THE RELATIONSHIP BETWEEN EXCHANGE RATE REGIMES AND CAPITAL INFLOWS IN DEVELOPING COUNTRIES

Transcription

1 THE RELATIONSHIP BETWEEN EXCHANGE RATE REGIMES AND CAPITAL INFLOWS IN DEVELOPING COUNTRIES A Thesis submitted to the Faculty of the Graduate School of Arts and Sciences of Georgetown University in partial fulfillment of the requirements for the degree of Master of Public Policy in Public Policy By Wei Kuang, B.S. Washington, DC April 12, 2016

2 Copyright 2016 by Wei Kuang All Rights Reserved ii

3 I would like to express my sincere gratitude to Adam Thomas for his invaluable advice during this thesis project and to Jeffery Mayer for his time and edits. Many thanks, Wei Kuang iii

4 THE RELATIONSHIP BETWEEN EXCHANGE RATE REGIMES AND CAPITAL INFLOWS IN DEVELOPING COUNTRIES Wei Kuang, B.S. Thesis Advisor: Adam Thomas, Ph.D. ABSTRACT The surge of capital inflows from advanced economies to emerging markets has triggered ongoing discussions of the determinants of this phenomenon. One major focus of academic interest is a country s exchange rate regime. This paper investigates the relationship between exchange rate regimes and capital flows to 41 developing countries from 2000 to I hypothesize that, all else equal, a fixed exchange rate is associated with moderately larger volumes of capital inflows, compared with a flexible exchange rate. However, using annual country-level data and a fixed effects model, I find no statistically significant relationship between exchange rate regimes and capital inflows in developing countries. Possible reasons for this result include omitted variable bias and insufficient data, and I suggest that future research focuses on these limitations. iv

5 TABLE OF CONTENTS Introduction... 1 Background... 2 Literature Review... 5 Conceptual Framework...8 Data and Methods Descriptive Statistics...15 Regression Results...16 Discussion Appendix References v

6 INTRODUCTION International capital flows are the movement of money between countries for trade, investment and business operations purposes. They most often take the form of foreign direct investment (FDI), portfolio investment, and financial derivatives. 1 Capital inflows are often considered to be a driver of a country s economic growth (Ghosh et al., 2014). For developing countries in particular, foreign capital inflows can help finance investment and stimulate stock markets, which promote higher incomes and domestic consumption (Boudias, 2014). Since the start of the 1990s, capital flows from developed economies to emerging market countries, especially in Latin America and Asia, have grown rapidly, and during two periods in particular. The first period started in the early 1990s and ended in the wake of the 1997 Asian financial crisis; the second period began in the early 2000s and lasted until the great recession in This international funding from the private sector has contributed significantly to the economic growth of developing countries (Balakrishnan et. al., 2013). The surge of capital inflows to developing countries has generated debate about causes and effects. While large capital inflows are generally believed to stimulate economic growth, volatile and high-volume capital outflows can create economic instability (Boudias, 2014). As a result, for the past two decades, most recipient countries have adjusted their exchange rate policies to smooth shifting patterns of international capital inflows (Ahmed and Zlate, 2014). 1 A list of capital inflows measures used for this study and their definitions is included in Table 1 on page 13. 1

7 Using annual country-level panel data, this paper aims to explore the relationship between exchange rate regime flexibility and capital inflows in 41 developing countries from 2000 to The structure of the paper is as follows: in the next section, I provide a brief history of exchange rate regimes; I then synthesize the relevant literature, discuss my conceptual framework, hypothesis, data, and multiple regressions. Lastly, I present my research findings and discuss the policy implications of these findings. BACKGROUND A fixed exchange rate policy is a regime in which a currency s value is pegged to the value of another currency or commodity, while a flexible exchange rate policy is a regime in which a currency s value is allowed to vary in response to changes in foreign exchange markets. For instance, in a flexible exchange rate regime, if demand for the dollar goes up while the supply of dollars remains unchanged, holding all else constant, the market value of the dollar will also go up. On the other hand, if the exchange rate is fixed, under the same scenario, the value of the dollar remains unchanged. The gold standard was an earlier version of modern-day fixed exchange rate regimes in which a currency s value is fixed to the value of gold. (Stephey, 2008) At the beginning of the twentieth century, most advanced economies pegged their currencies to the value of gold. However, this practice ended with the Great Depression when many analysts concluded that the gold standard had been a cause of the crisis. Soon after World War II, the victors established a system of generally fixed, but occasionally flexible, exchange rates known as the Bretton Woods system, applying to 44 participating 2

