Capital Flows, Exchange Rate Flexibility, and the Real Exchange Rate
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 2 https://isni.org/isni/0000000404811396, International Monetary Fund

Contributor Notes

This paper analyzes the impact of capital inflows and exchange rate flexibility on the real exchange rate in developing countries based on panel cointegration techniques. The results show that public and private flows are associated with a real exchange rate appreciation. Among private flows, portfolio investment has the highest appreciation effect-almost seven times that of foreign direct investment or bank loans-and private transfers have the lowest effect. Using a de facto measure of exchange rate flexibility, we find that a more flexible exchange rate helps to dampen appreciation of the real exchange rate stemming from capital inflows.

Abstract

This paper analyzes the impact of capital inflows and exchange rate flexibility on the real exchange rate in developing countries based on panel cointegration techniques. The results show that public and private flows are associated with a real exchange rate appreciation. Among private flows, portfolio investment has the highest appreciation effect-almost seven times that of foreign direct investment or bank loans-and private transfers have the lowest effect. Using a de facto measure of exchange rate flexibility, we find that a more flexible exchange rate helps to dampen appreciation of the real exchange rate stemming from capital inflows.

I. Introduction

Policymakers often seek to attract external resources on the assumption that they will finance savings gaps and promote growth and economic development (Dornbusch, 1998). However, evidence of the growth potential of capital account openness is mixed (Kose et al., 2006). Moreover, significant increases in capital inflows can make the financial system more vulnerable and overheat the economy. Lending booms, which often follow increased capital inflows, increase financial system vulnerability (a) by exacerbating maturity mismatches between bank assets and their liabilities, and in some cases mismatches between the currencies in which banks lend and borrow, and (b) through associated asset price bubbles. Macroeconomic overheating can be provoked by accelerated economic growth and inflation, and particularly by appreciation of the real effective exchange rate (REER).

The loss of competitiveness caused by an appreciated real exchange rate is one of the main negative consequences associated with capital inflows (Calvo, Leiderman, and Reinhart, 1993; Bandara, 1995; Edwards, 1998; Agenor, 1998; Lartey, 2008). Where the exchange rate regime is flexible, real appreciation of the exchange rate is due to appreciation of the nominal exchange rate. Where the exchange rate is fixed, real appreciation is due to a rise in inflation after the money supply increases. Appreciation of the real exchange rate undermines competitiveness, widens the current account deficit, and increases vulnerability to a financial crisis. Significant appreciation could lead to a sudden drying up of capital flows, causing an abrupt adjustment of the current account. Beyond its negative effect on investment, significant appreciation of the real exchange rate could thus create major problems for macroeconomic management.

The surge in external financing to developing countries, particularly private flows, over the last decade and up to the current financial crisis sheds some light on the “transfer problem”. The “transfer problem” refers to the impact of capital inflows or outflows on the domestic economy—which is captured mainly through the real exchange rate.

The spectacular rise in private flows in recent years was driven by foreign direct investment (FDI) and current private transfers, mainly remittances. While commercial bank loans constituted the main component of private capital flows to developing countries in the mid- 1980s, later FDI and remittances replaced bank loans, particularly in low-income countries. Portfolio investments have been a significant part of private capital flows to emerging countries since the 1990s. These changes in the landscape of capital flows to developing countries underline the importance of reassessing the transfer problem, looking particularly at the components of private flows.

While most studies analyze the effect of aggregated capital inflows on the real exchange rate, this paper proposes a comprehensive analysis of the impact on it of different forms of private capital flows (FDI, portfolio investment, bank loans, and private transfers). Moreover, developing countries use a variety of macroeconomic tools to dampen real appreciation of their exchange rates caused by capital inflows, such as exchange rate flexibility (IMF, 2007). This paper questions whether such a policy is effective. Using a sample of 42 developing countries for 1980–2006, we apply the pooled mean group estimator that allows short-run heterogeneity while imposing long-run homogeneity on the real exchange rate determination across countries. The results show that aggregated capital inflows as well as public and private flows are associated with real exchange rate appreciation. Among private flows, portfolio investment has the highest appreciation effect—almost seven times that of FDI or bank loans. Private transfers have the least effect. The de facto measure of exchange rate flexibility allows us to conclude that a more flexible exchange rate could effectively dampen the real appreciation stemming from capital inflows.

