international financial integration and

Globalization of the world economy hasbeen driven by a variety of forces: risingtrade in goods, increasing internationalcapital flows, greater technological

spillovers, and growing labor mobility. This chap-ter examines the increase in capital movementand, in particular, the impact of international fi-nancial integration on developing countries.Views on this issue have changed since the early1990s, when international financial deregula-tion, and the associated capital flows from indus-trial to developing countries, appeared to be fu-eling faster growth and catch-up. A series offinancial crises, starting with industrial countriesin the European Exchange Rate Mechanism(ERM) in 1992 and 1993 but then moving on ina significantly more virulent form to developingcountries with the 1995 Tequila crisis in Mexico,the Asian crisis of 1997–98, and the Russian andLatin American crises in 1998–2000, have madeclear that international capital flows have risks aswell as benefits.

Opening the financial market to the rest ofthe world is a complex and often long-drawn outprocess that involves lifting restrictions on for-eign direct investment (FDI) flows and long- andshort-term financial instruments. The bulk ofthis chapter assesses the net benefits to output ofinternational financial market integration, focus-ing on the channels through which capital flowsfrom industrial to developing countries might af-fect economic activity in the recipient countries.The remaining part of the chapter describes thelessons learned from sequencing. On the posi-tive side, less restrictive capital controls andhigher international capital flows can increaseinvestment possibilities, create technologyspillovers, and deepen domestic capital markets.However, they can also create instability by leav-ing countries open to sudden reversals in capitalflows. This risk can increase if domestic macro-economic and financial policies are weak or in-

consistent, or if financial systems are not suffi-ciently developed to cope with large capitalflows.

The chapter first documents the increase in fi-nancial integration over the last three decades,using two complementary measures of capital ac-count liberalization, one of which has not beenwidely applied in previous analysis. These meas-ures indicate that the industrial countries liberal-ized their capital accounts early on whereas inthe developing countries the shift toward capitalaccount liberalization in general was slow, butquickened in the early 1990s. These indicatorsare then used to examine whether countrieswith more open capital accounts have had a bet-ter economic performance than countries withmore restrictive capital accounts. The analysis re-veals a weak relationship between growth andcapital account liberalization and, like otherstudies, has difficulty finding evidence of a sig-nificant relationship. That said, work using thenew measure of openness suggests that it is pos-sible to identify a significant impact from capitalaccount liberalization on investment and finan-cial development (as well as positive spilloverfrom FDI to growth in countries with adequatehuman capital). The effect of a “typical” liberal-ization through these two channels is estimatedto increase growth by !/2 percent a year or more.But, opening up the capital account can also en-tail costs if it is not sequenced and implementedappropriately, particularly when the institutionalframework is weak. This helps to explain theweaker direct relationship between liberalizationand growth.

Prior to analyzing these issues in more detail,some definitions of key concepts are needed.First, in this chapter, international financial lib-eralization—or opening the capital account—will be simply referred to as liberalization, whilethe level of such integration will be referred toas openness. Both concepts refer to the degree

145

CHAPTER IVINTERNATIONAL FINANCIAL INTEGRATION ANDDEVELOPING COUNTRIES

to which restrictions on financial transactionsbased upon citizenship or residency, or relatedto cross-border transactions, have been elimi-nated.1 Second, domestic financial liberalization,which relates to the reduction of distortions inthe domestic financial system, will be referred toas domestic liberalization. While both forms ofliberalization are important and often go handin hand, the focus of this chapter is on the ef-fects of international financial liberalization.