8 countries (Chakravarty, 2011). The Bretton Woods system, for the first time, created an international standard to exchange one currency to another (Stephey, 2008). Under the Bretton Woods system, countries were generally expected to maintain fixed exchange rates but were allowed to adopt more flexible rates in extraordinary times, such as an economic recession. In an effort to facilitate international trade, 44 member countries agreed to peg their currencies to the US dollar. The United States reassured their allies by linking the US dollar to gold, with the value of a dollar set equal to 35 ounces of bullion (Stephey, 2008). However, this system collapsed in 1973, when the United States terminated convertibility of the US dollar to gold. Since then, countries have been free to choose their own exchange rate policies (Chakravarty, 2011). Some countries have adopted flexible exchange rates in order to achieve monetary independence, when others have sacrificed monetary independence by choosing a fixed exchange rate (Stephey, 2008). 2 Given the enormous capital flows to emerging markets in recent decades, economists have attempted to identify optimal exchange rate policies for emerging countries. Supporters of fixed exchange rate regimes have argued that this system provides stability and transparency because the value of the currency is generally easy to predict (Lane, 1995). A particularly attractive feature of pegged regimes is that, in principle, by reducing speculation and devaluation risk (the exchange rate is fixed in a pegged economy), a country can make investors more likely to invest in its market 2 According to the impossible trinity theory in macroeconomics, a country cannot have a fixed exchange rate, an open capital account, and an independent monetary policy at the same time (Ghosh, 2014). This implies that, once a country s capital account is open under a fixed exchange rate regime, it loses monetary policy autonomy. A country is considered to have an open capital account if it allows for the free flow of capital into and out of a country. Many developing countries tightly control their capital accounts. Monetary policy is the macroeconomic policy of a country s central bank. It involves management of the money supply and interest rates and are economic policy tools that central banks use to achieve macroeconomic control over consumption, inflation, and liquidity. 3

9 compared with markets that operate under flexible rate regimes. Thus, capital inflows tend to increase when countries adopt pegged regimes (Aizemen, 1992). However, a pegged regime can also lead to unstable capital outflows, and a peg is dangerous when the currency that is the basis for the peg (e.g., the US dollar) appreciates drastically, because the pegged economy loses monetary policy autonomy with the free flow of capital (Ohno and Shimuzu, 2015). In a pegged economy, the solution to this dilemma is to maintain a system of strong capital controls i.e. strong controls on the amount of money going into and out of a country (e.g. an annual overseas cash transfer limit of USD 100,000 per person in China)-and thus maintain an independent monetary policy. Nevertheless, countries adopt this approach at a cost, namely reduced capital inflows due to a tightly controlled capital account (Chakravarty, 2011). Economists favoring flexible exchange rates in developing countries claim that developing countries should adopt a more liberalized, market-based policy in order to attract consistent capital inflows (Mundell, 1963). Those economists argue that countries enjoy monetary independence under a floating rate, because they can buy and sell their currency to adjust its value. However, export-oriented economies will generally benefit more from a fixed exchange rate regime, since the rate of return is predictable and without concerns about currency fluctuation (Obstfeld and Rogoff, 1995). In sum, then, the question as to the optimal exchange rate regime for emerging markets is still unresolved. This paper contributes to this discussion by analyzing the relationship between developing countries exchange rate regimes and their capital inflows. 4

10 LITERATURE REVIEW This section reviews economic theory regarding the effects of various exchange rate policies and examines the existing empirical findings on the relationship between exchange rate regimes and capital inflows in developing countries. Economic Theory Most theoretical papers predict that emerging markets will attract more capital inflows if they adopt fixed exchange rates. To study the relationship between exchange rate regimes and FDI, Aizenman (1992) develops a macroeconomic model with an assumption that investors are risk neutral. He concludes that a fixed exchange rate regime is more attractive to FDI, compared with a flexible exchange rate regime. More specifically, Aizenman theorizes that, holding constant the degree of volatility in terms of economic shocks, a fixed exchange rate regime is associated with higher returns to investment, relative to a flexible exchange rate regime. 3 The logic behind his conclusion is that under the assumption of free entry into global markets, the behavior of investors depends on expected profits. Thus, because fixed exchange rate regimes are predicted to produce less uncertainty than more liberalized regimes, this policy approach is predicted to be more likely to attract foreign investors. Magud et al. (2011) and Mundell (1963) highlight two additional reasons why fixed exchange rate regimes may attract more capital inflows. First, fixed exchange rates can lower transaction costs, since one currency s value is fixed against another currency at all times, thus encouraging cross-border investment. Second, investors can take advantage of interest rate differentials (IRD) at minimum risk in a fixed exchange rate 3 In economics, a shock is an unexpected event that affects an economy. An example of a positive shock is technological advancement, and an example of negative shock is a financial crisis. 5