The rest of this paper is organized as follows: Section II stresses potential heterogeneity by type of capital flow and discusses the role played by the exchange rate regime. Section III describes the main trends and composition of external financing for developing countries. Section IV presents the pooled mean group estimator and the dataset. Section V analyzes the results, and Section VI draws conclusions.

II. Composition of Capital Inflows, Exchange Rate Regime, and The Real Exchange Rate

Edwards (1989, 1994), Williamson (1994), Hinkle and Montiel (1999), Edwards and Savastano (2000), and Maeso-Fernandez, Osbat, and Schnatz (2004) provide comprehensive surveys of the extensive literature on determinants of the real exchange rate. A number of studies look at the impact of capital flows on the real exchange rate—the transfer problem. Capital inflows generate higher demand for both tradables and nontradables and lead to a higher relative price of nontradables and to appreciation of the real exchange rate. This is necessary so that domestic resources will be diverted to production of nontradables to meet the increased demand.

As the next section will establish, net capital inflows to developing countries have increased dramatically since the 1980s, with private flows having growing influence, particularly in middle-income countries. The impact on the real exchange rate holds. As Lane and Milesi- Ferretti (2004) put it, based on an econometric analysis of 48 countries, net external liabilities go hand in hand with depreciation of the real exchange rate. Lee, Milesi-Ferretti, and Ricci (2008) show that higher net foreign assets cause the real exchange rate to appreciate. In other words, external capital allows expenditure to exceed income, generating excess demand for nontradables. This effect has to be qualified, however, for at least two reasons: (1) The real exchange rate can be affected differently depending on the composition of capital inflows, and (2) the type of exchange rate regime may sway the effect of capital inflows on the real exchange rate.

A. Composition of Capital Inflows and the Real Exchange Rate

The current specification of the exchange rate determination model suggests that all capital inflows have similar impact on the exchange rate. This is questionable. Although there is as yet little evidence, some authors have recently, with mixed results, hypothesized specific impacts, highlighting the role of official flows, FDI, and remittances.

Official flows are generally associated with appreciation of the real exchange rate (Kasekende and Atingi-Ego, 1999; Bulir and Lane, 2002; Prati, Sahai, and Tressel, 2003; Lartey, 2007; Elbadawi, Kaltani, and Schmidt-Hebbel, 2008). Other studies, however, do not find that public flows cause a real appreciation (Li and Rowe, 2007; Hussain, Berg, and Aiyar, 2009; Mongardini and Rayner, 2009). For Cerra, Tekin, and Turnovsky (2008) foreign aid leads to a real exchange rate appreciation only if it enhances productivity in the tradable sector. Where foreign aid is channeled to improve productive capacity in the nontradable sector, the authors find evidence of a real depreciation.

The impact of official flows on the real exchange rate also depends on how they are used. Assuming that a significant part of official flows is targeted to enlarge basic infrastructure to meet the Millennium Development Goals, the relative contribution of domestic consumption to global expenditure should be considered as an important factor in exchange rate evolution. Where there are supply constraints, capital inflows associated with higher consumption put more pressure on the relative price of domestic goods than capital inflows associated with higher investments, which have significant imported goods content.2

FDI leads to less credit and money expansion because it is less, and only briefly, intermediated through the local banking system. The inflation potential of FDI may thus be lower than that of commercial bank loans. FDI flows could be related to investment in imported machinery and equipment; these imports do not suffer from constraints in local supply capacity and thus have almost no appreciation effect. The spillover effects of FDI may also improve local productive capacity through transfer of technology and managerial know-how (Javorcik, 2004). As countries with a better investment climate attract more FDI (Kinda, 2008, 2010), local productive capacity could improve before FDI flows, reducing pressure on the real exchange rate. FDI is also a more stable flow than bank lending and portfolio investment.

Appreciation of the real exchange rate due to FDI is less than appreciation due to more volatile private flows that do not necessarily increase productive capacity, such as portfolio investments (Lartey, 2007). The number of studies dealing with the impact of private flows on the real exchange rate is limited and the results are mixed, as evidenced by Athukorala and Rajapatirana (2003). Lartey (2007) finds that FDI causes the real exchange rate to appreciate but the aggregate “other capital flows” does not. Saborowski (2009) suggests that in developing countries capital inflows, particularly FDI, lead to a real appreciation of the exchange rate.

