- Email: [email protected]
Spillover effect of US monetary policy to ASEAN stock markets: Evidence from Indonesia, Singapore, and Thailand
Spillover effect of US monetary policy to ASEAN stock markets: Evidence from Indonesia, Singapore, and Thailand Lu Yang, Shigeyuki Hamori PII: DOI: Reference:
S0927-538X(13)00091-7 doi: 10.1016/j.pacfin.2013.12.003 PACFIN 667
To appear in:
Pacific-Basin Finance Journal
Received date: Accepted date:
29 June 2013 5 December 2013
Please cite this article as: Yang, Lu, Hamori, Shigeyuki, Spillover effect of US monetary policy to ASEAN stock markets: Evidence from Indonesia, Singapore, and Thailand, Pacific-Basin Finance Journal (2013), doi: 10.1016/j.pacfin.2013.12.003
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
T
Spillover effect of US monetary policy to ASEAN stock markets: Evidence from Indonesia, Singapore, and Thailand Running head: Spillover effect of US monetary policy to ASEAN stock markets
RI P
Lu Yang Graduate School of Economics, Kobe University 2-1 Rokkodai, Nada-Ku, Kobe 657-8501, Japan
SC
E-mail: [email protected]
Shigeyuki Hamori
MA NU
Tel.: +81 080 4016 4406
Faculty of Economics, Kobe University
2-1, Rokkodai, Nada-Ku, Kobe 657-8501, Japan Email: [email protected]
Abstract
PT
ED
Tel: +81-78-803-6832
AC
CE
In this paper, we investigate the spillover effect from US monetary policy to selected ASEAN stock markets by employing Markov-switching models. Based on univariate Markov-switching models, we confirm the existence of two distinct regimes for both US monetary policy and the stock markets. By applying multivariate Markov-switching models, we find that US interest rates have a negative effect on the selected ASEAN stock markets during economic expansion periods. However, this kind of effect disappears during economic crisis periods. Our empirical results indicate that the spillover effect from US monetary policy influences the ASEAN stock markets only during the tranquil period. These results have important implications for the transmission mechanisms of asset price, such as the credit channel, trade channel, and balance sheet channel.
Keywords: Markov-switching models, Spillover effect, Excess liquidity, Monetary policy. JEL codes: C22 E44 E52 G15 Acknowledgements: We are grateful to the anonymous referee for the many helpful comments and suggestions. The research performed by the second author is in part supported by the Grant-in-Aid of the Japan Society for the Promotion of Science.
ACCEPTED MANUSCRIPT 1. Introduction
AC
CE
PT
ED
MA NU
SC
RI P
T
Over the past decades, the stock markets in the Association of Southeast Asian Nations (ASEAN) have increasingly matured as they integrated into the world capital markets. However, most countries in the ASEAN are still emerging economies. Since most of them have the properties of a small open economy, their stock markets are more easily affected by changes in US monetary policies. Traditional economic theory suggests a relationship between stock market performance and information (e.g., Fama et al., 1969; Mitchell and Mulherin, 1994). Shocks (information) from changes in monetary policy play an important role in the stock market since it is designed to impact the macro-economy, which in turn affects the stock market indirectly. As the world economy globalizes and world financial markets integrate, the shocks from developed markets like the US affect other markets through various transmission mechanisms such as the credit channel, balance sheet channel, and trade channel. The main objective of this paper is to analyze the impacts of the US interest rate and excess liquidity on the ASEAN stock markets. Kim and Nguyen (2009) investigated the spillover effect from the U.S. Federal Reserve System’s (FED) and the European Central Bank’s (ECB) target interest rate news on the market returns and return volatilities of 12 stock markets in the Asia-Pacific. They found that a majority of stock markets showed significant negative returns in response to unexpected rate rises, and return volatilities for these markets were higher in response to the interest rate news.1 Even though our paper uses a totally different frequency of data and methodology, the contributions of Kim (2003, 2009) showed that the US interest rate had a direct impact on the ASEAN stock markets. Methodologically, most existing studies focusing on excess liquidity analysis2 are based on vector autoregression (VAR) models. Bagliano and Morana (2012) found that financial disturbances were transmitted to foreign countries through US house and stock price dynamics, as well as excess liquidity creation. Moreover, they found that that the trade channel was the key transmission mechanism of real shocks. However, Brana (2012) found that excess liquidity at the global level had spillover effects on output and price levels in emerging countries but had little impact on real estate, commodity, and share prices in emerging countries. In contrast to previous literature, we analyze the spillover effect based on state-dependent models.3 Specifically, we select the 3-month Treasury bill rate as the Federal fund rate and return on US stock price index as the measure of excess liquidity. There are abundant studies that refer to excess liquidity measurement and effects. Generally, excess liquidity can be identified by rising asset prices (Belke et al., 2010; Belke et al., 2013). Since asset prices are good indicators of general price in global financial markets, in this paper, we utilize them to measure excess liquidity effect. Particularly, we select US stock market to be our measure of the excess liquidity effect and to be the transmission intermediary. Moreover, by employing state-dependent models, we can investigate the spillover effect in different regimes based on this backdrop. In this paper, we focus on three questions. First, we determine whether the monetary policy 1 2 3
For an earlier discussion of this issue, refers to Kim (2003). For the theoretical analysis, refer to Agénor and Aynaoui (2010). For applications of Markov-switching models, refer to Chan et al. (2011) and Simo-Kengne et al. (2013).
ACCEPTED MANUSCRIPT
RI P
T
in the US influences the ASEAN stock markets in this paper. Second, if so, we determine whether the spillover effect differs across the regimes? Finally, we also determine whether there are differences of empirical results between the ASEAN countries? The remainder of this paper is organized as follows. Section 2 discusses the methodology we used in this paper. Section 3 describes the data and statistical issues. Section 4 provides the empirical results, and Section 5 concludes. 2. Specifications of Markov-switching models
MA NU
SC
In this paper, the multivariate Markov-Switching Intercept Autoregressive Heteroscedasticity (MSIAH) model (Guidolin and Timmermann, 2006; Ang and Timmermann, 2012) is applied to analyze the linkage between US monetary policy and the ASEAN stock markets. The model can be defined in a general form as follows: (1)
PT
ED
where yt refers to a matrix including the return on equity index, return on gold price index, and 3-month Treasury bill rate, which we examine. is a vector of means in state St, and is a matrix of autoregressive coefficients in state St. Assuming the unobservable state-dependent parameter St follows an irreducible ergodic M-state Markov process with a transition matrix, we have the following transition matrix:
CE
(2)
AC
where Further, we assume that the residuals follow a normal distribution for all regimes:
where
(3) and
is a
variance-covariance matrix conditional on St:
(4)
Finally, the conditional distribution of yt based on state St and past information can be expressed as (5)
where N = 3 is the number of variables in the system with the estimated joint distribution.