11 market. For instance, consider that an investor borrows USD10,000 and converts these funds into Hong Kong dollars, allowing the investor to purchase a Hong Kong bond. If the purchased bond yields 8% while the equivalent U.S. bond yields 4%, then the IRD equals 4%, the difference between the two. The IRD is thus the degree to which such an investor can expect to profit if we assume a fixed rate of exchange between the US dollar and the Hong Kong dollar. Empirical Studies While the conceptual logic in favor of fixed exchange rates seems compelling, the empirical evidence is less so. Magud et al. (2011) focus on periods of large capital inflows, and use three different panel datasets to study the relationship between exchange rate regimes and capital inflows. The three panels include seven Latin American countries from 1993 to 2002; five Asian countries from 1990 to 1997; and 13 European countries from 1999 to For all three panels, and using Ordinary Least Squares (OLS), fixed effects, and Generalized Least Squares (GLS) specifications, the authors find that exchange rate flexibility does not affect capital inflows to emerging economies. Moreover, using two different panels for 12 countries from 1980 to 2010 and 22 countries from 1994 to 2008, Boudias (2014) estimates a similar set of models and finds that exchange rate regimes do not impact capital inflows. Thus, neither study supports the theory that developing countries with fixed exchange rates regimes attract more capital inflows. In addition, Ghosh et al. (2014) use annual data for 56 emerging market countries from 1980 to 2011 and find that, holding other factors constant, during surge periods, countries with fixed exchange rates experience capital inflows that are 3% of GDP larger 6

12 than countries with more flexible exchange rate regimes. 4 Ghosh et al. (2014) use different controls than Boudias (2014) and Magud et al. (2011), and a larger sample for a longer period of study. Moreover, Ghosh et al. (2014) and the other two papers also use different specifications. These methodological differences may explain the differences in their findings. Other contributing factors In their analysis of monthly panel data from January 1988 to September 1992 on US capital flows to nine Latin American and nine Asian countries, Chuhan et al. (1998) find that the lower US interest rates, the slow growth in the US economy, and the surge of domestic economic development in recipient countries (Asia and Latin America) are all positively related to capital inflows. Moreover, Taylor and Sarno (1997) study the determinants of capital flows (portfolio investment only), from the United States to Latin America and Asia from 1988 to Their Result suggests that US interest rates were a stronger predictor of portfolio flows to these regions than the slowdown of the US economy or and growth rates of emerging countries. In addition, Reinhart and Reinhart (2008) study 181 countries from 1980 to 2007, and find that large net capital flows are positively associated with the incidence of banking and currency crises. Forbes and Warnock (2012) find that decreases in global risk (e.g., economic uncertainty) are positively and significantly related to disruptions in capital inflows and negatively related to surges in capital inflows. I draw on these findings when developing the list of control variables for my analysis. 4 In Ghosh et al. (2014), surge periods are identified when a country-year observation has net capital flows in the top 30th percentile of either the country's own distribution, or the full sample's distribution. 7

13 Contribution to the literature My research aims to contribute to existing literature by analyzing a broader range of country-level data for 41 developing countries. Most papers examine the relationship between exchange rate regimes and capital inflows (Boudias 2014; Magud et al., 2011) in less than 30 countries. Analyzing a larger sample of countries should allow for a more unbiased estimate of the relationship between exchange rate regimes and capital flows in developing countries. CONCEPTUAL FRAMEWORK I hypothesize that, all else equal, a fixed exchange rate regime is associated with moderately larger volumes of capital inflows, compared with a flexible exchange rate regime (Boudias 2014; Magud et al., 2011; Ghosh et al., 2014). My analysis will also account for macroeconomic and financial factors that may influence capital inflows into a country. Figure 1 identifies eight factors that may influence capital inflows. These factors are grouped into three categories: exchange rate regime, macroeconomic factors, and financial factors. 8

14 Figure 1. Factors that contribute to capital inflows Exchange Rate Regime Macroeconomic Factors: Gross Domestic Product GDP Growth Rate External Debt/GDP Trade/GDP Financial Crisis Financial Factors: Interest Rate Differentials Capital Account Opennes Index Capital In;lows Macroeconomic Factors A country s Gross Domestic Product (GDP) and real GDP growth rates capture its level of economic development (Chuhan et al., 1998). The ratio of external debt to GDP measures the debt level of a country, and trade (exports plus imports) as a share of GDP captures trade openness of a country (Boudias, 2014). GDP growth is positively associated with capital inflows because a growing economy is more attractive to investors, while a high level of external debt has been found to deter capital inflows because countries that are heavily indebted have a higher default risk and are therefore less attractive to foreign investors (Magud et al., 2011). Financial crises may also deter capital flows due to instability in the economy (Reinhart and Reinhart, 2008). 9