Remittances (private transfers), can be assimilated to private capital inflows; their impact on the real exchange rate depends on whether they are pro- or countercyclical. On the one hand, remittances act as a buffer, helping to smooth consumption, if they increase when the recipient economy is suffering an economic downturn (Lueth and Ruiz-Arranz, 2007; Chami et al., 2008). In this case they help to keep recipient economies stable by compensating for foreign exchange losses due to macroeconomic shocks. These countercyclical remittances do not have much effect on the real exchange rate.

On the other hand, remittances for investment purposes3 can be procyclical, exacerbating macroeconomic overheating and driving the real exchange rate to appreciate more. In some developing countries, for instance, procyclical remittances spent on real estate have increased input prices, giving rise to construction booms. When most of the remittances are spent on traded goods (imported consumer durables, for instance), their effects on the real exchange rate tend to be weaker (Chami et al., 2008). Although the effect of private transfers or remittances on the real exchange rate is generally suggested in theory, in fact the empirical results are mixed (Chami et al., 2008). Bourdet and Falck (2003), Amuedo-Dorantes and Pozo (2004), Montiel (2006), and Saadi-Sedik and Petri (2006), among others, find that remittance inflows cause the real exchange rate to appreciate. Others, like Izquierdo and Montiel (2006) and Rajan and Subramanian (2005), are not able to conclude unequivocally that remittances are associated with real exchange rate appreciation.

B. Exchange Rate Regime and the Real Exchange Rate

The relation between the real exchange rate and capital inflows can be seen as depending on the choice of the exchange rate system. In the 1970s the academic debate on this issue focused exclusively on a binary choice between floating or fixed exchange rates. Although such a duality has analytic convenience, the reality today is much more complex, as suggested by the distinction between de jure and de facto classifications, which expands the number of regime categories.

With a fixed exchange rate, capital inflows potentially increase inflation. The scope of these pressures depends on whether inflows are driven by autonomous factors or by an increase in domestic money demand and also on the policy response to the inflows. In a number of countries, a surge in capital flows led to a credit boom when monetary authorities failed to sterilize them. There higher money supply and inflationary pressures spread within the economy, contributing to an increase in the relative prices of nontradables. A sterilization policy can dampen real appreciation, but recognition of the “perils of sterilization” (Calvo, 1991) led to doubt about its long-term sustainability. Indeed, when the exchange rate is fixed, a sterilization policy leads to higher interest rates and to additional capital inflows. Moreover, holding foreign assets with lower interest rates than domestic ones generates quasi-fiscal losses for central banks, leading them to give up the policy in the medium or long term.

With a floating exchange rate, capital inflows lead to an appreciation of the nominal exchange rate, enhancing a fall in the relative prices of imported goods and a shift away from the consumption of nontradables. Exchange rate flexibility ensures that monetary policy is somewhat independent of capital inflows. By introducing uncertainty, a more flexible exchange rate could discourage short-term speculative flows and reduce financial system vulnerability, particularly when supervision and regulation are poor (Calvo, Leiderman, and Reinhart, 1996; Lopez-Mejia, 1999). Hence, a flexible exchange rate regime would penalize the capital flows that generate the most real appreciation. However, a pure flexible exchange rate could be a problem if the rate resulting from all types of capital inflows differs from the long-term equilibrium rate. Appreciation of the nominal exchange rate may have a significant impact on the real sector, necessitating central bank interventions to limit perverse effects and costly reallocations of productive resources within the economy.

Although some monetary instruments might prevent the undesired real economic effects of a nominal appreciation of the exchange rate with a managed floating exchange rate system, it is difficult to go against market forces for long. That is also true for a fixed system, the efficiency of which is conditional on the possibility that the monetary authorities will neutralize inflows of external assets. Intermediate regimes could offer some flexibility. In countries with an intermediate exchange rate regime, authorities aim for a specific level of nominal exchange rate and monetary aggregate. Reserve accumulation then becomes a policy instrument. Holding to a specified nominal exchange rate with intervention by accumulating more reserves lowers the pressure on the nominal exchange rate and may raise inflation. By contrast, small-scale interventions, with authorities accumulating fewer reserves, can raise pressure on the nominal exchange rate and lower inflation.

