. International Journal of Economics and Financial Issues ISSN: 2146-4138 available at http: www.econjournals.com International Journal of Economics and Financial Issues, 2017, 7(2), 437-447. International Journal of Economics and Financial Issues | Vol 7 • Issue 2 • 2017 437 Exchange Rate Movements, Stock Prices and Volatility in the Caribbean and Latin America Andre Yone Haughton1, Emma M. Iglesias2*†1 1Department of Economics, University of the West Indies Mona, Kingston 7, Jamaica, 2Department of Applied Economics, II. Facultad de Economía y Empresa, University of A Coruña, Campus de Elviña, A Coruña, 15071, Spain. *Email: emma.iglesias@udc.es ABSTRACT We analyze the interrelationship between stock prices and exchange rates in the only two Caribbean countries with stock market and floating exchange rates: Jamaica and Trinidad and Tobago. We also study the same four Latin American countries as in Diamandis and Drakos (2011). Using their model, our results show a very mild relationship between both variables in Jamaica, Trinidad and Tobago, Argentina and Brazil, but we cannot find any relationship in the other countries as in Diamandis and Drakos (2011). However, when we extend their model including a generalized autocorrelation conditional heteroskedasticity (component to examine the impact of volatility, our results changed drastically: Stock prices significantly impacted the exchange rate in the tranquil sub-period and the full period for Jamaica, over all three periods for Trinidad and Tobago and in the tranquil period for Argentina, Mexico and Chile. This shows the importance of incorporating volatility explicitly in the model. Our results have the policy implications that governments in the previous countries should try to prevent a currency crisis by stimulating economic growth and the expansion of the stock market to attract capital inflow as in Lin (2012). Keywords: Exchange Rates, Stock Prices, Volatility JEL Classifications: F31, G01 † Authors are very grateful for the useful comments received at the World Finance Conference in 2016. The second author is very grateful for the financial support from the Spanish Ministry of Science and Innovation, project ECO2015-63845-P and from Xunta de Galicia, project ED431C2016/007. 1. INTRODUCTION Research on the interaction between exchange rate and stock prices has received more attention since the recent global financial crisis. Researchers in the Unites States and Asia mainly, have been trying to estimate the direction of causality between both variables. According to Lin (2012), there are two main theories surrounding their interaction: (1) The one proposed by Dornbusch and Fisher (1980) with the “flow oriented” models of exchange rates, which looks specifically at the balance of trade between countries. Theoretically, exchange rate fluctuations influence the output and hence competiveness of firms. If firms are more competitive this has a direct positive effect on its stock prices, since stock prices represent future cash flow streaming for a company. (2) The second approach, the “stock oriented” model of exchange rate proposed by Frankel (1983) and Branson (1993), states that advances in the stock market affect exchange rate through the liquidity and the wealth effects. A decrease in stock prices reduces the wealth of local investors, which lowers their demand for money. Then banks react by lowering interest rates which dampens capital inflows, reducing the demand for local currency and therefore depreciates the local currency. Since domestic and foreign assets are not perfect substitutes in the portfolio balancing effect, as investors adjust their portfolio ratio of domestic to foreign assets in response to changes in economic conditions, the exchange rate responds accordingly. Evidence of either theory is not uniform across countries as various studies employing a range of different techniques revealed varying results. Most of these studies have been focused on North America, Europe, Asia, and Latin America to a lesser extent but none have investigated this issue in the Caribbean, from the best of our knowledge. Recently, Hassanain (2017) has analyzed the case of the Gulf Cooperation Council. Our objective is to analyze the interrelationship between the stock market and the exchange rates in the two Caribbean countries that Haughton and Iglesias: Exchange Rate Movements, Stock Prices and Volatility in the Caribbean and Latin America International Journal of Economics and Financial Issues | Vol 7 • Issue 2 • 2017438 have