15 Financial Market Factors Interest rate differentials (IRD) measures the profit margin that an investor is expected to receive in a foreign market. IRD thus explain the incentive for investors to invest in a foreign currency. A higher IRD is associated with an increase in the foreign country s capital inflows (Taylor and Sarno, 1997). A country s capital account openness index measures its financial and capital freedom; I expect a higher index to be associated with an increase in capital inflows (Chin and Ito 2006). 5 DATA & METHODS This study analyzes country-level annual data on capital inflows and exchange rate regimes for 41 developing countries over an eleven-year period, from 2000 to I collect data on capital inflows from a dataset on foreign assets and foreign liabilities constructed by Philip Lane and Gian Maria Milesi-Ferretti. 6 I include all countries for which data on capital inflows are available. There are massive gaps in the data before 2000, and data are unavailable for years after I measure capital inflows in terms of portfolio investment (debt and equity) and financial derivatives. I exclude FDI in my measure of capital inflows because FDI is more stable in the short run than portfolio investment and financial derivatives, and thus FDI does not vary much relative to the change in exchange rate regimes. Studying capital inflows net of FDI thus allows me to focus on the most volatile component of capital inflows (Boudias, 2014). In addition, I construct information on exchange rate regimes from the coarse classification data series 5 The capital account openness index was developed by Chinn and Ito (2006) 6 See Lane and Milesi-Ferretti (2007) 10

16 constructed by Carmen Reinhart and Kenneth Rogoff. An increase in the index suggests a more flexible exchange rate regime. 7 It is difficult to classify exchange rate policies because, in many cases, there is a substantial difference between the regimes that countries use in practice (de facto exchange rate regimes), and the regimes that governments announce to adopt (de jure exchange rate regimes) (Chakravarty, 2011). 8 The coarse classification measures de facto rather than de jure regimes (i.e., the actual exchange rate attributable to domestic policies). 9 Hence, it is a more accurate measure of the real exchange rate regimes, and changes in exchange rate regimes measured by the coarse classification will more likely capture changes in capital inflows. The coarse classification ranges from one to five (e.g., de facto pegged regimes are assigned a value of one, while freely falling regimes are assigned a value of five). Table 4 in the Appendix reports the full set of coarse classification categories used in my model. 10 As discussed in the previous section, I also control for macroeconomic and financial factors that are plausibly related to exchange rate policies and capital inflows. I use the same set of control variables as Boudias (2014) and Magud et al. (2011). I obtain data on Gross Domestic Product (GDP), GDP growth rates, exports, and imports from the World Bank s World Development Indicator dataset. I obtain data on external debt as percent of GDP from a dataset constructed by Lane and Milesi-Ferretti (2007). IRD are the difference between a second country s domestic interest rate and the US federal funds 7 See Reinhart and Rogoff (2002), and the update from Ilzetski, Reinhart and Rogoff (2008) 8 A de facto exchange rate policy is the actual exchange rate policy observed in a country, as compared with a de jure exchange rate policy, which is the exchange rate policy announced by a country s government. 9 See Ilzetzki et al. (2008) for more details on the coarse classification 10 The original coarse classification has values range from one to six; none of my country-year observations had a value of six. 11

17 rate. I obtain US federal funds rate from the International Financial Statistics database published by the International Monetary Fund (2013), and I take a second country s domestic interest rate information from the World Bank s World Development Indicators. I take my capital account openness index measure from a dataset created by Menzie Chinn and Hiro Ito (2006). I obtain information on systemic bank crises from a dataset created by Luc Laeven and Fabián Valencia at IMF (2012). In order to analyze the relationship between exchange rate arrangements and capital inflows, I estimate a regression model with country and year fixed effects. My fixed effects specification controls for all country-level characteristics that are fixed over time and all time-varying characteristics that are fixed across countries. For instance, differences in geography, natural endowments, climate, ethno-linguistic characteristics, and unchanging political and legal systems between countries will be captured by the country dummies, while technological development, the US federal funds rate, and US economic conditions will be captured by the year dummies. I lag my key independent variable by one year in order to ensure that my estimates are not affected by reverse causality (in which exchange rate regimes are affected by capital inflows). My estimated equation is: k_flow it = β + β ERR 0 1 it-5 + β GDP + β GDP_growth 2 it 3 it + β debt + β IRD + 4 it 5 it β trade + β crisis + β kaopen + α 6 it 7 it 8 it i + γt + µ it where i is the country index, and t is the year index; α i represents country dummies γ t represents the year dummies, and µ it is the error term. The dependent variable, capital inflows, is measured as a percentage of GDP. Table 1 provides definitions for all variables included in my model. 12

