-1,0) CAR is –1.71% (Z-statistic –3.74). The announcement period (-1,0) CAR for bonds is –0.52%, (t-statistic –1.99). Because of infrequency of trading with bonds, the negative CAR accumulation continued after the announcement. Many debt issues do not trade daily, and our methodology tracked the pattern of abnormal return accumulation until the first day of trading after the immediate event window (-1,0).

The significant negative abnormal return accumulation for senior securities provides strong support for the IRH rather than the ISH. The change in anticipated distributions of cash flows associated with the announcement is clearly substantial if the value of the prior-claim senior securities is reduced by a sufficient magnitude as to generate a significant decline in their average value.

While the average CARs were negative for both equity and senior securities, a closer examination of the distribution of outcomes revealed that for both types of securities, while most experienced negative revaluations, a substantial subset enjoyed positive revaluations. For each type of securities, 8 of the 26 (-1,0) CARs were positive. Moreover, in 15 of the cases both were negative, and in 5 both positive. This is reflected in a regression of.bond abnormal returns on the equity abnormal returns which was significant at the 1% level.

In the 15 cases where both types of securities experienced negative CARs, the transactions were characterized by outsiders selling large blocks of shares, and institutional sellers. These firms were typically those with low or no dividend, were seeking to be acquired, had made prior acquisitions, had experienced downgrades of their debt, and in many instances has received qualified audit reports. In short, they had the profile of negative performers on several dimensions.

In the 5 cases where abnormal returns were positive for both types of securities, the transactions were typified by primarily insiders selling small portions (e.g. 10%) of their holdings and still retaining half of their pre-sale holdings. Personal portfolio balancing was the likely reason. The firms generally had increasing dividends, and/or stock splits. These situations generally profile a more positive scenario. These transactions can generally be interpreted as insiders of firms performing well rebalancing personal portfolios, but still retaining substantial holdings of the firm’s equity.

CONCLUSION

The respective attributes of firms where both debt and equity securities experience the same consequences of the announcement of secondary offerings lends further substantial support to the IRH in contrast to the ISH. Secondary offerings are balance sheet neutral events in that they do not affect the resources available to the firm, or the financial leverage. They do increase the floating supply of equity but not that of debt. In this context, we interpret our findings as strong support for the Information Release Hypothesis.

CO-MOVEMENTS OF TURKISH EQUITY MARKET WITH
U.S. AND EUROPEAN EQUITY MARKETS

INTRODUCTION

Low correlation between the world’s equity markets is often presented as evidence in support of the portfolio gain to investors from international diversification. Because of high correlation between the world’s developed equity markets, investors have turned their attention to the emerging equity markets to achieve better international portfolio diversification. The Turkish equity market is one of the world’s largest and fastest growing emerging equity markets. The objective of this paper is to study the diversification benefits of including Turkish stocks in U.S. and European international equity portfolios.

DATA AND METHODOLOGY

The total value of stocks traded in the Turkish equity market was quite small until 1989. It was only $13 million in 1986, $118 in 1987, and also $118 in 1988. The total value of stocks traded increased sharply from $773 million in 1989 to about $38 billion in 1996. The study covers the 1989-1996 period.

Monthly U.S.-dollar index returns are used in the analysis for the 8-year period from January 1989 through December 1996. ISE National-100 monthly U.S-dollar index returns are used for the Turkish equity market. These data were obtained from the Web site of the Istanbul Stock Exchange. MSCI monthly U.S-dollar index returns are used for the U.S. and European equity markets. These data were obtained from the MSCI monthly publications. Morgan Stanley Capital International computes monthly U.S.-dollar index returns for the following twelve European counties: Austria, Belgium, Denmark, France, Germany, Italy, the Netherlands, Norway, Spain, Sweden, Switzerland, and the U.K. Our study covers all twelve of these European countries.

FINDINGS

The correlation coefficients for the 1989-1996 period indicate that the Turkish equity market is most closely correlated with the Austrian equity market (r=0.39). The correlation coefficients with the other twelve equity markets in the sample are quite low. The correlation coefficient is zero (r=0.00) with the Dutch and British equity markets. The correlation coefficient with the U.S. equity market is negative (r=-0.13). These figures indicate that the Turkish equity market is an excellent prospect for international diversification with the U.S. and European equity markets.

We computed average correlation coefficients for the fourteen equity markets by finding the average of each equity market’s correlation coefficients with the other thirteen equity markets. A high correlation coefficient would imply that the country is not a very attractive investment opportunity to international investors for diversification purposes. A low average correlation coefficient would indicate that the country provides good diversification benefits to international investors. The countries with the highest average correlation coefficients are France (r=0.51), Germany (r=0.49), the U.K. (r=0.49), and the Netherlands (r=0.49). The countries with the lowest average correlation coefficients are the U.S. (r=0.35), Denmark (r=0.32), Italy (r=0.31), and Turkey (r=0.09). Among the fourteen countries, Turkey is the best prospect for international diversification.

To study the changes in international portfolio diversification benefits over time, we divided the 1989-1996 period into two consecutive 4-year sub-periods and compared these two sub-periods. Several previous studies determined that international portfolio diversification benefits decreased considerably after the 1987 international stock market crash compared with the pre-crash period [see Meric and Meric (1997 and 1998)]. Our findings in this study indicate that international diversification benefits have been gradually increasing since the 1987 crash. The average correlation coefficient for the fourteen equity markets is only 0.33 in the 1993-1996 period compared with 0.44 for the 1989-1992 period.

We used Box’s M test to determine if the correlation matrix of the fourteen equity markets changed significantly from the 1989-1992 period to the 1993-1996 period. The test statistic indicates that there are significant changes in the correlation of the markets over time.

We used principal components analysis to study the co-movement patterns of the fourteen equity markets during the 1989-1996 period. There are three statistically significant principal components. The U.S., U.K., German, French, Swiss, Dutch, and Belgium equity markets dominate the first principal component. This principal component explains 46.6 percent of the total variation in the index returns matrix. The Swedish, Norwegian, Danish, Italian, and Spanish equity markets make the greatest contribution to the second principal component. This principal component explains 10.3 percent of the total variation in the index returns matrix. Since the Austrian and Turkish equity markets are closely correlated and they have quite different movement patterns than the other twelve markets, they have the highest factor loadings in the third principal component. However, this principal component can explain only 7.5 percent of the total variation in the index returns matrix.

We also applied principal component analysis to each of the two 4-year sub-periods separately and used Box’s M test to determine if there was a significant change in the variance-covariance matrix of the fourteen equity markets from the first 4-year period to the second 4-year period. The test statistic indicates that the co-movement patterns of the markets changed significantly from the 1989-1992 period to the 1993-1996 period.

PRIVATIZATION AND VALUATION

Moustafa F. Abdel-Magid, Simon Fraser University

In recent years a number of countries have transferred property rights and ownership of previously state owned enterprises to private investors. The methods used to privatize vary among countries industries. In general, the methods commonly used are: (1) floating shares; (2) auctions; (3) concessions. Privatization is facilitated by international financing through the International Finance Corporation (IFC) and international capital markets. Since privatization is expected to grow in scope world-wide, continued analysis and discussion of its problems are timely and useful.

When privatization is contemplated, governments are usually fearful of a political backlash if the sale turns out to be grossly underpriced. Investors, on the other hand, are usually worried about overpaying for the assets acquired. Fair pricing of the privatized enterprise is paramount. Fair pricing requires proper valuation of the assets and liabilities of the entity being privatized. In countries with well-functioning market economies such as Japan, and the United Kingdom, valuation of privatized industries has been relatively smooth. There are well established methods of asset valuation such as discounted cash flow, capitalization of earnings, net realizable value, and adjusted net book value, that work well in a market economy. Moreover, the information generated by the accounting systems of privatized companies in market economies are reliable and provide a basis for fair pricing. Methods of pricing used assets are also available as supplementary techniques. Countries with well functioning capital markets can also make use of their experience with pricing of initial public offering (IPO).

Valuation for privatization has proven more problematic in countries with economies in transition. Countries like Poland, Hungary, the Czech Republic, Russia, and China have no established markets and the accounting systems of privatized companies were totally inadequate to generate information useful for proper valuation. The valuation problem is also made complex by the fact that privatization is usually accompanied by the need to re-design the product, retooling of factories, downsizing and retraining, extensive market research, and serious currency problems. High uncertainty of the future conditions of these economies makes matters more difficult.

This paper is prepared to stimulate discussion and clarify these issues.

The insights gained can facilitate future privatizations.

AN INQUIRY ON THE FACTORS CONTRRIBUTING
TO THE ECONOMIC CRISIS IN TURKEY

Following the global crisis which started in East Asia in 1997 and imposing its effects on both the developing and developed countries turned the attention of the researchers to assessing the early indicators of the crises, to explore the reasons underlying crises, to analyze the contagious characteristic of the crises among economies and even to formulating a new system for a safer capital flow network. Reviewing the abundant number of researches this paper will confine itself to define the crisis and then determining the factors contributing to such a condition for the case of Turkey.

This paper starts with the definition of "crisis". Taking the stock exchange as a good indicator of a crisis, we take 1.5 standard deviation outlier threshold and beyond an indicator of crisis. Assigning the numeral 1 for the outliers and the numeral 0 for the values which fall into the threshold we construct a crisis index. Such an index leaves us no room but applying a binary dependent variable model as an estimation technique. Hence, the paper chooses to employ probit model as an estimation process.

Following the guidelines stated above the paper aims to search for the factors which affect the crisis condition through January,1986, the date which marks the start of the operation in Istanbul Stock Exchange, and November, 1998, the date marking upper bound of data availability. The model to be used in the analysis could be put forward as:

D (Crisis_i,t) = b ₀ + q (L)_i,t + e _i,t

D (Crisis_i,t) = 1 if crisis exists

= 0 if there is no crisis

while q (L)_i,t stands for the information set of a number contemporaneous and/or lagged control regressors; such as industrial production index, real interest rates, inflation, credit growth, foreign exchange basket of 1USD+1.5DEM, M2/International Reserves ratio and growth of domestic debt stock of the Treasury. On the other hand, e _i,t, stands for the normally distributed disturbance term which represents the omitted influences that affect the probability of a crisis in Turkey.

Applying the probit model releases the preliminary results that a rise in the industrial production or depreciation of TL vis-à-vis the foreign currency basket contribute negatively to the probability of a crisis and a rising burden of the domestic debt of the Treasury increases the probability of a crisis. The research is important since, first, it defines and marks the instance of crises and second, it quests to reveal what variables have contributed and what other variables have served to sooth a crisis condition in Turkey within the period under consideration.

BANK PRIVATIZATION IN TURKEY: ISSUES AND SUGGESTIONS

This paper offers a critical assessment of bank privatization in Turkey by examining the role of State owned banks in Turkish banking sector and political, legal, and managerial characteristics in the privatization process. Despite Turkey’s commitment to privatization of its state owned enterprises including its state owned banks, its success with privatization has been negligible. The state owned banks, despite their inefficiencies and spoiled management, still hold a major share in banking industry. Their privatization is urgent.

Justifications for the privatization of state owned banks are as follows: State owned banks have completed their historical mission. They have developed inefficient and slow bureaucratic mechanisms within heavily regulated governmental sector. They have been charged with activities that are additional or out of their core competencies. They have been subject to political interventions particularly in the areas of extending credits, collecting loans, and staffing. Their top managers have been appointed and changed on the basis of political favoritism by frequently changed governments. The banks have chronically experienced the erosion of their capital.

As a result, 1994 Privatization Program included privatization of six state banks: Sumerbank A.S., Etibank Bankcilik A.O., Sekerbank, T.A.S., TOBANK, T. Sinai Kalkinma Bankasi, Cay Bank A.S. Furthermore, the privatization of Halk Bank, Emlak Bank was considered.

Nevertheless, so far the major banks privatized include TOBANK (state shares sold), Sumerbank and Etibank Bankacilik A.O. (accomplished after a second time) while the privatization of Turk Ticaret Bankasi, the most recent case, has ended up with a scandal that led to the fall of the government.

Thus, it necessary to look at the privatization process closely and reform the current system before pursuing further privatization of the state owned banks.

The paper develops a conceptual framework in which political, legal, economic and financial, and managerial dimensions of bank privatization can be studied. Politically, there has been a lack of support for privatization despite the promises made by the late governments in recent years. Coupled with political instability in the country, the privatization program has been frequently interrupted. A consensus on privatization between the coalition partners has never been achieved. Legally, although a privatization law was enacted in 1994, it has been challenged a number of times before the Constitution Court and thus the privatization requires an effective legal framework. Economically and financially, state owned banks suffer from non-performing loans and insufficient capitals. Finally from a management perspective, state owned banks have been managed poorly because of primarily political appointments, lack of leaderships, and overstaffing.

The paper identifies current issues and then provides recommendations to deal effectively with the current crises in privatizing Turkish banks. Drawing from the relevant literature and privatization experiences in other countries, it advances some measures on structure and management of Turkish bank privatization.

Current Issues

The most pressing issue is a wide spread corruption or unethical conduct in the privatization process. The Turkish bank privatization is marked with such unethical practices as favoritism in transferring the ownership of state owned banks from public to private. Especially the most recent case, the privatization of Turk Ticaret Bankasi, illustrates the point. The government of Yilmaz including Primer Minister and State Minister in charge of economic affairs was accused of having a private deal with the person who won the bidding. After revelation of this information, the Yilmaz government fell after Parliament's vote of lack of confidence. This is, however, not an isolated incident. There is common belief that every government in power manipulates privatization process in order to transfer the state owned banks to their supporters or to receive some bribes from prospective buyers. Of course such a prevailing belief damages the privatization process.

The second most important issue is that slow and inefficient procedure in privatization process. Because of bureaucratic inertia and the difficulty in obtaining political consensus, it usually takes a very long time to prepare bidding files and evaluate the bids received from various prospective buyers. Longer the process, more problems arise.

The third problem with the bank privatization is that the lack of understanding or appreciation of the benefits of privatization in public. Since a privatization may lead to some layoffs of employees, both labor unions, some pressure groups, and employees resist any privatization attempts. Privatization is considered a source of increasing unemployment and only as a vehicle to provide some benefits to the partisan alliances.

Recommendations

In order to eliminate the corruption and unethical conduct, a transparency in the privatization should be provided. The whole process should be designed clearly without leaving any room for misunderstanding and misinterpretations. All constituents must know the whole process and results. Bidding evaluations should be made open, before public, for example before media.

It is also necessary to speed up the privatization process by preparing enterprises, designing procedures and their announcement, and finally choosing the best candidate. Timing is important in privatization process. It is extremely important to check the background of participants to make sure that questionable characters do not involve in the process.

Turkish governments committed to privatization should make special efforts to spell out the benefits of privatization and build alliances to realize the objectives of the privatization. They should also reform the current privatization law so those opponents to privatization cannot defeat the process by appealing to the Constitution Court. It requires a great deal of public relations in promoting privatization by educating the media, interest groups, members of parliament, and the public in large.

Today effective realization of bank privatization is a crucial for Turkish economy. A systematic approach and measures recommended in this paper will help to accomplish this formidable task. It requires commitment, perseverance, strategic formulation and implementation.

IN SEARCH FOR A LINK BETWEEN PERFORMANCE ANDOWNERXHIP STRUCTURE:
THE CASE OF BANK PRIVATISATION IN ITALY AND FRANCE

Marco Giorgino and Elena Magnani, Politecnico di Milano, Italy

Most governments adopt privatisation programs with concrete objectives in mind. One such goal is to improve the operating and financial performance of the former state-owned bank (or company) by exposing it to market forces. Almost all governments expect that privatisation will: increase the firm’s profitability; increase its operating efficiency; cause the firm to increase its capital investment spending and to increase its output (Megginson, Nash and Randenborgh, 1994).

During the Eighties the academic literature on the performance of state-owned enterprises was voluminous, but the few empirical analyses of privatisation itself that had been published (Bailey, 1986; Kay and Thompson, 1986; Yarrow, 1986; Bishop and Kay, 1989; Wortzel and Wortzel, 1989) were far from conclusive.

Most recent theoretical and empirical studies offer stronger support for the dual proposition that private firms outperform state-owned enterprises (Boardman and Vining, 1989; Masera, 1997) and that privatisation itself increases the operating efficiency of the divested firms (Galal, Jones, Tandon, and Vogelsang, 1992; Barca, 1993).

This study investigates the effects of privatisation on operating and financial performance in newly privatised Italian and French banks. Six cases are analysed, being the total number of deals realised from 1987 until now in those countries. The small number of cases allows to make also in-depth reflections about the reasons that brought to certain results.

Our sample data collection procedure was to mail a request for information to each of the banks. We requested each bank to send us the offering prospectus from their divestment share issues, as well as the annual reports for the three years prior and subsequent to share issue, as well as for the year of privatisation itself. We used also supplementary information from business magazines.

We compare the pre- and post-privatisation operating performance of privatised banks. Table 1 details the economic characteristics we examine for measure changes resulting from privatisation. It also presents and defines the preferred and alternative empirical proxies we employ in our analysis (12 balance sheet ratios). The index symbols A and B in the predicted relationship column stand for after and before, respectively.

Selected proxies are initially calculated for each bank, starting from data drawn from its annual report. They are calculated three years before through three years after privatisation. We try to avoid considering a change in performance reflecting a generalised trend as due to the privatisation, by using a comparison with a control group. Thus, we calculate for each proxy variable the under- over-performance of each privatised bank in comparison with the reference banking system referring to three, two, one years before and one, two, three years after the deal(j _{3YB, j 2YB,} j _1YB, j _1YA, j _2YA, and j _3YA). Then, we calculate six indexes (m _{3YB, m 2YB,} m _1YB, m _1YA, m _2YA, and m _3YA,) as an average of the over-/under-performance calculated for each bank. Lastly, a time series analysis is performed through the comparison between the average value of under- or over-performance after (AVG_A) and before (AVG_B) the transaction. Two synthetical indexes are calculated (J and y ), in order to catch the overall effect, and the analysis is split for each country to grasp the differences between Italy and France.

Results of the analysis show that the aggregated profitability (ROE) of privatised banks increases passing from before to after the transactions. It remains always superior to the profitability of the control groups, but, on average, before the transaction privatised banks performs a little better that the system while after they perform strongly better. This is more evident for France than for Italy. ROI gives more evidence of the profitability improvement that from before to after increases nearly five times. Again, France shows negative initial values and its trend is better than Italy, even if the over-performance of privatised banks after the transaction is higher in Italy than in France.

The aggregated profitability increase is hardly understandable. As a matter of fact, both the net profitability over interest-bearing activities (NII_TA) and the profitability over non-interest-bearing activities (BI_TA) do not perform well or at least show strongly different trends. BI_TA strongly decreases in Italy, even if the experimental group continues to perform better than the control group. On the contrary, the proxy value increases in France so that the global trend is (little) positive. NII_TA also generally decreases, due to a decrease in the value of the proxy in Italy and an increase in France where the values remain however negative. This situation can be easily explained through the fact that after privatisation the banks are free to give credit after their own policies, so that previously compulsory loans to public enterprises and bodies are given up. This results in a decrease of the proportion of total assets invested in loans. As a matter of fact, L_TA reduces, even if this value results from different situations in Italy and France: starting from similar values they reach completely opposite values, negative for Italy and positive for France.