floating exchange rates and stock markets: Jamaica and Trinidad and Tobago. We also study the same four Latin American countries as in Diamandis and Drakos (2011): Argentina, Brazil, Chile and Mexico. We follow Lin (2012) by analyzing the relationships between the exchange rate market and the equity market during the tranquil (2002-2008) and during crisis (2008-2012) periods and we also use the autoregressive distributed lag (ARDL) model bounds test approach proposed by Pesaran et al. (2001). Diamandis and Drakos (2011) claim that the type of exchange rate regime being operated in the particular country will influence the long run relationship between both variables. All the countries in this study operate a float or managed float exchange rate regime. The relationship between exchange rate and stock prices have a tendency to be greater during crisis periods as returns in asset markets are lower and volatility are higher (Lin, 2012; Guo et al., 2011). Therefore, we extend Diamandis and Drakos (2011) and Lin (2012) studies to incorporate a generalized autocorrelation conditional heteroskedasticity (GARCH)(1,1) component as in Bollerslev (1986) in the ARDL framework to take account of the impact of risk in the model. Note that the importance of volatility is made clear in many contexts nowadays also in the Caribbean countries (see for example Mapp and Wiston (2015) who analyze the impact of the informal economy on the volatility in the Caribbean countries). Our objective in this paper is to show that volatility must be modeled explicitly also when analyzing the relationship between stock prices and exchange rates. Gordon and Pettiford (2016) have also demonstrated how accounting for ARCH effects can be a crucial factor to establish relationships among macroeconomic variables. Like in Asia, North America and Europe, the Global Financial Crisis in 2008 resulted in an immediate decline in the stock prices (sp in Figure 1) in the Caribbean and Latin American Countries (Figure 1). Notice the behavior is similar for the countries in each group, and also stock prices in Latin America tend to be higher (corresponding to larger economies) than the Caribbean. Exchange rates in our Caribbean and Latin America countries depreciated as a result of the financial crisis (Figure 2). The shift Figure 1: Stock prices in Jamaica, Trinidad and Tobago, Argentina, Brazil, Chile and Mexico for Jamaica in particular is more obvious given that exchange rates are higher and appear to be less stable than the other countries (Figure 3). We therefore expect a stronger relationship between the exchange rate market and the stock market in Jamaica. Even though Caribbean and Latin American countries have floating exchange rates, their central banks can intervene the market to curtail rapid depreciations. In the event of a sudden decline in the value of the domestic currency, the central bank might increase interest rates to attract foreign investors thereby increasing the supply of foreign currency or sell reserves to maintain currency stability. If interest rates are already high (such is the case in Jamaica), further increases will reduce economic activity thereby reducing productivity of firms and therefore the value of their stock. Jamaica level of reserves has not been enough to keep the currency stable. To account for this in our analysis, we follow Lin (2012) by including net international reserves (NIR) and interest rate variables in our model. This will improve our results and correct for omitted variable biases. The following section provides a brief overview of the current literature. 2. LITERATURE REVIEW Research on the interrelationship between exchange rates and stock prices has been carried out for a variety of countries using various techniques which have produced varying results. In Latin America, Diamandis and Drakos (2011) analyze the long run relationship and short run dynamics between exchange rates and stock prices as well as the impact of exogenous shocks on four countries: Argentina, Brazil, Chile, and Mexico using cointegration techniques and Granger causality tests. They found no significant long run relationship between stock prices and exchange rates for each country. However, after incorporating the US stock market, their results show that stock prices and exchange rates are positively related, with the US stock market facilitating the transmission between the two in these countries. The interaction is independent of the sample choice but Hansen and Johansen (1993) instability tests show that