Chapter 1 Introduction
- International Monetary Fund. Independent Evaluation Office
- Published Date:
- August 2005
The 1990s witnessed a large swing in global private capital flows (Figure 1.1). Net private flows to developing countries, for example, grew from less than $100 billion in 1990 to well over $200 billion in 1995. The subsequent years, however, saw an equally substantial reversal of these inflows, which caused several emerging market economies to experience severe capital account crises. The volume of private capital flows to developing countries remained subdued through the early 2000s.
Figure 1.1.Private Capital Flows to Developing Countries1
Source: IMF database.
1 Portfolio investment flows, other private investment flows, and foreign direct investment to all developing countries, Israel, and Korea. Excludes government borrowing.
Against this background, there has been a major debate over the actual and potential role of the IMF in encouraging countries to open their capital accounts1 and any possible associated increase in their vulnerability to crisis. Within the broader debate over the increasing importance of international capital flows in the world economy,2 some have alleged that the IMF, in concert with some major shareholder governments, had encouraged member countries to liberalize their capital accounts prematurely without ensuring that adequate institutions and prudential regulations were in place.3 Others argue that rapid liberalization, with insufficient attention to sequencing and establishing the appropriate preconditions, has been responsible for much of the financial instability and economic distress experienced by many emerging market countries.4
The role of the IMF has been particularly controversial because capital account liberalization is an area where there is little professional consensus (see Box 1.1). In this context, Eichengreen (2001) has noted that the views favoring liberalization emerged with a surprising degree of certitude in advance of (and in the absence of) definitive evidence. The IMF’s role has been controversial for another reason: Although current account liberalization is among the IMF’s official purposes outlined in its Articles of Agreement, it has no explicit mandate to promote capital account liberalization. Indeed, the Articles give the IMF only limited jurisdiction over the capital account (see Chapter 2, “The Legal Basis,” for details). Nevertheless, the IMF has given greater attention to capital account issues in recent decades, given the increasing importance of international capital flows for macroeconomic stability and exchange rate management in many countries. In view of these facts, an independent assessment of how the IMF has addressed capital account issues seems warranted.
Box 1.1.The Debate on the Benefits of Capital Account Liberalization
The theoretical rationale for capital account liberalization is based primarily on the argument that free capital mobility promotes an efficient global allocation of savings and a better diversification of risk, hence greater economic growth and welfare (Fischer, 1998). An opposing view has held that there is considerable information asymmetry in international financial markets, so that free capital mobility—especially when significant domestic distortions exist—does not necessarily lead to an optimal allocation of resources (Stiglitz, 2000 and 2004). Between these two opposing positions is the view that, while there are benefits to be gained from liberalization, the magnitude of the gains is relatively small.1 While the idea that free capital mobility enhances economic welfare is an appealing concept to many economists, there has been surprisingly little empirical evidence to date to either support or refute conclusively such a view.
Recent empirical work has addressed this issue from the standpoint of the effect of capital liberalization on economic growth (see Edison and others, 2002, for a survey). Unfortunately, the debate remains inconclusive because such empirical studies inherently involve a joint test of the effect of liberalization on growth and the particular method of quantifying the degree of liberalization or effectiveness of capital controls. This problem is common to all empirical studies in this area.2 As it turns out, empirical results are sensitive not only to the quantitative measure of capital controls but also to the choice of sample and methodology. For example, while Quinn (1997) finds a positive association between capital account liberalization and economic growth, Grilli and Milesi-Ferretti (1995) and Rodrik (1998) fail to find any such relationship. This ambiguity may reflect the role of institutions (for example, the rule of law), macroeconomic stability, and other factors in determining the effect of liberalization on growth (Arteta and others, 2001; Eichengreen and Leblang, 2002).3 On the other hand, studies that have more narrowly focused on stock market liberalization have found a positive impact on growth (for example, Henry, 2003).1 For example, Gourinchas and Jeanne (2004) use a calibrated neoclassical model to show that, for a typical developing country, the welfare gains from switching from financial autarky to perfect capital mobility is about 1 percent permanent increase in domestic consumption.2 Another common problem is the endogeneity of capital controls, which makes it difficult to disentangle the effect of capital controls per se from that of the macroeconomic and international environments within which they are introduced.3Prasad and others (2003) also consider the effects of financial integration on consumption smoothing and find little evidence to indicate the benefits of liberalization. See Stiglitz (2004) for commentaries on this work.
