15, Vol. 1, No. Recent Comovements of the Yen-US Dollar Exchange Rate and Stock Prices in Japan Chikashi Tsuji Professor, Faculty of Economics, Chuo University 7-1 Higashinakano Hachioji-shi, Tokyo 19-393, Japan E-mail: mail_sec_low@minos.ocn.ne.jp Received: September, 15 Accepted: October, 15 Published: October 5, 15 doi:1.59/wjbm.v1i.1 URL: http://dx.doi.org/1.59/ wjbm.v1i.1 Abstract This paper examines the recent relations of the yen/us dollar exchange rate and stock prices in Japan. Using bivariate Bayesian Vector Autoregressive (VAR) models, we derive several interesting findings as follows. First, 1) our analyses by Bayesian VAR models clarify that recently, the daily lags of the yen/dollar exchange rate series statistically significantly explain the evolution of the Nikkei 5, Nikkei 5, Japan Securities Dealers Association Quotation (JASDAQ), and Tokyo Stock Price Index (TOPIX) Core 3 stock index prices in Japan. Second, ) our impulse response analyses reveal that Japanese stock prices clearly respond to the yen/dollar exchange rate changes in the recent years whilst the exchange rate changes little respond to the stock prices in Japan. As above our results demonstrate, recently, the past yen/dollar exchange rate time-series much more affect the evolution of the Japanese stock prices whilst the past Japanese stock price series little affect the yen/dollar exchange rate changes. Moreover, 3) analyzing the time-varying correlation coefficients between the yen/dollar exchange rate changes and Japanese stock returns, we also find the large increases in the contemporaneous correlations between the exchange rate and stock returns in the recent years in Japan. Keywords: Bayesian VAR model, Comovement, Exchange rate, Stock price 19
15, Vol. 1, No. 1. Introduction Recently, the yen has been highly depreciated than before while the Japanese stock prices have generally increased. Then how are the correlations between the yen/dollar exchange rate changes and Japanese stock price series recently? Did the relations between them change in the recent years? In order to answer the question by using the actual time-series data, this paper examines the recent relations between the yen/us dollar exchange rate and several stock prices in Japan. As for methodology in our analyses, we apply Bayesian Vector Autoregressive (VAR) models to the time-series data. Our interesting findings are as follows. First, 1) our analyses by Bayesian VAR models clarify that recently, the daily lags of the yen/dollar exchange rate statistically significantly explain the price movements of the Nikkei 5, Nikkei 5, Japan Securities Dealers Association Quotation (JASDAQ), and Tokyo Stock Price Index (TOPIX) Core 3 stock indices in Japan. Second, ) our impulse response analyses reveal that the yen/dollar exchange rate series strongly affect the stock price changes in the recent years whilst the exchange rates show little clear response to the shock in stock prices in Japan. Moreover, 3) our analyses of the time-varying correlation coefficients between the yen/dollar exchange rate changes and stock returns in Japan also derive the clear evidence of the higher contemporaneous correlations between stock returns and the exchange rate changes in the recent years. As for the rest of this paper, Section reviews existing studies; Section 3 introduces our data and variables; Sections describes our methodology; Section 5 explains our results; and Section presents our conclusions.. Literature Review We here review very recent related studies. First, Litsios (13) suggested that the UK equity market significantly affected the nominal values of pound sterling effective exchange rate in the long-run from 19 to 11. Tsagkanos and Siriopoulos (13) estimated the relations between stock prices and exchange rates in the US and EU during the recent financial crisis from to. Śmiech and Papież (13) tested the causality among fossil fuel prices, exchange rates, and the German stock index by using weekly data from October 1 to June. Dellas and Tavlas (13) examined exchange rate regimes and asset prices and they suggested that the responses of asset prices to various shocks slightly differed across regimes. Using international data during the banking crisis from 7 to 1, Caporale et al. (1) examined the relations between stock market prices and exchange rates in the US, UK, Canada, Japan, the euro area, and Switzerland. Du (1) tested a hypothesis that persistent evolution of exchange rates is a distress risk and a Merton s (1973) state variable; and this study derived the supportive evidence for the hypothesis. Ho and Huang (15) tested the nonlinear relations between stock price indices and exchange rates of Brazil, Russia, India, and China. Abouwafia and Chambers (15) tested the linkages among monetary policy, exchange rates, and stock prices in the Middle East region. Gelman et al. (15) analyzed the relations between real exchange rates and capital flows. Kal et al. (15) examined the international links among stocks, bonds and exchange rates. Using major emerging market data, Aktuğ (15) tested the interactions among financial assets including exchange rates.
