Full terms and conditions of use:

Transcription

1 This article was downloaded by: [Oakland University] On: 12 June 2013, At: 07:31 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK Applied Economics Publication details, including instructions for authors and subscription information: Stock prices and the effective exchange rate of the dollar Mohsen Bahmani-Oskooee a & Ahmad Sohrabian b a Department of Economics, The University of Wisconsin-Milwaukee, Milwaukee, Wisconsin, b Department of Finance, Real Estate and Law, College of Business Administration, California State Polytechnic University, Pomona, 91768, USA Published online: 28 Jul To cite this article: Mohsen Bahmani-Oskooee & Ahmad Sohrabian (1992): Stock prices and the effective exchange rate of the dollar, Applied Economics, 24:4, To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

2 Applied Economics, 1992, 24, Stock prices and the efective exchange rate of the dollar MOHSEN BAHMANI-OSKOOEE* and AHMAD SOHRABIANt *Department of Economics, The University of Wisconsin-Milwaukee, Milwaukee, WI and 'Department of Finance, Real Estate and Law, College of Business Administration, California State Polytechnic University, Pomona, Pomona, CA 91768, USA The literature on the relation between stock prices and exchange rates is very poor and includes few studies that have argued that exchange rate changes do effect stock prices. By relying on the portfolio approach to exchange rate determination, it is argued that a change in stock prices could also have an impact on exchange rates, i.e. there could be a two-way relationship between exchange rates and stock prices. Granger concept of causality as well as cointegration technique are employed to support this conjecture. The empirical results show that there is bidirectional causality between stock prices measured by S&P 500 index and the effective exchange rate of the dollar, at least in the short-run. The cointegration analysis reveals that there is no long-run relationship between two variables. I. INTRODUCTION Changes in stock prices is perhaps the only news that one could hear about every day. News media often report that stock prices fell because of a bad economic and political news or rose because of an encouraging economic and political event. Thus, one area in finance literature includes studies that have tried to outline the major determinants of stock prices. Since it is not easy to quantify the political events, most studies have concentrated on economic variables for which data are readily available. The most important and perhaps the first macroeconomic variable that has been identified as a major determinant of stock prices is the money supply (e.g. Palmer, 1970; Homa and Jaffee, 1971; Hamburger and Kochin, 1972; Cooper, 1974; Rozeff, 1974; Rogalski and Vinso, 1977, 1978; Tanner and Trapani, 1977; Grossman, 1981; Urich and Wachtel, 1981; Sorensen, 1982; Cornell, 1983; Pearce and Roley, 1983; Wong, 1986). Note that there is a consensus that it is the unanticipated component of money supply that affects the stock prices and not the anticipated component. Other macrovariables that are said to influence the stock prices, have been identified by different authors. Pearce and Roley (1985) found a significantly negative relation between stock prices and discount rate changes. Schwert (1981) investigated the stock market reaction to the monthly CPI inflation rate announcement.' Although real economic activity (announced or level) has been identified as another factor, no significant effect has been established, say, by Pearce and Roley (1985) or Darrat (1990). None of the studies reviewed above, have indicated the possibility of any significant relation that may exist between exchange rates and stock prices. The literature on this aspect of the topic is poor and includes only a few studies with mixed results. Franck and Young (1972) investigated the effects of six exchange rate changes and found no uniform pattern of stock price reaction of multinational firms to exchange rate realignments. Ang and Ghallab (1976) found that although stock prices did not reflect an anticipation of devaluation, they adjusted rapidly. As for the current floating period, Aggarwal (1981) found that trade-weighted exchange rate of the dollar had a significantly positive effect on the stock prices indicating that revaluation of the US dollar raised the stock prices.' However, when Soenen and 'For channels by which inflation surprises may effect the stock prices see Fama and Schwert (1977) and Pearce and Roley (1985). 2Notice that Aggarwal used an aggregate index of stock prices. Therefore, his conclusion holds for multinational as well as for domestic firm's stock prices. Indeed, he has provided theoretical arguments as to why a change in exchange rate will change the stock prices of multinational as well as domestic firms Chapman & Hall 459