Trends in Capital Account Liberalizationand Capital Flows

Generally speaking, the increasing financiallinks across countries, and especially between in-dustrial and developing countries, have been as-sociated with the liberalization of both interna-tional and domestic financial markets. Thissection describes the trends in capital accountliberalization and the associated rapid growth ofinternational capital flows.2

Before reviewing the experience over the last30 years, two lessons from history are worth not-ing. First, at a general level, openness to tradeand, to a somewhat lesser extent, open financialmarkets have been associated with higher growthand greater convergence across regions. The col-lapse of international trade and capital flows inthe interwar period involved slow growth and lit-tle convergence. By contrast, the period since1950, which has seen rapid opening of trade andfinancial markets, has had high rates of growthand some convergence, although progress wasmost rapid in the early part of the period, whentrade links were growing fastest.3 Second, the lib-eralized capital markets that were a feature of

the pre-1914 classical gold standard supportedrelatively larger capital flows than today, but hadinternational financial crises that were, if any-thing, somewhat less frequent and less disruptiveof output than those since 1973 (Bordo and oth-ers, 2001). The relatively greater stability beforeWorld War I appears to reflect the large degreeof credibility of domestic policies associated withthe gold standard, greater ties in terms of her-itage between Britain (the major lender) andmany of the important borrowers, and the factthat much of the money was used to finance rail-roads and other infrastructure, which was easilymonitored and directly used in producing ex-ports and hence earning foreign currency.

Since 1970, the experience of developingcountries with capital account liberalization hasdiffered dramatically from that of the industrialcountries. Nevertheless, researchers face a com-mon problem in both industrial and developingcountries when attempting to measure financialopenness (see Box 4.1). Liberalization involves acomplex process of reducing controls across awide array of assets. There are large differencesin the nature, the intensity, and the effectivenessof the various impediments governments haveput into place to limit the movement of cross-border capital flows. Consequently, there are of-ten large gaps between the ideal set of indicatorsand the measures that are available in practice(see Eichengreen, 2001).

This chapter uses two indicators derived fromdifferent methodologies to measure capital ac-count liberalization. Most formal empirical workanalyzing the impact of capital account liberal-ization has used a restriction measure based onthe restrictions on capital flows as reported to

CHAPTER IV INTERNATIONAL FINANCIAL INTEGRATION AND DEVELOPING COUNTRIES

146

1It should be noted that the use of the words “capital account” is inconsistent with the terminology in the 1993 System ofNational Accounts, which switched the name from capital account to financial account. To avoid confusion, this chapteruses the older and more generally understood term.

2Most of the analysis in this chapter is based on a sample of 55 developing countries listed in Table 4.2 (as well as, insome cases, 20 industrial countries). Because of data limitations, the reported regression analysis reports results for 38 de-veloping countries. Given the long period over which the analysis is conducted, several countries currently defined as in-dustrial (Cyprus, Israel, Korea, and Singapore) are included with developing country group. The following countries wereexcluded: countries with a population below 500,000 people, Heavily Indebted Poor Countries (which mostly receive offi-cial flows), and transition economies (partly due to lack of data). For recent studies of the HIPCs and of the transitioneconomies, see the World Economic Outlook May 2000 and World Economic Outlook October 2000, respectively.

3See World Economic Outlook, May 2000; and O’Rourke (2001).

TRENDS IN CAPITAL ACCOUNT LIBERALIZATION AND CAPITAL FLOWS

147

Assessing international capital account liberal-ization is complex, largely reflecting the difficultyin quantifying the extent and the effectiveness ofcapital controls. No fully satisfactory measure isavailable, which puts a premium on examiningthe results from alternative indicators. This chap-ter uses two measures derived from differentmethodologies to measure the effectiveness ofcapital controls. The first, used in most previousanalysis of liberalization, is a restriction measure,constructed as an on/off indicator of the exis-tence of rules/restrictions that inhibit cross-bor-der capital flows or discriminate on the basis ofcitizenship or residence of transacting agents.The second and complementary indicator of cap-ital account integration used in this chapter (andonly recently available) is a measure of opennessthat is based on the estimated stocks of gross for-eign assets and liabilities as a ratio to GDP.

The measure of restrictions comes from theInternational Monetary Fund’s Annual Report onExchange Arrangements and Exchange Restrictions(AREAER).1 This publication provides a writtendescription of current controls and, since 1967,on/off indicators of the presence or absence ofrestrictions on the current and capital account.Most studies focus on the indicator measuring re-strictions on payments for capital transactions,the only measure that focuses on the capital ac-count.2 In 1996, however, the presentation waschanged to a new system, which contains moredetail on capital account restrictions, but is notbackwardly compatible with the earlier indicator.

