Information about Europe Europa
Chapter

VI How Does Financial Globalization Affect Stability and Growth?

Author(s):
Giovanni Dell'Ariccia, Paolo Mauro, Andre Faria, Jonathan Ostry, Julian Di Giovanni, Martin Schindler, Ayhan Kose, and Marco Terrones
Published Date:
December 2008
Share
  • ShareShare
Information about Europe Europa
Show Summary Details

Financial globalization has been argued to affect many aspects of economic performance—including long-run economic growth, the propensity to experience growth upturns or downturns, the sustainability of growth spells, the volatility of economic growth, the frequency of economic crises, and the depth and duration of output drops in the aftermath of crises. This section focuses on financial globalization's effects on three of these aspects, namely: macroeconomic volatility, crisis propensity, and economic growth1

A number of underlying mechanisms are likely to be involved in the transmission of financial globalization to economic volatility and growth:

  • Financial sector development. Well-developed domestic financial markets may be instrumental in moderating boom-bust cycles that could be triggered by sudden stops in financial flows (Aghion and Banerjee, 2005) and in efficiently allocating foreign financial flows to competing investment projects, thereby promoting economic growth (Aoki, Benigno, and Kiyotaki, 2006). Furthermore, access to international markets is not available to all members of society, and underdeveloped domestic financial systems may prevent the pooling of risk across agents (Levchenko, 2005).
  • Institutional quality. Better institutional quality helps to shift the composition of financial flows toward FDI and portfolio equity, thereby enhancing growth and macroeconomic stability benefits (Becker and others, 2007). Bordo and Meissner (2007) suggest that countries with stronger institutions (in addition to well-developed financial markets and prudent macroeconomic policies) enjoyed greater economic growth benefits from financial integration during the 1870–1913 period.
  • Sound macroeconomic policies. In the absence of a sound macroeconomic policy framework, international financial integration may lead to excessive borrowing and debt accumulation, thus increasing vulnerability to crisis.
  • Trade integration. A high degree of trade openness seems to be associated with fewer sudden stops and current account reversals. Trade integration may also facilitate recoveries from financial crises and mitigate their adverse growth effects (Edwards, 2005; and Calvo, Izquierdo, and Mejía, 2004).

Volatility and the Frequency of Crises

Following the Asian crisis, a presumption emerged in some policy circles that financial globalization would tend to exacerbate macroeconomic volatility in emerging market and developing countries, and increase vulnerability to sudden stops. The academic literature, however, has found generally inconclusive results on the issue (Kose and others, 2006). Empirical evidence presented below suggests that the relationship between financial integration and macroeconomic volatility (proxied here by consumption volatility) depends on a country's domestic financial development and the quality of its institutions, consistent with a “thresholds” view of the effects of financial integration.

Indeed, in the panel regression results reported in Table 6.1 and Figure 6.1, the estimated slope coefficient on de facto financial integration is positive and significant for countries with relatively weak perceived institutional quality and a relatively low degree of domestic financial development, whereas the impact is not significantly different from zero for countries with stronger institutions and more developed domestic financial systems.2 Equivalently, the positive relationship between financial integration and consumption volatility holds for countries with relatively poor institutional quality and low financial sector development; for countries over a certain threshold, the relationship is neutral and may even turn negative (more integration implying less volatility).

