III Financial Integration and Economic Growth
- Ayhan Kose, Kenneth Rogoff, Eswar Prasad, and Shang-Jin Wei
- Published Date:
- September 2003
Theoretical models have identified a number of channels through which international financial integration can help to promote economic growth in the developing world. It has proven difficult, however, to empirically identify a strong and robust causal relationship between financial integration and growth.
Potential Benefits of Financial Globalization in Theory
In theory, there are a number of direct and indirect channels through which embracing financial globalization can help enhance growth in developing countries. Figure 3.1 provides a schematic summary of these possible channels. These channels are interrelated in some ways, but this delineation is useful for reviewing the empirical evidence on the quantitative importance of each channel.
Figure 3.1.Channels Through Which Financial Integration Can Raise Economic Growth
Augmentation of Domestic Savings
In principle, North-South capital flows benefit both groups. They allow for increased investment in capital-poor countries while providing a higher return on capital than is available in capital-rich countries. This effectively reduces the risk-free rate in the developing countries.
Reduction in Cost of Capital Through Better Global Allocation of Risk
International asset-pricing models predict that stock market liberalization improves the allocation of risk (Henry, 2000a; and Stulz, 1999a and b). First, increased risk-sharing opportunities between foreign and domestic investors might help to diversify risks. Second, the increased ability to diversify risk, in turn, would encourage firms to take on more total investment, thereby enhancing growth. Third, as capital flows increased, the domestic stock market would become more liquid, which could further reduce the equity risk premium, thereby lowering the cost of raising capital for investment.
Transfer of Technological and Managerial Know-How
Financially integrated economies seem to attract a disproportionately large share of FDI inflows, which have the potential to generate technology spillovers and serve as a conduit for passing on better management practices. These spillovers can raise aggregate productivity and, in turn, boost economic growth (Borensztein, De Gregorio, and Lee (1998); MacDougall (1960); and Grossman and Helpman (1991b)).
Stimulation of Domestic Financial Sector Development
It has already been noted that international portfolio flows can increase the liquidity of domestic stock markets. Increased foreign ownership of domestic banks can also generate a variety of other benefits (Levine, 1996; Caprio and Honohan, 1999). First, foreign bank participation can facilitate access to international financial markets. Second, it can help improve the regulatory and supervisory framework of the domestic banking industry. Third, foreign banks often introduce a variety of new financial instruments and techniques and also foster technological improvements in domestic markets. The entry of foreign banks tends to increase competition, which, in turn, can improve the quality of domestic financial services as well as improve allocative efficiency.
Promotion of Specialization
The notion that specialization in production may increase productivity and growth is intuitive. Without any mechanism for risk management, however, a highly specialized production structure will produce high output volatility and, hence, high consumption volatility. Concerns about exposure to such increases in volatility may discourage countries from taking up growth-enhancing specialization activities; the higher volatility will also generally imply lower overall saving and investment rates. In principle, financial globalization could play a useful role by helping countries to engage in international risk sharing and thereby reduce consumption volatility. This point will be taken up again in the next section. Here it should just be noted that risk sharing would indirectly encourage specialization, which, in turn, would raise the growth rate. This logic is explained by Brainard and Cooper (1968), Kemp and Liviatan (1973), Ruffin (1974), and Imbs and Wacziarg (2003). Among developed countries and across regions within given developed countries, there is, indeed, some evidence that better risk sharing is associated with higher specialization (Kalemli-Ozcan, Sørensen, and Yosha, 2001).
Commitment to Better Economic Policies
International financial integration could increase productivity in an economy through its impact on the government’s ability to credibly commit to a future course of policies. More specifically, the disciplining role of financial integration could change the dynamics of domestic investment in an economy to the extent that it leads to a reallocation of capital toward more productive activities in response to changes in macroeconomic policies. National governments are occasionally tempted to institute predatory tax policies on physical capital. The prospect of such policies tends to discourage investment and reduce growth. Financial opening can be self-sustaining and constrains the government from engaging in such predatory policies in the future, since the negative consequences of such actions are far more severe under financial integration. Gourinchas and Jeanne (2003a) illustrate this point in a theoretical model.
