The global economy has become substantially more financially integrated over the past three decades. Average de facto financial globalization (measured, as discussed in Box 2.1, by gross external assets and liabilities as a share of GDP) has approximately tripled since the mid-1970s. Experience has differed by income group: the worldwide increase in financial globalization has been driven mainly by high-income countries, where financial integration has accelerated since the early 1990s (Figure 3.1). Although low- and middle-income countries have also become more financially integrated, average increases have been more moderate. Regionally, many countries in developing and emerging East Asia as well as in Eastern and Central Europe have displayed relatively large increases in international financial integration—sixfold and threefold, respectively, on average, compared with a twofold increase in the low- and middle-income countries as a whole.
Gross External Assets and Liabilities by Income Group
(In percent of GDP)
Source: Lane and Milesi-Ferretti (2006).Notes: Based on a sample of 74 countries (see Appendix 1) for which data on de facto financial globalization and de jure capital controls are available for the entire sample period. Income groups are according to the World Bank definition. The graph depicts unweighted averages of countries' ratios of the sum of external assets and liabilities relative to GDP.Increasing financial integration among OECD countries has been characterized by two-way, or “diversification,” asset trade—large gross holdings of assets and liabilities that have resulted in a relatively small net external position (Table 3.1).1 In contrast, for other countries, net liability positions are relatively large. The data also suggest that the composition of external assets and liabilities has shifted away from debt instruments over the past decade, though debt remains—across income groups and regions—the largest component of external liabilities (Figure 3.2).2 FDI inflows have gained importance in many low- and middle-income countries, whereas portfolio equity finance has increased substantially in several high-income countries.
Composition of Gross External Assets and Liabilities, 1975 and 2004
(In percent)
Source: Lane and Milesi-Ferretti (2006).Notes: Based on a sample of 32 high-income, 31 middle-income, and 11 low-income countries. For each year and income group, the pie charts depict the shares of each type of external assets plus liabilities in total external assets plus liabilities. Group averages are unweighted.Gross and Net External Positions, 2004
(In percent)
Gross and Net External Positions, 2004
(In percent)
| External Position | |||||||
|---|---|---|---|---|---|---|---|
| Gross | Net | Grubel-Loyd Index | |||||
| High income | 531.5 | 44.7 | 92 | ||||
| OECD | 462.1 | −13.5 | 97 | ||||
| Non-OECD | 664.5 | 156.4 | 76 | ||||
| Middle income | 151.3 | −45.8 | 70 | ||||
| Low income | 119.3 | −49.3 | 59 | ||||
Gross and Net External Positions, 2004
(In percent)
| External Position | |||||||
|---|---|---|---|---|---|---|---|
| Gross | Net | Grubel-Loyd Index | |||||
| High income | 531.5 | 44.7 | 92 | ||||
| OECD | 462.1 | −13.5 | 97 | ||||
| Non-OECD | 664.5 | 156.4 | 76 | ||||
| Middle income | 151.3 | −45.8 | 70 | ||||
| Low income | 119.3 | −49.3 | 59 | ||||
De Jure Financial Openness
Legal (de jure) controls on capital account transactions—a policy variable—are potentially important determinants of de facto financial globalization.3 Over the past three decades, most countries have relaxed de jure controls on the capital account, though the process of liberalization has slowed since the mid-1990s. This broad trend is apparent for both the relatively liberalized and the relatively nonliberalized countries, though liberalization efforts took place somewhat earlier in the former group than in the latter group (Figure 3.3, left panel). About one-half of the countries in the sample are currently considered fully open to capital flows, up from under one-third in 1975. While liberalizations were the dominant trend over the period, about 10 percent of the countries in the sample tightened their controls, often in response to crises. The capital controls index developed in this paper indicates that 17 countries not fully open in 1995 had fully opened their capital accounts by 2005, while only 4 countries opted to fully close their capital accounts between 1995 and 2005.4
Capital Controls by Financial Openness and Income Group
Sources: IMF, Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER); and IMF staff calculations.Notes: Based on a sample of 74 countries for which data on de facto globalization and de jure capital controls are available for the entire sample period. The graph depicts unweighted averages of countries' capital controls, using the IMF's binary capital controls indicator (based on the AREAER's pre–1995 methodology). Countries in the left panel are categorized according to the 1975–2005 mean of their capital controls variable: the cutoff for liberalized versus nonliberalized is the sample mean. Countries in the right panel are grouped according to the World Bank definition (see Appendix I).Although the level of controls appears to be inversely related to a country's per capita income, countries in all income groups—on average—have relaxed capital controls over the past three decades (Figure 3.3, right panel). Liberalizations were pervasive among OECD countries—many of which moved from a highly restricted financial account position in 1975 to being fully open by 2005, while among emerging market and developing economies there were regional differences. Many countries in Eastern Europe and Latin America liberalized their financial accounts—owing, in a number of cases, to prospective accession to the European Union (EU) or bilateral or regional trade agreements (IEO, 2005, p. 32; and Árvai, 2005). In contrast, several countries in East Asia and the Middle East tightened capital controls, and most countries in sub-Saharan Africa maintained financial account restrictions. Several high-income oil-exporting countries also introduced new restrictions during the 1990s.