15, Vol. 1, No. Table 1. Descriptive statistics of the changes in terms of the yen/dollar exchange rate and stock returns in Japan Panel A. Descriptive statistics for the period from April 1, 1 to November 13, Mean Maximum Minimum Standard deviation Skewness Kurtosis Mean Maximum Minimum Standard deviation Skewness Kurtosis EXLR NK5LR NK5LR.51 3.351.119.593 1.1 9.975 JASLR.51.11 1.5.95 3.7..3 5.53 11.153 1.37 1.37 11.7 TC3LR.59 5.199. 1.51.53 7.939.3 7.1975 1. 1.1793 1.7 1.17 Panel B. Descriptive statistics for the period from December, to August 1, 15 Mean Maximum Minimum Standard deviation Skewness Kurtosis Mean Maximum Minimum Standard deviation Skewness Kurtosis EXLR NK5LR NK5LR.59 3.191.737..39.519 JASLR. 5.31 5.5511 1.15.17.5.111.57 7.597 1.395.5559 5.37 TC3LR.99 5.35.179 1.35.1.35.3.973 7.9 1.55.599.9 Notes: Descriptive statistics for the analyzed variables are shown in this table. EXLR denotes the percentage log changes of the yen/dollar exchange rate, NK5LR means the percentage log return of the Nikkei 5 stock index, NK5LR means the percentage log return of the Nikkei 5 stock index, JASLR denotes the percentage log return of the JASDAQ stock index, and TC3LR represents the percentage log return of the TOPIX Core 3 stock index in Japan. 1
15, Vol. 1, No. 3. Data This section describes our data and variables. We use an exchange rate and four Japanese stock index series. Specifically, EXLR denotes the percentage log changes of the yen/us dollar exchange rate, NK5LR means the percentage log return of the Nikkei 5 stock index, NK5LR means the percentage log return of the Nikkei 5 stock index, JASLR denotes the percentage log return of the JASDAQ stock index, and TC3LR represents the percentage log return of the TOPIX Core 3 stock index in Japan. We also use the level variables of the above five variables. Namely, EX denotes the level variable of the yen/dollar exchange rate, NK5 means the level variable of the Nikkei 5 stock index, NK5 means the level variable of the Nikkei 5 stock index, JAS denotes the level variable of the JASDAQ stock index, and TC3 represents the level variable of the TOPIX Core 3 stock index in Japan. Further, our analyzing sample periods are firstly 1) from April 1, 1 to November 13, and secondly ) from December, to August 1, 15. All data for our analyses are from the QUICK Corp. The descriptive statistics of the above five daily time-series of log change (return) variables are shown in Table 1. This table suggests the following variable characteristics. First, 1) the mean values of the five variables are all negative for the period from April 1, 1 to November 13, and they are all positive for the period from December, to August 1, 15. Second, ) the standard deviations of the five variables are slightly higher for the period from December, to August 1, 15 than those for the period from April 1, 1 to November 13,. We also show the standard correlation coefficients among EXLR, NK5LR, NK5LR, JASLR, and TC3LR in Table. Table clearly shows that the standard correlation coefficients between the changes of the yen/us dollar rate and the four Japanese stock returns are much higher for the period from December, to August 1, 15 than those from April 1, 1 to November 13,.. Methodology