A country’s willingness to undertake financial integration could be interpreted as a signal that it is going to practice more friendly policies toward foreign investment in the future. Bartolini and Drazen (1997a) suggest that the removal of restrictions on capital outflows can, through its signaling role, lead to an increase in capital inflows. Many countries—including Colombia, Egypt, Italy, Mexico, New Zealand, Spain, the United Kingdom, and Uruguay—have received significant capital inflows after removing restrictions on capital outflows.15
On the surface, there seems to be a positive association between embracing financial globalization and achieving economic development. In general, industrial countries are more financially integrated with the global economy than developing countries. Thus, embracing globalization is apparently part of being economically advanced.
Within the developing world, MFI economies grew faster than LFI economies over the last three decades. From 1970 to 1999, average per capita output rose almost threefold in the group of MFI developing economies, six times greater than the average increase experienced by LFI economies. This pattern of higher growth for the former group applies over each of the three decades and also to both consumption and investment growth.
There are two problems, however, with concluding from this data pattern that financial integration has a positive effect on growth. First, this hypothesis may not be able to stand up to closer scrutiny. Second, these observations may reflect only an association between international financial integration and economic performance rather than a causal relationship. In other words, these observations do not rule out the possibility that there is reverse causation: countries that manage to enjoy robust growth may also choose to engage in financial integration even if financial globalization does not directly contribute to faster growth in a quantitatively significant way.
To obtain an intuitive impression of the relationship between financial openness and growth, Table 3.1 presents a list of the fastest-growing developing economies during 1980-2000 and a list of the slowest-growing (or, rather, the fastest-declining) economies during the same period. Some countries underwent financial integration during this period, especially in the latter half of the 1990s.16 Therefore, any result based on total changes over this long period should be interpreted with caution. Nonetheless, several features of the table are noteworthy.
Per Capita GDP1
Per Capita GDP1
|2||Korea, Rep. of||234.0||Yes||Niger||–37.8||No|
|7||Hong Kong SAR||114.5||Yes||Venezuela||–17.3||Yes/No|
Growth rate of real per capita GDP, in constant local currency unit.
Growth rate of real per capita GDP, in constant local currency unit.
An obvious observation that can be made from the table is that financial integration is not a necessary condition for achieving a high growth rate. China and India have achieved high growth rates despite somewhat limited and selective capital account liberalization. For example, although China became substantially more open to foreign direct investment, it was not particularly open to most other types of cross-border capital flows. Mauritius and Botswana have managed to achieve very strong growth rates during the period although they are relatively closed to financial flows.
The second observation that can be made is that financial integration is not a sufficient condition for a fast economic growth rate either. For example, Jordan and Peru became relatively open to foreign capital flows, yet their economies experienced negative rather than positive growth, during the period. Nonetheless, Table 3.1 also suggests that declining economies are more likely to be financially closed, though the direction of causality is not clear.
This way of looking at countries with extreme growth performances is only informative up to a point; it needs to be supplemented by a comprehensive examination of the experience of a broader set of countries using a more systematic approach to measuring financial openness. To illustrate this relationship more broadly, Figure 3.2 presents a scatter plot of the growth rate of real per capita GDP against the increase in financial integration over 1982–97. There is essentially no association between these variables. Figure 3.3 presents a scatter plot of these two variables after taking into account the effects of a country’s initial income, initial schooling, average investment-to-GDP ratio, political instability, and regional location. Again, the figure does not suggest a positive association between financial integration and economic growth. In fact, this finding is not unique to the particular choice of time period or country coverage, as is reflected in a broad variety of other research papers on the subject.