18 Table 1. Definitions of Variables and Data Sources Variable Definition Source Dependent Variables Variable consists of two subcomponents: Capital Inflows (Percent of GDP) Exchange Rate Regime (ERR) (coarse classification system) Portfolio Investment (debt and equity): Portfolio investments are defined as cross-border investment positions and transactions that involve debt and equity. Financial Derivatives: a financial derivative is a contract that originates its value from an underlying asset. Key Independent Variable The coarse classification system assigns countries a score from 1 to 6, where an increase in scores reflects a more flexible ERR. Lane & Milesi- Ferretti (2007) Reinhart and Rogoff (2002); updated by Ilzetzki et al. (2008) 13

19 Table 1 Continued Variable Definition Source Control Variables Gross Domestic Product (Millions of USD) GDP Growth Rate (percent change from year ago) The annual gross market value of final goods and services in the economy. The rate at which a country's GDP changes from year t-1 to year t. World Development Indicator (WDI) WDI External Debt/GDP (percent of GDP) The portion of a country's debt that is owned by foreign lenders. The debt here includes the external debts of government and private debt issued. Lane & Milesi- Ferretti (2007) Trade/GDP (percent of GDP) Systemic Banking Crisis Interest Rate Differentials Capital Account Openness Index Exports are sales of goods and services to a foreign country; imports are purchases of foreign manufactured goods and services in the buyer's domestic market. Exports plus imports are the sum of trade. A dummy variable indicating whether in a given year a country goes through a systemic banking crisis or not. The difference between a second country s domestic interest rate and the US federal funds rate. An index measuring a country s degree of capital account freedom, which is created using information on restrictions about cross-border financial transactions reported in the IMF s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). WDI and UN Comtrade Database Laeven and Valencia (2012) International Financial Statistics (2013); WDI Chinn and Ito (2006) 14

20 DESCRIPTIVE STATISTICS Table 2 provides descriptive statistics for all country-years included in my analysis. 11 The average annual capital inflow as a proportion of GDP is.197%, with a range from zero (Bangladesh in 2002) to 5.75% (Mauritius in 2009). 12 Coarse classification of exchange rate regime values range from one to five, and have a mean of countries changed their exchange rate regimes during the period of study, with Lithuania changing its regime four times. Exchange rate policy in Argentina, Macedonia, and Turkey was changed three times, while Malta, Latvia, and China each changed their exchange rate policies twice. The mean annual GDP level is USD 269 billion, and GDP values range from USD 3.44 billion (Macedonia in 2001) to of USD 5,930 billion (China in 2010). The mean GDP growth rate is 4.52%, with a range from -17.7% (Latvia in 2009) to 17.3% (Kuwait in 2003). 11 In the database from which I took my dependent variable, there was a substantial gap in information on portfolio debt (a key measure of capital flows) for selected countries from 1996 to These years are therefore not included in my analyses. Additionally, some or all country-year observations for nine countries were dropped due to missing data on portfolio debt (UAE, Vietnam, Sri Lanka, Qatar, Oman, Lebanon, and Kenya, all from 2000 to 2010; Nigeria from 2000 to 2005; and Serbia from 2000 to 2007). Additionally, seven countries are missing data on portfolio debt for 2000: Morocco is missing data for years 2000 and 2001; three countries (Pakistan, Macedonia, and Cyprus) are missing data for 2000 to 2002; and China is missing data for 2000 to These countries are retained in my analysis, but country-year observations are dropped for the years that contain missing data. Data on domestic interest rates (deposit interest rate/money market rate) for Tunisia, Poland, Pakistan, and South Korea come from IMF s International Financial Statistics database, due to missing data for these countries in the World Bank s World Development Indicators database. For trade, Lithuanian data were taken from the UN Comtrade Database because four years of data were missing from the original source (WDI). Among country-years that are not missing data for this variable, estimates from the two sources match closely. My sample size is 430 (41 countries*11 years - 21 country-year observations containing missing values). 12 Mauritius is an island of Sub-Saharan Africa. Its economy is highly dependent on tourism and international trade. In the past decade, the Mauritian government has liberalized its capital market. One of the country s most common forms of businesses is offshore banking (Country overview at the World Bank). It is reasonable to assume that people transfer money into and out of Mauritius with considerable frequency, which contributes to the volatility in that country s capital inflows and outflows. 15

21 Table 2. Descriptive Statistics Variable Mean Std. Dev. Min Max Dependent Variable Capital Inflow (%of GDP) Independent Variable Exchange Rate Regime Macroeconomic Factors GDP (USD Million) 269, ,690 3,437 5,930,393 GDP Growth Rate (%) Imports (%of GDP) Exports (%of GDP) Trade (%of GDP) Debt (%of GDP) Systemic Banking Crisis Financial Factors Capital Account Openness Interest Rate Differentials (%) N=430 REGRESSION RESULTS Table 3 presents my regression results. Model (1) is a simple OLS regression that includes all control variables, but does not include country or year fixed effects. Model (2) introduces country fixed effects; model (3) excludes country-fixed effects and includes year fixed effects; and model (4) is a fully specified regression with both country and year fixed effects and all control variables. Model (5) explores the relationship between exchange rate regime flexibility and capital inflows using a dichotomous measure of flexibility. An exchange rate regime is defined as flexible in this dichotomous measure if its exchange rate regime index is bigger than two. In model (6), the exchange rate regime index is lagged for one year to ensure that my estimates are not affected by 16