Table 6.1.Impact of Financial Integration on Consumption Volatility
Private CreditInstitutional Quality
(1)(2)
Financial integration0.03**0.06***
(0.01)(0.02)
Terms of trade volatility0.13***0.12**
(0.04)(0.05)
Trade openness0.04***0.03**
(0.01)(0.02)
ln(initial income per capita)0.03**0.02**
(0.01)(0.01)
Financial integration* private credit−0.02*
(0.01)
Private credit (percent of GDP)0.01
(0.02)
Financial integration* institutional quality−0.37***
(0.12)
Institutional quality (divided by 100)0.01
(0.15)
R2 adjusted0.140.14
N8176
Threshold1.1515.85
Sources: IMF, International Financial Statistics; and IMF staff estimates.Notes: Estimated by panel fixed effects (country and decade dummies) over 1965–2004, subject to data availability. The dependent variable is the standard deviation of the growth rate of consumption per capita over each decade. Financial integration is defined as total liabilities as percent of GDP. Dummy variables are included for each decade, but the estimated coefficients are not reported, for the sake of brevity. Robust standard errors are reported in parentheses. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively.
Sources: IMF, International Financial Statistics; and IMF staff estimates.Notes: Estimated by panel fixed effects (country and decade dummies) over 1965–2004, subject to data availability. The dependent variable is the standard deviation of the growth rate of consumption per capita over each decade. Financial integration is defined as total liabilities as percent of GDP. Dummy variables are included for each decade, but the estimated coefficients are not reported, for the sake of brevity. Robust standard errors are reported in parentheses. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively.

Figure 6.1.Financial Integration and Consumption Volatility

Sources: IMF, International Financial Statistics; and Political Risk Services, International Country Risk Guide.

Notes: Figure based on regression results reported in Table 6.1, which refer to the estimated impact of an increase in de facto financial globalization on consumption volatility, including an interaction effect for domestic financial development (or institutional quality). The solid line shows the impact (marginal effect) of an increase in total external liabilities on consumption volatility, as a function of the ratio of private credit to GDP (or institutional quality) at the different levels indicated along the horizontal axis. The dashed lines are the standard error bands around the estimated marginal effect. The histogram reports the percentage of countries in the sample at each given level of credit market development (or institutional quality) as of 2004, indicated along the horizontal axis. Institutional quality is the sum of three indices (law and order, bureaucratic quality, and absence of corruption), each of which ranges from 0 to 6.

Drawing on the regression results, it is possible to estimate thresholds for institutional quality and domestic financial development beyond which financial globalization's impact is no longer positive or no longer statistically significant. While the exact values of the thresholds need to be interpreted with caution, given the considerable uncertainty surrounding the estimates, based on average data over the period 2000–04, virtually all advanced countries and about one-third of emerging market countries meet the thresholds beyond which the estimated effect of financial integration on consumption volatility is insignificant. The developing countries in the sample are currently below the thresholds.

How large is the impact of financial globalization on consumption volatility for different groups of countries? One way to address this question is to hold the level of domestic financial development constant at a given level, and trace the impact of a change in financial globalization. Using this approach, for a country at the 25th percentile of the distributions of both financial development and financial integration (where average consumption volatility is about 6 percent), an increase in financial integration to the 75th percentile is associated with an increase in volatility of 1.4 percentage points. This effect becomes smaller and loses statistical significance as financial development increases. For example, for a country at the 75th percentile of financial development, the impact of financial integration on volatility is not statistically significant.

Turning now from volatility to crisis propensity, despite a widespread perception that financial globalization may lead to higher frequency of crises, existing empirical studies (surveyed in Kose and others, 2006) do not support the view that greater financial integration increases the likelihood of crisis. On the contrary, a majority of studies find that crises are, if anything, less frequent in financially open countries than in financially closed ones. This could of course be an outcome of self selection, in which countries less prone to crises will choose to open up, whereas more vulnerable countries might choose to remain closed. However, some studies suggest that, even taking into account the possibility that self-selection could result in estimation bias, the frequency of currency crises is not higher in more financially open countries (Glick, Guo, and Hutchinson, 2006).

Consistent with their role in the transmission of financial openness to macroeconomic volatility, thresholds also appear to influence the impact of financial openness on crisis propensity, with factors such as financial sector development, institutional quality, macroeconomic policy soundness, and trade openness playing key roles. Specifically, within a sample of countries with de facto open financial accounts (that is, above the median with respect to financial integration), countries above the median of the distribution for at least three of the four factors listed above experienced significantly lower crisis frequency between 1970 and 2004 compared with countries that were above the median for no more than two factors (Table 6.2).3 This suggests that threshold effects—at work in the case of the effects of financial globalization on macroeconomic volatility—also appear to be present in determining the interaction of financial integration and crisis risks.