Figure 3.2.Increase in Financial Openness and Growth of Real Per Capita GDP
Source: IMF staff calculations based on the data documented in Wei and Wu (2002b).
Notes: Capital account openness is measured as (gross private capital inflows + gross private capital outflows)/GDR. Coef = 0.002; Robust SE = 0.003; and t-statistic = 0.67.
Figure 3.3.Increase in Financial Openness and Growth of Real Per Capita GDP: Conditional Relationship, 1982-97
Source: IMF staff calculations based on the data documented in Wei and Wu (2002b).
Notes: Capital account openness is measured by (gross private capital inflows + gross private capital outflows)/GDR. Coef = 0.006; Robust SE = 0.004; and t-statistic = 1.42.
A number of empirical studies have tried to systematically examine whether financial integration contributes to growth using various approaches to dealing with the difficult problem of proving causation. Table 3.2 summarizes the 14 most recent studies on this subject.17 Three of these papers report a positive effect of financial integration on growth. However, the majority of the papers tend to find no effect or a mixed effect for developing countries. This suggests that if financial integration has a positive effect on growth, it is probably neither strong nor robust (see Box 3.1).18
|Study||Number of |
|Years Covered||Effect on |
|Alesina, Grilli, and Milesi-Ferretti (1994)||20||1950–89||No effect|
|Grilli and Milesi-Ferretti (1995)||61||1966–89||No effect|
|Kraay (1998)||117||1985–97||No effect or, at best, mixed|
|Rodrik (1998)||95||1975–89||No effect|
|Klein and Olivei (2000)||Up to 92||1986–95||Positive|
|Arteta, Eichengreen, and Wyplosz (2001)||51-59||1973–92||Mixed|
|Bekaert, Harvey, and Lundblad (2001a)||30||1981–97||Positive|
|Edwards (2001)||62||1980s||No effect for poor countries|
|O’Donnell (2001)||94||1971–94||No effect or, at best, mixed|
|Reisen and Soto (2001)||44||1986–97||Mixed|
|Edison, Klein, Ricci, and Sløk (2002)||Up to 89||1973–95||Mixed|
|Edison, Levine, Ricci, and Sløk (2002)||57||1980–2000||No effect|
Of the papers summarized in Table 3.2, the one by Edison, Levine, Ricci, and Sløk (2002) is perhaps the most thorough and comprehensive in terms of measures of financial integration and in terms of empirical specifications. These authors measure a country’s degree of financial integration both by the government’s restrictions on capital account transactions as recorded in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions and by the observed size of capital flows crossing the border, normalized by the size of the economy. The dataset in that paper goes through 2000, the latest year analyzed in any existing study on this subject. Furthermore, the authors also employ a statistical methodology that allows them to deal with possible reverse causality—that is, the possibility that any observed association between financial integration and growth could occur because faster-growing economies are more likely to choose to liberalize their capital accounts. After a battery of statistical analyses, that paper concludes that, overall, there is no robustly significant effect of financial integration on economic growth.
Why is it so difficult to find a strong and robust effect of financial integration on economic growth for developing countries, when the theoretical basis for this result is apparently so strong? Perhaps there is some logic to this outcome after all. A number of researchers have now concluded that most of the differences in per capita income across countries stem not from differences in capital-labor ratios but from differences in total factor productivity, which, in turn, could be explained by soft factors or social infrastructure such as governance, the rule of law, and respect for property rights.19 In this case, although financial integration may open the door for additional capital to come in from abroad, it is unlikely, by itself, to offer a major boost to growth. In fact, if domestic governance is sufficiently weak, financial integration could cause an exodus of domestic capital and, hence, lower the growth rate of an economy.
Box 3.1.Effects of Different Types of Capital Flows on Growth
The cumulative evidence from the literature does not offer clear-cut and robust support for the notion that capital flows generically provide a quantitatively big boost to economic growth. There have been several studies, however, that suggest that different types of capital flows may have different effects.