22 reverse causality. Model (7) explores the interaction between systemic banking crises and exchange rate regimes. Robust standard errors are reported beneath all coefficients. As previously discussed, my key independent variable- the exchange rate regime index- has values ranging from one to five, and an increase in the index suggests a more flexible exchange rate regime. The coefficients from model (1) indicate a positive but statistically insignificant relationship between exchange rate regime flexibility and capital inflows. However, this model likely suffers from omitted variable bias caused by not controlling for country-level characteristics that are fixed over time, and time-varying characteristics that are fixed across countries. In model (2), the inclusion of fixed country-level characteristics produces a coefficient that is slightly smaller in magnitude, and the estimate remains statistically in-differentiable from zero. This change is not surprising since we would expect that fixed country-level characteristics, such as natural endowments and unchanging political and legal systems to be correlated with both capital inflows and exchange rate regimes. The sign of the relationship between the countrylevel characteristics and exchange rate regimes is likely to be the same as the sign of the relationship between the country-level characteristics and capital inflows. Model (3) controls for time-fixed effects, but not for country-fixed effects. This model generates a slightly larger positive coefficient that is statistically significant. On the other hand, model (4), which includes both country and year fixed effects, generates a coefficient that is negative and statistically insignificant. This change is not surprising. The least biased estimates in my literature review suggest a moderate and negative correlation between exchange rate regime flexibility and capital inflows. By including country and year fixed effects, model (4) introduces the least bias into my analysis. 17

23 Model (5) measures exchange rate regime flexibility using a dummy that equals one if a country s exchange rate regime index is larger than two in a given year. An index of two is chosen as the threshold because it is the middle point, where exchange rate regimes, to some extent, start to float freely based on market activities. The coefficient for this variable is negative, though it is not statistically different from zero. In model (6), I lag my key independent variable, the exchange rate regime index, for one year to ensure that my estimates are not affected by reverse causality. The coefficient indicates a negative but statistically insignificant relationship between exchange rate regimes in the previous year and capital inflows in the current year. This result is consistent with all other specifications. Model (7) introduces an interaction term between my exchange rate regime measure and my systemic banking crisis dummy. Since in a systemic banking crisis, most governments adjust their macroeconomic policies to save their economies, one might hypothesize that the relationship between exchange rate regimes and inflows differ for countries going through systemic banking crises in given years. For country-years that are not going through a systemic banking crisis, a one unit increase in a country s exchange rate regime index is associated with a.85% decrease in capital inflows as a share of GDP. However, the coefficient is not statistically significant. For country-years going through a systemic banking crisis, a one-unit increase in a country s exchange rate regime index is associated with a 2.73% (-0.85%+3.58%) increase in capital inflows as a share of GDP. However, the F-test indicates that there is no statistically significant relationship between exchange rate regimes and capital inflows in countries that are going through a systemic banking crisis. 18

24 In sum, my findings suggest that there is a moderate positive relationship between exchange rate regime flexibility and capital inflows when country-level characteristics are not controlled for. However, there is no statistically significant relationship between exchange rate regime flexibility and capital inflows in any other case. The inclusion of country and year fixed effects decreases the extent of omitted variable bias in my estimates, and generates a slightly smaller coefficient on exchange rate regime index, relative to models that do not include country and year fixed effects. 19

25 Table 3. Regression Results Dependent Variable: Capital Inflows (% of GDP) (1) (2) (3) (4) (5) (6) (7) Exchange Rate Regime Index ** (0.0220) (0.0155) (0.0179) (0.0239) (0.0296) GDP (logged) * (0.0140) (0.0765) (0.0314) (0.152) (0.151) (0.163) (0.152) GDP Growth Rate ( ) ( ) ( ) ( ) (0.0100) (0.0118) ( ) Trade (% of GDP) *** *** ( ) ( ) ( ) ( ) ( ) ( ) ( ) Debt (% of GDP) ( ) ( ) ( ) ( ) ( ) ( ) ( ) Systemic Banking Crisis (0.0666) (0.0588) (0.0298) (0.0372) (0.0362) (0.0548) (0.106) Capital Account Openness *** (0.0125) (0.0590) ( ) (0.0538) (0.0534) (0.0774) (0.0568) 20 Interest Rate Differentials ( ) ( ) ( ) ( ) ( ) ( ) ( ) Exchange Rate Flexible (0.0509) Exchange Rate Regime* Systemic Banking Crisis (0.0391) Exchange Rate Flexibility lagged for 1 year (0.0309) Constant (0.179) (0.726) (0.270) (1.754) (1.737) (1.936) (1.775) Observations R-squared Year FE? No No Yes Yes Yes Yes Yes Country FE? No Yes No Yes Yes Yes Yes F-statistics and p-value for joint hypotheses Crisis and Exchange Rate Regime 0.54 (0.585) Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1