The proportion of the banking income explained by the net interest income (NII_BI) decreases. Again, there are differences between Italy and France: in particular, the performance of French banks after the privatisation is strongly negative. The proportion of fees and other revenues over banking income (FOR_BI), on the contrary, increases, and more than compensates the decrease in NII_BI. Here, country specific trends are less evident, as Italy is almost static while France improves performances. Operating costs increase, in comparison both with banking income (OC_BI) and with total assets (OC_TA). However, in this last case, trends in Italy and France are the same, while, comparing costs with a flow-value (as is banking income), trends diverge. This generalised increase in operating costs does happen even if the productivity of banks increases. This is not evident if we measure productivity through the total assets-per-employee ratio, but if we consider the banking income-per-employee ratio, we can see that it witnesses an increase. This result, however, derives mainly from the performance of Italy rather than of France.

At last, we find evidence that after the privatisation banks increase their leverage. However this results from a strong decrease of capitalisation (E_TA) in Italy, where privatised banks pass from an over-performance over the control group to an under-performance, and an increase in France. However, also in this country the degree of capitalisation of privatised banks remain lower than the system.

The empirical analysis seems consequently not to completely correspond with the predictions of the theory that states that public-owned companies are often less profitable and less efficient than private companies. However, the analysis witnesses a strong restructuring of privatised banks after the deals. Both their strategy and their organisational structure are changed in order to meet the requirement of shareholder value maximisation. This results in contradictory empirical findings, as the operating achievements are influenced by important investment, which may affect the short-term performance.

SHORT AND LONG TERM IMPACT OF OPTION LISTING ON UNDERLYING SECURITIES:
HONG KONG EVIDENCE

The effect of option introduction on the returns, risk and liquidity of the underlying security on Hong Kong Stock Exchange was examined. Ross (1986) and Grossman (1988) argues that options can be useful in attaining efficiency in competitive equilibrium by expanding the set of contingencies covered by the marketed assets. Conrad (1989) finds that the introduction of options causes a permanent price increase in the underlying stock beginning a few days before the start of trading on the U.S. market, Watt, Yadav and Draper (1992) find that option listing results in a temporary price increase on the UK market.

The introduction of options may have different effects on the volatility of the underlying shares. Two studies by the Chicago Board Options Exchange (CBOE 1975 and 1976) and Skinner (1989) find that a statistically significant decline in price volatility occurred following the options listing date. Fedenia and Grammtikos (1992) shows that options listing significantly affects the bid-ask spreads on the underlying stocks. Skinner (1989) finds that stock market trading volume increases after options are listed on firms’ stock. This study examines the impact of listing of options on returns, risk and liquidity of the underlying security on Hong Kong Stock Exchange.

Empirical design

In 1993, the Stock Exchange introduced the European style Hang Seng Index Options which can only be exercised on the expiry day. Up to 28 February 1998, there are totally 2 index options and 16 stock options traded in Hong Kong Stock Market.

For empirical analysis, stock returns, high and low prices and trading volume data are obtained on a daily basis for a 520-day window around the option listing date from the Datastream Database.

Price Impact

We apply the market model. Using the market model to measure the normal return, the sample abnormal return is

For statistical testing purposes, we also calculated the ratio of the cumulative abnormal returns (CAR) to its estimated standard deviation for each event j.

Risk Impact

A standard deviation of return, beta and high-low statistic as measures of risk, and also employ trading volume and bid-ask spread as measures of liquidity of underlying stock. For the empirical analysis, we calculate the values of each of the variables of interest – standard deviation, beta, high-low statistic, trading volume and bid-ask spread in both the pre-listing and post-listing periods for each stock that has option.

We conduct both the paired t test and the Wilcoxon Signed Rank test on the cross-sectional series of pre-listing and post-listing values of these parameters. This study also uses the modified EGARCH model that captures structural change in the stock volatility because GARCH fails to capture the negative asymmetry apparent in many financial time series. The EGARCH model ameliorates this problem by allowing the standardized residual as an MA regressor in the variance equation, while preserving the estimation of the magnitude effect. Additionally, the ARCH/GARCH approach to modeling changing volatility precludes the testing of Black’s (1976) leverage effect. The tendency for negative shocks to be associated with increased volatility is captured in the ARCH/GARCH class of models. The EGARCH model developed by Nelson (1991) is given by

where a₁,...,a_p, b₁,...,b_q, , and denote parameters

Findings and concluding remarks

This paper examines the effect of option introduction on the returns, risk and liquidity of the underlying security on Hong Kong Stock Exchange. The results of impact on return suggest that event of options listing in Hong Kong does not play any role in the return of underlying stocks as found by Conrad (1989). Empirical results of impact on risk document that standard deviation decreases in the short-term period and daily high-low statistic increases in long-term period, but fail to disclose any change in systematic risk. We use trading volume and bid-ask spread as measures of liquidity. We find the market-adjusted trading volume decreases after the option listing in the long-term period. This is consistent with the theory that the option listing may cause trading to be diverted from the underlying stock to the corresponding option and thereby decrease liquidity. For both the unadjusted and market-adjusted average bid-ask spread, the paired test and the Wilcoxon test indicate that the hypothesis that the pre and post liquidity is the same can not be rejected at any reasonable level of significance in both short term period and long term period. The combined evidence suggests that the listing of options on Hong Kong Stock Exchange appears to have little impact on the return, risk and liquidity.

DAY OF THE WEEK EFFECTS:
RECENT EVIDENCE FROM NINETEEN STOCK MARKETS

Ozgur Berk KAN, Old Dominion University
Asli BAYAR, Bilkent University

 

This study presents international evidence for the existence of the day of the week effects for a recent period of time from the perspectives of domestic and global investors. The daily effects are analyzed in stock market returns denominated in both local currency and dollars for nineteen countries. The sample covers the period July 1993 to July 1998. A daily pattern in stock markets is observed for fourteen countries in local currency returns and for twelve countries in dollar returns.

The observed daily patterns differ for local and dollar returns, the latter being exhibiting lower daily means and higher standard deviations compared to the former. In local currency terms, a pattern of higher returns is observed around the middle of the week, Tuesday and then Wednesday; and a lower pattern is observed towards the end of the week, Thursday and then Friday. In dollar terms, a higher pattern occurs around the middle of the week, Wednesday and then Tuesday; and a lower pattern is observed towards the end of the week, Thursday and then Friday. The lower patterns are more apparent in both cases. Volatility is the highest on Mondays in both local and dollar returns. In local currency returns, volatility is the lowest towards the end of the week, Thursday and Friday, whereas the lowest volatility of dollar returns is observed on Tuesdays.

The lowest coefficient of variation values are respectively observed on Tuesdays and Wednesdays for local and dollar returns while the highest values appear towards the end of the week, Thursday and Friday, for both local and dollar returns.

Our empirical results detect significant and different daily patterns of mean returns and their volatility in local currency and dollar terms. The empirical results have useful implications for international portfolio diversification.

THE EFFECTS OF FOREIGN INVESTMENT
ACTIVITIES ON FIRM VALUE

FORECASTING INTEGRATED WORLD STOCK MARKETS
USING INTERNATIONAL CO-MOVEMENTS

Markowitz approach to portfolio diversification indicates that today, the global investor can earn potential gains from international diversification rather than domestic diversification as long as returns in different countries are less correlated than those in domestic markets. Another approach in investigating international co-movements is to focus on price discovery in world markets. Naturally, co-integration and error correction modeling provides a useful framework for analyzing price adjustments in internationally linked markets.

It appears that previous empirical studies on the relationship between world stock markets do not provide consistent results. The reasons for the inconsistent results are numerous including the choice of markets, different sample periods, different frequency of observations, and the different methodologies employed. The focus of previous studies also creates problems with interpretation of results. Most studies are concerned with integration versus segmentation of markets as indicators of the degree of international diversification for the global investor.

The major contributions of this paper are as follows. In this study, the degree of market integration is investigated in order to forecast national markets according to their international co-movements. The focus of the paper is different from previous research that investigates market integration for global diversification. Besides, we attempt to maintain a research framework whereby a coherent data-base is used, to include all the emerging markets as classified by the IFC. The data frequency is weekly for all countries and the co-integration methodology is employed to examine the interrelationship of the major world stock returns.

This paper aims at forecasting stock returns in emerging markets using their interrelations to other stock exchanges including world leaders and counterparts in their regions. For that purpose first, we examine international co-movements in stock prices by employing the Engle-Granger two-step cointegration technique. We determine the intra-continental and inter-continental co-movements of stock prices and group the national markets accordingly. Next, we forecast each national stock market according to the lead-lag structures and the transmission between the markets. Forecast performance of Error Correction Model (ECM) and Vector Autoregressive Models (VAR) will be compared for all national markets.

The results reveal that all the national markets are cointegrated with the world leaders and with other emerging markets grouped according to their geographical proximity. This result is important in terms of its implications for the global investor. Accounting for the information embodied in the long-run equilibrium relationship, short run dynamics can be examined to see the process by which the national indexes return to their equilibrium states.

The results of the forecasting exercise are not very promising however. For longer forecast horizons, non of the models pass the parameter constancy tests. For shorter forecast horizons only the Latin American and the Far Eastern markets pass the parameter constancy tests. For those countries mixed results were obtained as to better forecast errors from ECM and VAR models. Latin American and Far Eastern markets have distinguishing characteristics among the emerging markets. They are more established in the sense that international awareness about those markets are high and they have been attracting international investors for a longer time period. Also, in terms of listed companies, trading volumes and market capitalization these stock markets are in better terms than their counterparts in Europe and Asia. It can be argued that they are thus, better integrated with the world and with each other, in terms of information and capital flows. Therefore their behavior could be better forecasted.

INTERNATIONAL EVIDENCE ON THE DAY OF THE
WEEKEFFECT ON STOCK MARKET VOLATILITY

Halil Kiymaz, University of Houston-Clear Lake
Hakan Berument, Central Bank of Turkey

The presence of calendar anomalies in stock market returns has been documented extensively in finance literature. The most common anomalies are the Weekend Effect or the Day of the week Effect and the January Effect.

Stock markets in the United States and other countries demonstrate differences in distribution of stock returns in each of the day of the week. The day of the week effect in the equity markets, which shows that the average returns on Monday is significantly less than the average return over the other days of the week was documented by Cross (1973), French (1980), Gibson and Hess (1981), Lakonishok and Levi (1982), Keim and Stambaugh (1984), Rogalski (1984). Studies on the day of the week effect are not limited to equity markets in the US. The "day-of-the-week effect" anomaly has been investigated and reported for both international equity markets and other financial markets. Jaffe and Westerfield (1985a), (1985b), Aggarwal and Rivoli (1989), Chang et al. (1993), Kato and Schallheim (1985), Athanassakos and Robinson (1994), and Dubois (1996) show that the distribution of stock returns varies by the day of the week internationally. Furthermore, the day of the week effect is found to be present in T-Bill market (Flannery and Protopapadakis (1988)), in commodity and stock futures market (Cornell (1985), Dyl and Maberly (1986), Gary and Kim (1987), in foreign exchange market (Corhay et al. (1995)).

With the existence of the day of the week effect patterns investors may to take advantage of irregularities by designing trading strategies. For a rational financial decision-maker returns constitute only one part of decision process. The other aspect, which is taken into the account when one makes investment decision, is risk or volatility of returns. If investors can specify a certain pattern in volatility, then it would be easier to make investment decision based on both return and risk. In fact, this would give investors another tool to design profitable strategies.

The purpose of this paper is to investigate the day of the week effect in stock market volatility of major international stock markets, namely Canada, Germany, Japan, UK and US. None of the existing studies has investigated the day of the week effect in stock market volatility in the conditional variance specification framework. This paper fills the void by examining volatility and the day-of-the-week effect in major international stock markets.

The data includes daily prices of TSE-Composite (Canada), DAX Index (Germany), Nikkei-225 (Japan), FT-100 (UK), and NYSE-Composite (US) from October 25, 1989 to December 25, 1997. There are 2154 daily observation for Canada; 2102 for Germany; 2133 for Japan; 2133 for the UK, and 2133 for the US.

We employ three different types of specifications for the return and volatility equations. The first one incorporates the day of the week effect only into the return equation. The second and third ones incorporate the day of the week effect into both the return and volatility equations. While second one uses GARCH (1,1) for the volatility equation and the third model is established after considering a class of conditional variance specifications.

The empirical findings indicate that the day of the week effect in both return and volatility equations are present. For Canada, Germany and Japan, there is statistically significant higher volatility on Monday compare to Wednesday. For Germany and Japan, volatility on Tuesday and Thursday is lower than that of Wednesday. Both Canada and US however, experience higher volatility on Thursday and Friday than on Wednesday. Overall, the lowest volatility is observed on Wednesday for Canada, on Tuesday for Germany, Japan, and UK and on Monday for US. The highest volatility occurs on Monday for Canada, Germany and Japan and on Friday for UK and US. Moreover, the leverage effect is also observed for Japan, UK and the US. For these countries, volatility increases more when the innovation is negative than positive. Hence, volatility increases more when the agents over estimate the return -- when the innovation is negative -- then underestimate the return -- when the innovation is positive. These findings are in the line with those of French et al. (1987). Finally, the coefficients of the conditional standard deviation of the return equation are positive for all countries under consideration, indicating that investors would certainly want to be compensated for the riskier assets. Hence, there is positive risk premium.

EMPIRICAL PERFORMANCE OF THE CZECH AND HUNGARIAN
INDEX OPTIONS UNDER JUMP

Previous empirical option papers which rely on the Black-Scholes (1973) model often show that the value of an option rarely matches exactly to the market price at which it is traded on the exchange. The main reason for this discrepancy may be that the Black-Scholes model uses too many ''simple unrealistic assumptions'' (Black, 1988). All the strong restrictions implied by the Black-Scholes model are known to be wrong in its detail and formal statistical rejections of the null would tell us no more than we already know. The more interesting question is, how wrong or right is it? In other words, what empirical implications arise for option pricing if an investor ignores the existence of distributional abnormalities, such as jumps and time varying volatility, in the underlying processes that could invalidate the Black-Scholes analysis? The objective of our paper is to address these questions using the newly created derivative instruments by the Austrian Futures and Options Exchange (OTOB).

The Central European Clearing Houses and Exchanges (CECE) under the OTOB have created product lines which focus only on East European Index options, and began trading these products since March 1997. In this paper, we analyze only the Czech (CTX) and Hungarian (HTX) index options since they provide the longest time series and, so far, the two most liquid products. And further, these two markets provide two distinct and different characteristics: One is bearish (CTX) and the other is bullish (HTX). The CECE options products are designed for an investor who ''wishes to receive positive returns from the foreign market, but wants to be certain that those returns are meaningful when translated back into his own (domestic) currency.'' (Reiner, 1992 page 148) Consequently, for this investor, it is the product of the index valued in foreign currency and the exchange rate at expiry that is important. There are generally two types of foreign index linked options: One depends on a ''fixed'' exchange rate, and the other belongs to a ''flexible'' exchange rate group. In this paper, we investigate the empirical implications of the European CTX and HTX call options that have strike prices in domestic currency (i.e. in U.S. dollars) when jumps and time-varying volatilities are present in the underlying processes.

This paper is motivated by two factors. First, the underlying processes exhibit leptokurtosis (''fat tails'' ) in their return distributions. As noted first by Mandelbrot (1963) and Fama (1965), this is one of the undisputed stylized facts in any financial market. Just name a few, Bollerslev, Chou, and Kroner (1992), Bates (1996b), and Duffie and Pan (1997) all give evidence of leptokurtosis in return distributions for various financial markets around the world. Consequently, if these options are based other than the Black-Scholes model, which critically depends on the log-normal distribution of the terminal stock price, then it would be interesting to observe operationally significant discrepancies between the models that incorporate ''fat tails'' and Black-Scholes, especially when the markets for these underlying equities are extremely thin and have high transaction costs.

Second, there is no empirical literature on the option model that includes both the foreign index and jump features for the products that we are investigating. There are numerous empirical and theoretical option papers on the Poisson jump-diffusion model based on Merton's (1976a) model. For example, Ball and Torous (1985), and Jorion (1988) follows the original Merton (1976a) and assume that jump risk is idiosyncratic and thus diversifiable. Thus, a portfolio formed with the stock and the option with proportions chosen to eliminate the Brownian risk will earn the risk-free rate as before even though the jump risk remains. The others in their models of jump diffusion (Bates, 1991; and Amin, 1993), stochastic volatility jump diffusion (Bates, 1996a, b), and stochastic volatility interest rate jump diffusion (Bakshi, Cao and Chen, 1998) all assume that the jump process is systematic and hence should be priced. Trautmann and Beinert (1995) analyze both cases for the German Stock Exchange (DAX). There are also voluminous work on foreign index derivatives based on Margrabe's (1978) paper. For example, Reiner (1992), Dravid, Richardson and Sun (1993), and Wei (1995) solve for the value of various foreign index options given European exercise; Craig, Dravid, and Richardson (1995) provide evidence of market efficiency using foreign based derivatives; Dravid, Richardson and Sun (1994) analyze the Yen/Deutsche Mark warrants which are U.S. dollar denominated.. To analyze the specific products mentioned in this paper, we utilize the combination of works by Merton's (1976) jump-diffusion model and the foreign index option model addressed in Reiner (1992).

We find that, when making pair-wise and combined comparisons between the pure diffusion process and one which includes either a Poisson jump process or a time varying variance, the statistical evidence lend support for the Poisson jump process to describe the data. Consequently, taking the Poisson jump as the underlying process for our option pricing model, we find that the differences in option pricing for the Czech and Hungarian index options arise when the jump model is compared to the model without the jumps. Using the estimated parameters, approximately four-fifth of 2.4 percent (for CTX) and 3.4 (for HTX) percent underpricing biases reported for the short term at-of-the-money call options can be explained by the Jump option pricing model. However, we find that these pricing errors are quite small. Consequently, we question whether these mispricings can be operational when the underlying markets for the traded derivatives are illiquid and have transaction costs in the range of 3-6 percent per trade.

CROSS CORRELATIONS AND PREDICTABILITY OF STOCK RETURNS

Recent studies, such as Lo and MacKinlay (1990), have shown that small stock returns can be partially predicted by the past returns of larger stocks. The cross correlations are asymmetric in the sense that returns to small stocks are correlated with lagged returns on large stocks, but lagged returns for small stocks do not help predict returns to large stocks. The existence of this lead-lag relationship between large and small stocks raises questions about market efficiency and to date, two studies have examined whether trading rules can exploit the predictability inherent in cross correlations. McQueen, Pinegar, and Thorley (1996) devised a trading rule that yields annualized abnormal returns of 6.8%, while Knez and Ready (1996) generated trading rule profits of up to 21% annually, but they noted that such profits might be reduced to zero given realistic transaction costs.

