This section describes empirical studies of the link between capital account liberalization and macroeconomic performance.34 It emphasizes measurement issues and the need to recognize the endogeneity of controls, and reviews the statistical evidence on the macroeconomic effects of capital controls and capital account liberalization.
Measurement Issues
A significant obstacle to cross-country studies of capital account liberalization is the absence of a clear measure of the degree of liberalization and the intensity of controls. Most studies have relied on dummy variables constructed from information provided in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (International Monetary Fund, various issues). The most frequently used proxy is a bivariate index of restrictions on payments for capital transactions. Unfortunately, this variable does not measure the intensity of controls, and it primarily captures restrictions on capital outflows (because it refers to resident-owned funds only).35 To overcome the first shortcoming, some studies have also used information on separate exchange rates for some or all capital account transactions (multiple currency practices) and surrender of export proceeds, either separately or combined into a single index of the intensity of controls. A few studies (see, for example, Quinn, 1997; Tamirisa, 1998) have drawn on the information contained in the country-by-country descriptions of foreign exchange restrictions to construct more nuanced indices. Still others have used onshore-offshore or covered interest rate differentials to infer the effectiveness and intensity of controls (see Dooley 1996b for a survey). However, reflecting problems of data availability for many countries, measures based on interest differentials have been used mainly in studies employing high-frequency data for industrial countries. Most of them focus on the effectiveness of controls rather than on their determinants (see Appendix IV).
Determinants of Restrictions
The incidence of controls is not random. Insofar as countries with particular macroeconomic and financial characteristics are especially prone to adopt controls, there is the danger that observers may incorrectly interpret those characteristics as “effects” of controls, when in fact the causality runs in the other direction. It is thus important to analyze the decision to adopt capital account restrictions.
A limited number of studies taking a political economy approach have examined the conditions that make the maintenance of controls or the liberalization of the capital account more likely. Epstein and Schor (1992) are representative of this genre. Using dummy variables from the Annual Report on Exchange Arrangements and Exchange Restrictions for the Organization for Economic Cooperation and Development (OECD) countries, they find that restrictions on capital account transactions are more likely to be imposed in countries with strong left-wing parties and where the central bank is not independent. Alesina, Grilli, and Milesi-Ferretti (1994) and Grilli and Milesi-Ferretti (1995) adopt a similar measure of controls; the first study uses panel data for OECD countries, and the second considers a wider sample of industrial and developing economies. Both find that countries with a fixed or managed exchange rate, low per capita incomes, and high ratios of government consumption to GDP are more likely to maintain controls. Countries with more independent central banks and balanced current accounts are less likely to maintain such controls. Although their measure of the presence of controls does not capture the intensity of the policy, Alesina and Milesi-Ferretti (1995) show that the same results obtain when the capital controls dummy is replaced with a broader measure of restrictions, including multiple exchange rates, surrender of export proceeds, and current account restrictions.
Johnston and others (1998) construct a more detailed measure of controls that exploits the disaggregated information provided in the most recent issue of the Annual Report on Exchange Arrangements and Exchange Restrictions.36 They find that the intensity of capital controls is negatively correlated with economic development and positively correlated with the level of tariff barriers, the black market premium, and the volatility of the exchange rate. No attempt is made, however, to infer the direction of causality. Quinn and Inclán (1997) construct measures of financial openness that combine proxies for current and capital account restrictions and test for the importance of political and structural determinants of openness. They document the trend toward international financial liberalization and find that financial openness is higher in countries with more independent central banks and lower in countries with left-wing governments.37 Finally, Lemmen and Eijffinger (1996) use the onshore-offshore interest differential for 11 OECD countries to proxy for the intensity of controls and find that these are positively related to the domestic rate of inflation, the degree of political instability, and—somewhat surprisingly—the level of investment.38
Macroeconomic Effects
A useful starting point for reviewing this literature is Eichengreen, Rose, and Wyplosz (1996b). These authors ask whether in a sample of industrial countries, there are differences in the behavior of key macroeconomic and policy variables in the presence or absence of capital controls. They separate periods of speculative turbulence, defined as episodes characterized by abnormally large declines in foreign exchange reserves, increases in interest rates, and/or a depreciation of the exchange rate, from tranquil periods. Differences between crises occurring in the presence or absence of controls are evident in the behavior of inflation, money growth, and trade imbalances, all of which are higher in crises occurring in the presence of controls. Such differences are even more noticeable for the observations from tranquil periods: greater real overvaluation, larger budget and trade deficits, and faster growth of money and credit are evident in countries with controls. These findings can be interpreted in two (not incompatible) ways: first, that capital controls have a significant impact on macroeconomic policies and outcomes; and second, that they are imposed to support, at least temporarily, policies that would be unsustainable under free capital mobility.