The Scope of the Evaluation
The evaluation seeks to (1) contribute to transparency by documenting what in practice has been the IMF’s approach to capital account liberalization and related issues; and (2) identify areas, if any, where the IMF’s instruments and operating methods might be improved, in order to deal with capital account issues more effectively.5 The issues addressed in the evaluation cover not only capital account liberalization but also capital flow management issues, including particularly the temporary use of capital controls. We evaluate the IMF’s actual approach to these issues, not necessarily its official policy. Indeed, as will become clear, it is difficult to argue that the IMF had a firm formal policy on the issues we address—at least not during the period covered by the evaluation. In evaluating the IMF’s approach, we rely primarily on country-based analysis. We will try to identify, for example, what policy advice the IMF gave in the context of a specific country at a specific point in time. Although context is important, the focus remains on the role of the IMF. We make no judgment on the underlying policies adopted by country authorities.
Given the lack of consensus in the academic and official policymaking communities, there is no universal set of criteria against which the IMF’s approach to capital account issues can be assessed. Rather, we take a pragmatic approach to evaluation by asking the following questions about the IMF’s policy advice:
- Was there any difference between the IMF’s general policy pronouncements and the advice it gave to individual countries?
- Was the IMF’s policy advice operational? Was it based on solid evidence?
- How did the IMF’s advice change over time? Did this change keep pace with available evidence—that is, did the IMF learn as new evidence emerged (see, for example, Box 1.2)?
- Did the IMF give similar advice to countries in similar situations?
- Was the policy advice on the capital account set in a broader assessment of the authorities’ macroeconomic policies and institutional framework?
Box 1.2.Effectiveness of Market-Based Capital Controls: Evidence from Chile
From around 1996, there began to emerge a substantial body of empirical research on the effectiveness of Chile’s capital inflow control, which was introduced in 1991 in the form of an unremunerated reserve requirement (URR). We discuss available evidence to provide background to subsequent discussions on the policy advice given by the IMF on such controls.
This measure required a designated share of certain capital inflows to be deposited with the central bank at zero interest for a designated period of time (see Box 2.2 for details of how the system worked). Although different studies came to different conclusions, by 1999, the sense of the literature—though evidence was often weak—was that (1) the URR allowed domestic interest rates to be somewhat higher; (2) it lengthened the maturity of capital inflows; (3) it had only limited effectiveness, if any, in reducing the volume of total inflows; and (4) it had little or no effect on the real exchange rate (see Nadal-De Simone and Sorsa (1999) and Gallego and others (2002) for a review of the literature).
Most of these studies, however, contain serious methodological problems, making it difficult to accept any conclusion with confidence. For example, most used net inflows as the dependent variable, but government actions (including outflows liberalization, debt prepayment, and debt conversion programs) reduced the net inflows independently of the URR by increasing the outflows. Along with the liberalization of capital outflows, there were also changes in the administrative regulation of capital inflows. Likewise, the operation of the URR itself changed over time, as the authorities tried to close loopholes and increase effectiveness by widening its coverage and raising its rate. The exclusion of certain short-term flows (such as trade credits) may also have biased the results of many of the studies, given the substitutability that existed between transactions subject to the URR and those that were not. For these and other reasons, Nadal-De Simone and Sorsa (1999) concluded that it was “premature to point at the Chilean experience as supportive of the effectiveness of controls on capital inflows.”
Some of these methodological problems have been addressed in a more recent study by Gallego and others (2002). The study extends previous research by considering the endogeneity of the URR (the central bank may tighten the URR in response to changes in the strength of capital inflows), the effect of administrative controls on capital flows, and using a much longer sample period covering 1989–2000. A significant contribution of the study is its consideration of the URR’s “effective cost,” which incorporates both “tax effectiveness” and “cost” and essentially measures how binding the URR was. The findings of this study broadly support the conclusions of previous work: (1) the URR temporarily allowed domestic interest rates to increase relative to international rates;1 (2) it had no significant effect on the real exchange rate; (3) it significantly reduced the volume of capital inflows, though the effect diminished over time; and (4) it unambiguously changed the composition of capital inflows in favor of longer maturities. The less ambiguous effect of the URR on total inflows is new, but is supported by the findings of other recent research (see, for example, Le Fort and Lehman, 2003; Ffrench-Davis and Tapia, 2004).1 This, however, was achieved probably at the expense of increasing the cost of capital to smaller domestic firms with limited access to international capital markets (Forbes, 2003; Gallego and Hernandez, 2003).