In addition to past stock returns, a large body of literature has shown that macroeconomic variables and stock market fundamentals also predict stock returns. Connor (1995) categorizes models designed to capture these sources of predictability as statistical factor models, macroeconomic factor models, and fundamental factor models. For a pooled cross sectional time series of U.S. stock returns for 1985 - 1993, he finds that macroeconomic variables contain no marginal explanatory power when added to either fundamental or statistical factor models. In contrast, Lo and MacKinlay (1990) hypothesize that macroeconomic information impacts large companies first and is transmitted with a lag to smaller companies. If this hypothesis is correct, with the "right" set of macroeconomic variables as predictors, the proper lag structure and functional form, any economically significant prediction from cross correlations should be eliminated. Following this argument, one would expect macroeconomic variables to forecast small stock returns better than statistical models involving cross correlations, which is the opposite of Connor’s (1995) findings.

This study examines the relative importance of cross correlations versus macroeconomic variables in models that forecast returns for portfolios of U.S. small stocks. Unlike previous studies that examine predictability within sample, comparisons between these two sources of predictability are made using out-of-sample tests. Following an approach developed by Pesaran and Timmermann (1995), various models are fitted within sample and tested for one-month-ahead out-of-sample predictability. The models are updated monthly using a rolling 120-month estimation window. Small stocks are purchased and held as long as one-month-ahead portfolio returns are predicted to be positive, while the risk-free asset is held whenever the forecast for excess stock returns (returns above the risk free rate) is negative. Base-case forecasting models are developed for both macroeconomic variables and cross correlations. Then, lagged large stock returns and macroeconomic variables are included in the same model to determine the marginal contribution of each source of predictability. The models are judged on the basis of directional forecast accuracy and trading rule profits before and after the inclusion of trading costs.

The base case involving only macroeconomic variables (Model A) provides a 53.81% directional accuracy, versus a 52.94% directional accuracy for a buy and hold strategy. In absence of transaction costs, Model A provides abnormal returns of .431% per month. The best of the models that include information about both cross correlations and macroeconomic variables (Model E) yields 55.28% directional forecasting accuracy and abnormal returns of .516% per month, in absence of transaction costs. The addition of cross correlations to the macroeconomic variables adds a 1.47 percentage point improvement in directional forecast accuracy and a .085 percentage point increase in abnormal returns (at zero trading cost) over Model A.

Such results can be compared to models based upon cross correlations (lagged large stock returns) alone. The best of these models (Model F) gives similar directional forecast accuracy to Model A, but trading rule profits are only .215% per month. Comparing these models, the marginal contribution of macroeconomic variables is .321% (.516-.215) versus .085% (.516-.431) for cross correlations. While cross correlations add little to profitability generated by macroeconomic variable models, macroeconomic variables add significantly to the abnormal returns generated by cross correlations alone. This situation could arise if cross correlations serve as a proxy for omitted lagged macroeconomic variables.

Encompassing tests are another way to judge the relative out-of-sample importance of macroeconomic variables models versus statistical models with cross correlations. Donaldson and Kamstra (1996, p.57) note that a model, such as our Model A, should be preferred to another model, such as Model F, if A explains what F can not explain and F can not explain what A can not explain. They show that a formal test for encompassing between any two models, such as A and F, involves regressing the forecast error from Model A on the forecast from Model F to see if Model F can explain what Model A can not explain. Then the forecast error from Model F is regressed on the forecast from Model A to see if Model A can explain what Model F can not explain. For models A, E, and F, we find that both models A and E encompass Model F (cross correlations alone) at the 5% significance level. Model A is not encompassed by either Model E or Model F and similarly Model E is not encompassed by either Model A or Model F. This test confirms earlier results that macroeconomic variables are more important for out-of-sample forecasts than cross correlations. In fact, it suggests that the marginal contribution of cross correlations is not statistically important in distinguishing between Models A (macroeconomic variables alone) and Model E (macroeconomic variables + cross correlations).

Adding trading costs of .25% per trade, or .5% roundturn, which Berkowitz, Logue, and Noser (1988) found to be the average trading cost faced by large institutional investors, reduces abnormal returns for the best model (Model E) to .457% per month. If trading costs are as high as 3% per trade, or 6% roundturn, as suggested by Knez and Ready (1996), then none of the models provide positive abnormal returns. However, Keim and Madhavan (1995) calculate that trading costs of small stocks for large investors are 1.35% - 2.68% (2.7% - 5.36% roundturn costs). For 1.35% trading costs, Models A and E provide monthly abnormal returns of .159% and .197%, which decline to .028% and .044% for 2% trading costs. Such returns seem achievable, but our results need not constitute a violation of market efficiency. Trading costs were higher and the technology needed to exploit this predictability may not have been available during the earlier years of our data set.

In conclusion, both macroeconomic variable and cross correlation models give similar out-of-sample directional forecast accuracy, but macroeconomic variables are preferred using encompassing tests and on the basis of trading rule profits. Cross correlations only marginally improve upon the forecast accuracy and trading rule profits generated by models using macroeconomic variables alone, while adding macroeconomic variables to cross correlation models substantially increases abnormal returns. Cross correlations add to macroeconomic variable predictability in some periods, but not in other periods. The likely reason is that macroeconomic variables included in the best model are generally stable from month to month, but they do occasionally change. During periods when macroeconomic relationships are changing, cross correlations probably pick up changing market conditions faster than lagged macroeconomic variables alone.

THE PECKING ORDER THEORY AND AN EMPIRICAL INVESTIGATION
FOR TURKISH EVIDENCE^*

Tülin Akkum, Istanbul University

 

A group of theoretical studies for explaining the determinants of capital structure is based on the asymmetric information about firm value between firm management and investors in capital markets. These studies developing different approaches assume that firm management (and existing shareholders) have superior information about firm’s assets and investment opportunities as compared to outside potential investors.

Some of these theoretical approaches predict that firm management can convey their special information to capital markets by their financing decision and dividend policy choices. These models indicate that increasing the proportion of debt in the capital structure or increasing dividends will effect firm value positively and value of the firm will increase.

Other approaches predict that capital structure can be designed for not taking suboptimal investment decisions emanating from information asymmetry. The Pecking Order Theory introduced by Stewart C. Myers (1984) and Nicholas S. Majluf and Stewart C. Myers (1984), has been the most accepted approach for determining long term financing choices among these studies. The Theory is based on the fact that if investors are less informed than firm insiders about the value of the firm, equity may be mispriced by the market. Due to asymmetric information, common stock prices will decline upon an announcement of new equity issues and firms may forego positive NPV projects in case they finance them with new equity shares. This underinvestment problem can be avoided if firms can finance their new projects using a security which is not so severely mispriced (undervalued) by the market. Therefore, internal funds or debt will be preferred to equity.

The Pecking Order Theory which predicts long term financing strategies, drives the firms to follow a hierarchy of financing. The Theory states that firms prefer internal equity (retained earnings) to external financing. If financing requirements for investments exceed retained earnings, debt financing is preferred to equity financing and firms issue new equity only as a last resort.

The Theory has been tested in numerous empirical studies. The inverse relations between profitability ratios and financial leverage and empirical findings of event studies investigating price reactions to financing decisions and dividend policies of firms have all supported the predictions of the Theory. Other studies undertaken for testing the Theory directly showed that the financing hierarchy predicted by the Theory has been followed by firms in developed countries and it has been superior to other theoretical models, such as Static Trade-Off Theory and Agency Theory.

The purpose of this study is to empirically test whether the Pecking Order Theory has been valid in Turkey. The study spans the period of 1988-1996 and a sample of 44 manufacturing firms quoted at the Istanbul Stock Exchange.

In accordance with the predictions of the Theory, the relations between investment decisions, financing decisions and dividend policies of the firms have been tested by three hypotheses using correlation analyses and multiple regression analyses. All hypotheses test the relevance of the model integrally. They are largely parallel to those in the studies of Jonathan Baskin (1985, 1989) and David E. Allen (1993, 1994).

First hypothesis is for testing financing decisions of the firms. Debt ratios were regressed against profitability ratios in current and previous periods and growth ratios for significant negative and positive relations, respectively.

Second hypothesis is for testing both financing decisions and dividend policies. Debt ratios were regressed against dividend payments in previous periods for significant positive relations between these variables. Multiple regressions also included the independent variables of the first hypothesis for financing decisions. In a sense, tests of the second hypothesis also support the empirical findings of the first hypothesis.

Third hypothesis is for testing the effects of dividend policies on investments and growth. Growth rates of the firms for the whole period were regressed against dividend payments in previous periods for significant negative relations between these variables. Other control variables that may influence growth, namely profitability ratios, financial leverage and firm size were also included in the multiple regressions.

Empirical findings of all multiple regressions consistently show that the Pecking Order Theory had not been valid in Turkey in the 1988-1996 period. None of the expected relations could be obtained between above mentioned variables, except the significant negative relations between financial leverage and profitability ratios in the last years of the research period. With these limited findings, it is impossible to conclude that debt ratios represent the cumulative need for financing of investments and firms follow the pecking order hierarchy.

Another purpose of this study is to determine the financing hierarchy of Turkish firms where the Pecking Order Theory proved not to be valid. Additional correlation and regression analyses were performed to find out the long term financing strategies of the firms.

Fourth hypothesis is for testing the effect of capital increases on investments and it may be considered as the alternative of the third hypothesis. Growth rates of the firms were regressed against capital increases with rights offer of the whole period for significant positive relations between these variables. Profitability ratios of previous periods representing internal funds and firm size were also included as control variables in multiple regressions.

Empirical findings support all the expected relations between the variables. Growth rates vary positively with capital increases which support the fourth hypothesis. They also have significant positive relations with profitability ratios of previous periods. Finally, growth rates vary inversely with firm size.

The results of this research confirm that Turkish firms finance their investments with internal equity. If external financing is required, equity financing is preferred to debt financing in contradiction with the Pecking Order Theory.

DEBATES OVER INTERNATIONAL COMPETITIVE ADVANTAGE: AN ASSESSMENT

It is undeniable that one of the most important issues currently on the agendas of strategists in firms and policy makers in governments is a through understanding of the sources of competitive advantage. Prof. Porter's highly influential book The Competitive Advantage of Nations, whose second edition has recently been published, has undoubtedly enlarged our understanding of competitive advantage. Apart from that, his research has also stimulated further applications of the framework as well as a lively debate over the issue of competitiveness. This debate is of special importance both for practitioners and management scholars since it significantly improves the existing understanding of the role of a nation in shaping the environment in which firms in particular industries create and sustain international competitive advantage.

In addition to the ten nations included in the original work, the framework offered by Porter to explain the sources of international competitive advantage has also been applied to other countries (e.g. Canada, New Zealand, Norway) and regions (e.g. Massachusetts). Most of them have been studied by the project teams headed by Porter himself, and they have been largely validating. Moreover, the majority of nations studied are developed countries.

Inspired by these facts, I applied the diamond framework to Turkey, a middle income developing country. Accordingly, after identifying the internationally competitive industries and clusters by using Porter’s methodology, I examine in detail five Turkish industries, namely, glass, construction, leather clothes, automobiles and the flat steel industries. The findings are generally supportive of Porter, meaning that the diamond framework works in a developing country setting. The results suggest, however, some major areas (e.g. domestic rivalry and the role of government) where one or more of the Turkish cases contradict the Porter hypothesis. The Turkish automobile industry, on the other hand, challenges Porter’s treatment of inward foreign direct investment. In this article, I integrate the results of this research to the debates over competitiveness in general and Porter's diamond in particular, and provide an evaluation and discussion of the key issues that should be reconsidered regarding the sources of competitive advantage.

The main purpose of the paper is to contribute towards a better understanding of the determinants of competitive advantage. Since an attempt to provide a complete discussion of all relevant issues here is simply not realistic, however, I instead focus on the six major areas that have been the subject of intense dispute regarding international competitiveness. These are namely, the national competitiveness debate, the relationship between international competitiveness and national culture, the disputes over the most appropriate geographic unit of analysis, disagreements on locating the source of advantage when there is substantial multinational involvement in an industry, the ideal role of government in the process of creating and sustaining competitive advantage, and the relative importance attributed to domestic versus international competition. In this abstract, I introduce the main focuses of these debates, on which I would like to elaborate further in the full paper, integrating the results of Turkish case studies.

The National Competitiveness Debate: There have been ongoing discussions concerning the highly diverse uses of the word ‘competitiveness’, especially concentrating on the question of whether it can be used at the national level (See, for instance, Krugman, 1994; and the responses to Krugman by various authors in Prestowitz, Thurow, Scharping, Cohen, & Steil, 1994). This debate also reveals a common misinterpretation of Porter’s work, who actually believes that we must abandon the whole notion of a ‘competitive nation’ (Porter, 1990).

The Role of National Culture: Another vital issue, that is the impact of national culture on the competitive advantage of a country, is raised by van den Bosch and van Prooijen (1992) who believe that Porter pays very little attention to it. Porter (1992: 178) argues that "advantage grows out of national and even local circumstances in the diamond, one of which is culture. The importance of cultural factors only reinforces the notion that a firm's home base remains crucial even in global competition" (Porter, 1992: 178). According to him, however, the influence of culture on competitive advantage is an indirect one since it acts through the determinants outlined in the diamond, rather than on its own.

The Geographic Unit of Analysis: The argument that culture is a national, regional, or even local phenomenon, when thought together with the high geographic concentration of internationally competitive industries Porter observed in his study, prompts a question regarding the relevant geographic unit of analysis. Porter's emphasis on the local environment has been widely criticized. Dunning (1993: 12), for instance, emphasizes the importance of globalization and integration in several parts of the world. Regarding the EU, he argues that national diamonds will have to be replaced by 'supranational diamonds' to be able to capture the true competitive advantages of the Community. Jacobs and Jong (1992: 239-46), on the other hand, argue that there is a type of dialectic relationship between divergence and convergence, and appreciate Porter's idea that globalization paradoxically leads to more emphasis on local conditions and consequently creates an opportunity for firms to take advantage of them. Yet others (e.g. Hodgetts, 1993; Rugman & D'Cruz, 1993; Rugman & Verbeke, 1993; Rugman, 1991) share the idea that double and/or multiple-linked diamonds may reflect the sources of competitive advantage better than Porter's single diamond framework for the smaller nations that are highly interdependent with one or more of the triad blocks (i.e. Europe, North America, Japan). These debates, of course, reflect the broader ones on regionalisation and globalization in general.

Locating the Source of Advantage: The Role of MNE: Another key issue I would like to discuss in the paper relates to the problem of locating the source of advantage when there is substantial foreign ownership and/or multinational involvement in a competitive industry, another highly criticized part of Porter’s study. Many critics are of the opinion that, for the MNE, determinants of the diamond are sourced all over the world, whereas, according to Porter, they are created within a nation, which constitutes the home base for that particular MNE. In his view, competition can be global but sources of advantage are local (Porter & Amstrong, 1992).

The Role of Government: Another interesting topic concerns the indirect role Porter attributes to government in the diamond framework. Relating to the debate in the area as well as the results from the Turkish case studies, I discuss the role of government in creating the context, which shapes the business environment. In the internationally competitive Turkish glass industry, which is capital-intensive, for instance, the role of the Turkish government has been rather direct.

Domestic versus International Competition: Similarly, in light of the information from the Turkish cases, I advance the debate on the importance attributed to domestic rivalry in Porter's study. Specifically, the Turkish glass and flat steel industries pose a challenge to Porter's ideas since we have two highly internationally competitive industries, where there is virtually no domestic rivalry.

CAN FUND PORTFOLIOS REDUCE RISK?: THE ROLE OF
INTERNATIONAL FUNDS

This paper explores the diversification benefits of equity fund portfolios within and across different fund types. Mutual funds have been long touted as providing a diversification vehicle for small investors. The continual quest for introducing new fund types with different investment objectives has resulted in investors forming fund portfolios. With funds typically including well more than the ten to forty stocks required to be well diversified, it would be interesting to find that diversification benefits can be attained through combinations of funds, or mutual fund portfolios. In this paper, we investigate the impact of adding international equity funds to a portfolio of domestic funds on both time-series and terminal wealth standard deviation. We find that over the January 1990 to March 1996 period, substantial reductions in terminal wealth standard deviation could have been attained by combining randomly selected funds in a portfolio. We also find that adding international funds to domestic fund portfolios results in a reduction in time-series standard deviation. However, the benefit of reduced standard deviation is swamped by low returns for most international fund types.

INTRODUCTION

Portfolio diversification has been extensively studied in finance literature. With few exceptions, this work has examined the impact of adding securities to a portfolio on its standard deviation. The evidence consistently shows that by the time 10 to 40 stocks have been added virtually all company specific risk has been effectively diversified away and portfolio standard deviation has been significantly reduced (see Evans and Archer (1968), Tole (1982), and (Statman (1987)). These results have been shown to hold for either domestic or international stock portfolios (Solnik (1974)). With segmented markets and imperfect correlation between domestic and foreign economies, international diversification appears to hold promise.

O’Neal (1997) extends this work by exploring diversification issues related to portfolios of mutual funds. Using simulation analysis, he focuses on mutual funds categorized as either Growth or Growth and Income by Morningstar Inc. He finds little evidence of reductions to time-series standard deviation (TSSD) by adding funds to mutual fund portfolios. O’Neal does however demonstrate substantial reduction in terminal wealth standard deviation (TWSD), as well as several measures of downside risk. For the five-year horizon he shows a reduction of greater than 80 percent (75 percent) in TWSD as the number of Growth (Growth and Income) funds in the portfolio was increased from 1 to 30. Reductions in corresponding TSSD were 8.7 percent and 10.6 percent, respectively.

O’Neal limits his analysis to fund portfolio diversification within the same investment objective category (either Growth, or Growth and Income). It seems reasonable to question whether his results hold for all fund types and whether the same relationships hold when diversifying across fund types. Perhaps, combining funds with different investment objectives may provide even more dramatic diversification benefits. This study explores the impact of diversification within and across fund types on time-series and terminal wealth standard deviation, with special emphasis on the impact of international diversification.

DATA AND METHODOLOGY

Morningstar OnDisc categorizes mutual funds by investment objective. All funds with investment objectives of Growth, Equity-Income, Growth & Income, Small Stock, Aggressive Growth, or International with fund inception dates prior to 1/1/1990 were included in the sample. International funds include those designated by Morningstar as Global, Foreign, Europe, Pacific, and Diversified Emerging Markets. Since Global funds include equity of both foreign and U.S. companies, we consider this group separately. From this point on International funds other than Global are referred to as Foreign funds. Monthly fund returns were collected for each of the 614 domestic, 74 foreign, and 28 global funds over the 1/1990 to 3/1996 period.

Of the 716 funds in the study, roughly 86 percent are domestic funds, 10 percent are international funds, and 4 percent are global funds. The domestic fund sample includes funds with growth, equity income, growth and income, small company, and aggressive growth objectives. In order to test the impact of the number of funds in a portfolio on its volatility, funds were selected randomly to form multiple fund portfolios in the following categories: growth, equity income, growth and income, small company, aggressive growth, foreign, and global funds.

Portfolios were randomly selected across all domestic fund types, domestic and global, and domestic and foreign fund types. This analysis is especially useful to compare the impact of diversifying across domestic funds with differing objectives to the impact of diversifying internationally.