and Models This section documents our analyzing methodology and models. In order to examine the dynamic linkages of the yen/dollar exchange rate and Japanese stock index prices, we first compute and compare the time-varying correlation coefficients with -days windows. Second, we estimate four kinds of bivariate Bayesian VAR models with level variables and examine the impulse response functions derived from the VAR models. More concretely, the models we estimate in this paper are the bivariate Bayesian VAR() models of 1) EX and NK5; ) EX and NK5; 3) EX and JAS; and ) EX and TC3. We can summarize all our VAR models as the equations (1) and () as below. p q t 1 h 1 1, h t h j 1 1, j t j 1, t (1) y y z, p q t r 1, r t r s 1, s t s, t () z y z. In the above models, we employ the lag orders p = and q = to maintain the consistency of our analyses. Further, in the above equations, y is the yen/dollar exchange rate and z means
15, Vol. 1, No. one of the four Japanese stock prices: the Nikkei 5 stock index, the Nikkei 5 stock index, the JASDAQ stock index, or the TOPIX Core 3 stock index. Table. Correlation coefficients among the changes of the yen/dollar exchange rate and returns of stock indices in Japan Panel A. Correlation coefficients for the period from April 1, 1 to November 13, EXLR NK5LR NK5LR JASLR TC3LR EXLR NK5LR NK5LR JASLR TC3LR 1..375***.37***.199***.3517*** 1..975***.79***.9*** 1..71***.935*** 1..*** 1. Panel B. Correlation coefficients for the period from December, to August 1, 15 EXLR NK5LR NK5LR JASLR TC3LR EXLR NK5LR NK5LR JASLR TC3LR 1..57***.***.39***.*** 1..97***.9***.9395*** 1..***.939*** 1..1*** 1. Notes: This table shows standard correlation coefficients for our two sample periods and they are those among the log changes of the yen/dollar exchange rate and the log returns of four kinds of stock indices in Japan. Regarding our variables, EXLR denotes the percentage log changes of the yen/dollar exchange rate, NK5LR means the percentage log return of the Nikkei 5 stock index, NK5LR means the percentage log return of the Nikkei 5 stock index, JASLR denotes the percentage log return of the JASDAQ stock index, and TC3LR represents the percentage log return of the TOPIX Core 3 stock index in Japan. In this table, *** denotes the statistically significant rejection of the null hypothesis that the absolute value of the correlation coefficient is zero at the 1% level. 3
1//1 1//17 1/1/ 11/1/ 11/3/1 11/5/7 11//3 11/1/ 11// //9 /5/1 /7/17 /9/1 13/3/ 13/5/ 13/7/5 13/1/ 13//1 1// 1/5/ 1/7/1 1/9/17 1/11/7 15//9 15//1 15// 1//1 1//17 1/1/ 11/1/ 11/3/1 11/5/7 11//3 11/1/ 11// //9 /5/1 /7/17 /9/1 13/3/ 13/5/ 13/7/5 13/1/ 13//1 1// 1/5/ 1/7/1 1/9/17 1/11/7 15//9 15//1 15// 1//1 1//17 1/1/ 11/1/ 11/3/1 11/5/7 11//3 11/1/ 11// //9 /5/1 /7/17 /9/1 13/3/ 13/5/ 13/7/5 13/1/ 13//1 1// 1/5/ 1/7/1 1/9/17 1/11/7 15//9 15//1 15// World Journal of Business and Management 15, Vol. 1, No. Panel A. EXLR and NK5LR April 1, 1 to November 13, December, to August 1, 15 1...... -. -. -. -. 1...... -. -. Panel B. EXLR and NK5LR April 1, 1 to November 13, December, to August 1, 15 1...... -. -. -. -. 1...... -. -. Panel C. EXLR and JASLR April 1, 1 to November 13, December, to August 1, 15...... -. -..... -. -. -.
