Relationship Between Exchange Rates and Stock Prices: Empirical Evidence from Bahrain’s Financial Markets
Rizwan Tahir Ahmed Abdul Ghani
Abstract
Relationship between stock prices and exchange rates in Bahrain has been examined by using monthly data form January 1992 to October 2002. Co-integration and ECM models and Granger causality tests were used to determine the causal relationship between stock prices and exchange rates. The empirical results suggests long-run bi-directional causal relationship between stock prices and exchange rates (British Pond & Japanese Yen) and only uni-directional, from stock prices to exchange rate, causal relationship between them. There is, however, no evidence of uni or bi-directional causality between stock prices and exchange rate (German Mark) in short-run or long-run. The overall evidence support the goods market approach of exchange rate determination.
Keywords : Stock Prices, Exchange Rates, Causality Analysis Relationship Between Exchange Rates and Stock Prices: Empirical Evidence from Bahrain’s Financial Markets
1. INTRODUCTION
Two approaches have been developed in the literature to establish a relationship between exchange rate and stock prices. First approach, known as goods market approach, (e.g.
Dornbusch & Fischer (1980)) suggest that changes in exchange rates affect the competitiveness of a firm, which in turn influence the firm’s earnings or its cost of funds and hence its stock price. On a macro basis, then, the impact of exchange rate fluctuations on stock market would depend on both the degree of openness of domestic economy and the degree of the trade imbalance. Second approach, known as portfolio balance approach
(e.g. Frankel (1993)), stresses the role of capital account transactions. A rising stock market would attract capital flows which increases the demand for domestic currency and cause exchange rate to appreciate.
Although theories suggest causal relationship between exchange rate and stock prices, existing evidence indicates a weak link between them at a micro level. For US firms,
Jorion (1990, 1991), Bodnar and Gentry(1993) were unable to find a significant relationship and for Japaneses firms, He & Ng (1998) found that only 25 percent of their sample of 171 Japaneses multinationals has significant exchange rate exposure on stock returns.
On the macro level, Ma & Kao (1990) found that a currency appreciation negatively affects the domestic stock market for an export-dominant country and positively affects the domestic stock market for an import-dominant country, which seems to be consistent
2 with goods market theory. Ajayi and Mougoue (1996), using daily data for eight countries, show significant interactions between foreign exchange and stock markets, while Abdulla & Murinde (1997) document that a country’s monthly exchange rates tends to lead its stock prices but not the other way around. On the issue of causality between exchange rate and stock market, Ramasamy & Mathew (2001) noted that whether stock price movements cause exchange rate volatility or vice versa is country and time dependent.
Bahrain presents an example of a small open economy and considered as financial hub for the gulf region. Bahrain follows a pegged exchange rate system with its currency,
Bahraini Dinar (BD), being pegged against U.S. Dollar. It is reasonable to expect that exchange rate, vis a vis US dollar, may not fully response to stock price movements.
Therefore in this paper we have considered exchange rate between BD and some other
Major currencies, (Dutch Mark (DM), British Pond (GBP), Japanese Yen (YEN)), to seek evidence for any possible relationship between Bahrain’s stock market and major exchange rates.
II METHODOLOGY
Co-integration and error correction modelling technique involves three main steps.
Testing the relevant time series for stationarity (unit roots), testing for co-integration, and finally error-correction modelling. We shall use standard textbook1 notation to explain briefly the steps involved.
1 The discussion is based on Maddala (1988) “ Introduction to Time Series Analysis”.
3 A non-stationary time series Yt is said to be integrated of order d, [Yt ~ I(d)], if it achieves stationarity after being differenced d times2. To determine the order of integration, unit root tests have been developed. The most common test is known as Dickey-Fuller3 (DF) or Augmented Dickey-Fuller (ADF)4. To discuss the DF test, consider the model
Yt = β0 + β1t + ut ut = αut-1 + εt
where εt is a covariance stationary process with zero mean. The reduced form for this model is
Yt = γ + δt + β1t + αYt-1 + εt ------(1)
Where γ = β0 (1 – α) + β1α and δ = β1 (1 – α). This equation is said to have a unit root if α = 1. The DF test is based on testing the hypothesis α = 1 in (1) under the assumption that
εt are white noise errors. The test statistics are: k(1) = T(ά – 1) t(1) = (ά – 1)/SE(ά)
Since these statistics do not have a standard t distribution, the critical values for k(1) and t(1) are tabulated in Fuller (1976).