ACCEPTED MANUSCRIPT Incorporating the unobservable state variable St yields
T
(6)
(7)
SC
RI P
where M represents the number of possible regimes. The log likelihood function can be constructed as
(8)
AC
CE
PT
ED
MA NU
where T is the number of observations in the data set. The maximum likelihood method is used to estimate the parameters , , and for St = 1,…, M and the transition probability P. Particularly, Equation 1 is reduced to a simple VAR model when there is one regime (M = 1). For the univariate case, we also employ the Markov-Switching Intercept Heteroscedasticity = 0 for St = 1,…, M, to determine whether the more parsimonious (MSIH) model, where model can provide an adequate description of each variable. To find the best-fitting Markov-switching models, the Schwarz Bayesian Criterion (SBC) is selected. The best-fitting Markov-switching model is the one with the lowest SBC value. In addition, the linearity test of Davies (1977, 1987) is employed to justify the suitability of the state-dependent model. Moreover, the regime classification measure (RCM) of Ang and Bekaert (2002) is employed to evaluate the suitability of the state-dependent model. This measure is defined as
where the constant term 400 ensures the statistic ranges from 0 to 100, and refers to the smoothed regime probabilities conditioned on the full information set . A value of 0 implies a perfectly discrete two-regime model, while a value of 100 implies a perfectly integrated two-regime model. Therefore, a value of 50 usually serves as a benchmark. A value below 50 indicates that the two-regime model performs well, while a value above 50 indicates that the single-regime model works well.
3. Data description To investigate the impacts of the changes in US monetary policy on the ASEAN stock markets, we employ the return on equity index and 3-month Treasury bill rate as one system. Specifically, the US 3-month Treasury bill rate denotes the Federal fund rate (FED), return on equity bases on the Standard & Poor’s 500 (S&P 500) to measure the inflation overflow from excess liquidity, and return on equity bases on the Morgan Stanley Capital International (MSCI) stock price index for ASEAN countries. Instead of using the Fed fund rate directly, we employ the US 3-month Treasury bill rate to reflect the expectations of the market and the
ACCEPTED MANUSCRIPT
4. Empirical results
ED
MA NU
SC
RI P
T
movement of the interest rate. As to the ASEAN countries, we select Indonesia, Singapore, and Thailand for our research because they are the initial members of the ASEAN and therefore have greater data availability. Moreover, the difference between the stock markets of the developed country (Singapore) and the developing countries (Indonesia and Thailand) is also a concern. The sample period is from January 1990 to December 2012 based on monthly frequency. All the data is taken from Datastream. Table 1 reports descriptive statistics for the monthly returns (yt). Generally, the average returns for the stock markets are positive, indicating the long-term bull market for stocks in the ASEAN. The highest return and volatility are observed in the Indonesian stock market. The results of the Jarque-Bera (JB) test show that the null hypothesis of the normal distribution (unconditionally) is rejected in all cases. Table 2 reports a correlation matrix for the monthly returns (yt). Positive correlations are observed between US stock and Federal fund rate and between US and ASEAN stock market, while negative correlations are observed between ASEAN stock and Federal fund rate. These results are consistent with general economic theory. When the Federal Reserve cuts its interest rates, this is always positive news for the outside financial markets since the required rate decreases. The positive correlation between the US and ASEAN stock markets indicate that excess liquidity from the US has a positive correlation with the ASEAN stock markets. The raw data are plotted in Figure 1.
PT
4.1 Regimes for the marginal distributions
AC
CE
In the first step, we investigate whether there is evidence of regime-switching behavior for US monetary policy and the ASEAN stock markets. To do that, we treat each series individually by fitting a range of two-, three-, and four-regime MSIH and MSIAH models. By comparing the SBC values, we find the most appropriate Markov-switching model for each series. Finally, we evaluate the multi-regime model against the single-regime model based on the Davies and RCM test statistics.4 Table 3 reports the performances of the univariate Markov-switching models. Panel A summarizes the SBC values of the various univariate models in terms of fitting our variables. The results uniformly imply that the two-state model is better than the three- and four-state models based on the lowest SBC value for each series. Our empirical results with two specified regimes are roughly consistent with those obtained from real life markets according to the criteria of Pagan and Sossounov (2003). Moreover, the more parsimonious MSIH model is more appropriate in all cases except for the US 3-month Treasury bill rate. Panel B reports the Davies and RCM statistics. The Davies statistics are significant at 1% for all cases, indicating that state dependency exists in our model. Further, the RCM statistics for all cases are below 50, consistent with the existence of two regimes. Table 4 provides the estimation results of the two-state MSIAH model for the US 3-month Treasury bill rate and the two-state MSIH model for the other variables. For the Federal fund rate, we find that the mean yield is higher with higher volatility in regime 1 than that in regime 2. In contrast, we find that the mean returns are lower with higher volatility in regime 4
For the Markov-switching model, see Kim and Nelson (1999) and Bhar and Hamori (2004).
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
1 than that in regime 2 for the stock markets. The economic interpretation for these states is straightforward. Regime 1 represents the recession period for the economy when the low-risk assets such as bonds provide higher returns than the risky assets. For the US stock market, the mean return for both regimes provides positive returns while for the ASEAN stock markets, the case is contrary. For example, all the three stock markets have negative returns in regime 1 but positive returns in regime 2. Moreover, the volatility of the stock markets is lower in regime 2, which indicates the economies are experiencing stable growth. Table 4 also reports the estimated transition probabilities and their respective expected durations. Based on the results in Table 4, we conclude that the bull-market periods are much longer than the bear-market periods for ASEAN countries. Particularly, since the US 3-month Treasury bill rate is treated as the Federal fund rate, we can state that the monetary policy in the US experiences longer expansive periods than tight periods since the duration of regime 1 is longer. Meanwhile, the evidence from US stock returns tells us that excess liquidity plays an important role in pushing the asset price up. The reason behind this tendency can be inferred as follows. The state-dependent model works well in our specified model, suggesting that the growth rate of money supply is not constant. Especially, the high growth rate of money supply always brings excess liquidity, which pushes the asset price up easily. As demonstrated in our empirical results, the mean in regime 1 for the US stock market is lower and lasts longer than that in regime 2. Since regime 1 represents the crisis period for the US economy, the behavior of the Federal Reserve becomes clear, that is, it always provides liquidity to stimulate the economy. However, the capital always pursues a higher return and will flow to the outside markets, which in turn will lower the returns from the domestic market when the economy is not good. To demonstrate the transition of the market movement, we plot the transition probabilities for both regimes in Figure 2. Figure 2 shows the crisis periods (e.g., oil shock in 1991, Asian currency crisis in 1999, international financial crisis in 2008), indicated by the black line, are specified well in our model. Moreover, that the figure shows the international financial crisis has a smaller effect on the ASEAN countries than the Asian currency crisis did since the duration of regime 1 is shorter. Particularly, the longer duration of regime 2 in terms of the Federal fund rate indicates a bear market for the bond, which implies an excess liquidity in the US monetary system. In summary, our univariate results have a number of implications for understanding business cycles and market movements. First, we identify two states for each variable, in which case the movement of a stock market can be classified into either a bear or bull market. Second, our univariate results provide us the durations of specific regimes during the sample periods. Generally, the volatility is lower in the economic expansion regime, which is also more persistent for the ASEAN stock market, as indicated by the high positive mean value. In contrast, the US stock market experiences more crisis periods during our sample periods (which starts in 1990). Moreover, the monetary policy in the US also experiences a long series of expansive periods, especially after 2008, which makes the spillover effect from the excess liquidity an important issue. 4.2 Regimes for the joint distribution After demonstrating the regime switching of US monetary policy and the ASEAN stock