Table 6.2.Countries with De Facto Open Financial Accounts: Frequency of Crises, 1970–2004
Above the Median in at Least Three out of Four of the FactorsNBanking CrisesCurrency CrisesDebt CrisesSudden Stops
Yes230.610.570.220.7
No190.740.89***0.53***0.89*
Sources: IMF, International Financial Statistics (IFS), and IMF staff estimates based on the sources and definitions of sudden stops, and banking, currency, and debt crises described in Becker and others (2007, Appendix I). A country has a currency crisis if the following three conditions hold at some point during a calendar year: (1) devaluation/depreciation rate of at least 25 percentage cumulative over a 12-month period; (2) devaluation/depreciation rate by at least 10 percentage points greater than in the preceding 12 months; and (3) a minimum of three years since last crisis; this definition was applied using IFS data. Sudden stops in capital flows are defined as a decline in financial flows by 5 percentage points of GDP.Notes: The factors are financial sector development, institutional quality, macroeconomic policies soundness, and trade openness. Frequency of crises: fraction of countries that had at least one crisis during the sample period. N is the number of countries in each group. One-sided test of equality of means: *significant at the 10 percent level; **significant at the 5 percent level; ***significant at the 1 percent level.
Sources: IMF, International Financial Statistics (IFS), and IMF staff estimates based on the sources and definitions of sudden stops, and banking, currency, and debt crises described in Becker and others (2007, Appendix I). A country has a currency crisis if the following three conditions hold at some point during a calendar year: (1) devaluation/depreciation rate of at least 25 percentage cumulative over a 12-month period; (2) devaluation/depreciation rate by at least 10 percentage points greater than in the preceding 12 months; and (3) a minimum of three years since last crisis; this definition was applied using IFS data. Sudden stops in capital flows are defined as a decline in financial flows by 5 percentage points of GDP.Notes: The factors are financial sector development, institutional quality, macroeconomic policies soundness, and trade openness. Frequency of crises: fraction of countries that had at least one crisis during the sample period. N is the number of countries in each group. One-sided test of equality of means: *significant at the 10 percent level; **significant at the 5 percent level; ***significant at the 1 percent level.

Evidence based on case studies (summarized in Appendix III) also suggests that, among financially integrated countries, those with sound macroeconomic and fiscal policies and well-developed and regulated financial systems are noticeably less likely to face crisis. For countries that do not meet these preconditions, the case studies suggest that a gradual approach to liberalization—with appropriate sequencing of liberalization of capital controls and improvements in the domestic financial sector and macroeconomic framework—seems to reduce the likelihood of a crisis; external anchors (such as EU membership) are also associated with reduced crisis propensity. Overall, the case studies suggest that the likelihood of currency and debt crises following financial account liberalization is noticeably reduced when such liberalization is an element of a broader reform package, macroeconomic policies are sound, and external imbalances are limited.

Economic Growth

The theoretical presumption that financial globalization should raise economic growth is appealing and intuitive, yet a vast empirical literature relying on cross-country regressions has failed to identify robust evidence of such a relationship. This subsection considers first this macroeconomic evidence, and then turns to an emerging literature based on microeconomic evidence, which tends to find more significant effects of (de jure and de facto) financial globalization on economic growth or its proximate causes (such as improvements in economic efficiency or domestic financial development).

A survey of more than 40 empirical studies based on macroeconomic data and cross-country regressions concludes that the evidence of a link between financial integration and economic growth is not robust: while a few studies, mostly focusing on equity market liberalizations, find positive and significant effects, the majority of studies find insignificant effects, or results that do not hold up to changes in specification and country sample (Kose and others, 2006).4 This is corroborated by cross-country and panel regressions estimated by staff of economic growth on financial integration and a few other standard determinants, where the results appear to be fragile (Table 6.3). The apparent absence of robust evidence of a link between financial globalization and economic growth may not be surprising, in light of the well-known difficulties involved in finding robust determinants of economic growth in crosscountry or panel regressions. Nevertheless, it does raise the question of how to reconcile the theoretical promise of financial integration with the mixed/fragile empirical evidence. To address this question, three issues are considered.