Suppose that Yt ~ I(d) and Xt ~ I(d). Then Yt and Xt are said to be co-integrated if there
5 exists a β such that Yt - β Xt is I(d-b) and b > 0. Thus testing for co-integration one must make sure that both series are integrated of the same order in first step. Second step then
2 This definition is due to Granger (1986) and Engle and Granger (1987). 3 Dickey and Fuller (1979). 4 Engle and Granger (1987) has argued that ADF test allows for dynamics in the DF regression and consequently is over-parameterised in the first order case but correctly specified in the higher order cases. 5 An alternate approach to test for co-integration is developed by Johansen (1988, 1991). Johansen method is more appropriate for models with more than two time series.
4 involves estimationg the folloing co-integration equation by Ordinary Least Squares (OLS):
Yt = a0 + bo Xt + μt ------(2)
′ X t = a0 + bo Yt + μ t ------(3)
And testing for the stationarity of the residuals from Equations 2 & 3 to make sure that μt
′ and μ t are I(d-b), b > 0. If two variables are co-integrated, then the third step involves formulating the error- correction model (ECM) as follows:
n n (1 L)Yt C0 d 0 t1 eoi (1 L)Yti f oi (1 L) X ti t (4) i1 i1
n n (1 L)X C d e (1 L)X f (1 L)Y t 1 1 t 1 1i t i 1i ti t (5) i1 i1
′ where L is the lag operator and the error correction terms (ECTs) μt and μ t are the stationary residuals from co-integration Equations 2 and 3, respectively. According to the standard Granger causality test, X is said to Granger cause Y if foi’s are jointly significant. The inclusion of ECTs, however, provide additional channel through which the Granger causality could be detected. Thus, X is said to Granger cause Y, as long as
6 the ECT carries a significant coefficient even if foi’s are not jointly significant.
III DATA AND ESTIMATION
The data used in the study is monthly from January 1992 to October 2002, obtained from
Bahrain Monetary Agency (BMA). Data include monthly observations on Stock Price
Index (STIDX), Exchange rate between Bahraini Dinar & British pond (GBP), Exchange rate between Bahraini Dinar & German Mark (DM) and Exchange rate between Bahraini
Dinar & Japanese yen (YEN).
6 Granger (1988), p. 203
5 For estimation, we have used Regression Analysis of Time Series (RATS). For testing unit roots, we employed RATS procedure “URAUTO.SRC”. It runs a set of unit root tests in sequence attempting to classify the series based upon trend and unit root properties. We also employed procedure “PPUNIT.SRC” which computes the Phillips-
Perron (1988) modifications of the Dickey-Fuller test.
IV RESULTS OF ANALYSIS
The three steps of the co-integration technique and error-correction modelling, described above, are employed to investigate the relationship between stock price index (STIDX) and three major exchange rates, (GBP), (DM) & (YEN). We use log of these variables, such that their first differences could reflect the rate of change of each variable. The results corresponding to each step of the technique are reported in Tables 1, 2 and 3 respectively. Results reported in table 1 show that the calculated test values, in all specifications, are less than their critical values only for first differences of LOG STIDX,
LOG GBP, LOG DM & LOG YEN. This indicates that each variable is integrated of order one, i.e., I(1). If residuals from co-integration regressions are found to be integrated of order zero, i.e., I(0), stock price index and all three exchange rates would be considered co-integrated. Three Models, (Model 1: Stock Price Index & Exchange Rate
GBP; Model 2: Stock Price Index & Exchange Rate DM; Model 3: Stock Price Index &
Exchange Rate YEN) were estimated and results of six co-integration regressions (two for each model), are reported in table 2. The test values are less than their critical vlues.