3 460 M. Bahmani-Oskooee and A. Sohrabian Hennigar (1988) considered a different period, they found a significantly negative impact of revaluation on stock prices. The estimates reported by a few studies related to the effects of exchange rate change on stock prices mentioned in the preceding paragraph could all be biased due to the fact that there may exist feedback effects from stock prices to exchange rates which they have failed to recognize. Therefore, the purpose of this paper is to show that not only a change in exchange rate may cause a change in stock prices, but a change in stock prices could also cause a change in exchange rates. To this end, a theoretical discussion is provided; the methodology is introduced, the empirical results are reported. 11. THEORETICAL DISCUSSION interest rate and negatively to domestic interest rate. The role of the exchange rate is to balance the asset demand and supplies. Therefore, any change in the demand for and supply of assets will change the equilibrium exchange rate.4 Consider now an exogenous increase in domestic stock prices. This will result in an increase in domestic wealth. According to portfolio approach, the increased wealth will result in an increase in demand for money, thus, an increase in interest rates. High interest rates in turn, will attract foreign capital, resulting in appreciation of domestic currency. An alternative way of reaching this conclusion would be to say that an increase in domestic stock prices may result in an increase in foreign speculative demand on domestic assets. But since international investors have first to acquire domestic currency, the demand for domestic currency will increase, resulting in an appreciation of domestic currency. Aggarwal(l981) has argued that a change in exchange rate could change the stock prices of a multinational firm directly and those of the domestic firms indirectly. In the case of a multinational firm, a change in exchange rate will change the value of that firm's foreign operation which will be reflected in its balance sheet as a profit or a loss. Although there will be a change in the firm's assets as well as its liabilities, the net effect could be a profit or a loss. Once the profit or a loss is announced, that firm's stock prices will change. Now consider a domestic firm that exports part of its output. This firm will benefit directly from devaluation due to an increased demand for its output. Since higher sales usually results in higher profits, again the firm's stock prices will be affected. On the other hand if the firm is the user of the imported inputs, devaluation will raise its costs, therefore lower its profits. The news of a decline in profits may depress that firm's stock price^.^ Consider the effects of a change - in stock prices on 11. THE METHODOLOGY This section explains two methodologies that are used to establish the causal relationship between stock prices and the effective exchange rate of the dollar. The first relies on the Granger causality test and the second on a cointegration approach. According to Granger's method, a stationary time series, say, stock prices, SP, is said to cause another stationary time series, EX (exchange rate), if the prediction error of current EX declines by using past values of SP in addition to the past values of EX. Therefore, we need to start with the following autoregressive model, exchange rates. the way of jnferring the where is an error term. In estimating Equation 1, the impact of a change in stock prices on exchange rates is to question is how to choose an optimum number of lags on rely on the portfolio approach to exchange rate determinaeach variable. To this end the two-step procedure that tion. The essence of all portfolio models of exchange rate combines the Granger concept of causality with Akaike's determination is the notion that individuals allocate their FPE criteria is applied. According to Hsiao (1981, p. 91), in wealth among alternative assets that mostly include foreign the first stage the SP variable is excluded from Equation 1 and domestic money and foreign and domestic securities. It and the EX variable is treated as a one-dimensional autois usually hypothesized that the demand for domestic money regressive process. ~ h wries ~ o~autoregressions ~, are run by is inversely related to domestic as well as foreign interest varying the order of lags from 1 to M. The optimum number rates, whereas the demand for domestic securities is positof lags is selected by minimizing the FPE statistics defined ively related to domestic interest rate and negatively to as. --. foreign interest rate. By the same token it is assumed that the demand for foreign securities is positively related to foreign FPE(m)= [(T+m+ l)/(t-m- I)] [SSR(m)/T] (2) 3This argument shows that devaluation could either raise or lower a firm's stock prices depending on whether that firm is an exporting firm or it is a user of imported inputs. If it is involved in both activities, its stock prices could move in either direction. This will even be more true when an overall index of most stock prices is used to investigate the effects of devaluation on stock market. No wonder that Aggarwal(1981) and Soenen and Hennigar (1988) report conflicting results for different periods. 4For a nice intuitive explanation of the portfolio approach see Krueger (1983, pp ). Other related works are Dornbusch (1975), Frenkel and Rodriguez (1975) and Boyer (1977). For comparing the portfolio approach to other approaches such as monetary and traditional, see Dornbusch (1981).