Among countries that retained capital controls, on average outflows were somewhat more restricted than inflows while, in low-income countries, restrictions on short-term debt were more common than those on long-term debt (see Table 3.2 on p.7). It is also worth noting that controls on equity, and especially FDI, were brought down considerably between 1995 and 2005 across the membership, whereas controls on debt remained essentially unchanged, on average. More generally, in recent years, changes in the structure of capital controls have brought more countries in line with what has come to be referred to as the “integrated approach” (Box 3.1). According to the approach, countries should liberalize FDI inflows first; this should generally be followed by lifting controls on other long-term and nondebt flows, such as equity and outward FDI, before the liberalization of short-term flows and debt flows.5 In fact, as shown in Figure 3.4, both the number of countries with more liberal long-term than short-term flows, and the number of countries with more liberal nondebt flows, increased by 10–15 percent between 1995–97 and 2002–05.6
Patterns of De Jure Financial Openness, 1995–97 Versus 2003–05
(In percent of all countries)
Sources: IMF, Annual Report on Exchange Arrangements and Exchange Restrictions; and IMF staff calculations.Notes: Based on a sample of 73 countries with continuously closed (1) or open (0) financial accounts during 1995–2005. The first pair of bars shows the fraction of countries where, on average during 1995–97 and 2003–05, respectively, long-term debt flows were less restricted than short-term debt flows, while the second pair of bars shows the fraction of countries where nondebt (equity and FDI) flows were less restricted than debt (bonds and money-market) flows. These comparisons provide snapshot of the percent of countries whose capital controls structure was consistent (in 1995–97 and 2003–05, respectively) with the two aforementioned aspects of the integrated approach described in the textCapital Controls by Type, 1995–2005
Capital Controls by Type, 1995–2005
| All Countries | 1995–2005 Average | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Type of Control | 1995 | 2005 | Low income | Middle income | High income | ||||||
| Aggregate | 0.36 | 0.30 | 0.56 | 0.38 | 0.17 | ||||||
| Inflows | 0.32 | 0.26 | 0.50 | 0.33 | 0.16 | ||||||
| Outflows | 0.40 | 0.34 | 0.63 | 0.44 | 0.18 | ||||||
| Equity | 0.37 | 0.30 | 0.61 | 0.38 | 0.18 | ||||||
| Debt | 0.33 | 0.32 | 0.50 | 0.40 | 0.15 | ||||||
| Short term | 0.34 | 0.30 | 0.59 | 0.40 | 0.15 | ||||||
| Long term | 0.33 | 0.33 | 0.41 | 0.40 | 0.15 | ||||||
| FDI | 0.38 | 0.27 | 0.54 | 0.37 | 0.20 | ||||||
Capital Controls by Type, 1995–2005
| All Countries | 1995–2005 Average | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Type of Control | 1995 | 2005 | Low income | Middle income | High income | ||||||
| Aggregate | 0.36 | 0.30 | 0.56 | 0.38 | 0.17 | ||||||
| Inflows | 0.32 | 0.26 | 0.50 | 0.33 | 0.16 | ||||||
| Outflows | 0.40 | 0.34 | 0.63 | 0.44 | 0.18 | ||||||
| Equity | 0.37 | 0.30 | 0.61 | 0.38 | 0.18 | ||||||
| Debt | 0.33 | 0.32 | 0.50 | 0.40 | 0.15 | ||||||
| Short term | 0.34 | 0.30 | 0.59 | 0.40 | 0.15 | ||||||
| Long term | 0.33 | 0.33 | 0.41 | 0.40 | 0.15 | ||||||
| FDI | 0.38 | 0.27 | 0.54 | 0.37 | 0.20 | ||||||
The Integrated Approach to Capital Account Liberalization
As noted in the Independent Evaluation Office's report (IEO, 2005), the IMF's “integrated” or “sequencing” approach to capital account liberalization, developed in the late 1990s/early 2000s, appears to be widely accepted among IMF staff and underlies much of the institution's policy advice in this area. The approach considers capital account liberalization as part of a broader economic reform package encompassing the macroeconomic policy framework, the domestic financial system, and prudential regulation. The approach also emphasizes the importance of following a sequence of measures and reforms.1