some of their cointegrating coefficients are stable overtime. Empirically, the research is largely concentrated on more developed countries. Early research by Neih and Lee (2001) examine the dynamic relationship between stock prices and exchange rates for the G7 countries using basic cointegration tests and vector error correction models (VECM) from 1993 to 1996. This research did not account for dual causality between the variables and their findings suggest that there is no long run relationship between stock prices and exchange rate in the G7 counties. Muller and Verschoor (2006) examine how multinational firms in the US are affected by exchange rate fluctuations. They believe that currency movements are a major source of macroeconomic instability which affects a firm’s value; a situation they refer to as exchange rate exposure. Theoretically, they outline several reasons why the exchange rate/stock price interaction might be asymmetric. These include the asymmetric impact of hedging on cash flow, firms pricing to market strategies, asymmetry due to hysteric behavior, investors over reaction and mispricing errors and nonlinear currency risk exposure.1 1 For more in-depth analysis see Muller and Verschool (2006). Haughton and Iglesias: Exchange Rate Movements, Stock Prices and Volatility in the Caribbean and Latin America International Journal of Economics and Financial Issues | Vol 7 • Issue 2 • 2017 439 Also on the US, Vygodina (2006) uses a Granger causality test to investigate the relationship between stock prices and exchange rates controlling for the size of the firm from 1987 to 2005. The results found causality from large stock prices to US exchange rate but no causality from small stock prices. The results from the subsamples show that there might be evidence to support the claim that causality between the two variables is changing overtime. Some have also explored the issue in Asia. Pan et al (2007) used Granger causality tests and vector autoregressions to examine the dynamic linkages between exchange rate and stock prices in seven Asian countries: Hong Kong, Japan, Korea, Malaysia, Singapore, Taiwan and Thailand from 1988 to 1998. Their results showed significant causal relationships for Hong Kong, Japan, Malaysia and Thailand before the financial crisis. They also found evidence of causal relationships between the equity market to the foreign exchange market for Hong, Korea and Singapore. They also found causality from exchange rate to the stock market for all countries except Malaysia, while there is no causality from stock prices to exchange rates. They claim their results are robust to a variety a testing procedures including causality tests and variance decomposition. Yau and Nieh (2006) empirically investigate the new Taiwan dollar exchange rate against the Japanese yen on stock prices in Japan from 1991 to 2005. Granger causality test showed that no short run causal relationships existed between the variables for both countries. Also, the findings suggest there is no relationship between the exchange rate and the respective stock prices in the long run. Yau and Neih (2009) also examined the relationship between Japanese exchange rate and Taiwan stock market using a threshold error correction model proposed by Enders and Siklos (2001). Their findings suggest that there is a long run equilibrium relationship between the Taiwan dollar, the Japanese Yen and the stock prices of Japan and Taiwan, but asymmetry only exist for Taiwan as the effects of the Japanese exchange rate is symmetric. Zhao (2010) used monthly data from 1991 to 2009 to examine the dynamic effects between exchange rate and stock prices in China by employing a vector autoregressive approach (VAR) and a multivariate GARCH. Their results show that there is no definite long run relationship between the Chinese Renminbi real effective exchange rate and stock prices in China. They also found no spill over effects between the two variables. The paper goes a step further to examine the cross volatility effects between stock market and the exchange rate using likelihood ratio tests. The results show that there is volatility spill over effects from stock prices to exchange rate and from the exchange rate to stock prices. More recently, Tsai (2012) uses quantile regression to investigate the relationship between stock price index and exchange rate in six Asian countries: Singapore Thailand, Malaysia, Philippines, South Korea and Taiwan. The results supported a priori information that the two variables are