For each category, terminal wealth level is calculated by equally investing $1 in each fund in at the beginning of the 75-month holding period. At the end of each month returns are added to the beginning value of the portfolio, and the total is rebalanced to preserve the equal weighting across funds. Following O’Neal (1997) the number of funds in the portfolio, n, was varied from 1 to 30. The process is repeated 1,000 times for each n, resulting in 1,000 terminal wealth values. Volatility of terminal wealth is measured by calculating the standard deviation of the 1,000 terminal wealth values. For each simulation run, average monthly portfolio return, and time series standard deviation of monthly returns are also calculated. Average monthly returns and time series standard deviations are averaged across simulation runs.

RESULTS

Compared to reductions in volatility, the difference in mean portfolio return and terminal wealth is relatively small for each n-fund portfolio as the number of funds goes from 1 to 30. For each fund type, reduction in time-series standard deviation ranges from a low of 7.4 percent for Growth & Income funds to almost 12 percent for Global funds. In addition, virtually all of the diversification benefits are achieved with between 5 and 10 funds. A similar but much more dramatic story is told for terminal wealth standard deviation, with reductions of between 79 and 89 percent on average. Again, most of the impact is achieved with between 5 and 10 funds. Results on Growth and Growth & Income funds are consistent with O’Neal (1997).

However, we must guard against making overly optimistic claims regarding the benefits of diversification. At the heart of the concept of diversification is that reductions in volatility can be expected without sacrificing return and associated terminal wealth. Results indicate that mean returns vary widely across fund types and are especially low for international funds. To this end we explore the impact of increasing the number of funds in the portfolio on the coefficient of variation (CV), where CV equals the standard deviation divided by the mean, or in this case, risk per unit of reward. We calculate CVs for return and terminal wealth.

For within-fund type portfolios, curves for the CV of return are almost flat for each fund type as the number of funds increases, indicating little diversification benefits. On the other hand, CV of terminal wealth improves substantially as the number of funds in the portfolio increases Again however, most benefits are attained with from 5 to 10 funds.

Since international funds are not separately broken down into the same investment objective classes as domestic funds, care had to be taken in considering the impact of international fund diversification. Benefits accruing to diversification across domestic fund types must be filtered out in order to assess the true benefits attributable to international diversification. To this end, we repeated the simulations across all domestic fund types for portfolios of 1 to 30 funds. Then, we added either Global or Foreign funds to the pool of domestic funds and repeated the simulations. Results indicate that the reduction in time-series standard deviation increased from 9 to 15 percent for Foreign funds, a 67 percent improvement. The result for Global funds was less dramatic with a 22 percent improvement in the percent reduction of standard deviation. Clearly, international fund diversification appears to offer a potential risk reduction for investors concerned with time-series volatility. CVs of returns reveal, however, that potential reduction in time series standard deviation accrues at the expense of returns, as adding either Global or Foreign funds to domestic fund portfolios increases CVs, resulting in curves that virtually coincide.

International diversification also does not appear to have a dramatic impact on terminal wealth standard deviation. For terminal wealth standard deviation, the curve for domestic funds virtually coincides with the curves of domestic and Global, or domestic and Foreign funds. With both lower average monthly returns and terminal wealth, and no apparent diversification benefits, international fund diversification appears to be a questionable strategy for investors with long horizons and associated concern for volatility of terminal wealth.

CONCLUSION

This study demonstrates the diversification benefits associated with randomly selected fund portfolios. Regardless of whether funds are diversified within or across fund types, substantial reductions in terminal wealth standard deviation are possible with as few as five to ten funds. For either terminal wealth standard deviation or time series standard deviation, international fund diversification does not seem to provide incremental benefits compared to domestic diversification within or across fund types. Reductions in time series standard deviation from international diversification are swamped by lower returns resulting in virtually no improvement in CVs.

Should investors include international funds in their fund portfolios? Evidence indicates that there is virtually no benefit to adding international funds to fund portfolios. Investors can achieve diversification benefits by forming fund portfolios within and across domestic mutual fund types. It should be kept in mind, however, that results here are averages based on a thousand random draws. Use of an optimization routine might improve the odds of investors making good draws with efficient portfolio allocations.

DOES THE JAPANESE STOCK MARKET PRICE BANK RISK?
EVIDENCE OF BANK FAILURES

Efficient financial markets promote aggregate economic welfare by allocating financial securities to their most potentially productive uses as signaled by market prices. Numerous studies have explored the efficiency of financial markets in the U.S. and other countries. At least for the U.S., the evidence suggests that markets have priced securities reasonably well on the basis of their underlying risk characteristics, particularly where this information is publicly available.

This paper examines the efficiency of the Japanese stock market in pricing the value of banks. During the 1990s, the condition of the Japanese banking system has deteriorated significantly. By the late 1990s, many of the larger Japanese banks appeared to be market value insolvent and most of the remainder barely solvent. Many were kept in operation by implicit, conjectural government guarantees. It is thus of considerable interest to explore whether the Japanese stock market is able to price the securities of Japanese banks efficiently and what signals it is sending about the current and future financial condition of the banks. This is particularly important in light of the ongoing attempt by the Japanese government to resolve the insolvencies, and restructure and recapitalize the banking system. Whether financial markets can be relied upon to discipline banks and supplement or replace regulatory discipline will be an important determinant in designing any new regulatory structure in Japan.

This study examines the efficiency of the Japanese stock market by analyzing the effects of the failure of two Japanese banks--the Hyogo Bank, Ltd. in 1995 and Hokkaido Takushoku Bank, Ltd. in 1997--on the market valuation of surviving Japanese banks. These two bank failures are interesting and important events to analyze for a number of reasons. First, Hyogo Bank, although small compared to other Japanese banks, was the first commercial bank allowed to fail since the end of World War II. Hokkaido Takushoku Bank was the first major (one of the top 20) Japanese banks to be permitted to fail. Second, although depositors suffered no losses in either bank failure, there were significant differences in how the two failures were resolved. With Hyogo Bank, regulators employed the traditional Japanese resolution process and asked the bank’s major creditors and shareholders to forgive the loans they made to the bank and to provide the funds necessary to capitalize a new bank that would take over Hyogo’s operations. Thus, they absorbed all the losses. In contrast, the bad loans of Hokkaido Takushoku Bank were sold to the Deposit Insurance Corporation, which absorbed the losses, and the healthy assets of the banks were eventually taken over by two other banks. At the time, Hokkaido Takushoku failure was viewed by many market participants as the end of the traditional system and the beginning of a new era in Japanese financial history.

The evidence in the paper suggests that the Japanese stock market prices the relative risk characteristics of banks in periods of severe banking crisis and incorporates new information quickly and reasonably efficiently. There is little reason to suspect that market discipline cannot be used by bank regulators in Japan more extensively to supplement regulatory discipline and to promote a safer and more efficient banking system.

Hesna Genay, Federal Reserve Bank of Chicago

Although Japanese banks were never as profitable as European or U.S. banks, they grew rapidly in the 1980s, buoyed by a strong domestic economy and rapidly increasing asset prices. However, economic malaise, ever-increasing problem loans, and low core profitability have taken their toll on Japanese banks during the 1990s. Today, even the best performing banks are facing liquidity pressures, some are struggling to stay afloat, and two major institutions have been nationalized.

Although there is little doubt about the current weak condition of Japanese banks, their precise financial condition is a matter of debate. Differences between the disclosure, accounting, and regulatory rules in Japan and other industrial countries make it difficult to assess the exact condition of Japanese banks and compare them with other international banks. The analysis in this paper furthers our understanding of the performance of Japanese banks by examining their recent accounting and stock market performance. In particular, the paper compares the financial characteristics and performance of Japanese and U.S. banks; relates accounting and market returns on equity to bank characteristics (size, asset quality, asset composition, capital position, efficiency, and liquidity) and measures of economic activity; and examines the relationship between accounting and stock market performance directly.

The results indicate that, compared to U.S. banks, Japanese banks are less capitalized, invest more of their assets in loans and equity securities, and rely more on deposits to finance their assets. Furthermore, Japanese banks were less profitable than U.S. banks during the 1991-97 period.

In addition, the results indicate that accounting performance of Japanese banks is related to certain bank characteristics in line with expectations and the results of other banking studies. However, accounting performance is correlated with some of the bank characteristics and economic variables in puzzling ways. Specifically, ROE is negatively correlated with loan loss provisions and the ratio of net loans to total assets, indicating that banks with higher credit risk perform worse than other banks. Banks with greater investments in equity securities also perform worse than others. On the other hand, unlike banks in other countries, Japanese banks’ performance is not significantly related to their capital position. Furthermore, higher returns on the Tokyo Stock Exchange, which imply more favorable economic conditions, are negatively correlated with banks’ ROE. Moreover, the negative correlation between market returns and bank earnings is stronger for banks with greater equity investments. These results are in direct contrast to our expectations. Additional evidence suggests that these puzzling results may be due to banks’ accounting practices. Specifically, Japanese banks appear to increase their loan loss provisions when their core earnings and the returns on the market are high.

The discretionary accounting practices of banks, however, do not affect their stock returns, which are correlated with bank characteristics and economic activity in a manner consistent with expectations. In particular, stock returns are positively correlated with the market returns; but are negatively correlated with loan loss provisions, increases in business bankruptcies, and the amount of banks’ equity investments. These results suggest that although accounting practices of Japanese banks may distort their reported earnings, investors see through the veil of accounting numbers when valuing bank shares.

When stock returns of banks are directly related to their accounting performance, the results indicate a significant positive correlation between the two performance measures. However, the results also indicate that the relationship between these measures breaks down after 1994: In recent years, investors appear to put no significant weight on reported earnings when they value Japanese bank shares. These results may reflect an increased use of discretionary accounting by Japanese banks to manage their income or regulator capital. The results are also consistent with increased regulatory forbearance, where insolvent institutions are allowed to continue their operations. In the presence of regulatory forbearance and deposit insurance, as the condition of a bank deteriorates, more of its value is derived from the value of regulatory forbearance and option value of deposit insurance, and less from the value of assets in place. The anecdotal evidence suggests that both discretionary accounting practices and regulatory forbearance may account for the results reported in the paper.

IS RENEWABLE ENERGY TECHNOLOGY
ECONOMICALLY COMPETATIVE IN THE
MARKET PLACES?

New developments in renewable energy technology, particularly solar building-integrated photovoltaic (BIPV)electricity-generating technology, have the potential for accelerating the rate of environmentally sustainable economic development. In sunny developing countries, like Brazil, the shortages of electricity in urban areas and many gaps in rural electrification make the application of BIPV particularly supportive of economic growth, by greatly reducing urban commercial and residential buildings' operating costs, saving much of the capital costs of what would otherwise without BIPV, and providing the most economical and environmentally benign form of rural electrification. The paper investigates, from the market point of view, whether the non-polluting BIPV is economically competitive for private sectors to invest as well as for public policy makers to promote. We have been motivated in this effort by a dual concern for economic development and the environment, and by the theoretical conviction, increasingly substantiated empirically, that the advancement of both can occur cooperatively and simultaneously, without one advancing at the expense of the other.

We have focused our analyses and argument entirely on demonstrating and confirming quantitatively that environmentally benign renewable energy technologies, particularly BIPV in Brazil and other sunny countries, are mature technologies bearing no more technological risk than their predecessors, and that they are economically competitive in many markets sharing the above characteristics. Many others have made the environmental argument alone, assuming that environmental benefits even at net economic costs would be sufficient to motivate public action and private investment. We have seen that this has not often been the case, and so decide that we have to demonstrate the purely economic rationale for non-polluting renewable energy technologies substitution for non-renewable polluting ones.

Two analytical approaches, from both micro economic perspective as well as macro economic perspective, are used in the study. Economic impacts of a decade's implementation of BIPV are computed and predicted for the next decade with the use of both micro and macroeconomic models. The net impacts on GDP depend on the degree of BIPV implementation. Degree of implementation is the principal component of uncertainty in the range of forecasts, and is the dominant policy variable in Brazil and most other sunny countries. Long-term benefits from reduced environmental pollution and reduced generation of climate-changing greenhouse gases as a result of BIPV technology, and its other environmental, health, and social net benefits are not computed in the present models, but are expected to strengthen the findings. The findings demonstrate that substituting renewable, non-polluting indigenous solar energy resources of BIPV sheathing and roofing for conventional building sheathing and roofing and fossil fuel-generated energy sources for building lighting, heating, ventilating, and air conditioning does not cost jobs and economic growth, but on the contrary even in the short run of a few years contributes significantly to economic growth and job creation.

1. Micro Economic Analysis:

Building curtain wall sheathing of good quality glass, aluminum, or stainless steel costs about $15 per square foot. Solar electric power-generating PV curtain wall sheathing costs about $25 per square foot now, but is expected to cost only $15 per square foot in five to ten years - about the same as the conventional metal or glass sheathing and roofing it would replace. SMUD (Sacramento Municipal Utility District in California) reports $12 per square foot costs for PV panels, producing electric energy at a cost of $3.00 per watt (including $1.50 for PV panels and $1.50 for balance-of-system wiring costs) by year 2002. Solar architect Gregory Kiss of Kiss and Cathcart, New York, estimates that electric power generating capacity of BIPV will double from 6 watts per square foot to 12, while prices will drop from current $25 per square foot to $15 within ten years. It shows an improvement from a current $4 per watt to $1.25 per watt. By comparison, SMUD module costs will be $1.50/watt in 2002.

Using the above BIPV cost and electric power output estimates and assuming $0.10 per kWh grid power cost, for construction of a medium-size six-story (100' x 50' 6-story) sun-facing solar-powered office or apartment building with 10,000 square feet of BIPV sheathing and roof, located free of shadowing from 20 to 40 degrees North Latitude (Key West to Boston) or 20 to 35 degrees South Latitude (Rio and Sao Paulo to Montevideo or Santiago), the current payback time for investment in optimal solar BIPV sheathing and roofing is about six years (3 to 4 years in Rio where power costs 18 cents per kilowatt hour), with a competitive internal rate of return requiring little or no risk discounting. In ten years, given current trends in PV cost reduction and efficiency improvement, payback will be about two years (or one year in Rio). After payback, grid-connected solar buildings will have zero or near-zero annual electricity costs, greatly reducing operating and life cycle costs.

2. Macro Economic Analysis

Our findings for the macroeconomic competitiveness of BIPV in Brazil are in the followings. Assuming the new building construction rates of up to 10% per year, Brazil would have to roughly double its central grid power generation capacity in the next decade, if no BIPV implemented. That is, from roughly 270,000 GWh/yr. in 1996 (Geller et al, 1997) or roughly 300,000 GWh/yr. in Year 2000, to about 600,000 GWh/yr. in 2009. With BIPV, to take care of additional loads not reducible by BIPV, such as shadowed building sites or added electric equipment in the many (over half the total by 2009) older non-PV retrofitted buildings, only about a 20% increase in capacity would be required. The difference in 2009 generating capacity requirements, without BIPV (business as usual) and with maximum implementation of BIPV is the difference between 420,000 and 600,000, or 180,000 GWh/yr. This is the maximum potential cost avoided by full BIPV implementation. The possible impacts on real GDP growth are between 1% to 5% depending on the building construction rates as well as the average saving per square foot for BIPV. The microeconomic competitiveness for BIPV, i.e., the incentives for private investors to initiate any BIPV project, is still under investigation.

Geller, Howard, Gilberto de Martino Jannuzzi, Roberto Schaeffer, and Mauricio Tiomno Tolmasquim, 1997, The Efficient Use of Electricity in Brazil: Progress and Opportunity, American Council for an Energy-Efficient Economy, Washington, D.C.

INTERDPENDENCE AND DYNAMICS IN CURRENCY FUTURES MARKETS:
A MULTIVARIATE ANALYSIS OF INTRADAY DATA

Elyas Elyasiani, Temple University
Ahmet E. Kocagil, Pennsylvania State University

 

Financial theory suggests that markets where asset price dynamics exhibit detectable patterns such as trends, interdependencies, or time lags may offer attractive arbitrage opportunities for traders. These patterns are also of interest to financial economists and policy makers since they may signal potential market imperfection problems such as delays between information arrival and valuation process, and asset mispricing. Hence, an empirical examination of currency interdependence and currency dynamics provides valuable insight for it sheds light on various facets of information processing and operational traits of foreign exchange markets. The following questions are of special interest: (a) prevalence of interdependence and spillovers in currency markets, and their strength and direction, (b) informational efficiency, (c) speed of convergence to a new equilibrium upon arrival of new information in the market. In this study, econometric time series tools including cointegration (CI), impulse response functions (IRF), and variance decomposition (VD) are employed to address these issues.

Specifically, this paper investigates the dynamics and interdependencies in currency futures utilizing intraday data for six major foreign currencies: the British Pound, Deutsche Mark, Swiss Franc, Australian Dollar, Canadian Dollar, and Japanese Yen. Empirical results show that lack of cointegration among the foreign exchange futures is the prevailing mode of behavior. However, temporary deviations from the no-cointegration condition are detected with the number of cointegrating relationships showing a steady decline over the 1988-1996 period. This latter finding may be due to advancements in telecommunication technology, deregulation of financial markets, and relaxation of trade barriers and capital movements across countries.

Results of the impulse response functions reveal that currency markets are in general very efficient and absorb new information within 1/2 hour. Variance decomposition analysis is performed for each currency in order to explore the determinants of fluctuations and to ascertain the importance of internal dynamics relative to the shocks emanating from other currencies. The results indicate that the interdependence structure of the currency futures is asymmetric in nature. Namely, the time pattern, and the magnitude of the effect of a particular currency future on another, are dissimilar to how the latter affects the former. In other words, a leadership-followership format is in effect. Currencies displaying the highest degree of spillover are found to be those included in the European currency basket namely, the British Pound, the Deutsche Mark, and the Swiss Franc.

The advantages of this study are fourfold. First, the high-frequency cointegration tests used here can manifest capabilities for quick information transmission and information processing beyond the limits of tests utilizing daily data. Second, the multivariate framework employed allows for a more complete delineation of interactions among currencies and is able to pick up multicurrency dependencies and short-term dynamics that cannot be detected within a single-equation approach. Third, the sensitivity of cointegration findings is checked with respect to model specification, sample period, and time span of the horizon. Accordingly, the tests are performed for five different model specifications and six different lag structures for the full sample period 1984-1996, as well as for each 12-, 6-, and 3-month sub sample period. Fourth, cointegration test results are complemented with impulse response functions and variance decomposition, in order to gain additional insight on price dynamics and interdependencies. Given the possible shortcomings of cointegration tests, when used in isolation, in assessing informational efficiency of markets, the combined evidence based on cointegration, implicit response functions, and variance decomposition analysis presented in this paper yields a more persuasive and complete picture of market dynamics and information flow in currency markets in comparison with previous studies.

THE EFFECTS OF ELECTRONIC MONEY ON
BANKING SYSTEM & ITS APPLICABILTY IN TURKEY

Within the global payments market cash dominates among all transactions since it is the product that delivers worldwide acceptance for both retailers and consumers. Cash is the means of immediate physical transfer of value. Other payment methods typically require some form of clearing and settlement. At this point electronic money appears as an alternative to cash. It was developed to truly replicate the core features of cash and to be real alternative to traditional bank-notes and coins.

If the presence of electronic financial instruments like bank/credit cards and electronic fund transfer system are taken into consideration, it can be thought that electronic money (e-money) is not a new concept entirely. However, to accomplish the cashless society, it is seen that the new system must be formed by developing new instruments that can be used for small payments which form the 70 percent of all currency circulation. Because, when the payment instruments other than cash are used in small payments, then the shopping becomes impractical and high cost appear for the merchants.