Table 6.3.Financial Integration and Economic Growth
(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)
Initial income per capita (log)−1.04−1.09***−1.31***−1.44***−1.42−1.17***−1.03***−1.04***−1.03***−0.81***
(0.26)(0.29)(0.27)(0.28)(0.29)(0.3)(0.28)(0.28)(0.28)(0.26)
Average investment to GDP9.90***9.75***7.71**7.67**7.33**9.89***9.46**9.50**9.43**12.14***
(2.93)(3.65)(3.12)(3.07)(3.17)(3.56)(3.68)(3.66)(3.68)(3.26)
Years of schooling0.130.130.120.15*0.14*0.140.130.110.130.06
(0.09)(0.1)(0.08)(0.08)−0.08(0.1)(0.1)(0.09)(0.1)(0.08)
Population growth−31.50−34.10−63.89***−82.47***−67.14−36.55−32.41−30.55−32.33−28.92
(20.33)(23.1)(20.18)(19.69)(20.01)(23.74)(21.84)(21.94)(21.78)(17.43)
Africa dummy−0.63−0.62−0.65−0.53−0.66−0.61−0.63−0.66−0.63−0.59
(0.55)(0.57)(0.55)(0.54)−0.55(0.56)(0.56)(0.56)(0.56)(0.57)
Gross financial openness to GDP (stock)0.06*
(0.03)
Total inflows to GDP8.62**
(4.04)
Gross flows to GDP1.56***
(0.42)
Total outflows to GDP6.75
(5.31)
External assets to GDP0.19**
(0.09)
External liabilities to GDP0.02
(0.11)
FDI plus portfolio equity liabilities to GDP0.93
(1.12)
Debt liabilities to GDP0.01
(0.12)
De jure financial openness−0.84
(0.51)
Constant8.84***9.25***11.66***13.12***12.77***9.89***8.86***8.80***8.84***7.11***
(1.99)(2.28)(1.97)(2.12)(2.1)(2.35)(2.16)(2.09)(2.15)(1.88)
Observations91878686868787878784
R20.390.390.470.460.450.40.380.390.380.42
Sources: IMF, International Financial Statistics; Heston, Summers, and Aten, Penn World Tables Version 6.2; and IMF staff estimates.Notes: Cross-country regressions, ordinary least squares, 1975–2004. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively. Outliers, such as financial centers, are excluded from the sample; results are stronger when they are included.
Sources: IMF, International Financial Statistics; Heston, Summers, and Aten, Penn World Tables Version 6.2; and IMF staff estimates.Notes: Cross-country regressions, ordinary least squares, 1975–2004. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively. Outliers, such as financial centers, are excluded from the sample; results are stronger when they are included.
  • Composition. Unbundling financial globalization into different types of financial flow helps to uncover a relationship between financial integration and economic growth. Cross-country and panel regressions reported in Table 6.4 suggest that countries with a higher share of FDI in total liabilities tend to experience more rapid economic growth.5 The link is statistically and economically significant, and robust to variations in estimation technique. Concretely, keeping constant the stock of foreign liabilities, an increase in FDI by 10 percentage points of GDP (about the average of FDI in the sample) is associated with an increase in average growth of 0.3 percentage point. This evidence is consistent with many studies that have documented a positive impact of FDI on economic growth (e.g., Moran, Graham, and Blomström, 2005).
  • Thresholds. There is some evidence that the impact of financial integration on growth depends on factors similar to those governing the relationship between financial integration and volatility discussed above.6 Although the results are not particularly robust, financial integration appears to be beneficial for growth in countries that meet certain thresholds with respect to financial development, institutional quality, macro-economic policy soundness, and trade openness, but has potentially large negative effects in countries that do not.7 Such thresholds seem to be especially relevant for the effects of external debt accumulation on economic growth, and less relevant for FDI, whose effects on economic growth do not seem to depend on thresholds.
  • Indirect benefits. A growing body of empirical work suggests that financial liberalization has a positive impact on several variables that are associated with economic growth, even if their effects are difficult to detect in cross-country growth regressions:8
Table 6.4.Impact of FDI on GDP Growth
Initial income1.45
(1.46)
Schooling years−0.43
(0.31)
Population growth0.11
(0.45)
Investment (share of GDP)0.09
(8.55)
Government balance (share of GDP)14.24**
(5.64)
CPI Inflation–0.99***
(0.38)
Trade openness0.40
(0.69)
Private credit (share of GDP)−3.12
(2.81)
FDI and equity liabilities (share of GDP)3.00*
(1.74)
Total liabilities (share of GDP)−0.14
(1.11)
Sources: IMF, International Financial Statistics; and IMF staff estimates.Notes: System of generalized method of moments estimates on a panel of six five-year periods over 1975–2004. The dependent variable is the average growth rate of GDP per capita over each five-year period. The results are based on 410 observations (73 countries). Robust standard errors are reported in parentheses. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively.
Sources: IMF, International Financial Statistics; and IMF staff estimates.Notes: System of generalized method of moments estimates on a panel of six five-year periods over 1975–2004. The dependent variable is the average growth rate of GDP per capita over each five-year period. The results are based on 410 observations (73 countries). Robust standard errors are reported in parentheses. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively.