III. External Financing in Developing Countries

Aggregated total capital flow is the sum of public and private flows, using data from the World Economic Outlook. Private capital flows are the sum of four elements:

  • Direct investment in the reporting economy from abroad (FDI), including debt- creating liabilities to foreign investors and direct investment in the form of equity;

  • Portfolio investment (PIL), which is the sum of debt instruments issued by the domestic private sector (corporate bonds and other private debt securities) and foreign purchases of equities of domestic companies;

  • Current private transfers4 (PRT); and

  • Liabilities to foreign banks (LFB).5

  • Public flows are the sum of official loans (OL) and official current transfers (OT). Official loans are the sum of official liabilities: IMF credits (BFOLG); debt instruments, such as government bonds issued by the domestic public sector (BFPLDG); and debt forgiveness in the capital account, including relief granted by the IMF (BKFO). To get a more precise picture of the net resources effectively transferred in each country, interests paid on all debt (DSI) are deduced from the OL.6
    Net total external financing=FDI+PIL+LFB+PRTPrivate flows+BFOLG+BFPLDG+BKFODSINet Transfers on Debt+PUTPublic Flows

Using the estimates of total external financing, the following section reviews trends in the composition of external financing. Note that payment of interest on debt, which for presentation purposes is represented as “other capital inflows,” actually constitutes a capital outflow.

Private capital flows have steadily increased since the 1980s, but public flows have been decreasing. From less than 2 percent of GDP just after 1980, private flows reached more than 6 percent in 2005–06. The increase is even more dramatic in low-income countries, where for 2005–06 private flows represent almost 10 percent of GDP—well beyond the 1.5 percent when the 1980s began. In recent years, private capital flows have largely been dominated by FDI, followed by private transfers (remittances) and portfolio investment. In low-income countries the marked increase in private flows is mainly due to private transfers, which increased from less than 2 percent of GDP when the 1980s began to more than 6 percent in 2005–06. FDI in low-income countries also increased, from less than 1 percent of GDP in the 1980s to almost 4 percent in 2005–06.7

A favorable economic and investment climate, characterized by solid growth, moderate inflation, and sound infrastructure, facilitates productive activities that attract foreign investment. The factors driving the surge in remittances are more complex. The significant increase might be due to changes in the host or home country economy, reductions in transfer fees, or simply better-quality data (Dorsey et al., 2008).

Commercial bank loans have become insignificant, particularly in low-income countries, since the financial crises of the 1990s. Meanwhile, although portfolio investments have been negligible for low-income countries, they have been significant for emerging economies, especially recently (Figure A.1). Private flows have surpassed the public flows (grants and official loans) that were the main source of capital for low-income countries. While public flows plunged in all countries, there is an indication that grants were replacing loans in low- income countries, which is consistent with donor commitments.

Figure 2 illustrates the trends of the unweighted mean of the REER and of total capital inflows for the sample of 42 developing countries.8 On average, periods of reduced capital inflows are associated with depreciation of the REER and periods of increased flows with appreciation.

Figure 1.
Figure 1.

External Financing in Developing Countries

(In percent of GDP, sample of 42 countries analyzed)

Citation: IMF Working Papers 2011, 009; 10.5089/9781455211876.001.A001

Figure 2.
Figure 2.

The Real Exchange Rate and Capital Inflows

(Unweighted mean for a panel of 42 countries)

Citation: IMF Working Papers 2011, 009; 10.5089/9781455211876.001.A001

The following econometric analysis gives a clearer picture of the potential positive correlation between the REER and capital inflows shown in the graphical analyses.

IV. Econometric Model

Two estimation approaches are commonly used with dynamic panel data models. The first consists of averaging separate estimates for each group in the panel. According to Pesaran and Smith (1995), the mean group estimator provides consistent estimates of the parameter averages. It allows the parameters to be freely independent across groups and does not consider potential homogeneity between groups. The second is the usual pooled approach; examples are the random effects, fixed effects, and GMM methods. These models force the parameters (coefficients and error variances) to be identical across groups, but the intercept can differ between groups. GMM estimations of dynamic panel models could lead to inconsistent and misleading long-term coefficients, a possible problem that is exacerbated when the period is long (Pesaran, Shin, and Smith, 1999).