ACCEPTED MANUSCRIPT
SC
RI P
T
markets, we incorporate the US factors in our Markov-switching model to investigate the effect of the Federal fund rate and excess liquidity on the ASEAN stock markets. Following the same procedures, we first select the best-fitting MSIAH model based on the SBC values and evaluate it based on the Davies and RCM statistics. The results are given in Table 5. All the statistics suggest the two-state MSIAH model is the best-fitting model for all cases. Following the results of Pagan and Sossounov (2003), we believe the two-state model can provide sufficient analysis for our model. From the empirical results in Table 6, we can see that lower mean returns with higher volatility are observed in regime 1. The only exception is Singapore, for which the mean in regime 1 is negative. Same as in the univariate Markov-switching models, regime 1 represents the bear market, while regime 2 represents the bull market. More importantly, only the ) has a negative effect on the selected ASEAN stock markets Federal fund rate (
MA NU
during the economic expansion periods (regime 2). A relative decrease in the short-term US interest rate in the US has a positive impact on the ASEAN stock returns in the following month. However, this kind of effect disappears during economic recession periods (regime 1) because the authority may shut down the transmission channels during the crisis periods. Further, it may be difficult for the excess liquidity from the US to flow into the ASEAN countries. In addition, we find a positive effect from the US stock market in both regimes for are significant and positive. The degree of this kind all cases since all the parameters
ED
of linkage increases in regime 2 when the market is good. However, the lagged stock returns ( ) from the domestic markets are not well justified. For example, only the lagged
AC
CE
PT
stock return from Indonesia in regime 1 is positive and significant, while only the lagged stock returns from both Singapore and Thailand in regime 2 stay significant and positive. In sum, the spillover effect from US monetary policy plays an important role in determining the asset price in ASEAN stock markets. The transition probabilities for both regimes are plotted in Figure 3. Compared to the univariate system, the multivariate system provides us more robust results since the SBC value is lower than in the univariate system. In addition, the smoothed probability of both regimes in the multivariate system reflects the fluctuations of the stock market more reliably. It is clear from the figure that the duration for both regimes decreases as we input US factors into our models. Meanwhile, we can identify more specific periods of stock market comovements from the effect of US monetary policy. These periods tend to converge with the US monetary policy cycle. For example, a mini-crash such as that caused by the 9/11 terrorist attack may cause a market downturn that lasts for quite some time. In summary, our analysis identifies two states for our multivariate system. The crisis regime (regime 1) is characterized by relative higher volatility and lower stock returns. During these periods, we find that the spillover effect from US monetary policy on the selected ASEAN market does not matter. In contrast, the tranquil regime (regime 2) is characterized by relative lower volatility and higher stock returns. Furthermore, during these periods, there is strong evidence that the spillover effect from US monetary policy influences the selected ASEAN stock markets. Overall, even though the ASEAN countries experience almost the same regimes during our sample periods, the duration time for each period is different. Finally, we find that the tranquil period of the stock market in Singapore seems to be longer than the crisis period.
ACCEPTED MANUSCRIPT
5. Conclusion
AC
CE
PT
ED
MA NU
SC
RI P
T
In this paper, we investigate the linkage between the regime switching in US monetary policy and selected ASEAN stock markets. Specifically, we investigate the regime switching of the Federal fund rate (US 3-month Treasury bill rate), US stock market (S&P 500), and selected ASEAN stock markets based on univariate Markov-switching models. The two-state switching models are identified for all cases. Thus, we confirm the existence of two distinct regimes: periods of economic expansion and periods of economic decline. Based on the empirical results from the above analysis, we apply the multivariate Markov-switching models to investigate the effect of the Federal monetary policy for the different regimes on the ASEAN stock markets. We find that the Federal fund rate has a negative effect on the selected ASEAN stock markets during economic expansion periods. Moreover, we find that the ASEAN stock markets have a positive comovement with the US stock market in both regimes for all cases. However, the lagged stock returns on their own play only a small role in determining the future movement of stock markets in ASEAN countries. These results have important implications for the transmission mechanisms of asset price, especially from the US to small economies. For instance, the transmission mechanism differs by regime. The ASEAN stock markets are influenced by the Federal fund rate more easily in a bull market rather than in a bear market. Moreover, the spillover effect differs between the bull market and the bear market. Generally, the spillover effect has a greater impact in the bull market than in the bear market. Since capital control and exchange rate arrangement are always important issues for central banks in the developing world, our results show their stock markets are not easily influenced by the US dollar, especially during the times of economic crisis.
ACCEPTED MANUSCRIPT References
AC
CE
PT
ED
MA NU
SC
RI P
T
Agénor, P R, Aynaoui, K E (2010). Excess liquidity, bank pricing rules, and monetary policy. Journal of Banking and Finance, 34(5), 923-933. Ang, A, Bekaert, G (2002). Regime switches in interest rate. Journal of Business and Economics Statistics, 20(2), 163-182. Ang, A, Timmermann, A (2012). Regime changes and financial markets. Annual Review of Financial Economics, 4, 313-337. Bagliano, F C, Morana, C (2012). The Great Recession: US dynamics and spillovers to the world economy. Journal of Banking and Finance, 36(1), 1-13. Belke, A, Bordon, I G, Volz, U (2013). Effects of global liquidity on commodity and food prices. World Development, 44, 31-43. Belke, A, Orth, W, Setzer, R (2010). Liquidity and the dynamic pattern of asset price adjustment: A global view. Journal of Banking and Finance, 34(8), 1933-1945. Bhar, R and Hamori, S (2004). Hidden Markov Models : Applications to Financial Economics, Springer, Boston. Brana, S, Djigbenou, M L, Prat, S (2012). Global excess liquidity and asset prices in emerging countries: A PVAR approach. Emerging Markets Review, 13(3), 256-267. Chan, K F, Treepongkaruna, S, Brooks, R, Gray, S (2011). Asset market linkages: Evidence from financial, commodity and real estate assets. Journal of Banking and Finance, 35(6), 1415-1426. Davies, R B (1977). Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrika, 64(2), 247-254. Davies, R B (1987). Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrika 74(1), 33-43. Fama, E F, Fisher, L, Jensen, M C, Roll, R (1969). The adjustment of stock prices to new information. International Economic Review, 10(1), 1-21. Guidolin, M., Timmermann, A. (2006). An econometric model of nonlinear dynamics in the joint distribution of stock and bond returns. Journal of Applied Econometrics, 21(1), 1-22. Kim, S J (2003). The spillover effects of US and Japanese public information news in advanced Asia-Pacific stock markets. Pacific-Basin Finance Journal, 11(5), 611-630. Kim, C J and Nelson, C R (1999). State-Space Models with Regime Switching: Classical and Gibbs-Sampling Approaches with Applications, The MIT Press, Cambridge, Mass. Kim, S J, Nguyen, D Q T (2009). The spillover effects of target interest rate news from the U.S. Fed and the European Central Bank on the Asia-Pacific stock markets. Journal of International Financial Markets, Institutions and Money, 19(3), 415-431. Krolzig, H-M (1997). Markov-Switching Vector Autoregressions: Modelling, Statistical Inference, and Application to Business Cycle Analysis. Lecture Notes in Economics and Mathematical Systems. Springer-Verlag, Berlin. Mitchell, M L, Mulherin, J H (1994). The impact of public information on the stock market. Journal of Finance, 49(3), 923-950. Simo-Kengne, B D, Balcilar, M, Gupta, R, Reid, M, Aye, G C (2013). Is the relationship between monetary policy and house prices asymmetric across bull and bear markets in
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
South Africa? Evidence from a Markov-switching vector autoregressive model. Economic Modeling, 32, 161-171. Pagan, A R, Sossounov, K A (2003). A simple framework for analyzing bull and bear markets. Journal of Applied Econometrics, 18 (1), 23-46.