26 DISCUSSION This paper examines the relationship between exchange rate regimes and capital inflows in developing countries from 2000 to For the past two decades, economists have debated an optimal exchange rate regime for developing countries. My study adds perspective to this debate and provides useful information for policy makers. My findings are ambiguous. My main analyses show that, for the period , there is no statistically significant relationship between exchange rate flexibility and capital inflows in developing countries. However, when I control for time fixed effects, but not country-fixed effects, flexible exchange rate regimes appear to attract more capital inflows, compared with fixed exchange rate regimes. As shown in table 3, for model (3), a one-unit increase in the exchange rate regime index is associated with a 3.67% of GDP increase in capital inflows in developing countries. However, since I do not control for country fixed effects, it is possible that this result may be biased. This positive relationship may be caused by country level characteristics rather than exchange rate regime differences. However, it is also possible that this model (i.e., my model excluding country fixed effects) produces less biased estimates because 22 (e.g., Bahrain, Brazil, Bulgaria, Chile, Colombia) out of the 41developing countries included in my study did not change their exchange rate regimes between 2000 and 2010, and not controlling for country-fixed effects allows for greater variations in my dependent and independent variables in measuring the relationship between exchange rate regimes and capital inflows in these countries. Magud et al. (2011) and Boudias (2014) also include models in their analyses that do not control for country-fixed effects. 21

27 As mentioned in my literature review, Magud et al. (2011) and Boudias (2014) also find no relationship between exchange rate regime flexibility and capital inflows in developing countries. However, Ghosh et al. (2014) find a negative relationship between these two. My results differ from those of Ghosh et al. in part because our studies cover different periods. I study the period from 2000 to 2010, while Ghosh et al. analyze the period from 1980 to In addition, Ghosh et al. (2014) do not control for country and time fixed effects. The inclusion of fixed effects in my models may help to explain the difference between my results and theirs. My study has a number of limitations. For one thing, omitted variable bias likely influences my results. For instance, unlike Gosh et al. (2014), I do not include a control for the institutional quality index. 1 It is likely that institutional quality in a country is positively correlated with the country s capital inflows, since investors are generally risk averse. If the quality of contract enforcement, property rights, and shareholder protection are high, as measured by the institutional quality index, the country will attract more capital inflows. I speculate that institutional quality may be positively correlated with exchange rate flexibility, since in more liberalized countries, I observe more flexible exchange rate regimes. If my assumption holds, the omission of this variable may be exerting an upward bias on my coefficients. That means the coefficient in my results may be an overestimate of the coefficient that I would have observed if I had controlled for institutional quality. In addition, I only explore the relationship between exchange rate regime flexibility and capital inflows broadly. In other words, unlike other researchers in this 1 Gosh et al. (2014) include the institutional quality index that serves as a measure of corruption, and of the quality of legal and political systems 22

28 field (Magud et al., 2011; Boudias, 2014; and Gosh et al., 2014), I do not identify surge periods. 2 If I had identified and focused on surge periods, I would have been able to observe whether the relationship between exchange rate regimes and capital inflows differs between surge periods and non-surge periods. By definition, during surge periods, we observe greater variations in capital inflows. Thus, focusing on the surge periods can help improve understanding of the reason behind surges in capital inflows. These limitations suggest directions for future research. My analysis could also be extended in several ways. First, as more data becomes available, future empirical analysis could include a larger set of countries and a longer time period. In addition, researchers could examine the effects of exchange rate regimes on capital inflows, outflows, and net capital flows separately. Many countries experience large capital inflows and outflows at the same time, which may eventually affect the stability of their economies. Finding reasons behind such phenomena may help policy makers in these countries cope with this problem. Lastly, since my results also suggest that trade and capital account openness are likely to play a role in deciding capital flows to developing countries, I suggest that future research on capital inflows in developing countries place greater emphasis on these factors. My findings have modest, but potentially useful policy implications. My main analyses suggest that there is not a statistically significant relationship between exchange rate regime flexibility and capital inflows in developing countries from 2000 to This finding is consistent with Calvo and Mishkin s (2003) findings, which suggest that, to attract capital inflows and produce macroeconomic stability in emerging market 2 As noted above, surge periods are identified when a country-year observation has net capital flows in the top 30th percentile of either the country's own distribution, or the full sample's distribution. 23