  • Total factor productivity growth. Panel regressions estimated by staff (Table 6.5) suggest that total factor productivity growth (TFP) is positively and significantly associated with de jure financial openness. This result may be surprising, given the lack of robust evidence of a relationship between financial integration and economic growth, and little evidence of threshold effects impinging on the transmission of financial openness to TFP. One possible interpretation of these results is that financial openness enhances economic efficiency but has an unstable and seldom significant effect on factor accumulation, so that the ultimate effect on economic growth is difficult to pinpoint in the data.
  • Domestic financial sector development. Financial integration may catalyze domestic financial market development, through greater competitive pressures on financial intermediaries and movement toward international best practices in accounting, financial regulation, and supervision. Foreign ownership of banks may also facilitate transfer of technology and risk-management techniques (Goldberg, 2004; Levine, 2005; and Mishkin, 2006). As reported in Table 6.6, de jure financial openness and domestic financial sector development are significantly correlated, controlling for a range of other determinants. These results, moreover, appear to be robust across sample compositions and econometric specifications.
  • Macroeconomic policies. Financial integration may improve policy discipline and signal a country's commitment to sound policies (Bartolini and Drazen, 1997; and Gourinchas and Jeanne, 2005). Empirical studies suggest that countries with higher levels of financial openness experience lower inflation rates (Tytell and Wei, 2004; and Sen Gupta, 2007), though evidence is more mixed for fiscal policies (Garrett and Mitchell, 2001; and Kim, 2003).
Table 6.5.Financial Openness (De Jure) and Total Factor Productivity Growth
Fixed EffectsSystem−GMM
Initial total factor productivity−0.56***−0.25**
(0.08)(0.11)
Trade openness0.52***0.24
(in percent of GDP)(0.18)(0.24)
Financial openness (de jure)0.08***0.07**
(0.03)(0.03)
Population growth (in percent)-0.02−0.09***
(0.03)(0.02)
R20.45
Sargan test p-value0.25
AR1 test p-value0.02
AR2 test p-value0.10
Sources: Heston, Summers, and Aten, Penn World Tables Version 6.2; World Bank, World Development Indicators; IMF, Annual Report on Exchange Arrangements and Exchange Restrictions; and IMF staff estimates.Notes: Fixed effects and system of generalized method of moments (system-GMM) panel estimates. The dependent variable is the 10-year nonoverlapping growth rate of total factor productivity (TFP) over 1965-2005. Period dummies are included but not reported. In addition to the internal instruments, an emerging markets dummy has been used in the system-GMM. The results are based on 263 observations (68 countries). Robust standard errors (clustered by country) are reported in parentheses The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively. Outliers, such as financial centers, are excluded from the sample; results are stronger when they are included.
Sources: Heston, Summers, and Aten, Penn World Tables Version 6.2; World Bank, World Development Indicators; IMF, Annual Report on Exchange Arrangements and Exchange Restrictions; and IMF staff estimates.Notes: Fixed effects and system of generalized method of moments (system-GMM) panel estimates. The dependent variable is the 10-year nonoverlapping growth rate of total factor productivity (TFP) over 1965-2005. Period dummies are included but not reported. In addition to the internal instruments, an emerging markets dummy has been used in the system-GMM. The results are based on 263 observations (68 countries). Robust standard errors (clustered by country) are reported in parentheses The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively. Outliers, such as financial centers, are excluded from the sample; results are stronger when they are included.
Table 6.6.Financial Integration and Financial Sector Development
Fixed EffectsSystem-GMM
Ln private credit to GDP, lagged−0.53***−0.26***
(0.07)(0.06)
Ln real GDP per capita PPP0.38***0.14**
(0.1)(0.06)
Ln (1+ CPI inflation rate)–0.010.01
(0.06)(0.06)
Ln trade openness0.34**0.18**
(0.17)(0.09)
Financial account openness index0.21***0.19**
(0.07)(0.09)
Constant−2.05−0.58
(1.23)(0.64)
R20.34
Sargan test p-value1
AR1 test p-value0.01
AR2 test p-value0.9
Sources: IMF, International Financial Statistics; Financial account openness index (equal to 1 if country is classified as open and 0 if closed) constructed by staff based on the IMF's Annual Report on Exchange Arrangements and Exchange Restrictions; and IMF staff estimates.Notes: Panel of nonoverlapping five-year averages during 1975–2004. Dependent variable is change in the logarithm of private credit to GDP. Period dummies included but not reported. The results are based on 339 observations (59 countries). Robust standard errors clustered by country in parentheses. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively. In system-GMM estimation, all control variables enter as endogenous
Sources: IMF, International Financial Statistics; Financial account openness index (equal to 1 if country is classified as open and 0 if closed) constructed by staff based on the IMF's Annual Report on Exchange Arrangements and Exchange Restrictions; and IMF staff estimates.Notes: Panel of nonoverlapping five-year averages during 1975–2004. Dependent variable is change in the logarithm of private credit to GDP. Period dummies included but not reported. The results are based on 339 observations (59 countries). Robust standard errors clustered by country in parentheses. The symbols *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent level, respectively. In system-GMM estimation, all control variables enter as endogenous