Pesaran, Shin, and Smith (1999) propose an intermediate estimator that allows the short-term parameters to differ between groups while imposing equality of the long-term coefficients. The long-term movements of the REER and other macroeconomic fundamentals are expected to be identical from country to country but short-term movements are expected to be country-specific. The null hypothesis of homogeneity in the long-term coefficients can be verified with a Hausman test. The dynamic heterogeneous panel model of Pesaran, Shin, and Smith (1999) is an unrestricted error correction autoregressive distributed lag (ARDL) (p,q) representation.

Δyit=ϕiyi,t1+βixi,t1+j=1p1λijΔyi,tj+j=0q1δijΔxi,tj+μi+εit(1)

The cross-section units (countries) are denoted by i = 1, 2, …, N; t = 1, 2, …, T represent time periods; yit is the dependent variable; xit the matrix of regressors; μi the fixed effects; ϕi the coefficient on the lagged dependent variable; βi the vector of coefficients on the explanatory variables; λij the coefficients on the lagged first-differences of the dependent variable; and δij the coefficients on the first-differences of the explanatory variables and their lagged values. The disturbances, εit, are supposed to be normally and independently distributed across i and t with zero mean and variances σi2>0.

With φi < 0, there is a long-term relationship between yit and xit in the form:

yit=θixit+ηiti=1,2,,Nt=1,2,,T(2)

where θi=βiϕi represents the long-term coefficient, and the error terms of the long-term relationship (ηit) are stationary.

Considering the long-term relationship, equation 1 can be written as

Δyit=ϕiηi,t1+j=1p1λijΔyi,tj+j=0q1δijΔxi,tj+μi+εit(3)

The error correction term, ηi,t−1, is derived from the long-term equation (2), and the associated coefficient, φi, measures the speed of adjustment to long-run equilibrium.

By allowing short-term coefficients, intercepts, and error variances to differ between groups and by constraining long-term coefficients to be identical (θi=θ), the pooled mean group estimator of Pesaran, Shin, and Smith (1999) derives the parameters with the maximum likelihood technique. With the pooled likelihood estimators defined as ϕ^i, β^i, λ^ij, δ^ij, and θ^, the pooled mean group estimators are given by:

ϕ^PMG=i=1Nϕ^iN,β^PMG=i=1Nβ^iN(4)
λ^jPMG=i=1Nλ^ijN, j=1,,p1,δ^jPMG=i=1Nδ^ijN, j=0,,q1(5)
θ^PMG=θ^(6)

More specifically, the long-term relationship between the real exchange rate and macroeconomic fundamentals is given by the following relation:

REERit=θ0+θ1TOTit+θ2PRODit+θ3TRADEit+θ4CAPITALit+vit(7)i=1,2,,N   t=1,2,,T

where REERit is the real effective exchange rate; TOTit the terms of trade; TRADEit the ratio of exports and imports to GDP; PRODit the productivity gap; and CAPITALit the ratio of total external financing to GDP (see Table A.1. and Table A.2. for the list of variables and summary statistics).

The REER in the analysis is a CPI-based real exchange rate, defined as a weighted geometric mean of the bilateral nominal exchange rate and consumer price indices. An increase in the REER indicates an appreciation, and hence a potential loss of competitiveness. The REER of a country i is defined as:

REERi=NEERi×j=110(CPIiCPIj)wjNEERi=j=110(NBER)wj

With REERi representing the real effective exchange rate, NEERi the nominal effective exchange rate, and NBERi the nominal bilateral exchange rate of country i with regard to the currencies of country j, CPIi and CPIj denote the consumer price indexes of country i and country j, and wj is the weight of the j-th partner in the bilateral trade of country i for 1996–2003. The analysis considers the 10 main trade partners, excluding countries for which petroleum-related products represent at least 50% of exports9.

The productivity gap aims to capture the potential Balassa-Samuelson effect. It is defined as a country’s GDP per capita relative to the weighted average GDP per capita of its trading partners. The weights of the partner countries are similar to those used in constructing the REER. The Balasa-Samuelson effect assumes that productivity grows faster in tradable than in nontradable sectors. This results in higher wages in tradable sectors, which spill over to nontradable sectors and put upward pressure on wages. Since prices in tradable sectors are internationally determined and homogeneous across countries, higher wages in nontradable sectors result in a higher relative price for nontradables. This implies an increase in domestic inflation and an appreciation of the REER.