6 This suggests that all residuals are stationary and are integrated of order zero, i.e., I(0), which implies that stock price index and three major exchange rates are co-integrated.
The above analysis suggests that there exists a long-run relationship between stock price index and each one of three exchange rates. But in order to determine which variable
Granger causes the other and provides the short-run dynamics adjustment toward the long-run equilibrium, three error-correction (EC) Models were also estimated. Since EC model, as described by equations 4 and 5, involves lag variables, one must determine the optimal number of lags for each variable. An F-test was used for this purpose to select the appropriate specification in each case. The results for all three Models are reported in table 3.
The results indicates that there exists a long-run bi-directional causality between stock price index and exchange rate GBP (Model 1), and stock price index and exchange rate
YEN (Model 3). Because in each of these cases, error-correction term (EC-1) turns out to be highly significant. The negative coefficient of error-correction term suggest that currency appreciation negatively effects domestic stock market which seems to be consistent with goods market theory of exchange rate. In the short-run there is an evidence of only uni-directional causality from stock prices to exchange rate because stock price coefficients are found to be jointly significant in the exchange rate equations
(Model 1 & Model 3). This suggest a lead of stock prices on exchange rate in all Models except Model 2.
There is, however, no evidence of uni-directional or bi-directional causality between stock price index and exchange rate DM (Model 2), either in short-run or long-run.
7 V. CONCLUDING REMARKS
Monthly data form January 1992 to October 2002 was used to determine the causal relationship between stock prices and exchange rates by using co-integration and ECM models and Granger causality tests. The empirical results suggests long-run bi-directional causal relationship between stock prices and exchange rates (British Pond & Japanese
Yen) and only uni-directional, from stock prices to exchange rate, causal relationship between them. There is, however, no evidence of uni or bi-directional causality between stock prices and exchange rate (German Mark) in short-run or long-run. The overall evidence support the goods market approach of exchange rate determination.
8 Table No. 1
Results of DF unit root tests applied to the level and first differences of stock index and major exchange rate series
Series With Constant No Conclusion Constant and no Constant and Trend Trend and no Trend Log STIDX -3.20049 -2.76315 -0.04714 Series has unit root Log GBP -2.34834 -2.10506 -0.45999 Series has unit root Log DM -2.58593 -1.93807 -0.06091 Series has unit root Log YEN -2.84664 -2.22079 -0.11331 Series has unit root First Differences7 (1-L)Log STIDX -4.23564 - - Series stationary (1-L)Log GBP -6.05321 - - Series stationary (1-L) Log DM -4.52291 - - Series stationary (1-L) Log YEN -6.01001 - - Series stationary CRITICAL VALUES AT 5% SIGNIFICANCE -3.41000 -2.86000 -1.95000 LEVEL
7 The RATS procedure “URAUTO.SRC” conclude about stationarity of the series based on only one version which include constant and trend.
9 Table No. 2
Results of DF unit root tests applied to the residuals of Co-integration Equations
Co-integration Equation DF Test Values for Equation Residuals Conclusion Model 1
Log STIDX = f [Log GBP] -4.07564 Residuals Stationary
Log GBP = f [Log STIDX] -11.81597 Residuals Stationary Model 2
Log STIDX = f [Log DM] -22.53253 Residuals Stationary
Log DM = f [Log STIDX] -12.53519 Residuals Stationary Model 3
Log STIDX = f [Log YEN] -5.49961 Residuals Stationary
Log YEN = f [Log STIDX] -8.65191 Residuals Stationary
Test equation include constant (no trend) and critical value at 5% significance level is –2.86000
10 Table No. 3
RESULTS FOR ERROR-CORRECTION MODELS Model 1
DEPENDENT T- STATISTIC F-STATISTICS F-STATISTICS VARIABLE FOR EC-1 FOR FOR ∑(1-L)LOG ∑(1-L)LOG GBP STIDX (1-L)LOG STIDX -2.58547* 1.4118[12]*** 1.0000[12] (1-L) LOG GBP -2.69180* 28.8167[12]* 30.000[12]*
Model 2
DEPENDENT T- STATISTIC F-STATISTICS F-STATISTICS VARIABLE FOR EC-1 FOR FOR ∑(1-L)LOG ∑(1-L)LOG DM STIDX (1-L)LOG STIDX -0.62728 0.889[12] 1.1324[12] (1-L) LOG DM -1.40096 1.1609[12] 6.4828[12]*
Model 3
DEPENDENT T- STATISTIC F-STATISTICS F-STATISTICS VARIABLE FOR EC-1 FOR FOR ∑(1-L)LOG ∑(1-L)LOG STIDX YEN (1-L)LOG STIDX -2.83267* 1.1765[12] 1.000[12] (1-L) LOG YEN -2.31156** 1.2143[12]*** 0.6935[12] NOTES: EC-1 denotes lagged error-correction term. Number Includes in the brackets are the number of lags. *,**,*** shows significance at 1 percent, 5 percent, and 10 percent levels.