Other studies, while not distinguishing turbulent and tranquil periods, also focus on the effects of capital controls and liberalization on external variables, inflation, and growth. Johnston and Ryan (1994) test whether capital controls have significantly affected the level and composition of international capital flows in a sample of 52 industrial and developing countries in the 1980s and 1990s. They estimate the determinants of private capital flows and test whether there are differences between restricted and liberalized regimes. For industrial countries, they find that the dismantling of controls has been associated with significant changes in the volume and composition of private capital flows, but they find no evidence of such changes in developing countries, and attribute this last result to evasion.
Citing the experiences of the United Kingdom, Italy, New Zealand, and Spain, Bartolini and Drazen (1997a) and Labán and Larraín (1997) emphasize that the liberalization of outflows often increases inflows. Bartolini and Drazen (1997b) document the positive correlation between restrictions on outflows in developing countries and real interest rates in industrial countries: when industrial country interest rates decline, developing countries tend to remove restrictions on the capital outflows.39 This suggests that the liberalization of capital accounts in the 1990s has been sustained, at least in part, by the persistence of low interest rates in Japan and the United States.40
Alesina, Grilli, and Milesi-Ferretti (1994) and Grilli and Milesi-Ferretti (1995) find that countries imposing controls tend to have higher inflation and greater seigniorage revenue but lower real interest rates. They do not find correlation between capital controls and the rate of economic growth, however.41 Similar results are reported by Rodrik (1998), who finds for a sample of developing countries that the average rate of economic growth during 1975–89 is uncorrelated with the number of years countries had capital controls in place.42 He makes no attempt to control for endogeneity of restrictions, although he argues that, insofar as capital account liberalization generally occurs in periods of good economic performance, the fact that no relation is found may even signal a negative underlying effect of capital controls on growth.43
A limitation of all these studies is that the dummy variable for controls does not provide a measure of their intensity. As noted above, Quinn (1997) addresses this problem by constructing an index of financial and capital account openness for a set of 64 industrial and developing countries. After controlling for initial income, education, and political instability, he analyzes whether changes in this index and its components are correlated with economic growth. He finds a strong positive correlation between capital account liberalization and growth that is robust to changes in model specification. He also finds that capital account liberalization is positively correlated with the ratio of corporate tax revenue to GDP. However, a causal interpretation of these links is complicated by the fact that no effort is made to control for the endogeneity of capital account liberalization.44 Tamirisa (1998) also constructs an index of the degree of capital account restrictions for the year 1996 and finds that capital controls act as a barrier to trade in developing and transition economies.
A final set of studies focuses on whether restrictions on capital mobility limit a country’s ability to smooth consumption when hit by shocks to income (see, for example, Backus, Kehoe, and Kydland, 1992; and Obstfeld, 1994c). Lewis (1996, 1997), for example, tests whether official restrictions on international capital mobility help explain the lack of international risk sharing.45 Her tests are based on the idea that co-movements in consumption and output should be stronger in countries with tighter restrictions on capital flows, because residents of these countries will find it more difficult to smooth shocks to income by borrowing and lending internationally.46 She confirms that consumption and output growth co-vary more strongly in countries with restrictions on international capital flows than in countries without them and that these results hold for different foreign exchange restrictions individually (capital controls, surrender of export proceeds, and so forth). This finding suggests that consumers in countries with extensive international restrictions tend to be more liquidity constrained than consumers in countries with weaker international restrictions.
Financial Development and Growth
A crucial issue in analyzing the link between financial development and growth is measuring financial development itself. Among the measures used in the literature are the ratio of liquid liabilities of the financial system to GDP, the ratio of gross claims on the private sector to GDP, and the share of domestic credit intermediated by deposit banks.1 Studies focusing on the role of liquidity have also used the “value-traded ratio” (the ratio of total value of shares traded on a country’s stock exchanges to GDP) and the turnover ratio (the total value of shares traded on a country’s stock exchanges divided by stock market capitalization) (Levine and Zervos, 1998).
These studies generally conclude that there exists a strong link between financial development and growth (see Levine, 1997, for a survey.) This result is robust to different ways of measuring financial development and holds after initial income, human capital, measures of monetary, trade, and fiscal policy, and political instability are controlled for. Establishing the direction of causality is problematic, however, because financial development is itself affected by economic performance. King and Levine (1993a) adopt a post hoc, ergo propter hoc approach to this problem, confirming that financial development in a particular year is a good predictor of growth performance in subsequent years. Rajan and Zingales (1998) compare growth performance in sectors subject to different degrees of dependence on external finance. If finance stimulates growth, then finance-dependent sectors should grow relatively fast in countries where financial development is more advanced.2 They find evidence that sectors relying more heavily on external finance do indeed grow faster in countries with more developed financial systems.