29 countries, the development of stable financial and monetary institutions is more important than the choice of exchange rate policy. I agree with Calvo and Mishkin s conclusion regarding the policies most likely to affect capital inflows. Thus, I suggest with them that developing countries, seeking to promote capital inflows and macroeconomic stability also focus on monetary policies such as interest rates and money supply (Ahmed & Zlate, 2014, Taylor and Sarno, 1997). A more nuanced analysis of monetary policies and financial factors may shed further light on the determinants of capital inflows in developing countries. In summary, the limitations of my study prevent strong policy recommendations, while suggesting that future research seek to address those limitations. 24

30 APPENDIX Table 4. Exchange Rate Regimes Coarse Classification 1 No separate legal tender 1 Pre announced peg or currency board arrangement 1 Pre announced horizontal band that is narrower than or equal to +/-2% 1 De facto peg 2 Pre announced crawling peg 2 Pre announced crawling band that is narrower than or equal to +/-2% 2 De facto crawling peg 2 De facto crawling band that is narrower than or equal to +/-2% 3 Pre announced crawling band that is wider than or equal to +/-2% 3 De facto crawling band that is narrower than or equal to +/-5% 3 Moving band that is narrower than/ equal to +/-2% (allow for both appreciation and depreciation over time) 3 Managed floating 4 Freely floating 5 Freely falling 25

31 REFERENCES Ahmed, S., & Zlate, A. (2014). Capital flows to emerging market economies: a brave new world. Journal of International Money and Finance, 48, Aizenman, J. (1992). Exchange rate flexibility, volatility, and the patterns of domestic and foreign direct investment (No. w3953). National Bureau of Economic Research. Balakrishnan, R., Nowak, S., Panth, S., & Wu, Y. (2013). Surging capital flows to emerging Asia: facts, impacts and responses. Journal of International Commerce, Economics and Policy, 4(02), Boudias, R. (2014). Capital Inflows, Exchange Rate Regimes and Credit Dynamics in Emerging Market Economies (No ). Chakravarty, S. L. (2011). Exchange rate regime and monetary independence - the indian experience. Decision, 38(2), Retrieved from Calvo, G., & Mishkin, F. S. (2003). The mirage of exchange rate regimes for emerging market countries (No. w9808). National Bureau of Economic Research. Chuhan, P., Claessens, S., & Mamingi, N. (1998). Equity and bond flows to Latin America and Asia: the role of global and country factors. Journal of Development Economics, 55(2), Chinn, M. D., & Ito, H. (2006). What matters for financial development? Capital controls, institutions, and interactions. Journal of Development Economics, 81(1), Forbes, K. J., & Warnock, F. E. (2012). Capital flow waves: Surges, stops, flight, and retrenchment. Journal of International Economics, 88(2), Ghosh, A. R., Qureshi, M. S., Kim, J. I., & Zalduendo, J. (2014). Surges. Journal of International Economics, 92(2), Ilzetzki, E., Reinhart, C., & Rogoff, K. (2008). Exchange rate arrangements entering the 21st century: Which anchor will hold? Unpublished manuscript and data available online: Lane, P. July (1995) Determinants of Pegged Exchange Rates. Working paper. Columbia University. Lane, P. R., & Milesi-Ferretti, G. M. (2007). The external wealth of nations mark II: Revised and extended estimates of foreign assets and liabilities, Journal of international Economics, 73(2), Laeven, L., & Valencia, F. (2012). Systemic banking crises database: An update. Magud, N. E., Reinhart, C. M., & Vesperoni, E. R. (2011). Capital inflows, exchange rate flexibility and credit booms. Review of Development Economics, 18(3), Mundell, R. A. (1963). Capital mobility and stabilization policy under fixed and flexible exchange rates. Canadian Journal of Economics and Political Science/Revue canadienne de economiques et science politique, 29(04),

32 Reinhart, C. M., & Rogoff, K. S. (2002). The modern history of exchange rate arrangements: a reinterpretation (No. w8963). National Bureau of Economic Research. Reinhart, C. M., & Reinhart, V. R. (2008). Capital flow bonanzas: an encompassing view of the past and present (No. w14321). National Bureau of Economic Research. Obstfeld, M., and Rogoff, K. (1995) The mirage of fixed exchange rates. Journal of Economic Perspectives Ohno, S., & Shimizu, J. (2015). Do exchange rate arrangements and capital controls influence international capital flows and housing prices in Asia? Journal of Asian Economics. Stephey, M.J. (2008). "Bretton Woods System." Time. Time Inc., Taylor, M. P., & Sarno, L. (1997). Capital flows to developing countries: long-and short-term determinants. The World Bank Economic Review, 11(3),