4 Stock prices and the efectioe exchange rate of the dollar 461 where T is the total number of observations used in each regression, m is the number of lags, and SSR(m) is the associated sum of squared residuals. The optimum number of lags, say m*, is that which yields smallest FPE. In the second stage we move back to Equation 1 and treat the EX variable as a controlled variable with the order of lags set at its optimum level m*, and treat the SP variable as a manipulated variable. Once again by varying the order of lags on the manipulated variable from 1 to N, N twodimensional autoregressions are estimated. The optimum number of lags on the second variable, n*, is the one that minimizes the following two-dimensional FPE statistics: If FPE(m*, n*) in the second stage happens to be less than FPE(m*) from the first stage, then it is usually concluded that SP Granger causes EX. By going through exactly the same procedure and by interchanging EX and SP, we will also be able to detect causality from EX to SP. It should be indicated that most previous studies have relied on a Chow-test (F-test) in selecting the optimum number of lags of the manipulated variable. Throughout our empirical work, we also use the F-test in addition to FPE criteria to enhance our search for an optimum number of lags. More precisely, for each two-dimensional autoregressive equation we calculate the F-statistics. The largest and significant F-statistics or a decline in the two-dimensional FPE (m,n) statistics will govern the choice of optimum number of lags on the manipulated variable. The long-run equilibrium relationship between two variables is usually detected by the cointegration method outlined by Granger (1986) and Engle and Granger (1987). The essence of cointegration analysis is that if two variables are integrated of the same order d, they are said to be cointegrated if in the regression of one on the other, the residuals are integrated of any order less than d. IV. EMPIRICAL RESULTS This section tries to test the causality as well as cointegration between SP and EX variables using monthly observations over the period July 1973-December 1988 for a total of 186 observations from the US economy. SP variable is proxied by the values of Standard and Poor's Composite Index of 500 stocks (S&P 500) and the EX variable by the effective exchange rate of the d01lar.~ As indicated before, one of the requirements for the Granger test is that both time series must be stationary. Therefore, we first need to check for the stationarity of the time series involved, i.e. EX and SP series. To this end, the Augmented Dickey-Fuller (ADF) test is used. For a time series X, the ADF test is usually formulated by Equation 4 as follows: where L is the lag operator and wis an error term. The test is whether the estimate of d =O. To that end ADF test statistics is calculated by dividing estimate of d to its standard error. The cumulative distribution of the ADF test statistics is provided by Fuller (1976, p. 373).6 If the calculated statistics is less than its critical value from Fuller's table, then X is said to be stationary. The result of ADF test for SP and EX variables are reported in Table 1. Table 1 not only reports the results of ADF test for the level of each variable but also for the first differenced variables. Table 1 shows that the calculated ADF statistic is less than its critical value only when first differenced data are used. This indicates that while EX and SP themselves are not stationary, their first differences are, i.e. both series are integrated of order 1. Therefore, we need to use the first differenced variables to test for causality. As indicated in the methodology section, in the first stage of Granger causality test, (1 - L)EX, and (1 - L)SP, variables (L is the lag operator) are treated as a one-dimensional autoregressive process. By varying the order of lags from 1 to 24, the minimum FPE(m*) and the optimum number of lags for each variable are calculated and reported in Table 2. In the second stage each of the (1 - L) EX, and (1 - L)SP, variables are treated as controlled variables with the order of lags set at 1 and 2 respectively and the other variable is treated as a manipulated variable. By varying the order of lags on the manipulated variable from 1 to 24 the twodimensional FPE statistics are calculated as outlined by Equation 3. The results of this attempt are reported in Table 3. As can be seen from Table 3, the treatment of SP as a manipulated variable reduces the FPE of the controlled variable, EX, from to indicating that stock prices Granger cause exchange rates. However, treatment of EX as a manipulated variable does not reduce the FPE of the controlled variable SP. Thus by the FPE criteria, EX does not Granger cause SP. Do we obtain similar results if we rely upon the F-test? As indicated before, the F-test 'The data for S&P 500 are taken from Standard and Poor's Security Price Index Record: Statistical Service (various issues). The data for effective exchange rate of the dollar are from International Financial Statistics of the IMF (various issues). 6Dickey et al. (1986, p. 19) used only one lag of (1 - L)X, variable in Equation 6. However, there are indications in the literature that in order to assure that the residuals in Equation 4 are white noise, additional lags of (1 - L)X, must be added. The choice of additional lags is usually dictated by the level of significance of estimated lag coefficients. Throughout the paper, although we used four lags of first differenced term, we made sure that the reported results are not sensitive to this choice.