negatively related. More specifically, the negative relationship observed is more pronounced when exchange rates are extremely low or extremely high. This result is supported by the portfolio balancing effect in these two markets which outlines that an increase (decrease) in the returns on stock price index will result in an appreciation (depreciation) of the domestic currency via a decrease (increase) in the exchange rate. Tsai (2012) findings also suggest that the relationship is not homogeneous across countries and across market situations and the coefficients may vary since the portfolio balancing effect is not present all the time in every market. They explain that a significant impact of stock prices on exchange rates exist in time where large sums of capital enter or exit the market. Our research in this paper follows the method of Lin (2012) who examines the relationship between the exchange rate and stock prices in Asia’s emerging markets from 1986 to 2010. Using monthly data, the ARDL model proposed by Pesaran et al. (2001) was employed. This method is designed to account for structural breaks and data that are integrated of different orders. The results from the cointegration tests as well as the short run causality tests indicate that the co-movement between exchange rate and stock prices increases during times of economic crisis and it reduces when the economies are stable. These results correspond with the general literature on exchange rate spill over effects on the stock market. The results also show that most spill overs are in the channel from stock price shocks Figure 3: Exchange rates and stock prices in Jamaica Figure 2: Exchange rates Trinidad and Tobago, Argentina, Brazil and Mexico Haughton and Iglesias: Exchange Rate Movements, Stock Prices and Volatility in the Caribbean and Latin America International Journal of Economics and Financial Issues | Vol 7 • Issue 2 • 2017440 to exchange rates. In theory, economic slowdown reduces the value of companies stocks causing investors to withdraw their capital which reduces the demand for the domestic currency and out downward pressure on the exchange rate. Apart from the findings for aggregated data, Lin (2012) also examined the issue using industry level data, and the results indicated that the co movement is weak for export oriented industries such as IT for example. All in all, the findings suggest that the interrelationship between exchange rate and stock market is driven by changes in the capital account rather than changes to trade balance in these Asian countries. Our research estimates the interrelationship between stock prices and exchange rates using the same approach as in Diamandis and Drakos (2011) and Lin (2012) by employing the bounds cointegration tests in the ARDL model, since it provides meaningful long run results even if all the variables are not integrated of the same order. During crisis period, market returns are lower and volatility is higher as the correlation between assets prices tends to be greater see (see e.g., Climent and Meneu (2003) and Guo et al., 2011). This justifies our extension incorporating a GARCH(1,1) component in our ARDL framework as volatility (risk) is also a determining factor in the relationship between prices and exchange rates. The remainder of the paper is organized as follows: Section 3 outlines the data, data sources, the methodology employed and the results. Section 4 concludes. Appendix A contains some of the Tables A1-A6 and Figures A1 and A2. 3. DATA AND METHODOLOGY 3.1. Data We examine the interrelationship between the stock index and the exchange rate in the Caribbean and Latin America using monthly data from 2002 to 2012. Data on the share price index (sp), the exchange rate relative to the US dollar (fx), the money market rate (mm) and foreign reserves minus gold (r) are collected from the international monetary fund (IMF) international financial statistics. All data are transformed into logarithms. The data begins in January 2002 which is approximately the same time the asset bubble began to develop in the international asset markets. In this way we also minimize any effects of the Jamaican and the Mexican financial crisis of the 90’s. On the other hand, the reason to finish our sample size in 2012 is in order to remove the economic effects of the big increase in debt of the economy in Jamaica that happened in that year. In early 2010, the Jamaican Government asked the Jamaica debt exchange to retire high-priced domestic bonds and reduce annual debt