1//1 1//17 1/1/ 11/1/ 11/3/1 11/5/7 11//3 11/1/ 11// //9 /5/1 /7/17 /9/1 13/3/ 13/5/ 13/7/5 13/1/ 13//1 1// 1/5/ 1/7/1 1/9/17 1/11/7 15//9 15//1 15// World Journal of Business and Management 15, Vol. 1, No. Panel D. EXLR and TC3LR April 1, 1 to November 13, December, to August 1, 15 1...... -. -. -. -. 1...... -. -. Figure 1. Time-series Evolution of the Time-varying Correlation Coefficients between the Changes of the yen/us dollar Exchange Rate and Four Japanese Stock Index Returns. 5. Empirical Results First, Figure 1 shows the daily time-varying correlation coefficients with -days windows. In all panels, the correlations for the period from April 1, 1 to November 13, are on the left side and correlations for the period from December, to August 1, 15 are on the right side. Panels AD of Figure 1 demonstrate that all four Japanese stock return variables of NK5LR, NK5LR, JASLR, and TC3LR show much higher time-varying correlations with the yen/dollar exchange rate changes in our recent sample period. Next, estimation results of our four kinds of bivariate Bayesian VAR models are shown in Table 3. Specifically, in Table 3, Panel A presents the results for the period from April 1, 1 to November 13, and Panel B exhibits the results for the period from December, to August 1, 15. From Table 3, it is understood that, in Panel A, the lag variables of the exchange rate do not statistically significantly explain the future stock prices and the lag variables of stock prices do not statistically significantly explain the future exchange rate changes, either. On the other hand, in Panel B of Table 3, the first lag variables of EX statistically significantly and positively explain the four stock index prices, NK5, NK5, JAS, and TC3. In addition, Panel B of Table 3 shows that the first lag variables of NK5, NK5, and TC3 also present some statistically significant explanatory power for the time-series evolution of the EX. However, we note that all the statistically significant coefficients of these three stock price lag variables take very small values. Further, in order to judge the degrees of the above effects derived from the Bayesian VAR models, we depict mutual impulse response functions of EX, NK5, NK5, JAS, and TC3 in Figures and 3. Figure shows the responses for the period from April 1, 1 to 5
15, Vol. 1, No. November 13, and Figure 3 displays those for the period from December, to August 1, 15. Figures and 3 show that, in our first sample period, EX little responds to the shock in stock prices and only two stock prices of NK5 and NK5 respond to the shock in EX. As for our recent sample period, EX again little responds to the shock in stock prices while all four stock prices of NK5, NK5, JAS, and TC3 positively respond to the shock in EX strongly. Hence the above results clearly demonstrate that, recently, the past yen/dollar exchange rate time-series affect the evolution of the Japanese stock prices much strongly. On the other hand, the past series of the Japanese stock index prices little affect the evolution of the yen/dollar exchange rate in our two sample periods. Table 3. Estimation results of the bivariate Bayesian VAR models Panel A. Results for the period from April 1, 1 to November 13, Model 1 Model EX(1) EX() EX(3) EX() NK5(1) NK5() NK5(3) NK5() EX NK5 EX NK5.7939***.97***.1.7*.5.115.517..1915.1.9.1..1.111 1.5***. 9.95.51 15.599*.77 5.51.137 1.7.7.9***.711**.391..1.115.557 15.77**.5 EX(1) EX() EX(3) EX() NK5(1) NK5() NK5(3) NK5().79***.***.1.7*.39.11.53.17...1.1..17*. 1.7***..37.37 1..1..1.135.73.973***.53.3.9.371.73.795 17.91**.17 Adj.R.91.971 Adj.R.9.93 Model 3 Model EX(1) EX() EX JAS EX TC3.799***.135***.1.3.19..13 EX(1) EX().75***.95***.1.57..15*.5