ACCEPTED MANUSCRIPT Table 1 Summary statistics. Indonesia
Singapore
Thailand
0.0335 0.0375 0.0222 −0.0870 1.8975 14.32***
0.0051 0.0101 0.0438 −0.7673 4.5180 53.58***
0.0075 0.0060 0.1018 0.2464 5.9505 102.9***
0.0026 0.0040 0.0624 0.3195 4.8113 42.42***
0.0022 0.0048 0.1010 0.5193 5.3838 77.76***
T
US
RI P
Mean Median Std. Dev. Skewness Kurtosis JB test
FED
MA NU
SC
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns. *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively. Table 2 Correlations of monthly returns for each country.
AC
CE
PT
ED
Indonesia Thailand Singapore FED US Stock FED US Stock FED US Stock FED 1 1 1 US 0.0540 1 0.0540 1 0.0540 1 Stock −0.0703 0.0719 1 −0.1214 0.0417 1 −0.0298 0.2015 1 Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns.
ACCEPTED MANUSCRIPT
Singapore
Thailand
−3.502 −3.476 −3.434 −3.369 −3.291 −3.241
−1.831 −1.795 −1.760 −1.718 −1.572 −1.652
−2.763 −2.748 −2.684 −2.651 −2.539 −2.519
−1.808 −1.773 −1.752 −1.710 −1.577 −1.612
0.000 14.067
0.000 23.765
0.000 32.883
0.000 21.285
T
Indonesia
SC
Panel A: SBC Values 2-State MSIH −6.114 2-State MSIAH −9.807 3-State MSIH −7.048 3-State MSIAH −9.774 4-State MSIH −6.383 4-State MSIAH −9.791 Panel B: Davies and RCM Statistics Davies Test 0.000 RCM Statistic 14.408
US
MA NU
FED
RI P
Table 3 Performance measures for univariate Markov switching models.
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. We present SBC values to select the best-fitting Markov-switching models. The lowest SBC values (bold face numbers) indicate the best-fitting Markov-switching models. Davies test statistics are presented as p values. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns.
FED MSIAH 1.695 (5.426) −0.450 (0.462) 0.987*** (0.012) 0.997*** (0.03) 0.075*** (0.008) 0.001*** (0.000) 0.935 0.879 15.43 8.276
US
MSIH −206.2 (186.9) 174.5*** (58.41)
29.37*** (3.335) 5.017*** (0.834) 0.982 0.973 54.55 37.27
233.1 *** (50.27) 46.19*** (6.293) 0.951 0.977 20.33 44.16
Singapore MSIH −3.612 (109.3) 41.97 (36.38)
Thailand MSIH −211.3 (174.7) 111.9* (60.05)
AC
CE
Indonesia
MSIH 29.02 (43.77) 92.24*** (22.91)
PT
Model
ED
Table 4 Parameter estimates for univariate two-state Markov-switching models.
Duration 1 Duration 2
89.05*** (20.41) 17.77*** (3.106) 0.913 0.965 11.49 28.76
220.3*** (42.22) 49.99*** (7.044) 0.963 0.982 27.23 55.82
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. refers to the transition probability in regime 1(2), and the expected duration in regime 1(2) is calculated as . *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively.
ACCEPTED MANUSCRIPT Table 5 Performance measures for multivariate Markov-switching models.
−3.025 −2.922 −2.792
−1.808 −1.775 −1.731
0.000 31.89
T
Thailand
RI P
Panel A: SBC Values 2-State MSIAH −1.962 3-State MSIAH −1.865 4-State MSIAH −1.597 Panel B: Davies and RCM Statistics Davies Test 0.000 RCM Statistic 15.07
Singapore
SC
Indonesia
0.000 11.75
AC
CE
PT
ED
MA NU
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. We present SBC values to select the best-fitting Markov-switching models. The lowest SBC values (bold face numbers) indicate the best-fitting Markov-switching models. Davies test statistics are presented as p values. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns.
ACCEPTED MANUSCRIPT Table 6 Parameter estimates for multivariate MSIAH models. Indonesia
PT CE
Duration 1 Duration 2
Thailand
MA NU
152.3 (207.9) 239.9*** (92.31) −0.089 (0.094) −0.336 (0.522) 0.949*** (0.219) 0.132* (0.073) −1.138*** (0.265) 0.934*** (0.174) 125.4*** (17.68) 46.05*** (6.856) 0.775 0.828 4.459 5.838
RI P
SC
−11.73 (10.43) 10.08 (6.088) −0.065 (0.076) 0.152 (0.246) 0.773*** (0.097) 0.191*** (0.061) −0.385** (0.123) 0.963*** (0.099) 24.41*** (3.174) 23.02*** (3.110) 0.899 0.956 9.952 22.92
T
*
ED
144.2 (186.9) 236.8 (78.35) 0.281*** (0.064) −0.696 (0.477) 0.712*** (0.201) −0.067 (0.057) −0.624*** (0.195) 1.234*** (0.169) 91.54*** (9.992) 70.96*** (6.356) 0.787 0.756 4.694 4.089
Singapore
AC
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P GSCI for gold (gold), and MSCI stock index for each country. refers to the transition probability in regime 1(2), and . *, **, and *** represent the expected duration in regime 1(2) is calculated as significance at the 10%, 5%, and 1% levels, respectively.
MA NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig. 1. Time-series evolution of stock price index and Federal fund rate (FED) (3-month Treasury bill rate), respectively. The right-side of the y axis for the FED is measured in percentage. The sample period is from January 1990 to December 2008.
ACCEPTED MANUSCRIPT
(a) FED Rate- MSIAH model
1.00
T
0.75
RI P
0.50
0.25
0.00
SC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
(b) US Stock Market-MSIH model
1.00
MA NU
0.75
0.50
0.25
0.00
ED
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
(c) Singaporean Stock Market- MSIH model
1.00
PT
0.75
0.25
0.00
CE
0.50
1.00
0.75
AC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
(d) Indonesian Stock Market - MSIH model
0.50
0.25
0.00 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
ACCEPTED MANUSCRIPT (e) Tailandese Stock Market - MSIH model
1.00
0.75
T
0.50
RI P
0.25
0.00
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
AC
CE
PT
ED
MA NU
SC
Fig. 2. Smoothed probability of regime 1 (black line) and regime 2 (blue line) for univariate Markov-switching models. The sample period is from January 1990 to December 2008.
ACCEPTED MANUSCRIPT
(a) Indonesia
1.00
T
0.75
RI P
0.50
0.25
0.00
(b) Singapore
1.00
MA NU
0.75
SC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
0.50
0.25
0.00
ED
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
1.00
PT
0.75
0.00
CE
0.50
0.25
(c) Thailand
AC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Fig. 3 Smoothed probability of regime 1 (black line) and regime 2 (blue line) for multivariate MSIAH models. The sample period is from January 1990 to December 2008.
ACCEPTED MANUSCRIPT Highlight
T
We investigate the spillover effect from US monetary policy to selected ASEAN stock markets.
RI P
We confirm the existence of two distinct regimes for both US monetary policy and stock markets.
SC
We find that US interest rates have a negative effect on the selected ASEAN stock markets during the economic expansion periods.
AC
CE
PT
ED
MA NU
We find a negative effect of excess liquidity during economic recession periods on the Singaporean stock market.