A rise in the terms of trade is expected to cause the equilibrium REER to appreciate to the extent that it improves the trade balance—the income effect dominates the substitution effect. Trade openness also affects the prices of nontradables through income and substitution effects. More restrictions on trade have a negative effect on the prices of tradables through the income effect and a positive effect through the substitution effect, so the income effect is less likely to dominate (Edwards, 1988). It is thus expected that restricting trade will push down the price of tradables relative to nontradables, leading to appreciation of the equilibrium REER.

Assuming that all variables are I(1) and co-integrated, vit is supposed to be I(0) for all i and is independently distributed across t. With a maximum of one lag 10 for all variables, the equilibrium error correction representation of the ARDL(1, 1, 1, 1, 1) model is

ΔREERit=ϕiREERi,t1θ0θ1TOTitθ2PRODitθ3TRADEitθ4CAPITALitδ1iΔTOTitδ2iΔPRODitδ3iΔTRADEitδ4iΔCAPITALit+εit(8)

Since we are studying long-run relationships, the coefficients of primary interest are the θ. In the first part of the analysis, the interest variable, CAPITAL, will be disaggregated to assess the differential impact of each type of capital flow on the REER. Later, to assess the effectiveness of exchange rate policy as a hedge against real appreciation due to capital inflows, we will add into equation 8, the error correction equilibrium representation, an exchange rate flexibility variable and its cross-term (with capital inflow variable).

The measure of flexibility is then crucial. The spectrum of exchange rate regime choices is much more complex than suggested by the de jure classification. We approximate the flexibility of the exchange rate using an index based on the idea of exchange market pressure (EMP). The degree of exchange market pressure (EMP1) is derived from a relationship between the nominal exchange rate and relative foreign reserves:11

EMP1=%Δei,t/(%Δei,t+%Δfi,t),

where:

  • Δei,t=abs(eri,teri,t1eri,t1) erit is the nominal exchange rate of country i currency with the US dollar during year t; abs denotes the absolute value; and %Δei,t is the relative variation of the nominal exchange rate (Δei t) expressed as a percentage.

  • Δfi,t=abs(RESi,tRESi,t1)MBi,t1 RESi,t represents reserve assets, and MBi,t the monetary base in country i during year t.

In the hypothetical case of a pure floating system with no intervention on reserves (Δf = 0), the EMP index is equal to 1, reflecting maximum flexibility, with the exchange rate allowed to float freely. Changes in the EMP index reflect only changes in the nominal exchange rate. With a hard peg, the exchange rate is constant (Δe = 0) and the EMP index is equal to 0. Changes in the index reflect only changes in reserves through monetary authorities’ interventions. Intermediate cases indicate less exchange rate flexibility or more intervention in the foreign exchange market. More volatility of foreign reserves reduces the EMP. This suggests that the monetary authorities are using foreign reserves to limit variation in the12 nominal exchange rate. An alternative measure of the EMP index (EMP2) is to subtract the change of foreign exchange reserves from the change in nominal exchange rate as follows:

EMP2=Δei,tΔfi,t.

During previous episodes of capital inflows (before the debt and Asian crises), high flexibility of the exchange rate reflected large current account deficits. The wave of capital inflows that began early in the 2000s, however, is associated with a lessening of exchange rate flexibility, particularly in 2005, reflecting policy intervention with reserve accumulation (Figure 3).

Figure 3.
Figure 3.

Index of Exchange Rate Flexibility

Citation: IMF Working Papers 2011, 009; 10.5089/9781455211876.001.A001

The dataset consists of annual observations for 42 developing countries for 1980–2006. While data availability guided the choice of countries, the sample gives representative coverage of developing countries by including emerging and low-income countries as well as countries from the main developing regions.13

V. Econometric Results

Before presenting the results of the cointegration analysis, we first validate that the variables are nonstationary and cointegrated. Table A.3 presents the unit root tests on the REER and other variables. These tests confirm that almost all variables are nonstationary and could be considered as integrated of order one. As a second step, we test whether there is a long-term relationship between the variables of the baseline specifications. Following Pedroni (2000), various cointegration tests (Panel rho, Panel ADF, Group rho, Group ADF, etc.) confirm the existence of a cointegrating vector in all cases. The analysis focuses first on the effect of private capital flows on the real exchange rate and second on the importance of exchange rate flexibility.