EX(3) EX() JAS(1) JAS() JAS(3) JAS().*.7.17.5.337...9.1*.79.351*.755 1.5***.9..793..9 1.553***.9**..1.53.3.99 1.7***.9 EX(3) EX() TC3(1) TC3() TC3(3) TC3() World Journal of Business and Management 15, Vol. 1, No..*.55..5..17.1...339.9*.73 1.331***..11.39.7.7.935***.539.1193.17.991.13.59 5.7.75 Adj. R.93.97 Adj. R.9.973 Panel B. Results for the period from December, to August 1, 15 Model 1 Model EX(1) EX() EX(3) EX() NK5(1) NK5() NK5(3) NK5() EX NK5 EX NK5.57***.91***.7.7.939.1..**.31.1.7.1.11 1.79E5.79.3.35 9.91***.5 17.33.119.5.9599 3.39..155***.133***.1.5.9.131.515 31.5553**.397 EX(1) EX() EX(3) EX() NK5(1) NK5() NK5(3) NK5().53***.93***.7.9.353.15..3**.15..77.1...775.77.1393.173**.11 1.51*...99.19.577.593***.155***.7.1.957.13.91.33. Adj. R.99.9935 Adj. R.99.995 7
Model 3 Model EX(1) EX() EX(3) EX() JAS(1) JAS() JAS(3) JAS() World Journal of Business and Management 15, Vol. 1, No. EX JAS EX TC3.5***.**.11.177.9.11.57.15.1.17.5.1..73.513.9.37.13**.115.73.19.33.73.113.777.935***.11..5.799.17.3.153.7 EX(1) EX() EX(3) EX() TC3(1) TC3() TC3(3) TC3().53***.971***.5.31.39.17.359.5**.137.9.9.35*.7..791.3.37 1.355***.7.*.7.11.71.15.55.***.**.173.15.5.11.15 3.975.33 Adj. R.99.9937 Adj. R.99.9911 Notes: This table shows the estimation results of our four kinds of Bayesian VAR() models for the exchange rate and stock price series in Japan. In this table, EX denotes the level variable of the yen/dollar exchange rate, NK5 means the level variable of the Nikkei 5 stock index, NK5 means the level variable of the Nikkei 5 stock index, JAS denotes the level variable of the JASDAQ stock index, and TC3 represents the level variable of the TOPIX Core 3 stock index in Japan. Further, the adjusted R-squared value is denoted by Adj. R and ***, **, and * represent the statistical significance of the coefficients at the 1, 5, and 1% levels, respectively. Panel A. Response of NK5 to EX Panel B. Response of EX to NK5 1 1 1 1 1 1 1 1 1 1 1 1
15, Vol. 1, No. Panel C. Response of NK5 to EX Panel D. Response of EX to NK5 1 1 1 1 1 1 1 1 1 1 1 1 Panel E. Response of JAS to EX Panel F. Response of EX to JAS 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Panel G. Response of TC3 to EX Panel H. Response of EX to TC3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Figure. Accumulated Mutual Impulse Responses of Stock Prices and Exchange Rates: Results for the period from April 1, 1 to November 13,. Panel A. Response of NK5 to EX Panel B. Response of EX to NK5 1, 1, 1, 1 1 1 1 1 1 1 1 1 1 1 1 9
15, Vol. 1, No. Panel C. Response of NK5 to EX Panel D. Response of EX to NK5 1 9 1 1 7 1 5 1 3 1 1 1 1 1 1 1 1 1 Panel E. Response of JAS to EX Panel F. Response of EX to JAS 1 1 1 1 1 1 1 1 1-1 1 1 1 Panel G. Response of TC3 to EX Panel H. Response of EX to TC3 5 5 1 1 35 1 3 5 1 15 1 5 1 1 1 1 1 1 1 1 Figure 3. Accumulated Mutual Impulse Responses of Stock Prices and Exchange Rates: Results for the period from December, to August 1, 15.. Conclusions This paper examined the recent relations among the yen/dollar exchange rate and stock prices in Japan. Using bivariate Bayesian VAR models, we derived several interesting findings as follows. First, 1) our analyses by Bayesian VAR models clarified that recently, the first daily lag variables of the yen/dollar exchange rate were statistically significant in explaining the evolution of the Nikkei 5, Nikkei 5, JASDAQ, and TOPIX Core 3 stock price indices in Japan. In addition, ) our impulse response analyses revealed that, in the recent years, the yen/dollar 3