The integrated approach consists of the following 10 general principles: (1) capital account liberalization is best undertaken against a background of sound and sustainable macroeconomic policies; (2) financial sector reforms that support and reinforce macroeconomic stabilization should be given priority in implementation; (3) financial sector reforms that are mutually reinforced and operationally linked should be implemented together; (4) domestic financial reform should be complemented by prudential regulation and supervision and financial restructuring policies; (5) liberalization of capital flows by instruments and/or sectors should be sequenced to take into account concomitant risks—in general, long-term and non-debt-creating flows (especially FDI) should be liberalized before short-term and debt-creating flows; (6) the pace of reform should take into account the conditions in the nonfinancial sector; (7) reforms that take time should be started early; (8) reforms need to take into consideration the effectiveness of controls on capital flows in place at the time of liberalization; (9) the pace, timing, and sequencing of liberalization need to take account of political and regional considerations; and (10) the arrangements for policy transparency and data disclosure should be adapted to support capital account opening.
The evidence reported in this paper suggests that member countries have increasingly followed the integrated approach to liberalization. Taking a “snapshot” of countries' capital control structures, the extent to which countries follow the approach should be reflected in the share of countries with more controls on short-term debt than on long-term debt; and with more controls on debt than nondebt flows. As shown in (Figure 3.5, the degree to which countries' practice appears to conform to the approach has increased since the mid-1990s. More generally, as shown in Appendix III, most countries covered in the case studies have also liberalized FDI inflows early on, long-term before short-term flows, and nondebt flows before debt flows, particularly in the more recent period.
1 Eichengreen and others (1998); and Ishii, Habermeier, and others (2002).Countries' de facto financial integration has been influenced by their de jure financial account openness (Figure 3.5, top panel). First, during 1975–2004, de jure “liberalized” countries (defined as those with a lower-than-average index of capital controls over 1975–2005) had gross external assets and liabilities (relative to GDP) nearly twice as high as the non-liberalized countries (defined as those with a higher-than-average index of capital controls).7 Second, the “least liberalized” countries (those in the decile with the highest controls) saw smaller increases in de facto globalization than were experienced by countries with less restrictive regimes, though even the least liberalized countries did not isolate themselves completely from the trend toward greater de facto financial globalization—the ratio of their gross external assets and liabilities to GDP almost doubled over the period. Third, for countries that went from having above-average de jure restrictiveness during the first half of the sample period to below-average restrictiveness during the second half, de facto integration reached levels similar to those in countries that had been open throughout. Conversely, in countries that tightened controls during 1990–2005, financial integration converged to the lower and flatter trend of countries that had been closed throughout (Figure 3.5, bottom panel). These effects, it bears noting, portray the medium-run impact of highly durable characteristics of the capital control regime, rather than the impact of specific measures maintained for a relatively short time. On this latter issue, evidence from case studies suggests that when controls are re-imposed in countries that have experienced relatively liberal flows for a number of years, they tend to lose their effectiveness relatively quickly, especially where domestic financial markets are well developed (Obstfeld, 2007).