The main difficulty with the restriction meas-ure is that it does not take into account the in-tensity of the controls, or differentiate acrosstypes of restrictions. Recent work has addressedthis issue by assessing the intensity of enforce-

ment of the controls using the narrative informa-tion contained in the IMF’s AREAER. However,due to the labor-intensive nature of this work,such measures for developing countries are cur-rently only available for a few years in the 1970sand 1980s.3 Other studies have proposed restric-tiveness measures for a limited number of coun-tries that focus on controls of international salesand purchases of equities over recent years.4 Yetothers have used interest rate differentials andforward premium/discount to assess the degreeof capital mobility and capital account liberaliza-tion (see Frankel and MacArthur, 1988). The rel-atively small coverage of countries and time peri-ods limits these extended measures ofrestrictions for broad cross-sectional studies ofthe impact of capital controls, such as the back-ground study used for this chapter.

The second indicator of capital account liber-alization uses gross holdings of internationalassets and liabilities to measure openness to capi-tal transactions.5 This measure looks directly atone of the consequences of liberalization,namely larger ownership of foreign assets in thecountry and liabilities elsewhere. It applies thesame logic as similar variables used to measuredomestic financial depth, such as the stock ofcredit to the private sector as a ratio of GDP, andtrade openness, defined as the sum of exportsand imports as a ratio to GDP. Stocks of financialassets are a better measure of capital account

Box 4.1. Measuring Capital Account Liberalization

1The first work to employ this dataset was Grilli andMilesi-Ferretti (1995).

2Some researchers supplement this informationwith the additional external restriction measures avail-able in the AREAER; such as whether there is a sepa-rate exchange rate for current account and capital ac-count transactions, whether there are restrictions onpayments for current account transactions, whetherthere is the requirement of surrender or repatriationof export proceeds.

3See Quinn (1997). The data on developing coun-tries are limited to three years: 1973, 1982, and 1988.

4Henry (2000) and Bekaert and Harvey (2000)compile from a variety of sources dates that countriesliberalized their equity markets, while Edison andWarnock (2001) propose a measure that is based onrestrictions of foreign ownership of domestic equities,using the ratio of the market capitalization of thecountry’s Global index to the Investable index that iscompiled by the International Finance Corporation(IFC) in their Emerging Market Database that is nowmaintained by Standard and Poor’s.

5These openness measures have not been widely usedin the literature. O’Donnell (2001) and Chanda(2000) employ a similar openness measure to the onedescribed in the background study. Kraay (1998) de-fines openness using the sum of capital inflows andoutflows as a ratio of GDP.

the IMF by national authorities. This indicatordirectly measures capital controls, and countriesare either classified as open or closed. By its na-ture, however, this measure does not capture dif-ferences in the degree of liberalization: for ex-ample, a country might liberalize some, but notall, categories of their capital account, and, inaccordance with the restriction measure, it couldstill be labeled as closed. In addition, it is onlyavailable until 1995, when a new and more re-fined measure—not backward compatible—wasintroduced.

A second and complementary indicator ofcapital account integration is an openness meas-ure (described in Box 4.1), based on the esti-mated gross stocks of foreign assets and liabili-ties as a ratio to GDP.4 It is inspired by the use ofsimilar variables to measure domestic financialdepth, such as the stock of credit to the privatesector as a ratio to GDP. The openness measureis created by calculating the gross level of FDIand portfolio assets and liabilities via the accu-mulation of the corresponding inflows and out-flows.5 If this gross stock measure is high, it im-

plies that a country is open, in the sense that it isor has been experiencing significant private sec-tor flows to and from the rest of the world.

For both developing and industrial countries,international gross private capital flows havegrown markedly, but with considerable variationover time (Figure 4.1). The openness measurehas also increased over time, but in a more grad-ual and backward-looking fashion, as it measuresthe stock of assets (Figure 4.2). Clearly, however,the openness measure does not just capture therestrictiveness of capital controls, but also theimpact of all other factors influencing the levelof capital flows, such as the nature of domesticfinancial markets. In this respect, the opennessmeasure of liberalization is analogous to measur-ing the trade regime through calculating the ra-tio of exports and imports to GDP, while the re-striction measure of liberalization is similar tomeasures of trade restrictions in terms of tariffand nontariff barriers.