In asking the last two questions, in particular, we are not seeking to assess the IMF’s policy advice in individual countries against a specific yardstick of “appropriateness,” since, as already noted, there is no such agreed measuring rod in many circumstances. Rather, the aim is to collect evidence related to two common (and, to some extent, contradictory) criticisms of the IMF’s approach, namely that (1) it adopted a “one size fits all” approach in its policy advice, and (2) it was “inconsistent” by giving different policy advice to countries in broadly similar situations. In making such judgments, we faced a number of limitations—most notably that the rationale for particular policy advice was not always spelled out in the relevant staff reports.
The evaluation does not address the question of whether capital account liberalization leads to faster growth (or generates other benefits)—an issue on which the wider body of research evidence has not reached definitive conclusions—or whether the IMF Articles of Agreement should be amended to give the IMF an explicit mandate for capital account liberalization and jurisdiction on member countries’ capital account policies. Many aspects of these issues are not susceptible to evidence from the evaluation. In the case of the second issue, however, the evaluation does shed some light on whether ambiguities about the IMF’s institutional role with regard to capital account matters has affected its country work in this area.
Most of the country-based analysis will use a sample of emerging market economies, for which private capital flows have been important. This selection of countries is justified by the fact that emerging market economies received almost all of the private capital flows to developing countries in the 1990s and arguably required the close attention of the IMF. The largest 25 recipients, for example, accounted for as much as 90 percent of total cross-border investments during 1990–2002.6 Moreover, it is primarily for the possible role played in these countries that the IMF has been criticized.
It should be clearly stated at the outset that the choice of emerging market economies may serve to make the IMF’s role in capital account liberalization appear less significant than it actually was. From 1992 to 1997, for example, there was a significant reduction in the number of IMF member countries with capital controls, and much of this reduction was accounted for by low-income countries, including those in sub-Saharan Africa (Figure 1.2).7 It is possible that the IMF had a more direct role in encouraging capital account liberalization in some of these lower-income countries that relied on IMF financing.8 On the other hand, if the number of countries is weighted by GDP, there was a sharp rise in capital account restrictiveness in the early 1990s (reflecting the fact that a number of former socialist economies joined the IMF); for the period as a whole there was little change in the degree of capital account openness among the IMF’s developing country membership (Figure 1.3). This means that the sample, in which there is a greater representation of larger or higher-income developing countries, may well be biased toward those that tended to maintain some capital account restrictions (see below for the list of countries included in the sample).
Figure 1.2.Countries with Capital Controls1
Source: IMF, Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER).
1 Based on a one (controlled) or zero (not controlled) classification (covering all capital account transactions), as provided by the AREAER. There was a definitional change from 1997 to 1998.
Source: IMF, Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER).
1 Based on a one (controlled) or zero (not controlled) classification (covering all capital account transactions), as provided by the AREAER. There was a definitional change from 1997 to 1998. GDP shares are based on 1990–2000 averages.
2 The line would shift downward by about 5 percentage points if China and India were excluded.
The evaluation will pay particular attention to country experiences with capital account liberalization (in terms of speed, sequencing, and preconditions) and policy responses to capital flows, including the temporary use of capital controls, and the IMF’s role and advice in these areas. In discussing controls on capital outflows introduced in the context of a capital account crisis, it must be stressed that the focus will remain on issues specific to the capital account, and we will not consider broader crisis management issues (including, for example, private sector involvement and debt restructuring). Likewise, no attempt will be made to establish, in the context of a specific country, causality between capital account liberalization and a subsequent capital account crisis, although vulnerabilities to crisis created by a particular policy toward capital account liberalization may be noted.