11 REFERENCES
Abdalla, I.S.A. and V. Murinde, 1997, “Exchange Rate and Stock Price Interactions in Emerging Financial Markets: Evidence on India, Korea, Pakistan, and Philippines,” Applied Financial Economics 7, 25-35
Ajayi, R. A. and Mougoue, 1996, “On the Dynamic Relation between Stock Price and Exchange Rates, “Journal of Financial Research 19, 193-207.
Bodnar, G.M. and W. M. Gentry, 1993, “Exchange Rate Exposure and Industry Characteristics: Evidence from Canada, Japan, and USA,” Journal of International Money and Finance 12, 29-45
Dickey, D. A. and W. A. Fuller , 1979, “ Distribution of the Estimators for Autoregressive Time Series with a Unit Root,” Journal of the American Statistical Association 84, 427-31
Dornbusch, R. and S. Fischer, 1980, “Exchange Rates and Current Account,” American Economic Review 70, 960-71
Engle, R. F. and C. W. J. Granger, 1987, “Co-integration and Error Correction: Representation, Estimation, and Testing,” Econometrica 55, 251-76
Frankel, Jeffrey A., 1993, “Monetary and Portfolio-Balance models of the Determination Of Exchange rates” In Jeffrey A. Frankel on exchange rates, Cambridge, MA: MIT Press
Fuller, W. A., 1976, “Introduction to Statistical Time Series,” New York, John Wiley and Sons.
Granger, C. W. J., 1986, “Developments in the Study of Co-integrated Economic Variables,” Oxford Bulletin of Economics and Statistics, 48:3 213-28
Granger, C. W. J., 1988, “Some Recent Developments in a Concept of Causality,” Journal of Monetary Economics, 39, 199-106
He, J. and L. K. Ng, 1998, “The Foreign Exchange Exposure of Japanese Multinational Corporations,” Journal of Finance, 53, 733-753
Johansen, S., 1988, “Statistical Analysis of Co-integration Vectors,” Journal of Economic Dynamics and Control, 12, 231-254
Johansen, S. and K. Juselius. 1990, “Maximum Likelihood Estimation and Inference on Co-integration with Application to the demand for Money,” Oxford Bulletin of Economics and Statistics, 52, 169-209
12 Jorion, P., 1990, “The Exchange Rate Exposure of U.S. Multinationals,” Journal of Business, 63, 331-345 Jorion, P., 1991, “The pricing of Exchange Rate Risk in the Stock Market,” Journal of Financial and Quantitative Analysis, 26, 363-376
Ma, C. K. and G. W. Kao, 1990, “On Exchange Rate Changes and Stock Price Reactions,” Journal of Business Finance & Accounting, 17, 441-449
Maddala, M. D. 1988, “Introduction to Time Series Analysis,”
Phillips, P. and P. Perron, 1988, “Testing for a Unit Root in Time Series Regression,” Biometrica, 75, 335-46
Ramasamy, B. and M. Yeung, 2001, “The Causality Between Stock Returns and Exchange Rates: Revisited,” Research Paper Series,11, Division of Business and Management, The University of Nottingham in Malaysia
13