The various e-money products are still at a relatively early stage in their developments. Providers of products in various pilot projects and early nation-wide implementation indicate that potential exist for stored value cards and their network equivalents.

E-money concept which includes smart cards and digital cash is being searched by two types of researchers. First, by private companies which aim high earnings by providing a social benefit to the consumers. Secondly, by international organizations, governments and universities which all try to investigate positive and negative effects of e-money on the people and on whole economy.

E-money brings benefit as well as problems. One major advantages of e- money is its increased efficiency opening new opportunities, especially for small businesses. On the other hand it will encourage potentially the worsening the problems over taxation and money laundering.

It is possible that with the introduction of smart cards, private organizations will be able to issue electronic value outside regulated payment system. In this case it is important that consumers understand that stored value issued by a private organizations may not be as secure as currency or deposits issued by bank. If banks begin to create new money in the form of digital cash, there will be an opportunity for bankruptcies, the chain effect of which may easily lead to a virtual financial crises.

In developed countries and even in some developing countries, e- money concept is being investigated deeply while no serious research is being done except a few small applications in Turkey. Due to this reason , it is investigated e-money concept and its possible effects on Turkish banking system. Questions asked in this study are:

• Who will run the schemes in Turkey? Banks or other organizations?

• Do the technologies of e-money fall within existing banking regulations in Turkey?

• How do the technologies of e-money mesh with existing payment system in Turkey?

THE INFORMATION CONTENT OF THE TERM STRUCTURE OF INTEREST RATES

Petko S. Kalev and Brett A. Inder, Monash University, Australia

To explain the determinants of the shape of the term structureof nominal interest rates, researchers have employed a number of theories. The most cited theories, however, are the market expectations hypotheses (EH). A simple generalization of these theories demonstrate that the yield of the longer term interest rate is determined by an average of current and future yields associated with the shorter term interest rates plus a time-invariant term premium. Since the introduction of the efficient market hypothesis in finance during the 1960's, in general the expectations theory of the term structure of interest rates accommodates rational expectations (Muth 1961) as a necessary assumption. Different econometric models have been developed for testing the rational expectations hypothesis (REH), see inter alia Shiller (1979, 1990), Shiller et al. (1983), Campbell and Shiller (1987, 1991), Fama (1984), Fama and Bliss (1987), Hamilton (1988), Froot (1989) and Hall et al. (1992). Two main questions have been considered by researchers: 1) whether the REH of the term structure of interest rates holds, and 2) how much information about the future yields is contained in the current spot rates.

The empirical studies in general are not supportive to the REH, particularly when the term structure of the U.S. interests rates is considered. For a summary of the results on the predictive power of the spread on discount yields for U.S. Treasury securities, refer to both Table 1 in Rudebusch (1995, p. 249) and Table 1-4 in Roberds and Whiteman (1996, pp. 26-27) and the references therein. Based on U.S. and UK data, Driffill et al. (1997) further provide evidence against the REH, while Cuthbertson (1996), Hurn et al. (1995) and Taylor (1992) have employed UK yields. On the contrary, when the yield curve from other than U.S. and UK countries is examined, there is less empirical evidence against the REH, see for example, among others, Engsted (1996), Engsted and Tanggaard (1995), Estella and Mishkin (1997), Gerlach and Smets (1997), Hardouvelis (1994), Boero and Torricelli (1997).

A recent study by Johnson (1997) examines the validity of the REH of the term structure of interest rates based on postwar U.S. data from the McCulloch (1990) pure discount bond yields. Johnson considers tests based on the ex-post errors, expectations errors, built on the difference between the actual longer-term (n-period to maturity) yield rate and the equivalent one predicted by the REH from the shorter-term (one-period to maturity) yield rate. Although, the REH is rejected by other researchers, Johnson (1997) argues that a model, which he calls the ‘noise model’, might provide some useful approximation to the relationship, at least within the one month and other short-term discount bonds up to 12 months to maturity.

Similarly to Johnson (1997), this paper quantifies the expectations error and tests the level of falseness of the expectations hypothesis, that is, the strength of the departure of the yield curve from the expectations theory. However, there are some important differences. Firstly, our study considers the extended McCulloch and Kwon (1993) dataset of pure discount bonds. Next, we perform tests for the validity of the REH using zero coupon bond yields in a more general setting and on various maturities, starting from one month through to 60 months. Finally, we test the model by taking into account the serial correlation among the expectations errors under the REH. Johnson (1997, p. 1240) builds on previous research by Mankiw and Miron (1986), where one and two period to maturity interest rate series are under investigation. Of course, this is the only combination when the expectations error does not follow a moving average (MA) process. We stress that, if the time to maturity of the long-term bond, n, is more than two times the sampling interval, (eg. for interest rates sampled monthly m=1, n>2) then the expectations error possesses a

MA(n-2) structure.

In order to remove the MA(h-1) error structure, where the positive integer h gives the fraction between the forecast and the sampling interval respectively, the natural response is to apply generalized least squares (GLS) estimation. However, there are some pitfalls when one estimates a model with RE, see Flood and Garber (1980), Hansen and Hodrick (1980). To ensure consistent and asymptotically efficient estimation, we consider a forward filter (FF) as introduced by Hayashi and Sims (HS) (1983). Transforming both the dependent variable and the independent variables with the Hayashi-Sims forward filter (HSFF), we apply an instrumental variable (IV) procedure on the transformed regression equation using as instruments untransformed variables similar to those considered by Johnson (1997). Based on Hayashi-Sims' instrumental variable (HS-IV) estimation, our results indicate that a significant amount of information freely available to market agents is not incorporated in forming people's expectations. In contrast, if one uses OLS regressions, the information content in the expectations errors is significantly understated. Our results, in general, agree with the findings reported by many other researchers when the postwar US term structure is investigated, the REH is tested and consequently rejected.

INDETERMINACY OF LONG RUN EQUILIBRIUM VALUES
OF INTEREST AND PROFIT RATES

The purpose of this paper goes beyond the history of economic thought. The stagnating economic problems in many parts of the World call for a reconsideration of some basic concepts of economics, in the hope that a better understanding of them shall make a better guidance for economic policy.

Classical economists ( Ricardo ) considers the long run interest rate a real variable regulated by the profit rate on capital . The classical theory also argues that the interest rate is a measure of both the private and social marginal product of capital. Changes in the money supply are thought to be neutral with respect to affecting the interest rate in the long run. Keynes vehemently rejected the argument , for the MEC depends on investment and we must already know the rate of interest before we can calculate what that scale would be. Fisher ( 1930 ) defended the classical position , arguing that interdependency among variables does not necessarily imply indeterminacy as can be shown by a set of simultaneous equations. Fisher 's defense is flawed , for it deals with the problem as one of a single agent who tries to optimize saving and consumption decisions over two periods. An important point , however is that we have two problems in reality ;one of determinacy ,and another of causality.

An interest - based loan contract involves a peculiar kind of exchange. It implies the exchange of abstinence in return for a promise . But abstinence is hardly an object of exchange or a cause of reward, as Fisher himself explained . While the interest rate represents a contractual reward for the lender , it is not obvious what is the contractual reward to the borrower , if any.

The model in this paper shows diagrammatically that we can not know the position of the investment schedule unless we know the rate of net profit. But we can not know the rate of net profit unless we know the interest rate. But we can not know the interest rate unless we know the position of the investment schedule ( determined by net profit rate ) . Thus we are trapped in a circle. Mathematically , it is shown that the problem involves more unknowns than equations. Equilibrium can not be assumed to have existed ex anti . It is the thing to be proven to exist , if ever . This is the essence of Keynes' argument. But even if finance is to be done on the basis of profit sharing, the existence of interest rate renders the system indeterminate. The two modes of finance can not coexist for the purpose of system determinacy. The removal of the interest rate from the economy makes it determinate. ( JEL B12, E43 , G11 , P 43 )

A HIDDEN MARKOV CHAIN MODEL FOR THE TERM STRUCTURE OF
BOND CREDIT RISKS SPREADS

This paper provides a Markov chain model for the term structure and credit risk spreads of bond processes. It allows dependency between the stochastic process modelling the interest rate and the Markov chain process describing changes in the credit rating of the bonds by their mutual dependency on a hidden Markov chain. This Markov chain can be thought of as the underlying economic conditions. The model also allows a new interpretation of risk premia used in previous approaches. It also uses a linear programming approach to strip the bonds of their coupons in such a way as to guarantee there is no mis-pricing.

Corporate bond pricing models have been in existence for twenty-five years but it is only recently that a pricing model, which incorporates a firm’s credit rating as an indicator of the likelihood of default, has been developed. This is surprising since the rating of a company given by the major international credit rating agencies is the most widely available estimate of the credit risk involved in investing in the firm’s bonds. The first model of bond prices to incorporate credit ratings (Jarrow, Lando, Turnbull, 1997) assumed that the stochastic process describing the rating and possible bankruptcy of the firm was independent of the stochastic process giving future interest rates and hence the default-free bond prices. This paper presents a generalization of this model in which the two processes are dependent through their relationship with the stochastic process describing the state of the underlying economy. The model also generalizes the idea of risk premia adjustments by reinterpreting them as beliefs that the future of the rating and bankruptcy process is more extreme than it has been historically. This paper also introduces a procedure based on linear programming for stripping out the zero-bond prices for risky and riskless bonds in a way that guarantees there is no mis-pricing.

Models of bond prices take zero-coupon bonds as their basic entity, whereas most bonds have coupons which involve part payments during the life of the bond, as well as the redemption value to be paid on maturity. Thus there is a need to strip out the coupons and calculate what the market price of the bond implies about the value of a bond that will just pay 1 unit at time t. Some authors (Longstaff, Schwartz (1995 )) take the average bond price, coupon rate and maturity each month for over a given time period and fit a regression line. The data however will include the changes over time in market sentiment and so does not reflect the position at a given time. Jarrow, Lando and Turnbull (1997) split bonds into classes depending on their credit rating and their maturity. For each class the average market price and average yield were taken to be the values for bonds of that rating and maturity. Solving a triangular system of equations gave the zero-coupon bond prices. However, there was some mis-pricing of their bonds with their calculated zero-coupon bond prices not necessarily increasing as the credit rating improved nor decreasing as maturity increased. One can set up the problem of stripping out the coupons to get zero-coupon bond prices from bonds with coupons as a linear program. The model can include constraints that ensure rational pricing. In effect lower grade bonds have lower prices. We used data on US bond prices and credit ratings obtained from DATASTREAM and Standard and Poor (Standard and Poor 1997a, 1997b) respectively. 64 of the US Treasury Bonds which make up the DATASTREAM US yield curve data set in 1995 and 1996 were taken as the riskless bonds. Their market price on 3rd July 1996 was taken -the data being chosen as an example of a mid-week, mid-year, pre-holiday period. The set of risky bonds satisfied three criteria. They were in the DATASTREAM database of US industrial and US financial bonds; their market prices and S&P rating for 3rd July 1996 were available; there were no callable dates. The extra option that being callable gives a bond is more difficult to strip out of the price than the coupons. There were 178 such bonds in total (7 rated AAA, 24 rated AA, 61 rated A, 68 rated BBB, 12 rated BB, 6 rated B). DATASTREAM does not usually record the prices of C-rated speculative bonds but there were 8 bonds in the set that moved from C to investment grade or vice versa during the year (Standard and Poor 1997b) and hence we were able to obtain the 1996 market price when they were C-rated.

The use of a hidden Markov chain model for the term structure of credit risk spreads which further extending the ideas in Lando (19994), Jarrow and Turnbull (1995) and Jarrow, Lando and Turnbull (1997) proved to fit the bond prices more accurately. Novel features of our paper include the introduction of dependency between the rating process and the interest rate process through their joint dependency on a state of the economy process. The paper also provides a reinterpretation of the idea of risk premia introduced therein as the chance the markets view of the rating changes is more extreme than has been the case in the past. Finally, the utilization of linear programming provides a way of stripping the coupons for bonds in such a way as to minimize the mean absolute errors and at the same time ensure there is no miss-pricing of the zero-coupon bond prices

THE VALUE OF ANALYST RECOMMENDATIONS

Analyst recommendations and their impact on stock prices have long been of interest to individuals and institutional investors, as well as academic researchers. The specialized analyst provides a cost effective alternative to in house information processing, especially for individual investors and smaller capitalization and low-expense funds. Also, as a result of the expansion of the investment universe, both the number and scope of the analysts evaluating the stocks and reporting has increased. As in any area of subjective evaluation, analyst recommendations can differ significantly for the same stock, ranging from a strong buy to a strong sell.

The motivation of this study is to establish a methodology to evaluate the aggregate information produced by the analysts covering the investment universe and to study the potential of trading gains using investment strategies based on the advice given by the brokerage analysts. The study examines the investment value of the recommendations issued by the analysts and the factors that influence the level of the recommendations. This study uses two different data sets that cover a greater percentage of both the brokerage analyst and the investment universe compared to previous studies. The first data set is collected from the Bridge Financial Information System and originates from Technimetrics & Zacks Research. The evaluations were made by different analysts over the April 1997 to December 1997 period. The second data set originates from the S & P Ace system which includes 200 brokerage analysts' recommendations issued over the September 1995 to December 1997 period. In both data sets recommendations for each stock are classified into one of five groups: buy, accumulate, hold, avoid, and sell. The total number of analysts covering a stock is calculated by aggregating the number of analysts over the five groups.

The study consists of three major analysis. The first analysis finds recommendation levels for all the stocks using the given information. The second part of the study looks at different characteristics for a given stock and tries to find the factors that influence the stock's recommendation level. The final part focuses on the existence of a potential profitable trading strategy which uses the published analyst recommendation information.

Initial analysis of the data confirms the findings of previous research. The recommendation distribution is highly skewed. For all groups, avoid and sell recommendations are extremely rare. Two main reasons for the hesitance of analysts in issuing a negative sentiment are the relationship between the investment house and the company being the primary source of information for the analysts and reputational effects for the analyst. An analyst's reputation can be hurt more if he or she issues a sole and incorrect sell order as opposed to an incorrect buy order with the crowd. In conclusion, analysts appear to be herding on buy and accumulate groups and this fact seems to be the case in many different periods, regardless of the general market direction. Analysts also have a responsibility towards their investment clientele to provide timely and accurate information about their analysis, which sometimes may require the issuance of an unfavorable recommendation. Analysts do use the five different groups to signal negative sentiment without issuing an explicit sell advice. They may change the category the stock is in, i.e. from a buy to a hold, which practitioners interpret as an implicit sell signal. The measures used in this study to calculate recommendation levels are designed to incorporate such an effect since all of the measures calculate the aggregate information level.

The study gives the breakdown of the stock characteristics calculated for portfolios formed based on the four different measures. The analysis is repeated for the restricted case where the total number of analysts is required to be at least five. Compared to the average of the deciles, the buy portfolio has a much higher beta whereas the sell portfolio has a lower beta for all of the datasets. This indicates that the analysts tend to recommend stocks with higher market risk to be included in the portfolios. This may also be the result that recommended stocks have high past momentum and thus have high betas. Given the uptrend in the stock market during the sample period, this profile would be the expected choice of the analysts. Among the larger stocks, i.e. Dow Jones and S&P 500, analysts tend to recommend larger stocks. However, for the whole market, both the buy and sell portfolios are composed of small capitalization stocks. The buy portfolios generally consist of high market-to-book stocks whereas the sell portfolios include stocks that have low market-to-book values. The most influential characteristic is the return momentum. In all datasets, the buy portfolio has the highest return momentum and the sell portfolio, the lowest. Analysts seem to choose stocks that have already gone up in price substantially.

Although earnings momentum does not seem to have a clear cut distribution among the different deciles, buy groups generally have higher earnings momentum compared to the sell groups. Another very strong characteristic is the dividend yield. For all the datasets, buy portfolios have much lower dividend yields than the sell groups. Analysts tend to prefer low dividend yielding stocks. Finally, buy portfolios have a much higher strength (lower volatility) than sell portfolios. In conclusion, the main characteristics influential in a favorable analyst sentiment appear to be return momentum and dividend yield. Analysts tend to choose stocks that have already significantly gone up in value in the past six months. These stocks tend to be smaller in market capitalization relative to the overall market. In order to further study the effects of these characteristics in the selection process of the analysts, cross-sectional regressions were run on the data. The results indicate that return momentum and dividend yield seem to be the most important factors in the issuance of a positive sentiment by the analysts. These variables also affect the strength of the recommendation.

After the formation of the buy and sell portfolios, the returns and performance relative to the market indices are calculated. In line with previous research, most of the return to the buy-sell portfolio comes from the buy portfolio returns. The stocks recommended by the analysts seem to earn high returns. Moreover, the stocks that analysts shy away from earn very low returns. However, when compared to the market, the buy-sell strategy does not have a significant over or under performance. The restriction that requires the total number of analysts to be greater than five seems to increase the returns to the trading strategy. This can be explained by the fact that as the number of analysts covering the stock increases, the level of information produced also increases. As a result, the probability of recommending stocks that will perform in the future also increases.

The best performing strategies follow the restricted strength adjusted and reward-to-risk criteria. The buy-sell portfolio returns exceed the returns of the equally-weighted index but the results for the value-weighted index are mixed. The differences, though, are not statistically significant but this may as well be a result of the relatively short sample period. In conclusion, the stocks that the analysts recommend tend to have similar characteristics and these stocks tend to be fairly, but not excessively, compensated for these characteristics. Finally, the Fama-French three factor results support the hypothesis that an investor who follows the recommendations does well but no better than the market itself.

Stephen E. Christophe, George Mason University
Richard W. McEnally, The University of North Carolina, Chapel Hill

A persistent theme in the popular investments literature is that the stocks of U.S.-based multinational corporations (USMNCs) provide a viable vehicle for obtaining international portfolio diversification for investors who are unwilling or unable to hold shares of non-U.S. companies. For example, a recent edition of The Wall Street Journal (WSJ) contains an article that proclaims prominently: "Searching for overseas stock plays? Consider U.S. multinationals...(WSJ, June 1, 1993)." In the article, an investment strategist contends that U.S. firms with international operations are "more of a play on the global economy than on the domestic U.S. economy." The underlying rationale for this contention appears to be that USMNCs have substantial exposure to non-U.S. economies and thus their stock prices should behave in much the same way as stocks of companies domiciled in these foreign countries.

On its surface, the proposition seems somewhat unlikely, especially because its validity depends on a questionable implicit model of the relationship between economic exposure on the one hand, and investor returns on the other. In addition, the notion is not accepted universally by all practitioners, as evidenced by a subsequent WSJ article in which a number of different investment analysts contend that, while the stocks of USMNCs exhibit price swings that track quite well with swings in the U.S. market, they do not track closely with movements in foreign markets (WSJ, June 27, 1996).

Nonetheless, since many U.S.-based investors face either perceived or actual impediments to international investment, and since the benefits of international portfolio diversification are well-documented, the issue of whether USMNCs can provide indirect global diversification benefits is of considerable practical importance. Moreover, the only serious study of the topic (Jacquillat and Solnik, 1978) of which we are aware is quite dated, as it examined the return behavior of 40 European firms and 23 American firms during a period covering the late 1960s and early 1970s. They concluded that investing in the stocks of U.S. multinational firms was an ineffective method for obtaining international portfolio diversification.