In discussing capital account openness in specific countries, the evaluation will focus on de jure (as opposed to de facto) controls on capital transactions as defined by the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). Clearly, when one investigates the economic impact of capital account liberalization, one must define capital account openness in a way that has an operational content. Edison and Warnock (2003), for example, suggest such an operational measure that takes account not only of the existence but also of the intensity of capital controls.9 In this report, we are not asking questions about the economic impact of a particular control measure. Instead, we are more interested in knowing, for example, what the IMF said about removing or introducing a particular control measure. We are, therefore, focusing on de jure controls.
Sources of Evidence
The evaluation roughly covers the period 1990–2004 and uses two types of documentation. First, it will use Executive Board papers and minutes of discussions on systemic themes, including World Economic Outlook (WEO) and International Capital Markets Report (ICMR) or Global Financial Stability Report (GFSR) exercises,10 Occasional Papers, Working Papers, and various issues of the IMF Survey (for management speeches). The evidence gathered from these sources will be used to consider how the IMF viewed capital account issues over time, including whether it had a consistent approach and how effectively it adapted this approach in light of experience.
Second, the evaluation uses the IMF’s country documents, including staff reports for Article IV consultations and program reviews, internal briefing papers and back-to-office reports for staff missions, staff memorandums and notes prepared for particular country issues, the minutes of relevant Board discussions, and technical assistance reports. This evidence is drawn from four overlapping groups of countries:
- Countries for which only staff reports (and, in some cases, summings up of Board discussions) are used for the period 1990–2002. There are 15 countries in this category: Bulgaria, China, Croatia, Estonia, Israel, Lebanon, Peru, the Philippines, Poland, Romania, Russia, the Slovak Republic, Slovenia, South Africa, and Ukraine.
- Countries for which, in addition to staff reports, confidential internal documents are used for the period 1990–2002. There are 12 countries in this category: Colombia, Chile, the Czech Republic, Hungary, India, Latvia, Lithuania, Malaysia, Mexico, Thailand, Tunisia, and República Bolivariana de Venezuela.11
These first and second groups of countries, numbering 27, constitute the main sample upon which country analysis for 1990–2002 is primarily based (see below for the selection criteria).
- Countries that have requested technical assistance from the IMF on aspects of capital account liberalization, for which technical assistance reports are analyzed. There are 15 countries in this category: Belarus, China, Colombia, the Czech Republic, Hungary, India, the Islamic Republic of Iran, Kazakhstan, Latvia, Lesotho, Peru, Poland, Russia, Tanzania, and Tunisia.
- Countries with ongoing capital account issues for which confidential internal documents are used for the period 2003–04.12 There are 14 countries: Bulgaria, China, Colombia, Croatia, India, the Islamic Republic of Iran, Kazakhstan, Libya, Morocco, Romania, Russia, South Africa, Tunisia, and Venezuela.13
The 27 countries in the first and second groups are chosen on the basis of the size of portfolio capital flows (absolute or relative to GDP) during 1991–2002 (see Appendix 4), our own qualitative judgment of the degree of capital account openness in 1990, and the changes introduced during the 1990s. The list includes: (1) countries that significantly liberalized the capital account during the 1990s; (2) countries that either still maintain or have until very recently maintained significant controls on capital account transactions; and (3) countries that introduced measures to restrict capital account transactions over the period. The second group of countries was selected from this larger group for more in-depth examinations, based on our own judgment of the learning potential—the important criteria informing this judgment were diversity of experience and outcome—in order to make sure that we cover varied experiences with, and different stages of, capital account liberalization. Argentina, Brazil, Indonesia, and Korea were all important recipients of international private capital inflows during much of the 1990s but are not included in the sample, because the IEO’s earlier evaluations (IEO, 2003 and 2004) have already examined their relationships with the IMF (Box 1.3). This smaller sample, for example, allows us to take a closer look at the role of the IMF in countries that substantially eased restrictions on capital transactions during the early 1990s, the nature of IMF advice for countries that took a gradual approach to capital account liberalization, and the IMF’s views in the context of specific country experiences with capital controls.