servicing. However, debt continued to be a serious concern, forcing the government to negotiate and sign a new IMF agreement in May 2013 to gain access to approximately $1 billion additional funds. As a precursor, the government instigated a second National debt exchange in 2012. First we analyze the data across the full time period and later we split the data into two parts: (1) The first sub-sample from 2002:01 to 2008:08 (the so called tranquil period) where the asset bubble was developing. (2) The second sub-sample is taken from 2008:09 to 2012:02 (the crisis period). This will provide useful comparisons of the interrelationship between the variables before and after the announcement of the recent global financial crisis. The summary statistics are provided in Table 1 for the full sample 2002:01-2012:02, as well as each sub-sample periods. Figures A1 and A2 in Appendix A show the evolution of exchange rates in Latin America and the Caribbean. 3.2. Methodology and Results 3.2.1. Unit root tests To test for a long run relationship between the stock prices and exchange, the order of integration of each variable must first be examined. Through careful examination of the movements of the data for the countries of the Caribbean and Latin America overtime, it is visible that structural breaks may exists at different points in time (Figures 1-3 in Section 1). If such structural breaks Table 1: Summary statistics Variables Mean±SD Tranquil period (2002/01 to 2008/08) Crisis period (2008/09 to 2012:02) Full period (2002/01 to 2012/02) Stock prices Jamaica 0.151±0.048 −0.004±0.049 0.008±0.049 Trinidad and Tobago 0.012±0.031 −00.1±0.031 0.007±0.032 Argentina 0.019±0.071 0.013±0.093 0.016±0.079 Brazil 0.019±0.072 0.007±0.074 0.136±0.073 Chile −0.126±0.171 −0.134±0.123 −0.134±0.123 Mexico 0.017±0.049 0.010±0.063 0.014±0.054 Exchange rate Jamaica 0.005±0.009 0.004±0.015 0.005±0.012 Trinidad and Tobago −0.000±0.003 0.000±0.003 0.000±0.003 Argentina 0.009±0.066 0.008±0.014 0.009±0.054 Brazil −0.005±0.055 −0.003±0.047 −0.002±0.542 Chile −0.063±0.081 −0.069±0.185 −0.069±0.185 Mexico 0.001±0.020 0.004±0.042 0.003±0.030 Interest rate Jamaica −0.007±0.242 −0.056±2.370 −0.007±0.222 Trinidad and Tobago 0.012±0.171 −0.169±0.409 −0.049±0.287 Argentina −0.384±8.614 0.006±0.921 −0.244±6.928 Brazil −0.078±0.678 −0.073±0.375 −0.071±0.590 Chile −0.989±0.489 −2.590±0.579 −2.590±0.680 Mexico 0.007±0.549 −0.094±0.234 −0.026±0.465 Foreign reserves Jamaica 0.003±0.059 −0.003±0.075 0.000±0.064 Trinidad and Tobago 0.019±0.047 0.004±0.025 0.014±0.041 Argentina 0.015±0.062 0.001±0.019 0.009±0.054 Brazil 0.022±0.059 0.013±0.019 0.189±0.049 Chile −0.166±0.131 −0.069±0.101 −0.165±0.131 Mexico 0.009±0.022 0.010±0.037 0.009±0.027 All data is analyzed in logarithms. SD: Standard deviation Haughton and Iglesias: Exchange Rate Movements, Stock Prices and Volatility in the Caribbean and Latin America International Journal of Economics and Financial Issues | Vol 7 • Issue 2 • 2017 441 are not accounted for, it increases the likelihood of failing to reject the null of a unit root (e.g., Lin, 2012). Therefore, unit root tests are sensitive to the alternative to a trend break hypothesis. Consequently, along with the usual Augmented Dickey Fuller (1979) (ADF) unit root test, we also employ the Zivot and Andrews (1992) and the Clemente et al. (1998) unit root tests which both account for the presence of structural breaks in the data. The Zivot and Andrews (1992) allows for a structural break in the data set which may occur in the intercept or trend or both. The break is determined endogenously as the test supports various criteria for detection. The Zivot and Andrews (1992) can be specified in one of three general forms: Model A accounts for break in intercept only; Model B accounts for break in trend only and Model C accounts for break in both intercept and trend. Consider the following: Suppose the structural shift occurs at a period 1T(λ) and 0 otherwise; allows for break in intercept and, DTt=t−tλ if t>tλ, and 0 otherwise; allows for break in trend. ∆ is the difference operator and the error term εt~IID(0,σ 2) is assumed to be identically and independently distributed with mean zero and constant variance. The null hypothesis of the Zivot and Andrews (1992) test is that yt has unit root I(1) with no exogenous structural break against the alternative that series is a stationary I(0) with structural break at some unknown point in time. The minimum t-value for the Zivot and Andrews test is selected for the endogenously determined break point 1