In more recent years, however, it is possible that the diversification characteristics of USMNCs have changed as the global orientation of companies has increased and as the volume of international trade has expanded. Consequently, the issue is worthy of a new look. This paper provides such a look.

Since the potential benefits of a global diversification strategy are well known, they do not require substantial elaboration. In short, with a globally diversified portfolio, investors have been able (at least historically) to increase return at a given level of risk, or reduce risk while maintaining a desired level of return, or do both. These results arise from the comparatively low correlation of returns across different international equity markets

To investigate whether USMNCs can provide indirect access to the financial diversification benefits of a globally invested portfolio, we proceed initially by estimating a modified market model. The monthly returns of individual USMNCs are regressed on the monthly return on a U.S. market index, a Europe market index, and a Pacific market index. If USMNC stocks provide global diversification benefits, it should be evidenced by a significant co-movement between the USMNC's returns and the returns in foreign markets, and should result in statistically significant foreign market beta coefficients.

The sample of USMNCs considered in this study is drawn from the Forbes list of "The 100 Largest U.S. Multinationals" published in the July 17, 1995 issue. Forbes selects companies for inclusion in this list solely on their ranking on absolute dollar revenues abroad. In the interests of examining a manageable group of these 100 firms, we reduced this sample to the 50 firms with the highest proportion of foreign revenues relative to total revenues. Our supposition is that this subset of companies is the most likely to exhibit a significant correlation with the stock returns in foreign markets, if, indeed, any USMNCs exhibit this characteristic/tendency. Monthly return data for these 50 firms, and for domestic and foreign market indexes were collected for the 1986-1995 time period from the Datastream International data set. The results show that USMNCs exhibit a positive and statistically significantly U.S. stock market beta ranging from a minimum 0.280 to a maximum of 1.796. However, the foreign stock index betas for the USMNCs are not as impressive. Only three firms have statistically significant foreign betas associated with the Europe index, and only one firm has a statistically significant foreign beta associated with the Pacific index.

An extension of the empirical analysis considers how USMNC returns covary with individual foreign country stock market returns. The results show that breaking down the foreign region indexes into separate country-level indexes does little to enhance the case for USMNCs as vehicles for global diversification. Next, drawing on a methodology developed by Sharpe (1988, 1992) to investigate the management style of mutual fund managers, we use a variation of his approach to examine the regional sentiment of our equally weighted portfolio of multinational firms. We regress the monthly returns for our equally weighted portfolio of fifty USMNCs on the U.S., Europe, and Pacific market indexes while constraining the coefficients to sum to unity and also constraining them to be non-negative. Under this methodology, the return on the USMNCs can be interpreted as a pseudo portfolio constructed from the three stock indexes (U.S., Europe, and the Pacific), with the regression coefficients indicating the weighting of a particular index in the pseudo portfolio. The results of this estimation imply that the portfolio of the USMNCS effectively weights the returns of the U.S. stock market 98.9%, the Europe stock market 1.1%, and the Pacific stock market 0.0%. Consequently the overwhelming sentiment of the USMNC's returns is with the U.S. stock market

What all these findings confirm is that one cannot obtain a globally diversified portfolio by investing in the common stocks of U.S. multinational corporations. What then, must an investor do to obtain the potential gains from global diversification? And, in particular, is it necessary to transact through a foreign securities markets?

Fortunately, the answer to the latter question is no. It is possible to purchase securities that exhibit high correlations with foreign stock markets here in the U.S. through the purchase of American Depository Receipts (ADRs). We examine the return behavior of forty that have sufficient return histories available on Datastream International to estimate regressions over the 1986-95 time period considered in the preceding analyses. The results show that thirty-seven out of forty ADRs have statistically significant betas associated with their home country market return. Interestingly, nine ADRs have significant U.S. market betas.

In conclusion, the notion that the stocks of USMNCs can be used by U.S. investors to obtain global diversification benefits is not supported by the empirical evidence presented in this study. Rather, it seems that USMNC returns tend to move with the swings of their home market, irrespective of the location of the firm's operations. Fortunately, given the recent proliferation of new investment vehicles such as ADRs, international portfolio diversification seems readily available through other alternatives.

ESTIMATING BETA

Haim Shalit, Ben-Gurion University of the Negev
Shlomo Yitzhaki, Hebrew University of Jerusalem

The valuation of risky assets is one of the major research tasks in financial economics. The valuation question has led to the development of several Capital Asset Pricing Models, the most popular of which is the Sharpe-Lintner-Black mean-variance CAPM. In this model, the typical measure of asset riskiness is the beta, or the covariance between asset return and the market portfolio return. The covariance formula allows one to distinguish between systematic risk -- the risk common to all the assets in the portfolio-- and non-systematic risk, which can be eliminated by proper investment management.

All models that represent the investor as an expected utility maximizer characterize systematic risk using a covariance formula. The differences among the various models have to do with the exact specification of a formula for the covariance between marginal utility of wealth and asset return. Since this covariance cannot be observed, valuation models identify risk by a specific measure of variability, like the variance or the semivariance, and substitute for the latent covariance a covariance between the market portfolio and an asset's return.

Empirical asset-pricing models have attracted tremendous attention in finance, the goal being to assert or refute whether CAPM holds. The traditional technique used to estimate the risk-expected return relation consisted of two stages. In the first pass, betas are estimated from time-series. In the second pass, the relation between mean returns and betas is tested across firms or portfolios. The methodology has been subjected to many criticisms and improvements starting with Fama and MacBeth (1973) who introduced a rolling technique, following by proponents of maximum likelihood estimation such as Gibbons(1982), Stambaugh(1982), and Shanken (1992) to name a few. CAPM suffered a major setback from a series of papers by Fama and French (1992, 1993, 1995, 1996, and 1997) who claim that beta itself is not sufficient in explaining expected return. All these findings point to a major question: Is beta relevant in finance or merely mis-estimated?

In this paper, we do not address the issue of CAPM testing, but look mainly at the first pass regression and ask whether the standard procedure to estimate systematic risk is compatible with financial theory. In particular, we show how the regression technique used to measure beta risk is not robust with respect to wide market fluctuations. The sensitivity of beta to the presence of extreme observations can give rise to data mining and lead the way to find of peculiar relationships.

We argue that beta sensitivity can be traced to a combination of two factors:

(i) Incompatibility between the statistical methods used and financial theory. In particular, the Ordinary Least-Squares (OLS) regression estimator is based on a quadratic weighting scheme that tends to contravene the assumptions of risk aversion.

(ii) Probability distribution of market returns with "fat" tails; that is the data do not follow a normal distribution.

These elements make beta sensitive to market fluctuations and make OLS inappropriate for estimating betas.

We suggest alternative estimators for beta that are robust with respect to extreme fluctuations in the market return. In this sense, we follow the arguments raised by Chan and Lakonishok (1992) and Knez and Ready (1997) for the use of robust estimation procedures, but our rationale is different. By using trimmed regressions to seek robustness, crucial information regarding the behavior of securities returns with respect to market portfolio is removed for the sake of efficiency. The data that has been deleted may be considered by some investigators as the most valuable because it exhibits the range of returns distribution. We do not think that robustness should be attained at all cost and hence we do not seek robustness by using statistical methods that are less sensitive to wide fluctuations in general. Rather we seek to identify, according to economic theory, the relative weights that should be attached to different fluctuations. Adjustment of the weighting scheme according to economic theory allows improvement of the estimator for beta at low cost.

To document the magnitude of the sensitivity of beta to market fluctuations and to avoid any influence of small or unusual companies, we consider first the 30 firms in the Dow Jones Industrial Average (DJIA) and second, 20 portfolios built with the 100 largest traded firms. We use CRSP daily returns for a period of ten years from January 1984 through December 1993, a total of 2528 observations. Using these data, we conduct two experiments. In the first, the four best market performance observations, based on the S&P 500 Index are removed from the sample, and the betas are re-estimated. In the second test, the four best and the four worst observations of the market are deleted, a total of less than 0.3 percent of the entire number of observations.

When the four worst and the four best market returns are removed the betas of 7 firms (out of 30) change by more than 4 standard errors. Moreover, the betas of almost 25% of the firms change by less than one standard error. When only the four best observations are omitted, then the betas of 9 firms (30 percent of the firms) change by more than one standard error. The impression, however, one gets from the standard errors using the entire sample is that the probability of such occurrences is nil. These conclusions stand also for beta ranked portfolios which are used to alleviate the problem of sharp return fluctuations for individual firms. When the four worst and four best market returns are omitted, the betas of 7 portfolios (out of 20) change by more than 3 standard errors. When the four best observations are deleted, the betas of 9 portfolios change by more than one standard error, confirming what was exhibited with individual stocks.

These results indicate a great sensitivity to extreme market fluctuations, which raises doubts as to the robustness of the CAPM to choice of the sample period and specification of the model. This sensitivity exists both for upward and downward movements of the market. While sensitivity to extreme downturn market fluctuations can be justified by arguing extreme cases of risk aversion, it is not easy to explain sensitivity to extreme upward market movement.

In order to avoid redundant sensitivity in the estimation of beta, alternative estimators, based on the Gini index of variability, are proposed. The alternative , that are both more robust and better represent investors' risk aversion than the OLS estimators.

The paper compares the performances of those alternative estimators of beta and shows the conditions under which the alternative estimators should be preferred over the OLS.

THE SYSTEMATIC RISK OF AUSTRALIAN MULTINATIONAL CORPORATIONS

Previous research has provided conflicting views on whether the systematic risk of a domestic corporation (DC) changes when its scope expands to include overseas operations. The limited research in this area is primarily focused on large US-based multinational corporations (MNCs). Earlier evidence suggests that the systematic risk of US-based MNCs decreases when their cashflows originate from operations in different countries, implying the use of lower discount rates to evaluate overseas projects. However, the results cannot be generalized because they are based on small sample comparisons of MNCs and DCs due to matching constraints. Moreover, more recent evidence suggests just the opposite, i.e., an increased level of systematic risk corresponding with an increased level of internationalization. For example, Reeb, et al (1998) analyze two samples of 880 and 844 firms during 1987-96 corresponding to two different measures of internationalization. They find that the systematic risk of firms increases as their operations become more internationalized. Their results are robust to different measures of internationalization, over different periods and after controlling for size and leverage differences, as well as different market indices used to estimate systematic risk.

The purpose of this paper is to examine the relationship of systematic risk to the degree of internationalization of firms based in the relatively smaller market of Australia. Our analysis also focuses on previously unexamined issues such as whether differences in market size, industrial composition, institutional and competitive environments substantially affect the relationship between systematic risk and the degree of internationalization. The main research questions addressed in this paper are:

• Is the systematic risk of MNCs based in a smaller market also related to their degree of internationalization? How do these results compare with previous US-specific evidence?

• Is the relationship between systematic risk and the level of internationalization of Australian-based MNCs affected by confounding factors such as firm size, leverage and growth opportunities?

• Is the relationship between systematic risk and the level of internationalization affected by the industrial classification of MNCs? Is this relationship different for resource-based MNCs versus industrial MNCs?

• Is the relationship between systematic risk and the level of internationalization affected by equity market conditions? Is this relationship likely to be stronger during bullish or bearish markets?

This study covers the period 1992-98, which includes periods of bullish and bearish market conditions in Australia. Monthly data on stock returns and market index returns and the risk free rate are obtained from the Australian Graduate School of Management (AGSM) database. Annual data on total sales, total assets, total book value of debt and equity are obtained from the Connect 4 database.

Following previous research, and to facilitate comparisons, we use two measures to proxy for the degree of internationalization: (a) the ratio of foreign sales to total sales and (b) the ratio of foreign assets to total assets. The method used to estimate systematic risk is similar to that used by Reeb, et al (1998) to allow a direct comparison with their results. The method involves forming two datasets corresponding to the two proxies of internationalization. For each year, the systematic risk for firms in each dataset is estimated over a three-year period (the current year and two previous years). To minimize measurement errors, firms are grouped into portfolios of 15-20 firms, each according to their degree of internationalization in that year. Next, for each data set the following cross-sectional regression is estimated:

b p = a0p + a1p l p + a2p DTAp + a3p LNAp + a4p MBEp + up,

where b p is the equally-weighted portfolio beta, l p is the equally-weighted average degree of internationalization for portfolio p in each period, DTAp is the portfolio’s debt to total assets ratio, LNAp is the portfolio’s natural log of assets, MBEp is the portfolio’s market to book value of equity ratio, and up is the random noise term. The variables DTAp, LNAp and MBEp are included to control for differences in leverage, size and growth opportunities, respectively.

The null hypothesis tested is that a1p equals zero, if there is no relationship between systematic risk and the degree of internationalization. A significant positive (negative) a1p implies that internationalization increases (decreases) the systematic risk of the firm. To examine the relationship between systematic risk and industrial classification, the full sample is split into sub-samples of resource-based and industrial firms. The estimation process outlined above is then repeated. Next, a1p is estimated for each sub-sample to examine (i) whether the coefficient is significantly different from zero and (ii) whether the coefficients for the two sub-samples are significantly different from each other. To examine the relationship between systematic risk and the degree of internationalization under different market conditions, the above procedure is repeated by separating the sample into bearish and bullish market periods and comparing the estimates of a1p across these periods.

NOTE ON ASSET PROPORTIONS, STOCHASTIC DOMINANCE, AND 50 % RULE

This short note explores the demand problem in portfolio theory. The question which arises here is under what conditions is one asset more demanded than another by considering the more general case of interdependent assets.

In our framework, we use the same methodology as Hadar and Seo (1988) by considering a representative set of all risk averters. Our study provides conditions on the conditional distributions of the two assets which are sufficient for an asset to be more demanded.

Consider the two assets portfolio problem where X and Y denote the returns on the two risky assets. If k denotes the proportion of total wealth, , invested in X. A risk-averter agent solves the portfolio selection problem :

Where E is the expectations operator.

The optimal amount k* is the solution to :

We consider a representative set where an element is defined by :

is called a non-increasing unit step function and this new representative set is a proper subset of the set of positive non-increasing functions.

Recall the representative set used by Hadar and Seo (1988), it was a proper subset of the set of positive concave functions. We get the following theorem by using the same methodology as Hadar and Seo (1988) :

Theorem 1. Every agent in the new representative set prefers X to Y if and only if every agent with positive non increasing marginal utility function prefers X to Y.

This result enables us to use the new representative set in order to solve the demand problem.

Theorem 2. Let Z be the equally weighted index (). If for any given outcome of Z the cumulative conditional expected output of X up to this outcome is greater than the one of Y, then the percent of any investor’s portfolio invested in asset X will be greater than 50 % .

Mathematically, this can be expressed as :

where is the density function of Z.

Practically speaking , the last equation means that given the outcome of an equally weighted index, if the cumulative average of X is greater than the cumulative average of Y, then at least 50 % of the portfolio will be invested in X.

The integral in the equation is nothing more than the sum of the conditional expected outcomes of each asset weighted by the probabilities of the outcomes of the index.

We have characterized conditions under which the problem demand is solved in presence of dependent assets. The conditions hold on the distributions of the different assets rather than on the utility function.

CLOSE-END EQUITY FUNDS: BETTING ON DISCOUNTS AND PREMIUMS

Closed-end funds have grown in number rapidly since their inception in the 1980’s. Currently, hundreds are traded on stock exchanges in the United States, London and Hong Kong. They are closed-end mutual funds meaning they initially issue a fixed number of shares that trade on an exchange. Subsequently, there is no need to continuously issue or redeem ownership shares.

It is commonly observed that closed-end funds trade at large and variable discounts or premiums. This has puzzled the finance community since the net asset value of these funds and their share prices are two ways of evaluating the same portfolio of assets. A number of studies have attempted to explain these value discrepancies. The reasons given are barriers to arbitrage, partially segmented capital markets, irrational investors or asymmetric information.

This paper seeks to answer a different question concerning discounts and premiums on closed-end funds. The primary concern is whether or not an investor could have used the premiums and discounts to earn positive excess returns over a benchmark index.

Similar questions have been explored in the past. Thompson (1978) examined annual data on 23 U.S.-listed closed-end funds from 1940 to 1975. He showed that investors earned higher returns than a benchmark of U.S. stocks by purchasing shares of closed-end funds with discounts, and lower returns by purchasing shares with premiums. Richards, Fraser and Groth (1980) analyzed trading strategies on weekly data for 18 U.S. closed-end funds from 1970 to 1976. Each strategy bought funds at discounts greater than x percent and sold when discounts dropped to y percent (x>y>0). All produced higher returns than a benchmark index, but whether they produced higher risk-adjusted returns remained unclear. Anderson (1986) examined similar trading strategies with a different sample of 19 U.S. closed-end funds from 1965 to 1984. He found that many strategies would have enabled an investor to beat a benchmark index. As in the previous study, however, Anderson did not adjust excess returns for risk.

This paper is similar to the above-mentioned papers but exploits a new and extensive sample of U.S. and U.K.-listed closed-end funds. The sample contains 128 funds on January 4, 1991 and ends with 234 on September 4, 1998. Approximately half of the funds are U.S.-listed and the other half is U.K.-listed. The sample contains weekly discount and return data from each fund’s inception. Premiums and discounts are constructed by matching the share price on Friday to the net asset value reported on Wednesday, Thursday or Friday. All returns are computed on a total return basis in U.S. dollars correctly incorporating cash and stock dividends, rights offerings, scrip dividends.

The performance of a group of investment strategies versus a benchmark is explored. The investment strategies and their respective benchmarks are as follows:

Long Portfolios: Takes an equally weighted long position in N (N = 10, 20, 30, 40, 50) closed-end funds beginning with the most highly ranked fund (with the deepest discount). The benchmark portfolio is an equally weighted long position in all funds.

Short Portfolios: Takes an equally weighted short position in N (N =10, 20, 30, 40, 50, ALL) closed-end funds starting from the bottom of the ranking (from the fund that has the biggest premium). An equally weighted short portfolio of all funds, is the benchmark.

Combined Portfolios: These portfolios combine the long and the short portfolios described above. In particular, portfolios are formed with an equally weighted long position in N funds and an equally weighted short position in another N funds. This combined position that is neutral with respect to market risk is known as a "market neutral" strategy. CASH, an equally weighted long and short position in all funds, denotes the benchmark for the combined portfolio.

Each of the aforementioned strategies was backtested from January 4, 1991 to September 4, 1998 using information that an investor could have known at the time. The procedure for a long portfolio strategy is as follows. For a given holding period length k, number of securities N, transaction cost rate tc, an equally weighted portfolio of funds with discounts ranked in the top N is formed on January 4, 1991. This portfolio is held until the next rebalancing period, which begins k weeks hence. At that time, a new equally weighted portfolio of N funds is formed with discounts ranked in the new top N. The cost of trading from the old portfolio to new is subtracted from the k-week return on the old portfolio. This cost is the product of tc with the turnover actually realized on the portfolio. This process of portfolio formation and return evaluation continues until September 4, 1998. The procedure is similar for short portfolios. Combined portfolios are treated as separate long and short portfolios.