S0927-538X(13)00091-7 doi: 10.1016/j.pacfin.2013.12.003 PACFIN 667
To appear in:
Pacific-Basin Finance Journal
Received date: Accepted date:
29 June 2013 5 December 2013
Please cite this article as: Yang, Lu, Hamori, Shigeyuki, Spillover effect of US monetary policy to ASEAN stock markets: Evidence from Indonesia, Singapore, and Thailand, Pacific-Basin Finance Journal (2013), doi: 10.1016/j.pacfin.2013.12.003
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
T
Spillover effect of US monetary policy to ASEAN stock markets: Evidence from Indonesia, Singapore, and Thailand Running head: Spillover effect of US monetary policy to ASEAN stock markets
RI P
Lu Yang Graduate School of Economics, Kobe University 2-1 Rokkodai, Nada-Ku, Kobe 657-8501, Japan
SC
E-mail: [email protected]
Shigeyuki Hamori
MA NU
Tel.: +81 080 4016 4406
Faculty of Economics, Kobe University
2-1, Rokkodai, Nada-Ku, Kobe 657-8501, Japan Email: [email protected]
Abstract
PT
ED
Tel: +81-78-803-6832
AC
CE
In this paper, we investigate the spillover effect from US monetary policy to selected ASEAN stock markets by employing Markov-switching models. Based on univariate Markov-switching models, we confirm the existence of two distinct regimes for both US monetary policy and the stock markets. By applying multivariate Markov-switching models, we find that US interest rates have a negative effect on the selected ASEAN stock markets during economic expansion periods. However, this kind of effect disappears during economic crisis periods. Our empirical results indicate that the spillover effect from US monetary policy influences the ASEAN stock markets only during the tranquil period. These results have important implications for the transmission mechanisms of asset price, such as the credit channel, trade channel, and balance sheet channel.
Keywords: Markov-switching models, Spillover effect, Excess liquidity, Monetary policy. JEL codes: C22 E44 E52 G15 Acknowledgements: We are grateful to the anonymous referee for the many helpful comments and suggestions. The research performed by the second author is in part supported by the Grant-in-Aid of the Japan Society for the Promotion of Science.
ACCEPTED MANUSCRIPT 1. Introduction
AC
CE
PT
ED
MA NU
SC
RI P
T
Over the past decades, the stock markets in the Association of Southeast Asian Nations (ASEAN) have increasingly matured as they integrated into the world capital markets. However, most countries in the ASEAN are still emerging economies. Since most of them have the properties of a small open economy, their stock markets are more easily affected by changes in US monetary policies. Traditional economic theory suggests a relationship between stock market performance and information (e.g., Fama et al., 1969; Mitchell and Mulherin, 1994). Shocks (information) from changes in monetary policy play an important role in the stock market since it is designed to impact the macro-economy, which in turn affects the stock market indirectly. As the world economy globalizes and world financial markets integrate, the shocks from developed markets like the US affect other markets through various transmission mechanisms such as the credit channel, balance sheet channel, and trade channel. The main objective of this paper is to analyze the impacts of the US interest rate and excess liquidity on the ASEAN stock markets. Kim and Nguyen (2009) investigated the spillover effect from the U.S. Federal Reserve System’s (FED) and the European Central Bank’s (ECB) target interest rate news on the market returns and return volatilities of 12 stock markets in the Asia-Pacific. They found that a majority of stock markets showed significant negative returns in response to unexpected rate rises, and return volatilities for these markets were higher in response to the interest rate news.1 Even though our paper uses a totally different frequency of data and methodology, the contributions of Kim (2003, 2009) showed that the US interest rate had a direct impact on the ASEAN stock markets. Methodologically, most existing studies focusing on excess liquidity analysis2 are based on vector autoregression (VAR) models. Bagliano and Morana (2012) found that financial disturbances were transmitted to foreign countries through US house and stock price dynamics, as well as excess liquidity creation. Moreover, they found that that the trade channel was the key transmission mechanism of real shocks. However, Brana (2012) found that excess liquidity at the global level had spillover effects on output and price levels in emerging countries but had little impact on real estate, commodity, and share prices in emerging countries. In contrast to previous literature, we analyze the spillover effect based on state-dependent models.3 Specifically, we select the 3-month Treasury bill rate as the Federal fund rate and return on US stock price index as the measure of excess liquidity. There are abundant studies that refer to excess liquidity measurement and effects. Generally, excess liquidity can be identified by rising asset prices (Belke et al., 2010; Belke et al., 2013). Since asset prices are good indicators of general price in global financial markets, in this paper, we utilize them to measure excess liquidity effect. Particularly, we select US stock market to be our measure of the excess liquidity effect and to be the transmission intermediary. Moreover, by employing state-dependent models, we can investigate the spillover effect in different regimes based on this backdrop. In this paper, we focus on three questions. First, we determine whether the monetary policy 1 2 3
For an earlier discussion of this issue, refers to Kim (2003). For the theoretical analysis, refer to Agénor and Aynaoui (2010). For applications of Markov-switching models, refer to Chan et al. (2011) and Simo-Kengne et al. (2013).
ACCEPTED MANUSCRIPT
RI P
T
in the US influences the ASEAN stock markets in this paper. Second, if so, we determine whether the spillover effect differs across the regimes? Finally, we also determine whether there are differences of empirical results between the ASEAN countries? The remainder of this paper is organized as follows. Section 2 discusses the methodology we used in this paper. Section 3 describes the data and statistical issues. Section 4 provides the empirical results, and Section 5 concludes. 2. Specifications of Markov-switching models
MA NU
SC
In this paper, the multivariate Markov-Switching Intercept Autoregressive Heteroscedasticity (MSIAH) model (Guidolin and Timmermann, 2006; Ang and Timmermann, 2012) is applied to analyze the linkage between US monetary policy and the ASEAN stock markets. The model can be defined in a general form as follows: (1)
PT
ED
where yt refers to a matrix including the return on equity index, return on gold price index, and 3-month Treasury bill rate, which we examine. is a vector of means in state St, and is a matrix of autoregressive coefficients in state St. Assuming the unobservable state-dependent parameter St follows an irreducible ergodic M-state Markov process with a transition matrix, we have the following transition matrix:
CE
(2)
AC
where Further, we assume that the residuals follow a normal distribution for all regimes:
where
(3) and
is a
variance-covariance matrix conditional on St:
(4)
Finally, the conditional distribution of yt based on state St and past information can be expressed as (5)
where N = 3 is the number of variables in the system with the estimated joint distribution.
ACCEPTED MANUSCRIPT Incorporating the unobservable state variable St yields
T
(6)
(7)
SC
RI P
where M represents the number of possible regimes. The log likelihood function can be constructed as
(8)
AC
CE
PT
ED
MA NU
where T is the number of observations in the data set. The maximum likelihood method is used to estimate the parameters , , and for St = 1,…, M and the transition probability P. Particularly, Equation 1 is reduced to a simple VAR model when there is one regime (M = 1). For the univariate case, we also employ the Markov-Switching Intercept Heteroscedasticity = 0 for St = 1,…, M, to determine whether the more parsimonious (MSIH) model, where model can provide an adequate description of each variable. To find the best-fitting Markov-switching models, the Schwarz Bayesian Criterion (SBC) is selected. The best-fitting Markov-switching model is the one with the lowest SBC value. In addition, the linearity test of Davies (1977, 1987) is employed to justify the suitability of the state-dependent model. Moreover, the regime classification measure (RCM) of Ang and Bekaert (2002) is employed to evaluate the suitability of the state-dependent model. This measure is defined as
where the constant term 400 ensures the statistic ranges from 0 to 100, and refers to the smoothed regime probabilities conditioned on the full information set . A value of 0 implies a perfectly discrete two-regime model, while a value of 100 implies a perfectly integrated two-regime model. Therefore, a value of 50 usually serves as a benchmark. A value below 50 indicates that the two-regime model performs well, while a value above 50 indicates that the single-regime model works well.