Using data for the 1980s, De Mello (1999) reports evidence that FDI flows appear to promote economic growth in developing as well as Organization for Economic Cooperation and Development (OECD) countries. Borenzstein, De Gregorio, and Lee (1998) find that the positive effect of FDI can be detected when the recipient countries have a sufficiently high level of human capital.
FDI and other types of capital flows into developing countries started to pick up momentum in the 1990s, which makes it highly desirable to look at the evidence based on more recent data. Reisen and So to (2001) examine six types of capital flows: foreign direct investment, portfolio equity flows, portfolio bond flows, long-term bank credits, short-term bank credits, and official flows. They employ a dynamic panel regression framework to deal with potential endogeneity and missing variable problems and cover 44 countries over the period 1986–97. Of the six types of capital flows, only two, namely FDI and portfolio equity flows, are positively associated with subsequent economic growth rates.
Other studies have looked into the effects of different types of capital flows on domestic investment (and, hence, indirectly on growth). Bosworth and Collins (1999) analyzed such relationships using data covering 1979–95, focusing on variations within countries over time rather than variations across countries. These authors first removed the country means from the data and then regressed investment and savings shares on various forms of capital inflows (relative to GDP). They found that more FDI and bank lending are positively associated with increases in domestic investment. In contrast, the association between portfolio capital inflows and domestic investment, while positive, is not statistically significant. These authors made an attempt to deal with the possibility that capital flows are endogenous, meaning that capital flows and domestic investment can both be determined simultaneously by a common third factor.
The World Bank’s Global Development Finance (2001) replicated the Bosworth-Collins study using a dataset with more countries and a longer time period (1972-98). It found that the association between FDI (or other long-term capital inflows or bank lending) and domestic investment is stronger than between short-term debt and domestic investment. The association between portfolio capital and domestic investment is not statistically significant.
To summarize, across different recent studies surveyed here, FDI is one form of capital inflow that tends to be positively associated with domestic investment and domestic growth in a relatively consistent manner. Other forms of capital inflows could also have a positive relationship, but their effects tend to be less robust or less strong.
This logic can be illustrated using the results reported in Senhadji (2000). Over the period 1960 to 1994, the average growth rate of per capita output for the group of countries in sub-Saharan Africa was the lowest among regional groupings of developing countries. The difference in physical and human capital accumulation is only part of the reason why growth rates differ across countries. The gap in total factor productivity is the major element in explaining the difference in the growth rates.
Another possible explanation of why it is difficult to detect a causal effect of financial integration on growth is the costly banking crises that some developing countries have experienced in the process of financial integration. The results in Kaminsky and Reinhart (1999) suggest that a flawed sequencing of domestic financial liberalization, when accompanied by capital account liberalization, increases the chance of domestic banking crises and/or exchange rate crises. These crises are often accompanied by output collapses. As a result, the benefits from financial integration may not be evident in the data.20
It is interesting to contrast the empirical literature on the effects of financial integration with that on the effects of trade integration. Although there are some skeptics (Rodriguez and Rodrik, 2001), an overwhelming majority of empirical papers reach the conclusion that trade openness helps to promote economic growth. These studies employ a variety of techniques, including country case studies as well as cross-country regressions. In a recent paper that surveys all the prominent empirical research on the subject, Krueger and Berg (2002) conclude that “[v]aried evidence supports the view that trade openness contributes greatly to growth.” Furthermore, “[crosscountry regressions of the level of income on various determinants generally show that openness is the most important policy variable.”
Box 3.2.Do Financial and Trade Integration Have Different Effects on Economic Development? Evidence from Life Expectancy and Infant Mortality
As an alternative to examining the effect of openness on economic growth, this box asks whether trade and financial integration help to raise life expectancy and reduce infant mortality in developing countries and whether their effects are different.