The paper finds that long portfolios with deep discounts and frequent rebalancing outperform the benchmark and short portfolios with deep premiums, provided transaction costs are zero or low. This is expected and is motivated by the excess return potential in purchasing deeply discounted funds. Surprisingly, the paper finds that short portfolios with deep premiums and less frequent rebalancing outperform the benchmark and long portfolios, provided transaction costs are moderate to high. The later stems from the trade-off between the excess return potential of selling funds at premiums and lower turnover and transaction costs associated with portfolios of funds at premiums. In short, betting with the discount and frequent rebalancing works best when transaction costs are low and, conversely, betting against the premium and less frequent rebalancing works best when transaction costs are moderate to high.

MACROECONOMIC FACTORS AND SHARE RETURNS:
AN ANALYSIS USING EMERGING COUNTRY DATA

Over the last twenty years, a number of studies have examined, both theoretically and empirically, the relationship between asset prices and various economic factors. However, very few papers have investigated this relationship using emerging country data. One exception to this generalization is the work of Harvey (1995a,b). He analyzed the role of global factors in explaining the cross-sectional variation in emerging market returns. In particular, Harvey investigated the influence of the world-market equity return, the return on a foreign currency index, oil prices, world industrial production, and the world inflation rate. He found these global factors to insufficiently characterize returns in emerging markets. This finding is consistent with Harvey’s earlier evaluation of the ability of both global and local variables to predict emerging market returns. Specifically, Harvey found that local information variables, including currency, interest rates, dividend yield and lagged local returns, accounted for more than half of the predictable variance in the returns of emerging markets. This paper analyses the effects of economic variables on the monthly returns of thirteen emerging stock markets over the ten-year period 1987-1996: Chile, Greece, Hong Kong, India, Korea, Malaysia, Mexico, Philippines, Portugal, Singapore, South Africa, Thailand and Turkey.

The objectives of the paper are twofold. First, for each emerging market country, a principal components analysis is applied to a large set of domestic and world economic variables in order to reduce the dimensionality in the economic dataset to a limited number of core factors. The global variables studied include the world-market return, commodity prices, oil prices, US interest rates, world inflation and world industrial production, while the local variables include currency, inflation, interest rates, gross domestic product, money supply and the trade balance. Second, the possible influence of global and/or local information variables on the share returns of each emerging market is determined; the dominant principal components are extracted and used as inputs into a regression analysis to explain index returns. The results from the analyses suggest that gross domestic product, inflation, money supply and short-term interest rates are important local variables in all thirteen of the emerging markets considered, whilst world industrial production, world inflation, US interest rates and commodity prices are important global factors. Additionally, while global variables play a crucial role in explaining returns in some countries (Korea and Singapore), local factors are more important in other markets. For example, domestic factors only are important in explaining returns in two markets (India and Turkey), while the addition of the local variables to the world information variable set increases the proportion of returns explained in four markets (Greece, Mexico, Portugal and Thailand). By contrast, neither world nor local factors are significant in five countries (Chile, Hong Kong, Malaysia, the Philippines and South Africa).

EXCHANGE RATE FLUCTUATIONS  EXPOSED ON CLOSED-END FUND PERFORMANCE

The present study correlates the market prices of major Single Country Closed-End Funds (SCCEF) with the lagged Net Asset Values (NAV) of the funds in order to examine whether NAVs may lead market prices of SCCEFs.

NAV is computed by taking the total assets of the single country closed end fund (SCCEF) subtracting the liabilities and dividing the result by the number of shares outstanding. The NAV would change as a result of change in either the assets, liabilities, or the number of shares outstanding. The number of shares outstanding rarely change unlike that of open end funds. In closed end funds, once shares are issued they are traded in the secondary market as any other securities issued by corporations. The liabilities of SCCEF are minuscule in magnitude in comparison to the assets. Therefore, most of the changes that influence NAV come from the change in the value of the assets.

NAV is computed and adjusted by management of SCCEF on a weekly and recently (for some funds) on a daily basis. SCCEFs are made up of different securities within a specified single country. They are portfolios of securities where each security is traded in the issuing country while the portfolio is traded as SCCEF in another country, such as Hong Kong, U.K. and U.S. Similar to any other international portfolio investments the performance of SCCEFs is influenced by general market conditions, regional economic conditions, exchange rates (mainly between the country where the securities of the SCCEF are and the country where they are traded) and interest of investors on SCCEFs, among others.

Change in exchange rates between the countries where the securities are issued and where they are traded triggers change in NAV and market price of SCCEFs. The question that this study raises, is which one of these prices, the NAV or the market price, efficiently incorporate the change in exchange rates? If we assume that professionals manage the SCCEFs, these professionals measure and input the impact of currency translation, transaction and economic exposures. Unlike subsidiaries of multinational corporations where currency translation exposure is argued by some to be irrelevant when consolidated financial statements are prepared, we propose that it play significant role as economic and transaction exposures when it comes to SCCEFs. The assets of SCCEFs are mainly composed of securities of a single country. Change in exchange rates is going to affect the economic performance, the revenues and costs, of the firms that have issued the securities and thereby the prices of the securities within the country. Similarly, depending upon the strategy adopted by management on the level of turnover of securities that make up the SCCEF transaction exposure will also impact on the value of the assets. Finally, in the preparation of the balance sheet necessary to calculate the NAV, be it on a weekly or daily basis, the values in the balance sheet have to be recalculated for the effect of economic and transaction exposures as well as currency translation exposure. It is unlikely that small investors would populate closed end funds to go through this rigorous computation on a weekly or daily basis to efficiently price SCCEFs. Therefore, this study expect the NAV which is computed by management of SCCEF to lead in assessing, measuring and incorporating the influence of currency exposure rather than the market price.

In their 1997b study, Mulugetta and Mulugetta expected that investor sentiment would lead SCCEF prices to depreciate slower than NAV and, therefore, cause SCCEF's discounts to decrease or premiums to increase. The majority of Far East SCCEFs supported this expectation, indicating that widening premiums of Asian SCCEFs seem to be caused by overestimation of regional economic growth and underestimation of dollar appreciation.

The recent study by Mulugetta, Ghosh, and Mulugetta (1998a) has closely examined the movements of discounts/premiums of thirty-four SCCEFs during the Asian currency crisis period (July 1, 1997 to December 5, 1997) in comparison to the pre-crisis period (January 1, 1996 to June 30, 1997). The study identified two distinctive patterns: the "Southeast Asian" pattern unique to the funds in crisis and the "Latin American" pattern unique to the funds least affected by the crisis. The SCCEFs of Indonesia, Malaysia, Singapore, Thailand and Japan represented the first pattern, where the discounts significantly shrank (or the premiums grew) due to the faster depreciation of the net asset value (NAV) in comparison to the reduction in price. In contrast, the discounts of the Latin American funds widened during the crisis period due to faster growth of the NAVs in comparison to the market prices. Mulugetta, Ghosh and Mulugetta (1998b) also analyzed the Mexican SCCEFs during the Mexican currency crisis in comparison to the pre- as well as the post-crisis period. It revealed a pattern quite similar to the Southeast Asian pattern described above. During the crisis, the discounts shrank (or the premiums grew) due to the faster depreciation of NAV than the reduction in price. During the post-crisis period, on the other hand, the discounts grew due to the faster growth of the NAV than the price as expected.

The resent study examined twenty-one SCCEFs from January 1, 1994 to December 5, 1997. The databases were consisted of exact five-week days regardless whether the market was open or not. This data structure has allowed us to identify which day of the week SCCEFs' market prices may be mostly affected by the announcement of their net asset values. All data points were converted into natural logarithm first and then the change in the value from the previous data point was calculated. The Durbin-Watson test on the residuals of OLS (ordinary least squares) regressions detected insignificant serial correlation among residuals. Regression analyses have indicated that out of 21 funds, 13 funds had statistically significant, positive beta values associated with the changes in NAVs, which were lagged by one day. This implies that the NAV calculated on each Friday is most likely to affect the market price of that fund on the following Monday. In short, NAVs lead market prices, as expected. Furthermore, the analysis has showed that market prices of those funds that have daily NAV disclosure service seem to be affected more immediately by NAV announcements in comparison to those which do not provide such service. A more appropriate research design and more precise statistical tests should be developed in order to examine the impact of the daily disclosure of NAVs in comparison to the weekly disclosure in future studies.

CONVERGENCE AND MARKET SHARE IN THE EXCHANGE-TRADED AND
OVER-THE-COUNTER DERIVATIES MARKETS

Although there has been an upward trend in the growth of volume in both exchange-traded and OTC products, organized exchanges have seen their share of trading volume decline. Advances in financial engineering, combined with an increase in dealer-operated electronic trading systems, have led to huge increases in the volume and liquidity in OTC transactions. By participating in screen-based quote systems, for example, OTC dealers have created more unified markets, with component increases in liquidity deriving from better information exchange and more standardized transactions. Derivatives exchanges have responded by innovating their product line and by developing their own versions of electronic trading. Exchanges have introduced products whose characteristics can be customized, such as flex-options, attempting to bridge a traditional distinction between OTC products and exchange-traded contracts. In addition, as developers and promoters of their own electronic systems, such as GLOBEX and Project A, exchanges have been negotiating the listing of multiple assets, such as cash and derivative products, on their systems. As a case in point, the Chicago Board of Trade (CBOT), a futures exchange, registered a subsidiary, the Chicago Board Brokerage (CBB), as a broker of cash government securities. Via the CBB, CBOT members and member firms can trade Treasury securities and other OTC products via a screen-based trading and information system on the trading floor. These developments are notable because they typify a convergence of the exchange-traded market with the OTC market, both in terms of the products that are being offered and the microstructure of markets where they are traded. This article formulates a model that edifies the convergence of these markets, and sets forth propositions regarding the consequences of this type of innovation for investor costs and market share.

The main spirits of HSS are to choose the up and down movements of the binomial tree such that the unconditional mean and the conditional volatility of the approximated process approach their true value (see HSS's Lemma 1.), and to choose the conditional probabilities in a manner that the unconditional volatility converges to the true value (see HSS's Theorem 1.). This study first shows that the choice of the conditional probability in HSS is valid for Markovian process only. In contrast, we point out that the conditional probability depends on all the previous outcomes for non-Markovian process. Thus the HSS method are more suitable for modeling the random walk and constant volatility processes unless the number of stages of the binomial tree for each period is chosen properly.

To implement the HSS method in the multivariate case, this study suggests that if two random variables are too highly correlated, one must always keep the number of steps of the binomial process for the second variable far larger than that for the first variable. Moreover, one should construct the binomial tree for the underlying asset with large volatility first to obtain fast convergence of the option prices.

Furthermore, this study points out that the procedure in HSS for multiperiod, multivariate case is wrong. Suppose that (X, X₂) and (Y₁, Y₂) are multivariate lognormally distributed. The correct procedure to construct binomial processes is as follows:

1. Build up the X first and then Y₁, X₂ and Y₂. The vector of Y₁ is constructed using the conditional volatility of Y₁ given X; X₂ requires the conditional volatility of X₂ given both X and Y₁; and Y₂ requires the conditional volatility of Y₂ given X, Y₁, and X₂.
2. Compute the conditional probabilities q(y₁ | x1), q(x₂ | x₁, y₁), and q(y₂ | x₁, y₁, x₂).

We also show that the required inputs for the multivariate binomial approximations in HSS are also doubtful.

Finally, we adjust HSS's method to construct the multivariate binomial trees for the normally distributed processes. This is a very important extension for some finance models, for example Gaussian term structure models which are widely applied to value interest rate derivatives. The choice of the up, down movements, and the conditional probability in this case are analogous to the lognormal distribution case.

INFERRING FORWARD LOOKING FINANCIAL MARKET RISK PREMIA
FROM DERIVATIVES PRICES

This paper focuses on a topical and important area of theory and practice ie. The risk premium in financial markets. While there exists a vast amount of research into its behaviour, particularly in US markets, this is largely based on regression based techniques which do not capture well the dynamic and forward looking nature of the risk premium.

In this paper the time variation of the unobserved risk premium is modeled by a system of stochastic differential equations connected by arbitrage arguments between the spot equity market, the index futures and options on index futures. Although various processes for the dynamic of the risk premium may be considered, we motivate and analyze a mean-reverting form. The diffusion part is specified such that the risk premium remains positive. Since the risk premium is not directly observable, information on it is extracted using an unobserved component state space formulation of the system and filtering methodology. As an initial application of the methodology, the results from daily Australian market data from the SFE over a period of twelve months are presented. The small sample properties of the parameter estimates are also examined by bootstrapping the state space system. It is well known in the filtering literature that the estimation of state space system by Kalman filter is sensitive to the specification of the quantities such as initial state variables and the prior covariance matrix. The paper also carries out appropriate sensitivity analysis in this respect.

REGULATEE SHORT SELLING IN HONG KONG AND MALAYSIA:
RISK CHANGE AND REVALUATION EFFECTS

This paper reports findings from a new phenomenon in Hong Kong and in Malaysia, where limited approvals had been given in 1996 for institutional traders but not individuals to short sell shares of approved counters. Theory predicts that short selling helps to complete the market pricing process. Therefore, the systematic risk of the firms must decline for those approved under this form of regulated short sales. That would mean that the price reactions around the time of announcement/execution of short sales must lead to a one-time revaluation effect with a positive risk-adjusted return. Our investigation of this phenomenon in the Malaysian stock market provided initial results that shows price upticks. Data on Hong Kong market are being compiled for the same tests.

Introduction

Theory predicts that short sales of securities help to complete the price formation process. This prediction is consistent with the mainstream finance ideas on the desirability of short sales as a means of signalling to the market the informed traders’ lower price expectations in the future. Short sales are prohibited outright in all non-Japanese Asian markets. Traders engaging in short sales have been prosecuted by regulators and fined for such activities. After five years of reasonable liquidity improvements during the 1990s in the Hong Kong and Kuala Lumpur stock exchanges, the regulators had become more open to the idea of short sales as an instrument to reveal unfolding market trends. The authorities introduced new regulations to encourage limited short selling to gain experience so that short sales may be introduced in stages to cover the whole market.

Malaysian authorities introduced legislation permitting short sales to be conducted by approved sharebrokers and not individuals. Only the top 100 out of the 462 firms were to be included under this rule. The law was passed on 27 August, 1996. Approval of the Securities Commission was given for 50 counters on which short sales may be done initially. The approved counters started short sales three weeks later. Interviews of brokers conducted by us show that only 17 counters were actively engaged in this limited short sales even though 50 had been approved. Hong Kong also approved limited short sales a year earlier. We are in the process of collecting the details from this market.

Theory and Test Hypothesis

Finance theories (Portfolio Theory, CAPM, APT) give separate solutions for the capital market equilibrium line depending on whether the market is under short sales restriction. If no restrictions exist on short sales, trades clearing the market will lead, for example, to a normal parabola in the case of the Portfolio Theory of Markowitz (1958). Similarly, the solution to the equilibrium pricing in the CAPM (Sharpe, 1965) is one where the Capital Market Line will be extended further if short sales are permitted. Thus, in a market with no short sales restrictions, the market is assumed to have completeness in pricing, thus affording price formation to be unhindered. As such, there is a balanced pricing process in such a market. Thus, restrictions on short sales make market pricing incomplete, and thus the solutions are different when short sales are no longer forbidden.

This suggests that in a market with completeness in the pricing process, the shares, which are permitted to be offered in short sales, must become less risky. With less risk, the pricing process would be such that there must be a one-time price gain for the shares that are permitted to be short sold. This means that short sales completes the price discovery process, thus reducing the risk futures uncertainty, which in turn would mean that there ought to be abnormal return earned by the shareholders of the companies permitted to be short sold.

Data, Research Design and Hypotheses

The research design is very straightforward. There ought to be reductions in the risk of the stocks that were permitted to be short sold. Next, as a consequence of the reduced risk, there must be a revaluation effect that can be observed in the pricing of these stocks on and around the dates of approval/execution of short sales. These are the strategic hypotheses that we are testing. Two specific hypotheses can be tested. One is whether the risk parameter changes after the short sales are permitted. The other is whether there is a revaluation effect from the announcement/execution of short sales in the two markets. The revaluation will lead to earning of abnormal returns.

The data for the Malaysian market have been collected from the SCAN’s file of the exchange. Daily adjusted price data over 250 days before and 100 days after the event were collected for each of the 50 stocks permitted to be traded under the limited short sales regulations. Gross returns were computed from the price data. Risk-adjusted returns computed from the application of a Scholes-Williams (1977) procedure ensured that the risk parameters are robust and are unaffected from the effects of thin trading.

Next, we applied the event study methodology (Brown and Warner, 1985) using the procedure for the risk-adjustments suggested by the CAPM and the Market Model (Sharpe 1965; 1963). The average risk-adjusted returns were computed for the event dates, and for dates before and after the event. We are at the initial stage of testing and have not yet decided on how many days prior to and after the event these measures should be computed. We started the tests over –60 through 0 to +30 days as the initial test window. These returns were cumulated over the period, and the results are tested for significance. Since the event is one that is clustered on a single date, appropriate corrections will be made for the clustering effect to produce reliable results.

Expected Results

The preliminary results from these tests show that there is a revaluation effect during the test window. We are now screening the newspaper and conducting interviews to determine when the short sales actually began. In this market, it takes some time before approvals are given under any new regulations. Also, there is some time lag between the approval and the setting up of the procedures in the broker firms to execute the trades. We suspect that the actual effect will be clear if we could pinpoint the dates of execution of the trades in addition to the announced dates for the coming into operation of the new rule on short sale. In the case of Hong Kong, the same process of data collection and scrutiny is being followed.

In the final paper we propose to present at the FMA meeting, this paper will provide reliable results on this new phenomenon called regulated limited short sales of two well known Asian markets. Initial results from one market is encouraging, and these results are being refined. The results from the other more liquid Hong Kong market ought to be interesting as well. This is a new phenomenon, which requires careful investigation since the success or failure of this new policy may determine whether other markets will follow this deregulation that has taken place in two Asian markets. The results promise to be interesting.

EVIDENCE ON THE USE OF CURRANCY DERIVATIVES

Introduction

During the past decade the use of financial derivatives by industrial corporations has increased dramatically. Questions persist whether the theoretical reasons can explain the use of financial derivatives by these corporations. The theoretical reasons why firms use derivatives can be divided into five general categories. They are expected cost of financial distress, taxes, managerial incentives, expected costs of external financing and risk exposure (Nance, Smith and Smithson 1993).

A recent study by Howton and Perfect (1998) was one of the first to use the total outstanding notional principal value of derivative instruments as the dependent variable. Most other studies use a yes or no dependent variable. Yes if the firm reports using financial derivatives and no if the firm does not report using financial derivatives. Examples are the studies by Geczy, Minton and Schrand (1997) and Mian (1996). The Howton and Perfect study is one of the most recent in that the cross-sectional study is based on data for fiscal year 1994. And the accounting statements used are based on reporting requirements implemented by FASB 105.