3. Data description To investigate the impacts of the changes in US monetary policy on the ASEAN stock markets, we employ the return on equity index and 3-month Treasury bill rate as one system. Specifically, the US 3-month Treasury bill rate denotes the Federal fund rate (FED), return on equity bases on the Standard & Poor’s 500 (S&P 500) to measure the inflation overflow from excess liquidity, and return on equity bases on the Morgan Stanley Capital International (MSCI) stock price index for ASEAN countries. Instead of using the Fed fund rate directly, we employ the US 3-month Treasury bill rate to reflect the expectations of the market and the
ACCEPTED MANUSCRIPT
4. Empirical results
ED
MA NU
SC
RI P
T
movement of the interest rate. As to the ASEAN countries, we select Indonesia, Singapore, and Thailand for our research because they are the initial members of the ASEAN and therefore have greater data availability. Moreover, the difference between the stock markets of the developed country (Singapore) and the developing countries (Indonesia and Thailand) is also a concern. The sample period is from January 1990 to December 2012 based on monthly frequency. All the data is taken from Datastream. Table 1 reports descriptive statistics for the monthly returns (yt). Generally, the average returns for the stock markets are positive, indicating the long-term bull market for stocks in the ASEAN. The highest return and volatility are observed in the Indonesian stock market. The results of the Jarque-Bera (JB) test show that the null hypothesis of the normal distribution (unconditionally) is rejected in all cases. Table 2 reports a correlation matrix for the monthly returns (yt). Positive correlations are observed between US stock and Federal fund rate and between US and ASEAN stock market, while negative correlations are observed between ASEAN stock and Federal fund rate. These results are consistent with general economic theory. When the Federal Reserve cuts its interest rates, this is always positive news for the outside financial markets since the required rate decreases. The positive correlation between the US and ASEAN stock markets indicate that excess liquidity from the US has a positive correlation with the ASEAN stock markets. The raw data are plotted in Figure 1.
PT
4.1 Regimes for the marginal distributions
AC
CE
In the first step, we investigate whether there is evidence of regime-switching behavior for US monetary policy and the ASEAN stock markets. To do that, we treat each series individually by fitting a range of two-, three-, and four-regime MSIH and MSIAH models. By comparing the SBC values, we find the most appropriate Markov-switching model for each series. Finally, we evaluate the multi-regime model against the single-regime model based on the Davies and RCM test statistics.4 Table 3 reports the performances of the univariate Markov-switching models. Panel A summarizes the SBC values of the various univariate models in terms of fitting our variables. The results uniformly imply that the two-state model is better than the three- and four-state models based on the lowest SBC value for each series. Our empirical results with two specified regimes are roughly consistent with those obtained from real life markets according to the criteria of Pagan and Sossounov (2003). Moreover, the more parsimonious MSIH model is more appropriate in all cases except for the US 3-month Treasury bill rate. Panel B reports the Davies and RCM statistics. The Davies statistics are significant at 1% for all cases, indicating that state dependency exists in our model. Further, the RCM statistics for all cases are below 50, consistent with the existence of two regimes. Table 4 provides the estimation results of the two-state MSIAH model for the US 3-month Treasury bill rate and the two-state MSIH model for the other variables. For the Federal fund rate, we find that the mean yield is higher with higher volatility in regime 1 than that in regime 2. In contrast, we find that the mean returns are lower with higher volatility in regime 4
For the Markov-switching model, see Kim and Nelson (1999) and Bhar and Hamori (2004).
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
1 than that in regime 2 for the stock markets. The economic interpretation for these states is straightforward. Regime 1 represents the recession period for the economy when the low-risk assets such as bonds provide higher returns than the risky assets. For the US stock market, the mean return for both regimes provides positive returns while for the ASEAN stock markets, the case is contrary. For example, all the three stock markets have negative returns in regime 1 but positive returns in regime 2. Moreover, the volatility of the stock markets is lower in regime 2, which indicates the economies are experiencing stable growth. Table 4 also reports the estimated transition probabilities and their respective expected durations. Based on the results in Table 4, we conclude that the bull-market periods are much longer than the bear-market periods for ASEAN countries. Particularly, since the US 3-month Treasury bill rate is treated as the Federal fund rate, we can state that the monetary policy in the US experiences longer expansive periods than tight periods since the duration of regime 1 is longer. Meanwhile, the evidence from US stock returns tells us that excess liquidity plays an important role in pushing the asset price up. The reason behind this tendency can be inferred as follows. The state-dependent model works well in our specified model, suggesting that the growth rate of money supply is not constant. Especially, the high growth rate of money supply always brings excess liquidity, which pushes the asset price up easily. As demonstrated in our empirical results, the mean in regime 1 for the US stock market is lower and lasts longer than that in regime 2. Since regime 1 represents the crisis period for the US economy, the behavior of the Federal Reserve becomes clear, that is, it always provides liquidity to stimulate the economy. However, the capital always pursues a higher return and will flow to the outside markets, which in turn will lower the returns from the domestic market when the economy is not good. To demonstrate the transition of the market movement, we plot the transition probabilities for both regimes in Figure 2. Figure 2 shows the crisis periods (e.g., oil shock in 1991, Asian currency crisis in 1999, international financial crisis in 2008), indicated by the black line, are specified well in our model. Moreover, that the figure shows the international financial crisis has a smaller effect on the ASEAN countries than the Asian currency crisis did since the duration of regime 1 is shorter. Particularly, the longer duration of regime 2 in terms of the Federal fund rate indicates a bear market for the bond, which implies an excess liquidity in the US monetary system. In summary, our univariate results have a number of implications for understanding business cycles and market movements. First, we identify two states for each variable, in which case the movement of a stock market can be classified into either a bear or bull market. Second, our univariate results provide us the durations of specific regimes during the sample periods. Generally, the volatility is lower in the economic expansion regime, which is also more persistent for the ASEAN stock market, as indicated by the high positive mean value. In contrast, the US stock market experiences more crisis periods during our sample periods (which starts in 1990). Moreover, the monetary policy in the US also experiences a long series of expansive periods, especially after 2008, which makes the spillover effect from the excess liquidity an important issue. 4.2 Regimes for the joint distribution After demonstrating the regime switching of US monetary policy and the ASEAN stock
ACCEPTED MANUSCRIPT
SC
RI P
T
markets, we incorporate the US factors in our Markov-switching model to investigate the effect of the Federal fund rate and excess liquidity on the ASEAN stock markets. Following the same procedures, we first select the best-fitting MSIAH model based on the SBC values and evaluate it based on the Davies and RCM statistics. The results are given in Table 5. All the statistics suggest the two-state MSIAH model is the best-fitting model for all cases. Following the results of Pagan and Sossounov (2003), we believe the two-state model can provide sufficient analysis for our model. From the empirical results in Table 6, we can see that lower mean returns with higher volatility are observed in regime 1. The only exception is Singapore, for which the mean in regime 1 is negative. Same as in the univariate Markov-switching models, regime 1 represents the bear market, while regime 2 represents the bull market. More importantly, only the ) has a negative effect on the selected ASEAN stock markets Federal fund rate (