Box 1.3.Capital Account Liberalization in Indonesia and Korea
The role of capital account liberalization in the East Asian crisis of 1997 has been a major topic of discussion. An earlier IEO report (IEO, 2003) discusses the effectiveness of the IMF’s precrisis surveillance in identifying financial sector vulnerabilities created by capital account liberalization in Indonesia and Korea. Drawing on this report, we briefly review the IMF’s role in these two countries. The broad message is that IMF surveillance failed to assess fully the underlying risks but that it did not play a major role in formulating the particular capital account liberalization strategy adopted by the authorities.
Indonesia had removed most controls on capital outflows by the late 1980s, and is often cited as an example of a country that had liberalized its capital account before the current account. Indonesia, however, retained controls on various categories of capital flows throughout the 1990s. Almost all the liberalization measures taken from the late 1980s to the mid-1990s were related to the liberalization of direct investment inflows. Rapid capital inflows that began in 1990 took place against the background of financial sector liberalization and domestic capital market development. At the time of the crisis in 1997, a considerable number of controls remained on many types of capital account transactions.1 In response to the large capital inflows, the IMF staff advocated tight fiscal and monetary policies, greater exchange rate flexibility, accelerated structural and banking sector reforms, and even faster external debt repayment. The IMF did not push a particular path or pace of capital account liberalization. However, it underemphasized the risks of short-term capital inflows that were vulnerable to a sudden shift in sentiment, and did not fully appreciate the weakness of the banking sector and the vulnerability created by the country’s buildup of external debt.
In Korea, it was in the context of OECD accession that, in 1994, a Foreign Exchange System Reform Plan was announced to achieve full capital account convertibility in five years, in three stages (Kim and others, 2001; Cho, 2001). The process began first with the liberalization of capital outflows, followed by a gradual easing of restrictions on foreign investment in the domestic stock market and short-term trade-related borrowing. Notwithstanding these measures, however, Korea’s approach to capital account liberalization remained cautious. At the time of its OECD accession in 1996, Korea retained a number of reservations to the Code of Liberalization of Capital Movements, particularly regarding the liberalization of long-term capital inflows.2 The IMF staff was aware of the weak banking system but did not sufficiently appreciate the vulnerabilities created by the buildup of short-term external borrowing by weak, poorly regulated financial institutions. Its view on Korea’s particular choice of sequencing was that the speed of liberalization should be accelerated. The IEO report states that staff papers and Board discussions on Korea were “concerned primarily with the speed of liberalization (typically recommending a faster process)” and that “[issues] of sequencing and supervision were inadequately addressed in the surveillance process.” Further liberalization of the capital account proceeded in the context of a program supported under the 1997 Stand-By Arrangement.1 These included nonresident purchases of Indonesian shares; the sale or issue of money market instruments abroad by residents; the granting of commercial credits by nonresidents to residents; purchases of land by nonresidents; bank borrowing from abroad; and bank lending to nonresidents (Johnston and others, 1997).2 As a result, at the time of the 1997 crisis, controls of one type or another remained on such capital account transactions as: issues of foreign-currency-denominated securities by residents; purchases of local securities by nonresidents; purchases of money market instruments by nonresidents; external borrowing by banks; inward direct investments; and even some trade credits (Johnston and others, 1997; Kim and others, 2001).
In order to aid the evaluation, we have created an index of de jure capital account openness that utilizes a more detailed classification of capital account transactions than the simple 1/0 system. In particular, following Miniane (2004), we assign 1 (or 0) to each of the 10 categories of capital account transactions as reported in the AREAER when a restriction is present (absent), and express the sum in percentage terms. This index has been calculated for the 12 core countries to which we give closer attention (see Appendix 5).14 It turns out that 7 of these countries maintained moderate to extensive restrictions on capital account transactions almost consistently during 1990–2002; 2 countries had a largely open capital account; another 2 countries eased restrictions significantly in the late 1990s or early 2000s; and 1 gradually reduced restrictions over the period. As a result, when we look at the average index of capital account openness for these countries, we find that the index remained relatively high throughout the period, but observe a gradual decline in restrictiveness (Figure 1.4).
Figure 1.4.Average Capital Account Openness in 12 Sample Countries
Source: IEO estimates based on MFD data. See Appendix 5.