A. Composition of Capital Inflows and the Real Exchange Rate

Using the pooled mean group estimator, Table 1 presents the long-run coefficients that are of interest to us. With cointegration analysis, the potential endogeneity between the real exchange rate and the fundamentals does not affect the long-run coefficients: The adjustment term is always negative and significant, indicating that there is no omitted variable bias. Hausman tests do not reject the long-term homogeneity of coefficients at the 1 percent significance level. This result suggests that the pooled mean group estimator might be preferred to the mean group estimator that supposes heterogeneity in both short-term and long-term coefficients.

Table 1.

Composition of Capital Inflows and the Real Exchange Rate

article image
EC refers to the error correction term. Only long-run coefficients are reported.* significant at 10%; ** significant at 5%; *** significant at 1%All specifications include a maximum of one lag. Numbers in parentheses are absolute t-statistics. Numbers in brackets for the Hausman and co-integration tests are p-values. For co-integration tests, the null hypothesis is the absence of co-integration. The null hypothesis for the Hausman test is the restriction of the homogeneity of long-term coefficients.

First, the estimations present the impact of aggregated capital inflows on the REER. Next, the impacts of public and private flows on the REER are separately estimated, and the effects of different components of private capital flows on the REER are analyzed.

The results show that capital inflows are positively associated with the REER. A 1 percentage point increase in the ratio of total capital inflows to GDP implies an 0.13 percent appreciation of the REER. The real appreciation due to public flows is statistically higher14 (x2 (1) = 8.50 [0.003]) than the real appreciation due to private flows (Table 1, column 2). This could suggest that private flows are used more for investments that increase the productive capacity of the economy and public flows are more directed to government consumption, mainly in the nontradable sector.

The last column of Table 1 shows how different components of private capital flows affect the REER. Public flows still have a significant appreciation effect but a component of private flows, portfolio investments, cause the most real appreciation. The highest level of appreciation from portfolio investments is statistically significant compared to the effect of FDI (x2(1) = 26.7 [0.000]), private transfers (x2(1) = 46.4 [0.000]), and bank loans (x2(1) = 33.8 [0.000]). A 1 percentage point increase in the ratio of portfolio investments to GDP is associated with a 7.8 percent appreciation of the REER. Compared to other private flows, portfolio investments are more volatile and speculative—something generally associated with macroeconomic instability and no improvement of productivity.

The real appreciation stemming from FDI is about one-seventh of that induced by portfolio investments. FDI is a more stable flow than portfolio investment and increases productive capacity through transfers of technology and know-how. It is primarily for investment purposes and could lead to importation of new machinery and equipment, which has limited impact on the REER.

Loans from commercial banks are also associated with REER appreciation, to a degree statistically similar to the real appreciation due to FDI (x2(1) = 0.2 [0.65]). A 1 percentage point increase in FDI or bank loans leads the REER to appreciate by about 1 percent. One could expect bank loans to have a higher appreciation effect because they are more intermediated by the domestic banking system. The results suggest that bank loans could be directed, to some extent, to investment financing like FDI, improving productive capacity. In this case, the inflation potential of bank loans could be similar to that of FDI, even though spillover effects are not associated with bank loans.

Private transfers appear to have the least effect on REER appreciation: a 1 percentage point increase leads to just an 0.3 percent appreciation of the REER. This result could justify viewing remittances as more countercyclical: By helping households to smooth their consumption during difficult periods, remittances help keep the economy stable by avoiding an acute depreciation of the exchange rate that could follow losses of foreign exchange during a macroeconomic shock.

With respect to other macroeconomic fundamentals, terms of trade and trade openness are significant, with the expected sign. A 10 percent increase in the terms of trade appreciates the REER by almost 4 percent. More liberalized trade is associated with REER depreciation: A 10 percent increase in trade openness leads to a real depreciation of about 1 percent. These results are similar to those previously found in the literature (Chen and Rogoff, 2003; Cashin, Cépedes, and Sahay, 2004; Lee, Milesi-Feretti, and Ricci, 2008; Saborowski, 2009). The Balassa-Samuelson effect captured by relative GDP per capita is not always significant, although it has the expected sign. This could be because GDP per capita, though it is widely used, is a poor proxy for the Balassa-Samuelson effect. The results are not significantly different for low-income countries (Table A.4).

The speed of the adjustment reflected by the coefficient of convergence is about −0.2. The movements of the REER within a year correct about a fifth of the gap between the REER and equilibrium REER as determined by the fundamentals. Therefore, the half-life of an REER deviation from the long-term equilibrium value is about three years.