Gross External Assets and Liabilities by Levels and Changes in De Jure Financial Openness
(In percent of GDP)
Source: Lane and Milesi-Ferretti (2006).Notes: The graph depicts unweighted averages of countries' ratios of the sum of external assets and liabilities relative to GDP. The top panel is based on a sample of 74 countries for which data on de facto globalization and de jure capital controls are available for the entire sample period. Countries are categorized according to the 1975–2005 mean of their capital controls variable. The cutoff for liberalized versus nonliberalized is the sample mean; the most (least) liberalized countries represent the bottom (top) decile of the capital controls variable. In the bottom panel, six oil-producing countries are excluded. For each of the subperiods 1975–89 and 1990–2005, countries are categorized as liberalized (nonliberalized) if the mean of a country's capital controls variable is below (above) the sample mean for the subperiod. Countries switching from nonliberalized in 1975–89 to liberalized in 1990–2005 are labeled liberalizers, and vice versa for nonliberalizers.Beyond the relationship between the de jure regime and the overall level of de facto financial integration, there is also some evidence—for example, Eichengreen and others (1998)—that the structure of capital controls affects the composition of countries' external assets and liabilities. Indeed, other things equal, the evidence suggests that controls on portfolio equity and FDI are easier to enforce—and therefore more likely to be effective—than controls on debt and bank flows (Edwards, 1999). This evidence would seem to be broadly consistent with the observation that the share of FDI and equity in countries' external portfolios has increased during the past three decades, over the same period that de jure controls on FDI and equity were reduced compared with other types of controls.8
On the whole, the stylized facts in this section underscore the degree to which countries that have maintained controls in place for many years have experienced smaller increases in de facto globalization than countries that were always open. However, even the countries that maintained the strictest controls in the sample experienced some increase in financial integration, perhaps because trade in financial assets is closely associated with trade in goods, and it would have been too costly for these countries to isolate themselves from globalization in the broader sense. While durable aspects of the capital account regime seem to have long-term effects on financial integration, controls aimed at fine-tuning the level and composition of flows tend to lose their effectiveness relatively quickly, and may become increasingly difficult to enforce as countries' financial systems develop.
All cross-border financial holdings are included in the data presented in this paper: debt, bank loans, equity investment, and FDI. Existing data on cross-border holdings of assets and liabilities do not allow a clear-cut distinction between public and private positions. This distinction, even if possible, would in any case be blurred by past conversions of defaulted private obligations into public debt.
There is also evidence that the currency composition of emerging market debt is changing: the share of local-currency-denominated debt in marketable sovereign debt rose from 73 percent in 1996 to 82 percent in 2004 (IMF, 2006).
For the purposes of this paper, indices that measure controls on inflows and outflows separately, as well as controls on different categories of assets (equity, debt, and direct investment), have been developed for 91 countries for 1995–2005, drawing on the information in the Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER); see Appendix II and Schindler (2008) for further details. Long-term trends since 1975 draw on the AREAER's original binary index, which was extended to 2005 for the purposes of this paper. Shortcomings common to all indices based on the AREAER are that they do not capture differences in enforcement and the economic impact of controls across countries and time periods.
Countries with an aggregate capital controls index greater than 0.9 are here defined as fully closed, and those with an index less than 0.1 as fully liberalized. Using instead a definition based on the extreme values of the index (0.0 and 1.0), only 2 countries became fully closed, whereas 14 countries fully opened up.
The liberalization of some short-term flows into the banking system may be required at an early stage to foster the development of key domestic financial markets, notably the interbank money and foreign exchange markets. Suitable prudential measures in the banking system should be adopted in parallel.
This exercise takes a “snapshot” of whether a country's capital controls structure is broadly in line with the integrated approach, though this is only a rough indication of consistency, because the approach allows for deviations from the broad patterns being considered when warranted by country-specific circumstances. Also, the exercise does not examine whether individual countries have adhered to the sequencing of liberalization implicit in this approach. Árvai (2005), who examines liberalization efforts of eight EU accession countries, reports that sequencing was broadly in line with the integrated approach.
These results also hold when controlling for per capita income.
A more formal approach, based on panel regressions, however, does not find significant evidence linking the shift toward equity and FDI finance to changes in the structure of capital controls (Faria and others, forthcoming). It is possible that the cross-country variation in lifting controls on equity and FDI compared with other flows has been insufficient for its impact to be captured in regressions.