15, Vol. 1, No. exchange rate little responded to the past stock prices whilst stock prices strongly responded to the yen/dollar exchange rate changes. As above our examinations demonstrate, recently, the past yen/dollar exchange rate time-series affected the evolution of the Japanese stock prices much more strongly whilst the past Japanese stock price series still had little effect on the yen/dollar exchange rate changes. Furthermore, 3) by analyzing the correlation coefficients between the yen/dollar exchange rate changes and stock returns in Japan, we also found that, recently, daily contemporaneous correlations between exchange rate changes and Japanese stock returns largely increased. The above evidence observed particularly in the recent years in Japan may be mainly because of the recent quantitative easing monetary policy executed by the Bank of Japan. Thus further related analysis by using other methodology and other viewpoints is one of our future tasks. Acknowledgement I am particularly grateful to the repeated kind invitation from the journal to write to this journal. I also appreciate the Zengin Foundation for Studies on Economics and Finance for their grant-in-aid to this research. Further, I am grateful to the Japan Society for the Promotion of Science for their generous financial assistance to my research. Moreover, I thank the anonymous referees of this journal for their comments on this paper. Furthermore, I thank Jessica Caroline for the kindness to my paper. Finally, I deeply thank all the Editors of this journal for their kindness to my paper. References Abouwafia, H. E., & Chambers, M. J. (15). Monetary policy, exchange rates and stock prices in the Middle East region. International Review of Financial Analysis, 37, 1-. http://dx.doi.org/1.11/j.irfa.1.11.1 Aktuğ, R. E. (15). Empirical dynamics of emerging financial markets during the global mortgage crisis. Borsa İstanbul Review, 15, 17-3. http://dx.doi.org/1.11/j.bir.1.11.1 Caporale, G. M., Hunter, J., & Ali, F. M. (1). On the linkages between stock prices and exchange rates: Evidence from the banking crisis of 7-1. International Review of Financial Analysis, 33, 7-13. http://dx.doi.org/1.11/j.irfa.13..5 Dellas, H., & Tavlas, G. (13). Exchange rate regimes and asset prices. Journal of International Money and Finance, 3, 5-9. http://dx.doi.org/1.11/j.jimonfin.13.5. Du, D. (1). Persistent exchange-rate movements and stock returns. Journal of International Financial Markets, Institutions & Money,, 3-53. http://dx.doi.org/1.11/j.intfin.13.1.7 Gelman, M., Jochem, A., Reitz, S., & Taylor, M. P. (15). Real financial market exchange rates and capital flows. Journal of International Money and Finance, 5, 5-9. http://dx.doi.org/1.11/j.jimonfin.15.. 31
15, Vol. 1, No. Ho, L. C., & Huang, C. H. (15). The nonlinear relationships between stock indexes and exchange rates. Japan and the World Economy, 33, -7. http://dx.doi.org/1.11/j.japwor.15.. Kal, S. H., Arslaner, F., & Arslaner, N. (15). The dynamic relationship between stock, bond and foreign exchange markets. Economic Systems, forthcoming. http://dx.doi.org/doi:1.11/j.ecosys.15.3. Litsios, I. (13). Exchange rate determination and equity prices: Evidence from the UK. The Journal of Economic Asymmetries, 1, 115-. http://dx.doi.org/1.11/j.jeca.1.1. Merton, R. C. (1973). An intertemporal capital asset pricing model. Econometrica, 1, 7-7. Śmiech, S., & Papież, M. (13). Fossil fuel prices, exchange rate, and stock market: A dynamic causality analysis on the European market. Economics Letters, 11, 199-. http://dx.doi.org/doi:1.11/j.econlet..1.1 Tsagkanos, A., & Siriopoulos, C. (13). A long-run relationship between stock price index and exchange rate: A structural nonparametric cointegrating regression approach. Journal of International Financial Markets, Institutions & Money, 5, 1-11. http://dx.doi.org/1.11/j.intfin.13.1. Copyright Disclaimer Copyright for this article is retained by the author(s), with first publication rights granted to the journal. This is an open-access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3./). 3