B. Exchange Rate Regime and the Real Exchange Rate

Using the index of exchange rate flexibility based on EMP1, this study shows that a more flexible exchange rate helps to dampen REER appreciation stemming from capital inflows.15 The result is robust for low-income countries (Table 2).

Table 2.

Capital Inflows, Exchange Rate Flexibility, and the REER

article image
EC refers to the error correction term.All specifications include a maximum of one lag.* significant at 10%; ** significant at 5%; *** significant at 1%Numbers in parentheses are absolute t-statistics. Numbers in brackets for the Hausman and the co-integration tests are p-values. For co-integration tests, the null hypothesis is the absence of co-integration. The null hypothesis for the Hausman test is the restriction of long-term coefficient homogeneity.

As a robustness test, we present in Table A.5 the results obtained from regressions with an alternative measure of exchange rate flexibility. Ilzetzki, Reinhart, and Rogoff (2008) classify exchange rate regimes in 15 categories according to their de facto flexibility (Table A.6.). Whether we use this alternative measure or the index EMP2 defined above, the results confirm that exchange rate flexibility reduces real appreciation due to capital inflows.

Moreover, after the Asian financial crisis, developing countries, particularly in Asia, began to accumulate significant reserves for precautionary reasons. We control for changes in reserves that do not reflect management of exchange rate volatility. We thus define an additional measure of exchange rate flexibility using the difference between the level of reserves and their trend value, obtained using the Hodrick-Prescott method (Filtered Reserve). This allows us to capture changes in reserves due only to management of exchange rate volatility and not to other reasons, such as precautionary savings. The exchange rate flexibility index is also defined using the nominal effective exchange rate with each country’s top 10 trading partners, as in the definition of the REER. The results are robust with these alternative definitions of exchange rate flexibility (Table A.5). In all cases, exchange rate flexibility helps to dampen the real appreciation effect of capital inflows.

VI. Conclusion

This paper has analyzed the effect on the REER of different components of private capital inflows and has assessed the potential of exchange rate flexibility as a hedge against real appreciation.

Using the pooled mean group estimator (Pesaran, 1999), which considers long-term homogeneity in the behavior of the REER across countries while allowing for short-term heterogeneous shocks, we show that private and public capital inflows are associated with REER appreciation. Disaggregating private capital inflows shows that the appreciation effect of private flows differs by type of flow. Portfolio investments, which are more volatile, have the highest appreciation effect, followed by FDI and bank loans. Since these flows are potentially related to an increase in productive capacity, the real appreciation associated with FDI and bank loans is barely one-seventh of the real appreciation due to portfolio investments. Private transfers (mainly remittances) are the flows that have the least appreciation effect. This may suggest that remittances are more counter- than procyclical. Private transfers could help countries to offset the real depreciation of their exchange rate during periods of economic slowdown.

Countries often implement policies to reduce or avoid the loss of competitiveness associated with the REER appreciation that follows capital inflows. We have assessed the effectiveness of exchange rate flexibility policy, one of the main macroeconomic tools available to countries facing significant capital inflows. Using a de facto measure of exchange rate flexibility, we find that allowing the exchange rate more flexibility helps dampen real appreciation due to capital inflows. This result does not change significantly when alternative measures of exchange rate flexibility are used. The potential endogeneity bias calls for caution when interpreting this result as a causal relationship rather than a correlation.

When implementing policies to attract capital flows, developing countries should consider that a significant REER appreciation might destabilize macroeconomic management. Particular attention should be given to short-term flows, such as portfolio investments, which have a considerable real appreciation effect compared to other types of capital flow. Resisting nominal appreciation of the exchange rate through intervention in the foreign exchange market does not prove to be useful for avoiding a real appreciation. Allowing the exchange rate some flexibility would help to cure appreciation stemming from capital inflows and avoid a significant loss of competitiveness.

Capital Flows, Exchange Rate Flexibility, and the Real Exchange Rate
Author: Jean-Louis Combes, Patrick Plane, and Mr. Tidiane Kinda
  • View in gallery

    External Financing in Developing Countries

    (In percent of GDP, sample of 42 countries analyzed)

  • View in gallery

    The Real Exchange Rate and Capital Inflows

    (Unweighted mean for a panel of 42 countries)

  • View in gallery

    Index of Exchange Rate Flexibility