FASB Statement 119, Disclosure about Derivative Financial Instruments and Fair Value of Financial Instruments, which amends FASB 105 and 107 became effective after fiscal years ending December 15, 1994. This Statement requires disclosures about derivative financial instruments-futures, forward, swap, and option contracts, and other financial instruments with similar characteristics. More specifically, this Statement requires disclosures about amounts, nature, and terms of derivative financial instruments that are not subject to Statement 105 because they do not result in off-balance-sheet risk of accounting loss. It also requires that a distinction be made between financial instruments held or issued for trading purpose (including dealing and other trading activities measured at fair value with gains and losses recognized in earnings) and financial instruments held or issued for purposes other than trading. This statement also amends Statement 107 to require that fair value information be presented without combining, aggregating, or netting the fair value of derivative financial instruments with the fair value of non-derivative financial instruments. Additionally, the fair value information must be presented together with the related carrying amounts in the body of the financial statements, a single footnote, or a summary table in a form that makes it clear whether the amounts represent assets or liabilities.

Uniqueness

A literature search could not find any empirical studies of the use of currency derivatives that used accounting data that relies on requirements implemented by FASB 119. Thus, the unique feature of this research on the use of currency derivatives is that it uses fiscal year 1997 accounting statements. That is, accounting statements that have to be consistent with the requirements of FASB 119.

Data, Procedure and Results

The sample selection and regression methods used borrow from Geczy, Minton and Schrand (1977) and Howton and Perfect (1998). Following Geczy, Minton and Schrand (1997) the original sample is the Fortune 500 non-financial and non-utility firms in 1997. Additionally, the firm must have a 10K available on EDGAR, the firm cannot have experienced bankruptcy in the previous three years and has the necessary accounting data available on Compustat.

Following Howton and Perfect (1998) a Nested Tobit Model regression model is used. A Tobit model provides unbiased estimates when using a sample that contains firms that do not use currency derivatives. That is, these firms that do not use derivatives most likely differ as to their level of "non derivative use," but these differences are qualitative, not quantitative. A Tobit model is used when this type of qualitative information is present in the dependent variable.

The dependent variable is the total outstanding notional principal value of currency derivatives divided by total assets. Following both Geczy, Minton and Schrand (1997) and Howton and Perfect (1998) the following classes of independent variables are used: managerial incentives, expected costs of external financing, expected financial distress costs, taxes and risk exposure.

Preliminary results indicate a positive and usually statistically significant relationship between the use of derivatives and most proxies for financial distress costs, some proxies for managerial incentives, most proxies for taxes and most risk exposure measures.

Primary References

Geczy, Christopher, Bernadette A. Minton and Catherine Schrand, "Why Firms Use Currency Derivatives," The Journal of Finance, Vol. 52, No. 4, September 1997, pp 1323-1354.

Howton, Shawn D., Steven B. Perfect, "Managerial Compensation and Firm Derivative Usage: An Empirical Analysis," The Journal of Derivatives, Winter 1998, pp 53-64.

Mian, Shehzad L., "Evidence on Corporate Hedging Policy," Journal of Financial and Quantitative Analysis, Vol. 31 No. 3, September 1996, pp 419-439.

IMPACT OF NORMAL AND EXTRAORDINARY EARNINGS DISCLOSURES
ON STOCK PRICES IN HONG KONG

Motivation of Study

Prior to April 1986, the Hong Kong share market was very much a local market share trading was carried out in four different local exchanges. There were little formal regulations and institutional developments achieved in a coordinated manner. In April 1986, the Stock Exchange of Hong Kong was formally set up as the sole trading place for securities in Hong Kong. Alongside with this unification, a series of regulations and reforms were introduced to turn the Exchange into an orderly and fair market place for securities trading. After more than ten years of painstaking reforms and development, an appraisal of information efficiency of the Hong Kong share market is deemed timely and appropriate.

The standard of efficiency is benchmarked against an efficient market, in which prices always fully and instantaneously reflect relevant information. This study focuses on the issue of share prices' responding fully to accounting-based information released in the Hong Kong share market. This is complementary to our other study on the speed of price response to information, which together form a complete series of information efficiency studies on the Hong Kong stock market. The magnitude of price response is directly assessed by the earnings response coefficients (ERC), defined as a ratio of change in share price per unit change in earnings. Results of this study will add evidence to the research literature on ERC using the Hong Kong share market. Furthermore, by using a randomization process to segregate the effect of dividend information that is released contemporaneously with earnings announcements in the Hong Kong stock market, this study will demonstrate another way of operationalising the measure of ERC. This is particularly relevant for assessing information efficiency of other stock markets that possess similar institutional setup as the Hong Kong market.

Theory

In the context of capital markets, one of the roles of information disclosure is to facilitate the establishment of equilibrium prices, based on which investors make their investment decisions. While the measure of such information effect is usually difficult to define and transform into a pragmatic concept, an alternative can be sought by investigating a contextually relevant dimension of information usefulness in capital market: When information arrives, investors formulate and revise their beliefs about future outcomes. These are in turn reflected in price changes, hence an observable information effect. Since Ball and Brown’s (1968) path-breaking study, such analogy of value-relevance and usefulness of information has formed the basic building block of most capital market-based research.

With similar thought, this paper examines the information efficiency in the Hong Kong share market by equating value-relevance with usefulness of information. The measurement of the information effect will focus on the magnitude by which information induces price changes in market, which will be quantified by the measure of ERC. In articulating the research design for the subjects of interest, a systematic review of the broad theoretical foundation and development of the specific models are desirable. In particular, the notion of Efficient Market Theory, discussions of capital market-based earnings studies, and applications of ERC model are reviewed in the full paper. On the one hand, the review of all these theories and research paradigms is essential for laying a solid foundation for designing an information efficiency study that adds Hong Kong evidence to this line of research. On the other hand, the recapitulation of research development and the testing of selected models coming out from this review process will provide easy reference for information studies alike in future.

Data

Data used in this study include daily market indices and daily adjusted closing prices of listed shares in the Hong Kong share market, interim and final earnings announcements and dividend announcements. The sample period for the empirical analysis starts on 1 January 1988, and ends on 31 December 1996. The starting date is chosen so as to capture the Hong Kong market situation only after unification into one stock exchange in April 1986, and the dissipation of poor market sentiment after the October Crash in 1987.

Summary of Findings

Results show that on the whole, share prices respond significantly to favorable earnings information (i.e. increase in earnings as compared to previous year's). When there is no dividend information disclosed at either interim or final earnings announcements, distinct investors' response is only found in the final announcements, with a much gradual process of anticipating the information prior to the announcement. When dividend information is available, price responses are observed mainly in the interim announcements in the form of more abrupt price changes around the announcement dates. In short, all these have us believe that the Hong Kong stock market is informationally efficient in terms of completeness of price adjustment to accounting-based information.

FINANCE AND ENTREPRENEURSHIP INTERFACE

Entrepreneurship is an interdisciplinary field. It is a holistic and dynamic process. entrepreneurship can be viewed from different perspectives, but there are some commonalties: innovation, use of resources, creating wealth, and taking risks. The views developed within the realm of business administration namely strategic management, marketing, and finance, can significantly contribute to the better understanding of this dynamic process. Each discipline has a unique way of viewing Entrepreneurship, which remains relatively unaffected by perspectives of other disciplines. Many "uni-" rather than "multi-" disciplinary views of the field have been developed. However, the Entrepreneurship is inherently interdisciplinary. Valuation is integral to the relationship between entrepreneurial firms and the outside market. Furthermore, it helps reveal the impact of risk preference and financial goals on entrepreneurial decision making. An attempt could be made to link the traditional and modern financial concepts and to entrepreneurship.

In considering how the various fields of business administration and particularly finance might contribute to an interdisciplinary approach of entrepreneurship, first we need a definition of entrepreneurship. Therefore, the first objective of this paper is to synthesize various definitions of entrepreneurship. This paper will include the elements of entrepreneurship when a general consensus has been reached for their appropriateness in defining the field. The second objective of paper is to examine the key elements of modern finance discipline such as cash flow, return, and risk and to link them to conceptualization of entrepreneurship.

Gartner (1990) has done a factor analysis on a Delphi panel that explored the meaning attached to entrepreneurship, by 44 researchers and practitioners. His factor analysis of their responses produced eight common themes. Synthesizing various definitions of entrepreneurship, proposed in recent years, and the result of factor analysis, we can claim that the core idea and constructs behind entrepreneurship are: Opportunity Identification, Creation of Innovative Solutions, Risk-taking, and New Business Creation and Team Building. The discipline of finance can significantly contribute to better understanding of entrepreneurship.

The last few years have seen a growing of research into the specific institutions, principles, practices, and problems of small enterprise financial management. Financial management is concerned with understanding factors that determine the value of a business enterprise's uncertain cash flows over time, and with management of these factors in a normative sense. Thus, the basic dimensions of business activity on which the theory and practice of financial management focus are cash, time, and risk. The goal of financial management is to maximize the value of an enterprise to its owners, which is believed to be a function of the amount, timing, and risk of the cash-flows they ultimately receive (Brigham, 1992). Ang (1992) points out that a profitable small enterprise that does not depend to any significant extent on outside funding may tend to ignore current performance and focus resolutely on maximizing long-term value. The financial objectives of small enterprise may vary from those of growth or entrepreneurial enterprises. According to Petty and Bygrave (1993) the firm's financial objective is largely dependent on the stage of development of the small business. he also point out that for traditional small enterprises: "The concept of wealth maximization has reduced meaning, since there are so many exogenous considerations influencing the decisions, besides that of economics. The objective is not so much to create value, but to provide a "preferred" life style within the community. Even for the "successful" lifestyle firms, there is little in the way of value created beyond providing a living for the owner and his or her family."

In view of the above arguments it seems possible that the financial objective formally specified in the financial literature is an oversimplified and inaccurate statement of the real purpose of owner-managers of small enterprises. A casual review of entrepreneurship literature shows that there are numerous typologies of small enterprise which include traditional small enterprises versus growth or entrepreneurial enterprises, small enterprises of various legal forms in terms of sole proprietorships, partnerships, public companies listed on stock exchange, informal enterprises, minority enterprises , family enterprises, and franchised enterprises.

Given the vast diversity of small enterprises, the central question is "what should be included in the small enterprise financial objective"? McMahon and Stanger ( 1995) deduced the following conclusion from available empirical evidence on the broad motivations of small enterprise owner-managers: " Owner-managers of small enterprises are unlikely to have a single overriding aim in establishing and running their own businesses. Their intentions are apt to be numerous and complex. Many of the motives owner-managers have for being in small enterprise, and also the satisfactions they derive from this occupation, are unequivocally non-pecuniary. Owner-managers of small enterprises have considerable freedom to indulge their many and varied objectives, be they pecuniary or otherwise".

Any specification of a goal for small enterprise financial management obviously needs to be viewed in the broader context thus portrayed. Based on the literature review evidence, [Boyer and Roth (1978), Kao (1985) Welsh and White (1981), Walker and Petty (1986), Sahlman (1983), Rader (1987), Brigham (1992 and 1995), Brophy and Shulman (1992),McMahon, and Stanger (1995)], it appears that the financial objective of small enterprise owner-managers should include the dimensions of "Return", "Risk" "Liquidity", "Diversification", "Transferability", "Flexibility, "Control", and "Accountability" which are all likely to influence enterprise value to some degree.

It is a fundamental axiom of financial thought that expected return is a major parameter in financial decisions of all types. After reviewing empirical evidence on these issues, McMahon and Holmes (1989) form a general impression that small enterprise owner-managers typically seek to achieve a satisfactory level of short-term accounting profit. Boyer and Roth (1978) discovered that owner-managers are willing to sacrifice rate of return on their investment for non-pecuniary rewards such as control over their income and job security, pride in their achievements, self-actualization, and esteem in the community. Kao (1985) has questioned the usefulness and validity of profit maximization as a financial goal in small enterprises and has argued that, conceptually, profitability of such concerns should be measured as residual income after deducting non-pecuniary personal sacrifices made by owner-managers.

Both systematic and unsystematic risk is important to owner-managers of small enterprises. The nature, sources, and significance of systematic risk are well documented in the finance literature. Everett and Watson (1993) provide some tentative empirical support for the proposed significance of enterprise-specific sources of risk in financial management of small enterprises. In their study of systematic and unsystematic sources of risk in small enterprises located in managed shopping centers in Australia, Everett and Watson (1993, p. 13) conclude: "Regression models of failure against real trading bank interest rates have been constructed and reported to provide some indication of the likely effect of both systematic and unsystematic risk factors on small business failure rates in managed shopping centers. On average unsystematic factors appear responsible for approximately 80 percent of businesses that either "fail to make a go of it" or cease to "avoid further losses." Welsh and White (1981) emphasize the importance of cash flow in a small enterprise as follows: "A small business can survive a surprisingly long time without a profit. It fails the day it can't meet a critical payment. In a small company, the cash flow is more important than the magnitude of the profit or the ROI. Liquidity is a matter of life or death for the small business ".

The soundness of liquidity management has emerged as arguably the most critical influence on survival and financial well being in small enterprises. Walker and Petty (1986) identify two principal reasons why this should be the case, one of which is exogenous to the enterprise and the other endogenous: "The apparent difference in liquidity between large and small firms lends further support to the existing belief that working capital shortages are a common problem for small firms. The difference could be the result of at least two factors. First, the small firm's limited access to the capital markets may impose the need for more economy in the use of working capital. Second, the basic nature of the "entrepreneur" could have a beating upon the working capital decisions within the small corporation. If the managers of small firms are willing to assume greater risk, as experience would suggest, their attitude may well be reflected in the small firm's liquidity position". Thus, it would seem liquidity should be a matter of concern in the present context because cash is such a critically scarce resource in a small enterprise as a result of supply constraints, which do not exist to nearly the same extent for a large enterprise.

An owner-manager's consciousness of the risk this circumstance poses may cause him or her to be disposed towards capital investment opportunities that provide some potential for diversification of the enterprise's activities. According to McMahon, and Stanger (1995): "Contrary to the "independent project evaluation rule" of financial theory, the owner-manager may weigh this consideration quite favorably when evaluating a proposal and might be prepared to sacrifice return in order to achieve a degree of diversification".

The transferability of financial and human capital is likely to be a major preoccupation in family enterprises where succession by the subsequent generation demands attention. As far as financial decision making is concerned, the immobility of financial and human capital invested in a small enterprise may cause the owner-manager to favor outcomes promising higher returns and/or lower risks. Sahlman (1983) asserts that financial flexibility has value, but he recognizes this is not without its costs: "The fact that keeping financial reserves on hand, whether in the form of excess cash, unused debt capacity or lines of credit, is costly precludes all firms from maintaining unlimited flexibility". On asset investment and divestment in small enterprises, Pettit and Singer (1985) indicate that: "the flexibility of a smaller firm's operations may allow the manager to control and maintain firm risk more easily by moving resources from one productive process to another with changes in technology or economic conditions".

Flexibility can be most important in allowing investment decisions to be reversed should they prove to be mistaken or inappropriate to changed circumstances. Flexibility is similarly valuable in making financing and profit-distribution decisions. For each type of decision, the benefits of flexibility are particularly apparent when the decision-maker has limited expertise, as is often the case in small enterprise (McMahon, and Stanger, 1995).

The dependence of small enterprise value upon financial objectives pursued, decision-making, practices and control issues is discussed by Rader (1987): "Increasing value while maintaining control of the firm is the main task facing business owners. This is a difficult goal to achieve because managers and owners often perceive conflicting objectives and therefore use different methods and motivations for choosing various business strategies". Brigham (1992) also points out that: "There is value to being in control, and that value is not easily measurable. As a result, we often observe small businesses taking actions, such as refusing to bring in new stockholders even when they badly need new capital, that do not make sense when judged on the basis of value maximization but that do make sense when seen in the light of the personal objectives of the owner.

The potential significance of control to owner-managers of growth or entrepreneurial enterprises and neglect of this matter in the finance literature are highlighted by Brophy and Shulman (1992) as follows: "The conflict between growth and expected wealth maximization may be compounded by the issue of survival and the desire among entrepreneurial founders to remain in control of their corporate enterprise. The basic finance premise of maximizing shareholder wealth explains only some entrepreneurial behavior. Those firms unwilling to sacrifice control or maximize corporate growth are often ignored in the finance literature.

Many researchers have discovered the considerable importance of owner-manager independence as a motivation for entering small enterprise. Most notably, the Bolton Committee of Inquiry on Small Firms (1971) in the United Kingdom found that the need to attain and preserve independence is the most fundamental objective in the majority of small enterprises. An owner-manager is therefore unlikely to be indifferent to decision outcomes that would impose some degree of financial accountability to other stakeholders in his or her enterprise, especially when they are outside parties such as suppliers, customers, and financiers. External financing almost inevitably brings with it some form of monitoring arrangement intended to overcome difficulties posed by asymmetric access to financial information between those within the enterprise and the financiers concerned.

In conclusion, based on above arguments, it become imperative that the small enterprise financial objective reflect not only customary expected return and perceived systematic risk considerations, but also exposure to sources of unsystematic or enterprise-specific risk that typically exist in small enterprises arising from liquidity, diversification, transferability, flexibility, control and accountability considerations.

References

DETERMINANTS OF BANK PROFITABILITY IN LEBANON

David W. Peters, Elias A. Raad and Joseph F. Sinkey

Since 1992, the banking sector has been very profitable. However, there has been a significant variation in the rates of return earned by different banks in different years. This study empirically investigates the reasons for variation in return on equity using a pooled cross-sectional times-series sample of observed rates of return on equity for Lebanese banks from 1992 to 1997. Preliminary results indicate that rates of return on equity are positively influenced by market interest rates, the degree of financial leverage, size and the proportion of a bank's portfolio invested in Lebanese treasury bills. The positive relationship between return on equity and size provides a logical explanation for recent bank merger activity in Lebanon. The positive relationship between return on equity and banks' investments in treasury bills suggests that banks have been better off by investing in the high-yielding treasury bills issued by the Lebanese government than investing in loan portfolios.

Introduction

The years since the end of the civil dispute in Lebanon have been highly prosperous for the banking sector in Lebanon. The net return on average shareholders' equity for the banking sector in Lebanon was 88.33% in 1992. Since then the net return on average shareholders' equity has been 52.89%, 50.42%, 36.53%, 22.10% and 21.55% in 1993 through 1997 respectively. While some banks have earned returns far in excess of 100%, other banks have had significant losses and have gone out of business or merged with other banks. In each of those years, the standard deviations of returns on shareholders equity across the banking sector has exceeded the average return on shareholders' equity earned in the sector. Thus, while profitability has been high, the dispersion of profitability has been even higher. The purpose of this article is to explain the variation in bank profitability across time and between banks. Five important questions will be addressed in this article.

The primary reason for the high level of profitability has been the high level of interest rates. The vast majority of assets and liabilities of Lebanese banks are denominated in either Lebanese pounds or US dollars. Yields on treasury bills denominated in Lebanese pounds and deposits denominated in Lebanese pounds have been extremely high by international standards. The yield on 1-year Lebanese pound treasury bills was 25% at the end of 1992 and was approximately 15% at the end of 1997. The interest rates on loans and deposits offered by Lebanese banks denominated in US dollars have been much lower but have exceeded the interest rates on equivalent instruments offered by banks in the U.S.A. This article will examine what effect the level of interest rates has had on bank profitability in Lebanon over time.

There are vast differences in the equity / asset ratios of Lebanese banks. Some banks such as Banque Libanaise pour le Commerce have (equity / asset) ratios in excess of 10% at the end of