5 M. Bahmani-Oskooee and A. Sohrabian Table 1. Stationary test of EX and SP series using Augmented Dickey-Fuller (ADF) procedure Variable ADF test statistics Note: The critical value of ADF statistic at the usual 5% level from Fuller's Table (1976, p. 373) is for 100 observations and for 250 observations. Table 4 indicates that whereas in the case of EX as controlled variable, the optimum number of lags selected by both FPE criteria and F-test are the same, in the case of SP as controlled variable, they are not. Using the F-test, it appears that there is at least one model with two lags on SP and 14 lags on EX that yields significant F-statistics indicating that by this criteria, exchange rate also causes a change in stock prices. Therefore, using Table 4 the following models are selected: Table 2. The minimum FPE optimum number of lags when (1 - L) EX and (I - L)SP are treated as one-dimensional autoregressive process - Variable Min. FPE Optimum number of lags (1 - L)EX (1 - L)SP Table 3. The optimum number of lays on the manipulated variable and the minimum two-dimensional FPE (m, n) -- - Optimum lags of the Controlled variable Manipulated variable manipulated variable FPE (m, n) Table 4. The optimum number of lugs on the manipulated variable using F-test Optimum lags of the Controlled Manipulated manipulated F-statistics variable variable variable (DF) * Indicates that the calculated F-statistic is significant at the 5% level. should be caclulated for all 24 two-dimensional autoregressions. The optimum number of lags on the manipulated variables are the one that yields the largest F-statistics. Table 4 reports the results of this attempt. The full information results of Equations 5 and 6 are reported in Table 5. The size and significance of an individual lag coefficient in any autoregressive model may usually reflect the short-run effects of that variable. The long-run effects are usually inferred by the sum of the lag coefficients.' Considering the full-information estimates of Equation 5, Table 5 shows that the sum of the lagged coefficients of the (1 - L)SP variable is with a t-statistic of ' As for the sum of the lag coefficients of (1 - L)EX variable from estimates of Equation 6, a value of with the t-ratio of is obtained. Since these t-ratios are not significant, we can say that in the long-run there exists no relation between stock prices and the value of the dollar. What does the cointegration approach yield? Now that it has been established that there exists a dual causality between the EX and SP variables in the short-run but not in the long-run, we turn to the cointegration approach and determine whether further evidence can be obtained on the lack of a long-run relationship between effective exchange rate of the dollar and the US stock prices. Three distinct steps have to be taken. In the first step we must determine that both the EX and SP series are integrated of the same order. This has already been investigated and we have seen that both series are integrated of first order, i.e. EX, - I(1) and SP, - I(1). In the second step we estimate the cointegration equations where one variable is regressed on the other. The OLSQ estimates of cointegration equations with the standard error beneath each coefficient are reported in Equations 7 and 8. EX, = SP, (7) (2.84) (0.018) SP, = EX, (8) (36.64) (0.29) 'For more on this interpretation see Bahmani-Oskooee (1985, 1989). 'The t-ratio for the sum of the lagged coefficients is calculated by dividing the sum by the standard deviation of the sum. The standard deviation of the sum, in turn, is calculated from the variance-covariance matrix.

6 ~ Stock prices and the eflective exchange rate of the dollar Table 5. Full information estimates oj Equations 5 and 6 Coefficient -. ~ -on lag of Dependent variable (1- L) EX, Table 6. The ADF test applied to the residuals of Equations 7 ( E) and 8 (E') Variable ADF test statistics (1-L)EX,-1 See note to Table 1. Constant Numbers in parentheses are the t-ratios. Table 6 shows that the ADF calculated statistics for E and E' series are much bigger than their critical value from Fuller's (1976, p. 373) table indicating that the residuals are not stationary. However, when the first-differenced values of E and E' are used in an ADF test, as Table 6 shows, the ADF calculated statistics are much less than their critical values, indicating that the first-differenced residuals have achieved stationarity. Therefore, E-I(1) and E'-I(1). Since the residuals of the cointegration equations are integrated at an order that is the same as degree of integration of the EX and SP series themselves, we conclude that the two series are not, cointegrated. This indicates that there is no long-run relationship between two variables, a result similar to that obtained by checking the long-run effects of one variable on the other. V. CONCLUSION Few previous studies that have investigated the relation between stock prices and exchange rates, have postulated that a change in exchange rates effects the stock prices. They have reached this conclusion by simply regressing the stock prices on exchange rates. In