MA NU
during the economic expansion periods (regime 2). A relative decrease in the short-term US interest rate in the US has a positive impact on the ASEAN stock returns in the following month. However, this kind of effect disappears during economic recession periods (regime 1) because the authority may shut down the transmission channels during the crisis periods. Further, it may be difficult for the excess liquidity from the US to flow into the ASEAN countries. In addition, we find a positive effect from the US stock market in both regimes for are significant and positive. The degree of this kind all cases since all the parameters
ED
of linkage increases in regime 2 when the market is good. However, the lagged stock returns ( ) from the domestic markets are not well justified. For example, only the lagged
AC
CE
PT
stock return from Indonesia in regime 1 is positive and significant, while only the lagged stock returns from both Singapore and Thailand in regime 2 stay significant and positive. In sum, the spillover effect from US monetary policy plays an important role in determining the asset price in ASEAN stock markets. The transition probabilities for both regimes are plotted in Figure 3. Compared to the univariate system, the multivariate system provides us more robust results since the SBC value is lower than in the univariate system. In addition, the smoothed probability of both regimes in the multivariate system reflects the fluctuations of the stock market more reliably. It is clear from the figure that the duration for both regimes decreases as we input US factors into our models. Meanwhile, we can identify more specific periods of stock market comovements from the effect of US monetary policy. These periods tend to converge with the US monetary policy cycle. For example, a mini-crash such as that caused by the 9/11 terrorist attack may cause a market downturn that lasts for quite some time. In summary, our analysis identifies two states for our multivariate system. The crisis regime (regime 1) is characterized by relative higher volatility and lower stock returns. During these periods, we find that the spillover effect from US monetary policy on the selected ASEAN market does not matter. In contrast, the tranquil regime (regime 2) is characterized by relative lower volatility and higher stock returns. Furthermore, during these periods, there is strong evidence that the spillover effect from US monetary policy influences the selected ASEAN stock markets. Overall, even though the ASEAN countries experience almost the same regimes during our sample periods, the duration time for each period is different. Finally, we find that the tranquil period of the stock market in Singapore seems to be longer than the crisis period.
ACCEPTED MANUSCRIPT
5. Conclusion
AC
CE
PT
ED
MA NU
SC
RI P
T
In this paper, we investigate the linkage between the regime switching in US monetary policy and selected ASEAN stock markets. Specifically, we investigate the regime switching of the Federal fund rate (US 3-month Treasury bill rate), US stock market (S&P 500), and selected ASEAN stock markets based on univariate Markov-switching models. The two-state switching models are identified for all cases. Thus, we confirm the existence of two distinct regimes: periods of economic expansion and periods of economic decline. Based on the empirical results from the above analysis, we apply the multivariate Markov-switching models to investigate the effect of the Federal monetary policy for the different regimes on the ASEAN stock markets. We find that the Federal fund rate has a negative effect on the selected ASEAN stock markets during economic expansion periods. Moreover, we find that the ASEAN stock markets have a positive comovement with the US stock market in both regimes for all cases. However, the lagged stock returns on their own play only a small role in determining the future movement of stock markets in ASEAN countries. These results have important implications for the transmission mechanisms of asset price, especially from the US to small economies. For instance, the transmission mechanism differs by regime. The ASEAN stock markets are influenced by the Federal fund rate more easily in a bull market rather than in a bear market. Moreover, the spillover effect differs between the bull market and the bear market. Generally, the spillover effect has a greater impact in the bull market than in the bear market. Since capital control and exchange rate arrangement are always important issues for central banks in the developing world, our results show their stock markets are not easily influenced by the US dollar, especially during the times of economic crisis.
ACCEPTED MANUSCRIPT References
AC
CE
PT
ED
MA NU
SC
RI P
T
Agénor, P R, Aynaoui, K E (2010). Excess liquidity, bank pricing rules, and monetary policy. Journal of Banking and Finance, 34(5), 923-933. Ang, A, Bekaert, G (2002). Regime switches in interest rate. Journal of Business and Economics Statistics, 20(2), 163-182. Ang, A, Timmermann, A (2012). Regime changes and financial markets. Annual Review of Financial Economics, 4, 313-337. Bagliano, F C, Morana, C (2012). The Great Recession: US dynamics and spillovers to the world economy. Journal of Banking and Finance, 36(1), 1-13. Belke, A, Bordon, I G, Volz, U (2013). Effects of global liquidity on commodity and food prices. World Development, 44, 31-43. Belke, A, Orth, W, Setzer, R (2010). Liquidity and the dynamic pattern of asset price adjustment: A global view. Journal of Banking and Finance, 34(8), 1933-1945. Bhar, R and Hamori, S (2004). Hidden Markov Models : Applications to Financial Economics, Springer, Boston. Brana, S, Djigbenou, M L, Prat, S (2012). Global excess liquidity and asset prices in emerging countries: A PVAR approach. Emerging Markets Review, 13(3), 256-267. Chan, K F, Treepongkaruna, S, Brooks, R, Gray, S (2011). Asset market linkages: Evidence from financial, commodity and real estate assets. Journal of Banking and Finance, 35(6), 1415-1426. Davies, R B (1977). Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrika, 64(2), 247-254. Davies, R B (1987). Hypothesis testing when a nuisance parameter is present only under the alternative. Biometrika 74(1), 33-43. Fama, E F, Fisher, L, Jensen, M C, Roll, R (1969). The adjustment of stock prices to new information. International Economic Review, 10(1), 1-21. Guidolin, M., Timmermann, A. (2006). An econometric model of nonlinear dynamics in the joint distribution of stock and bond returns. Journal of Applied Econometrics, 21(1), 1-22. Kim, S J (2003). The spillover effects of US and Japanese public information news in advanced Asia-Pacific stock markets. Pacific-Basin Finance Journal, 11(5), 611-630. Kim, C J and Nelson, C R (1999). State-Space Models with Regime Switching: Classical and Gibbs-Sampling Approaches with Applications, The MIT Press, Cambridge, Mass. Kim, S J, Nguyen, D Q T (2009). The spillover effects of target interest rate news from the U.S. Fed and the European Central Bank on the Asia-Pacific stock markets. Journal of International Financial Markets, Institutions and Money, 19(3), 415-431. Krolzig, H-M (1997). Markov-Switching Vector Autoregressions: Modelling, Statistical Inference, and Application to Business Cycle Analysis. Lecture Notes in Economics and Mathematical Systems. Springer-Verlag, Berlin. Mitchell, M L, Mulherin, J H (1994). The impact of public information on the stock market. Journal of Finance, 49(3), 923-950. Simo-Kengne, B D, Balcilar, M, Gupta, R, Reid, M, Aye, G C (2013). Is the relationship between monetary policy and house prices asymmetric across bull and bear markets in
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA NU
SC
RI P
T
South Africa? Evidence from a Markov-switching vector autoregressive model. Economic Modeling, 32, 161-171. Pagan, A R, Sossounov, K A (2003). A simple framework for analyzing bull and bear markets. Journal of Applied Econometrics, 18 (1), 23-46.