1 While all these measures have shortcomings, more detailed measures of financial development are not available for enough countries. 2 At the same time, it is less likely that fast growth in these sectors is inducing financial development, and therefore the endogeneity problem in the estimation is reduced.Qualifications and Implications
These studies provide useful insights into the consequences of capital account liberalization. At best, however, they provide mixed support for the hypothesis that capital account liberalization has a positive impact on economic growth. Existing studies provide weaker evidence of a positive effect on growth for capital account liberalization in particular than for financial development more generally. (On the effects of financial development generally, see Box 4 on page 19.)
It is important to bear in mind several reasons why one might want to be cautious in drawing strong conclusions from these studies. First, insofar as existing studies have failed to fully sort out the direction-of-causality problem, the impact of capital account liberalization on growth may be disguised.47 Second, the dummy variables used in these studies may simply be too crude to capture the growth effects of capital account liberalization.48 Third, whereas the measures of capital controls used in these studies reflect mainly restrictions on capital outflows, much of the current policy discussion focuses on prudential measures likely to have their most immediate impact on capital inflows.49 In this respect, earlier empirical analyses may have only limited implications for the debate over the effectiveness of policies aimed at capital inflows.
Finally, the discussion so far says nothing about one potential macroeconomic effect of removing capital controls: the danger that liberalizing the capital account before the domestic financial system has been suitably strengthened can encourage volatile capital flows that create an environment conducive to the development of serious economic problems and, potentially, financial crises.
While a large number of Country-specific studies have examined the impact of capital account liberalization on the domestic financial system (see, for example, Diaz-Alejandro, 1985), and others have focused on the impact of capital account liberalization on economic growth, to our knowledge there exists no systematic cross-country study of the two-way interaction between domestic financial development and capital account liberalization on the one hand and economic performance on the other. The survey here necessarily reflects this limitation of the literature.
This is apparent in the data for countries that had controls on capital inflows, such as Germany in the early 1970s; Germany had no capital controls according to this measure.
Unfortunately, the nature of the data limits the analysis to a pure cross section for 1996.
This last correlation vanishes in the 1980s, however.
A possible explanation is that capital controls keep domestic interest rates low, thereby stimulating domestic investment.
Conversely, when interest rates rise in the industrial countries, controls in the developing world tend to be tightened.
It is important to emphasize the qualifying phrase “in part” when referring to the capital account, and not just because other factors could have been in play. In addition, the relationship does not hold so clearly for policy on capital inflows. That is, a number of countries—brazil, Chile, and Colombia among them—tightened their policies on at least some capital inflows during the same period of low interest rates.
After controlling for growth determinants such as initial income, the level of education, and the size of government.
After controlling for other growth determinants, among which an index of the quality of government institutions.
Rodrik also reports no (conditional) correlation between the number of years capital controls were in place and either the investment rate or the rate of inflation. A possible reconciliation of this last finding and those of Grilli and Milesi-Ferretti (1995) is that the impact of capital account liberalization on the rate of inflation is stronger in time series than in the cross section. This again raises the issue of endogeneity insofar as controls tend to be imposed or strengthened in periods of economic distress when growth is declining and inflation is rising.
Clearly, more research is needed to establish the generality of these results. Nevertheless, this study suggests that, to detect an impact of capital account liberalization on economic performance, one must move beyond binary dummy variables and construct measures that proxy, however imperfectly, the intensity of capital account restrictions.
Theoretical work by Obstfeld (1994a) and others suggests that the potential welfare gains from international risk sharing can be very large.
These tests are applications to international risk sharing of standard tests for the existence of liquidity constraints across consumers. See, for example, Campbell and Mankiw (1989) and Zeldes (1989).
As mentioned earlier, Rodrik (1998) argues that insofar as liberalization is associated with faster growth, simultaneity bias, to the extent that it remains, reinforces the skeptical view.
As already mentioned, these dummies do not provide a measure of the intensity of controls.
See, for example, Folkerts-Landau and Ito (1995). Both these considerations point to the need for complementing the research surveyed in this section with a careful study of recent experiences of countries with and without restrictions on capital inflows (see, for example, Calvo, Leiderman, and Reinhart, 1993; Schadler and others, 1993; and Férnandez-Arias and Montiel, 1996).