Organization of the Report
The rest of the report is organized as follows. Chapter 2 “General Policy and Analysis” reviews the legal basis for the IMF’s work on capital account issues, intellectual and operational developments within the IMF in this area from the early 1990s to the early 2000s, and how the issues were viewed in the IMF’s multilateral surveillance work. The following two chapters present an analysis of the IMF’s approach to capital account liberalization based on work in individual countries. Chapter 3 “Advice to Member Countries” assesses the IMF’s specific advice to member countries during 1990–2002 on capital account liberalization, macroeconomic and structural policies to manage large capital inflows, and the temporary use of capital controls. Chapter 4 “Ongoing Country Dialogue on Capital Account Issues” provides an overview and assessment of the IMF’s latest country work on capital account issues. Chapter 5 “Major Findings and Recommendations” summarizes major findings and suggests two broad recommendations to help improve the IMF’s operations in the area of capital account issues.
The main body of the report is followed by six appendixes. Appendix 1 “A More Detailed Assessment of Some Country Cases” provides a more in-depth analysis of how the IMF viewed capital account issues over time in the context of four countries with diverse experiences: the Czech Republic, Colombia, Tunisia, and Venezuela. Appendix 2 provides an overview of relevant staff research on capital account topics during 1990–2004. Appendix 3 reviews the IMF’s public communications on capital account issues, focusing on management speeches and other public statements. Appendix 4 summarizes quantitative indicators of capital flows in the main sample of 27 countries. Appendix 5 depicts the indices of capital account openness for the 12 core countries. Finally, Appendix 6 provides a list of people interviewed by the evaluation team.
Since the fifth edition of the IMF’s Balance of Payments Manual was published in 1993, the term used for statistical purposes has been the “capital and financial account.” However, this report follows the established practice, both within the IMF and in the academic literature, of using the term “capital account” to describe the subset of the balance of payments that covers all non-current international transactions.
The broad international interest in capital account issues that existed during the 1990s can be seen, for example, in the coverage given by successive issues of the UNCTAD’s Trade and Development Report (see, in particular, Chapter 5 of the 1999 report).
While such a policy was often referred to as part of the “Washington consensus,” full capital account liberalization was in fact not one of the 10 policy reforms that Williamson (1990) considered as forming the Washington consensus. The presumed consensus was not on liberalization of capital flows in general, but rather more specifically on that of foreign direct investment.
Academic proponents of these views are Desai (2003), Stiglitz (2000, 2002, and 2004), Wade (1998–99), and Wade and Veneroso (1998).
The IMF’s instruments include surveillance, technical assistance, and IMF-supported programs.
This figure does not include foreign direct investments. The countries in the sample used in this report account for about 40 percent of this total during 1990–97 (reflecting the exclusion of Argentina, Brazil, Indonesia, and Korea), but the share increases to over 50 percent during 1998–2002.
Judgment about the presence or absence of capital controls in each country is based on a one (controlled) or zero (not controlled) classification provided by the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). It should be noted that this binary classification does not take account of either the intensity or the number of controls.
A recent econometric study by Joyce and Noy (2005) shows that an extended IMF-supported program was statistically significant in explaining a country’s decision to remove capital controls in the 1990s, suggesting that low-income countries often liberalized the capital account in the context of IMF financial support.
See also Prasad and others (2003, pp. 6–8), for a discussion of the difference between “the existence of de jure restrictions on capital flows” and “de facto financial integration in terms of realized capital flows.”
In 2002, the Global Financial Stability Report replaced the International Capital Markets Report.
Brief field visits were made to receive the views of officials and other experts in several of these countries: Chile, Colombia, the Czech Republic, Hungary, India, Latvia, Mexico, and Tunisia. A list of interviewees is provided in Appendix 6.
These countries have been identified on the basis of a questionnaire sent to recent mission chiefs and interviews with IMF staff. The list of countries is not meant to be exhaustive.
Additional information on ongoing issues was obtained from interviews with senior IMF staff.
IMF (1999, pp. 83–96) offers a methodology of calculating an index of capital account restrictions based on even more detailed transaction categories and ranks 41 industrial, developing, and transition economies for 1996. See also Johnston and Tamirisa (1998). Because of data limitation, however, we instead follow—with some modifications—the methodology of Miniane (2004), who extends the indices to the pre-1996 period but for a smaller set of 34 countries and based on 10 categories of capital account transactions.