There are three motivations for studying these questions. First, as life expectancy and infant mortality are important dimensions of a society’s well-being, they are interesting objects to look at in their own right. Second, data on income level or growth come from national accounts, so all studies on economic growth have to make use of similar data sources. In comparison, vital statistics come from an entirely different data source (that is, birth and death records) and are typically collected by different government agencies. They therefore offer an independent and complementary check on the effect of openness on the livelihood of people. Third, to compare income levels or growth rates across countries, it is necessary to make certain purchasing power parity (PPP) adjustments to nominal income. However, existing PPP adjustments may not be reliable (Deaton, 2001). In contrast, the definitions of life and death are consistent across countries, so vital statistics have a higher degree of comparability than the data on poverty, income, or income distribution.
Data on 79 developing countries over the period 1962-97 are examined. This dataset covers all developing countries for which the relevant data exist and for which changes in infant mortality and life expectancy are not dominated by large-scale wars, genocides, famines, or major outbursts of AIDS. Panel regressions with country fixed effects and dynamic panel regressions are employed to account for other factors that may affect health and to account for possible endogeneity of the openness variables.
The results, summarized in Figure 3.4, suggest that the effects of trade and financial openness are different. There is no positive and robust association across developing countries between a faster increase in financial integration and a faster improvement in a society’s health. By comparison, there are several pieces of evidence suggesting that higher trade integration is associated with a faster increase in life expectancy and a faster reduction in infant mortality. For example, an 11 percentage point reduction in the average statutory tariff rate—approximately equal to one standard deviation of the change in the statutory tariff rate over the 1962–97 period—is associated with between 3 and 6 fewer infants dying per thousand live births, even after controlling for the effects of changes in per capita income, average level of female education, and other factors.Source: Wei and Wu (2002b).
Figure 3.4.Differential Effects of Financial and Trade Integration on Improvements in Health
Source: IMF staff calculations based on Wei and Wu (2002b).
Notes: On the vertical axes in the top two panels is log life expectancy at birth (in years). On the vertical axes in the bottom two panels is infant mortality, defined as the number of infants who die before reaching their first birthday per 1,000 live births. Tariff rate refers to average statutory tariff rate, and financial integration is measured by (gross private capital inflows + gross private capital outflows) / GDP.
The differential effects between trade and financial integration are echoed in other empirical research (see Box 3.2). As an alternative to examining the effects on economic growth or level of income, one can examine the effects of trade and financial openness on a society’s health status. Using data on 79 developing countries, Wei and Wu (2002b) report several pieces of evidence suggesting that a faster increase in trade openness—especially when measured by the reduction in tariff rates—is associated with a faster increase in life expectancy and a faster reduction in infant mortality, even after one takes into account the effects of income, institutions, and other factors. In contrast, greater financial integration is not associated with a faster-improvement in a society’s health status. This suggests that, as may be seen in the growth literature, trade integration appears to play a more beneficial role than financial integration in developing countries.
The contrast between financial and trade openness may have important lessons for policies. While there appear to be relatively few prerequisites for deriving benefits from trade openness, obtaining benefits from financial integration requires several conditions to be met. This is discussed in more detail in Section V.
It is useful to note that there may be a complementary relationship between trade and financial openness.21 For example, if a country has severe trade barriers protecting some inefficient domestic industries, then capital inflows may end up being directed to those industries, thereby exacerbating the existing misallocation of resources. Thus, there is a concrete channel through which financial openness without trade openness could lower a country’s level of efficiency.
Of course, the lack of a strong and robust effect of financial integration on economic growth does not necessarily imply that theories that make this connection are wrong. One could argue that the theories are about the long-run effects, and most theories abstract from the nitty-gritty of institution building, governance improvement, and other soft factors that are necessary for the hypothesized channels to operate. Indeed, developing countries may have little choice but to strengthen their financial linkages eventually in order to improve their growth potential in the long run. The problem is how to manage the short-run risks apparently associated with financial globalization. Financial integration without a proper set of preconditions might lead to few growth benefits and more output and consumption volatility in the short run.