Turning to the microeconomic, and especially firm-level, evidence, as well as event studies surrounding equity market liberalizations, a clearly beneficial impact of financial globalization on market capitalization, financial development, and the cost of capital is apparent (Bekaert, Harvey, and Lundblad, 2005; and Henry, 2006). Equity market liberalizations have also been found to reduce the cost of capital (Stulz, 1999) and to boost investment growth (Alfaro and Hammel, 2006). Relatedly, microeconomic studies (surveyed in Forbes, 2005a) have found that capital controls may impose significant efficiency costs, including through:

  • Lower international trade. Wei and Zhang (2007) present evidence suggesting that capital controls increase the cost of engaging in international trade even for those firms that do not intend to evade capital controls. A one-standard-deviation increase in controls on foreign exchange transactions reduces trade by the same amount as a hike in external tariffs by about 11 percentage points, according to their results. More generally, there is ample evidence from case studies that capital controls create incentives for circumvention through mis-invoicing.
  • Cost of capital. Capital controls are estimated to make it more difficult and expensive for small firms to raise capital (Forbes, 2005b). Moreover, multinational affiliates located in countries with capital controls face local borrowing costs that are about 5 percentage points higher than affiliates of the same parent company borrowing locally in countries without capital controls (Desai, Foley, and Hines, 2004).
  • Distortions. Economic behavior is likely to be distorted by capital controls, and resources and effort are wasted in seeking to circumvent controls. Moreover, a situation in which only some economic agents are able to evade controls may lead to an uneven playing field in which well connected firms—rather than the most efficient—survive. Beyond this, capital controls insulate domestic firms from competitive forces, and in some cases may even create a screen for cronyism and subsidies to politically connected firms (Johnson and Mitton, 2003).
  • Costs for the public administration. Significant administrative costs result from the need to monitor compliance with capital controls and, in many cases, to continually update the controls to close loopholes and limit evasion (Forbes, 2005a).