ACCEPTED MANUSCRIPT Table 1 Summary statistics. Indonesia
Singapore
Thailand
0.0335 0.0375 0.0222 −0.0870 1.8975 14.32***
0.0051 0.0101 0.0438 −0.7673 4.5180 53.58***
0.0075 0.0060 0.1018 0.2464 5.9505 102.9***
0.0026 0.0040 0.0624 0.3195 4.8113 42.42***
0.0022 0.0048 0.1010 0.5193 5.3838 77.76***
T
US
RI P
Mean Median Std. Dev. Skewness Kurtosis JB test
FED
MA NU
SC
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns. *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively. Table 2 Correlations of monthly returns for each country.
AC
CE
PT
ED
Indonesia Thailand Singapore FED US Stock FED US Stock FED US Stock FED 1 1 1 US 0.0540 1 0.0540 1 0.0540 1 Stock −0.0703 0.0719 1 −0.1214 0.0417 1 −0.0298 0.2015 1 Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns.
ACCEPTED MANUSCRIPT
Singapore
Thailand
−3.502 −3.476 −3.434 −3.369 −3.291 −3.241
−1.831 −1.795 −1.760 −1.718 −1.572 −1.652
−2.763 −2.748 −2.684 −2.651 −2.539 −2.519
−1.808 −1.773 −1.752 −1.710 −1.577 −1.612
0.000 14.067
0.000 23.765
0.000 32.883
0.000 21.285
T
Indonesia
SC
Panel A: SBC Values 2-State MSIH −6.114 2-State MSIAH −9.807 3-State MSIH −7.048 3-State MSIAH −9.774 4-State MSIH −6.383 4-State MSIAH −9.791 Panel B: Davies and RCM Statistics Davies Test 0.000 RCM Statistic 14.408
US
MA NU
FED
RI P
Table 3 Performance measures for univariate Markov switching models.
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. We present SBC values to select the best-fitting Markov-switching models. The lowest SBC values (bold face numbers) indicate the best-fitting Markov-switching models. Davies test statistics are presented as p values. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns.
FED MSIAH 1.695 (5.426) −0.450 (0.462) 0.987*** (0.012) 0.997*** (0.03) 0.075*** (0.008) 0.001*** (0.000) 0.935 0.879 15.43 8.276
US
MSIH −206.2 (186.9) 174.5*** (58.41)
29.37*** (3.335) 5.017*** (0.834) 0.982 0.973 54.55 37.27
233.1 *** (50.27) 46.19*** (6.293) 0.951 0.977 20.33 44.16
Singapore MSIH −3.612 (109.3) 41.97 (36.38)
Thailand MSIH −211.3 (174.7) 111.9* (60.05)
AC
CE
Indonesia
MSIH 29.02 (43.77) 92.24*** (22.91)
PT
Model
ED
Table 4 Parameter estimates for univariate two-state Markov-switching models.
Duration 1 Duration 2
89.05*** (20.41) 17.77*** (3.106) 0.913 0.965 11.49 28.76
220.3*** (42.22) 49.99*** (7.044) 0.963 0.982 27.23 55.82
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. refers to the transition probability in regime 1(2), and the expected duration in regime 1(2) is calculated as . *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively.
ACCEPTED MANUSCRIPT Table 5 Performance measures for multivariate Markov-switching models.
−3.025 −2.922 −2.792
−1.808 −1.775 −1.731
0.000 31.89
T
Thailand
RI P
Panel A: SBC Values 2-State MSIAH −1.962 3-State MSIAH −1.865 4-State MSIAH −1.597 Panel B: Davies and RCM Statistics Davies Test 0.000 RCM Statistic 15.07
Singapore
SC
Indonesia
0.000 11.75
AC
CE
PT
ED
MA NU
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P 500 (US), and MSCI stock index for each country. We present SBC values to select the best-fitting Markov-switching models. The lowest SBC values (bold face numbers) indicate the best-fitting Markov-switching models. Davies test statistics are presented as p values. The sample period is from January 1990 to December 2012 for a total of 276 monthly returns.
ACCEPTED MANUSCRIPT Table 6 Parameter estimates for multivariate MSIAH models. Indonesia
PT CE
Duration 1 Duration 2
Thailand
MA NU
152.3 (207.9) 239.9*** (92.31) −0.089 (0.094) −0.336 (0.522) 0.949*** (0.219) 0.132* (0.073) −1.138*** (0.265) 0.934*** (0.174) 125.4*** (17.68) 46.05*** (6.856) 0.775 0.828 4.459 5.838
RI P
SC
−11.73 (10.43) 10.08 (6.088) −0.065 (0.076) 0.152 (0.246) 0.773*** (0.097) 0.191*** (0.061) −0.385** (0.123) 0.963*** (0.099) 24.41*** (3.174) 23.02*** (3.110) 0.899 0.956 9.952 22.92
T
*
ED
144.2 (186.9) 236.8 (78.35) 0.281*** (0.064) −0.696 (0.477) 0.712*** (0.201) −0.067 (0.057) −0.624*** (0.195) 1.234*** (0.169) 91.54*** (9.992) 70.96*** (6.356) 0.787 0.756 4.694 4.089
Singapore
AC
Notes: 3-month Treasury bill yield for Federal fund rate (FED), S&P GSCI for gold (gold), and MSCI stock index for each country. refers to the transition probability in regime 1(2), and . *, **, and *** represent the expected duration in regime 1(2) is calculated as significance at the 10%, 5%, and 1% levels, respectively.
MA NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig. 1. Time-series evolution of stock price index and Federal fund rate (FED) (3-month Treasury bill rate), respectively. The right-side of the y axis for the FED is measured in percentage. The sample period is from January 1990 to December 2008.
ACCEPTED MANUSCRIPT
(a) FED Rate- MSIAH model
1.00
T
0.75
RI P
0.50
0.25
0.00
SC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
(b) US Stock Market-MSIH model
1.00
MA NU
0.75
0.50
0.25
0.00
ED
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
(c) Singaporean Stock Market- MSIH model
1.00
PT
0.75
0.25
0.00
CE
0.50
1.00
0.75
AC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
(d) Indonesian Stock Market - MSIH model
0.50
0.25
0.00 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
ACCEPTED MANUSCRIPT (e) Tailandese Stock Market - MSIH model
1.00
0.75
T
0.50
RI P
0.25
0.00
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
AC
CE
PT
ED
MA NU
SC
Fig. 2. Smoothed probability of regime 1 (black line) and regime 2 (blue line) for univariate Markov-switching models. The sample period is from January 1990 to December 2008.
ACCEPTED MANUSCRIPT
(a) Indonesia
1.00
T
0.75
RI P
0.50
0.25
0.00
(b) Singapore
1.00
MA NU
0.75
SC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
0.50
0.25
0.00
ED
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
1.00
PT
0.75
0.00
CE
0.50
0.25
(c) Thailand
AC
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Fig. 3 Smoothed probability of regime 1 (black line) and regime 2 (blue line) for multivariate MSIAH models. The sample period is from January 1990 to December 2008.
ACCEPTED MANUSCRIPT Highlight
T
We investigate the spillover effect from US monetary policy to selected ASEAN stock markets.
RI P
We confirm the existence of two distinct regimes for both US monetary policy and stock markets.
SC
We find that US interest rates have a negative effect on the selected ASEAN stock markets during the economic expansion periods.
AC
CE
PT
ED
MA NU
We find a negative effect of excess liquidity during economic recession periods on the Singaporean stock market.