this paper it is argued that it is possible for stock prices also to affect the exchange rates, i.e. there could be a two-way relationship between the two variables. Using the Granger concept of causality combined with Akaike's final prediction error criteria as well as the Chowtest, it was shown that there is a dual causal relationship between the stock prices measured by S&P 500 index and effective exchange rate of the dollar, at least in the short-run. Using the cointegration approach, we were unable to establish any long-run relationship between the two variables. REFERENCES Finally, we need to use the residuals of Equation 7, E, and those of Equation 8, E' to determine their order of integra- Once again this is done by using the ADF test for and E' as well as for their first differenced values. Table 6 reports the results of this practice. Aggarwal, R. (1981) Exchange rates and stock prices: a study of the U.S. capital markets under floating exchange rates. Akron Business and Economic Review, Ang, I. S. and Ghallab, A. (1976) The impact of U.S. devaluations on the stock prices of multinational cooperations. Journal of Business ~esearch, 4,

7 Bahmani-Oskooee, M. (1985) Devaluation and the J-curve: some evidence from LDCs. Review of Economics and Statistics, 67, Bahmani-Oskooee, M. (1989) Effects of the U.S. Government budget on its current account: an empirical inquiry. Quarterly review of Economics and Business, 29, Boyer, R. S. (1977) Devaluation and portfolio balance. American Economic Review, 67, Cooper, R. V. L. (1974) Efficient capital markets and the quantity thoery of money. Journal of Finance, 29, Cornell, B. (1983) Money supply announcements and interest rates: another view. Journal of Business, 6, Darrat, A. F. (1990) The impact of federal debt upon stock prices in the United States. Journal of Post Keynesian Economics, 12, Dickey, D. A., Bell, W. R. and Miller, R. B. (1986) Unit roots in time series models: tests and implications. ~meri'can Statistician, 40, Dornbusch, R. (1975) A portfolio balance model of the open economy. Journal of Monetary Economics, 1, Dornbusch, R. (1981) Monetary policy under exchange-rate flexibility. in International Trade and Finance Readings (Eds) E. Baldwin and J. D. Richardson, 2nd edn, Little, Brown, Boston pp Engle, R. F. and Granger, C. W. J. (1987) Co-integration and error correction: representation, estimation and testing. Econometrica, 55, Fama, E. F. and Schwert, G. W. (1977) Asset returns and inflation. Journal of Financial Economics, 5, Franck, P. and Young, A. (1972) Stock price reaction of multinational firms to exchange realignments. Financial Management, 1, Frenkel, J. A. and Rodriguez, C. A. (1975) Portfolio equilibrium and the balance of payments: a monetary approach. American Economic Review, 65, Fuller, W. A. (1976) Introduction to statistical time series, John Wiley, New York. Granger, C. W. J. (1986) Developments in the study of cointegrated economic variables. Oxford Bulletin of Economics and Statistics, 48, Grossman, J. (1981) The rationality of money supply expectations M. Bahmani-Oskooee and A. Sohr~bian and the short-run response of interest rates to monetary suprises. Journal of Money, Credit and Banking, 13, Hamburger, M. J. and Kochin, L. A. (1972) Money and stock prices: the. channels of influence. Journal of Finance, 27, Homa, K. E. and Jaffee, D. M. (1971) The supply of money and common stock prices. Journal of Finance, 26, Hsiao, C. (1981) Autoregressive modelling and money-income causality detection. Journal ofmonetary Economics, 7, Krueger, A. 0. (1983) Exchange-Rate Determination, Cambridge University Press, Cambridge, UK. Pearce, D. K. and Roley, V. (1983) The reaction of stock prices to unanticipated changes in money: a note. Journal of Finance, 38, Pearce, D. #. and Roley, V. (1985) Stock prices and economic news. Journal of Business, 58, Palmer, M. (1970) Money supply, portfolio adjustments and stock prices. Financial Analysts Journal, July/August, Rogalski, R. and Vinso, J. (1977) Stock returns, money supply and the direction of causality. Journal of Finance, 32, Rogalski, R. and Vinso, J. (1978) An analysis of monetary aggregates. Journal of Money, Credit and Banking, 10, Rozeff, M. (1974) Money and stock prices. Journal of Financial Economics 1, Schwert, G. W. (1981) The adjustment of stock prices to information about inflation. Journal of Finance 36, Soenen, L. A. and Hennigar, E. S. (1988) An analysis of exchange rates and stock prices-the U.S. experience between 1980 and Akron Business and Economic Review, Winter Sorensen, E. H. (1982) Rational expectations and the impact of money upon stock prices. Journal of Financial and Quantitative Analysis, 17, Tanner, J. E. and Trapani, J. M. (1977) Can the quantity theory be used to predict stock prices-or is the stock market efficient? Southern Economic Journal, 44, Urich, T. and Wachtel, P. (1981) Market response to the weekly money supply announcements in the 1970s. Journal of Finance, %, Wong, S. Q. (1986) The contribution of inflation uncertainty to the variable impacts of money on stock prices. Journal of Financial Research, 9,