To sum up, although policy advice on financial liberalization needs to consider whether countries meet certain thresholds that govern its impact, it also needs to take into account the impact of financial integration on countries' standing in relation to the thresholds, and the significant microeconomic costs of maintaining capital controls. This leads to a tension: on the one hand, liberalization for countries that do not meet the thresholds may amplify risks; on the other hand, liberalization may itself catalyze improvements in domestic financial development and macroeconomic policies, and reduce the distortionary costs of capital controls, perhaps engendering a virtuous circle in which ultimately the country will meet the necessary conditions to reap the full benefits of integration.

This section was prepared by Giovanni Dell'Ariccia, André Faria, Ayhan Kose, and Marco Terrones.
1It should be noted that, for a number of the empirical associations examined in this section, causality may run in both directions.
2This result is robust to estimation in a cross section of long-run averages, changes in country coverage, and sample period.
3The results are significant for currency crises, debt crises, and sudden stops, though not for banking crises. Results are robust to: splitting the sample on the basis of whether they meet 50 percent (or 100 percent) of the thresholds; excluding the advanced economies from the sample; defining countries as financially open if they are in the top tercile, instead of the top half; and using de jure, instead of de facto, measures of financial openness. Definitions and data sources for the various types of crises are in Becker and others (2007).
4Some studies have found positive and significant evidence for limited subsamples of countries, such as Eastern Europe (Abiad, Leigh, and Mody, 2007).
5In some instances, the distinction between FDI and non-FDI flows may be blurred in the data, in an environment where multinationals can to a large extent choose how to book transactions across branches or subsidiaries in different countries, for example to take advantage of tax or regulatory differentials. In terms of the empirical implementation, such features imply that both FDI and non-FDI flows are likely to be measured with error. It should be emphasized that this type of “measurement error” would tend to make it more difficult to establish a differential impact of FDI and non-FDI flows on growth. Taking this possible “attenuation bias” into consideration, the finding of a statistically significant difference between the impact of FDI and non-FDI flows is thus even more revealing.
6A number of empirical studies (surveyed in Kose and others, 2006) report evidence suggesting that preconditions with respect to domestic financial sector development, institutional quality, and trade openness need to be met for financial integration to have a beneficial impact on economic growth. Reliance on foreign capital (especially non-FDI forms of financing) has not been found to be positively associated with economic growth in a broad cross section of countries, though it has for a subsample consisting of advanced and transition economies. Prasad, Rajan, and Subramanian (2007) find that greater domestic financial development strengthens the favorable impact of foreign capital on economic growth.
7In particular, the significance of the results and the estimated thresholds beyond which the impact of financial integration is positive/negative are sensitive to changes in estimation technique and sample composition. Thus, further research is needed to make these findings applicable to policy analysis.
8Consistent with this view, while the coefficient on financial globalization is sometimes significant in the regressions reported in Table 6.3, such significance tends to disappear if the list of explanatory variables includes—as is the case in most empirical studies—measures of “collateral benefits,” such as domestic financial sector development, sound macroeconomic policies, and higher external trade. Beyond these effects, financial globalization may also affect the duration of growth spells—an effect that is difficult to capture in growth regressions—and, like trade openness, may improve institutional quality by creating constituencies for economic reform (Berg, Ostry, and Zettelmeyer, 2007; Johnson, Ostry, and Subramanian, 2006; and Rajan 2006).

    Other Resources Citing This Publication