What do the restriction measure and theopenness measure tell us about liberalizationover time? In industrial countries the behavior


148

openness than underlying flows, as they are lesssubject to temporary disturbances, although bytheir nature they tend to be a lagging indicatorof policy changes. In addition, various sub-com-ponents, related to different types of financialtransactions, can be analyzed.

The overall openness measure is computed byusing the stock measures of both assets and lia-bilities of FDI and portfolio transactions (thelatter encompassing equities and debt transac-tions).6 When interpreting this measure, it is

important to recognize that cross-border capitalmovements are influenced by a wide range ofpolicy outcomes and do not reflect barriersalone. For example, these measures are also af-fected by a range of policies and circumstancessuch as the stance of monetary and fiscal policy,the size of the domestic economy, and condi-tions in the rest of the world.

The background study for this chapter usesboth measures. In general, the results using therestriction measure are not particularly encour-aging. By contrast, as reported in the text, theopenness measure and its subcomponents ap-pear to provide more intuitively plausible resultsboth in the raw data and the econometricanalysis.

Box 4.1 (concluded)

6The underlying stocks are calculated by accumulat-ing the corresponding flows with appropriate valua-tion adjustments, and were constructed by Lane andMilesi-Ferretti (forthcoming).

4These underlying stock data were developed and described by Lane and Milesi-Ferretti (forthcoming). A similar meas-ure using the same underlying stock data has been considered by Chanda (2000) and O’Donnell (2001).

5Individual country data on stocks of bank loans are too fragmented to be included in the measure.

of the two measures is similar, and confirms thatindustrial countries have become considerablymore open over time. A particularly rapid de-cline in controls occurred during the 1980s,when the members of the European Community,now the European Union, liberalized capitalcontrols. Following this, there was a dramaticrise in cross-border capital flows. Based on theopenness measure, the top four most open in-dustrial countries have been: Canada, Nether-lands, Switzerland, and the United Kingdom.

In developing countries, the story is morecomplex. In general, both measures suggest aless dramatic shift toward liberalization andopenness than in industrial countries. The fol-lowing points are worth noting:

• For the developing countries as a whole, therestriction measure suggests that, after a pe-riod of liberalization in the 1970s, the trendtoward openness reversed in the 1980s.Liberalization resumed in the early 1990s,but the pace has been relatively slow; themeasure indicates that the current level ofthe indicator on average is only at the samelevel as it was in the late 1970s.

• The openness measure shows a modest de-cline in openness to capital flows during theearly 1970s, followed by a moderate increasein the 1980s, which accelerated sharply inthe early 1990s. While the most recent ac-celeration may reflect a more rapid liberal-ization than the relatively crude restrictive-ness measure suggests, it also reflects theincreasing openness of industrial countries,as well as the more general trend towardglobalization.

Within developing countries, there have beenimportant differences across regions:

• For Asia, there has been an upward trend inthe openness measure since the late 1970s,while the restriction measure has changedvery little.6 While it is probable that the re-


149

0

5

10

15

20

0

5

10

15

20 Africa Asia

Sources: IMF, International Financial Statistics; and IMF staff estimates. For country coverage, see Table 4.2.

0

5

10

15

20

0

5

10

15

20 Middle East and Europe Western Hemisphere

Gross FDI Portfolio and banks

0

5

10

15

20

0

5

10

15

20 Developing CountriesAdvanced Economies

1970 74 78 82 86 90 94 99

Figure 4.1. Gross Capital Flows(Percent of GDP)

Gross capital flows have risen over time, but are also volatile.

Developing Countries by Region1

1

1970 74 78 82 86 90 94 99

1970 74 78 82 86 90 94 99 1970 74 78 82 86 90 94 99

1970 74 78 82 86 90 94 991970 74 78 82 86 90 94 99

6Indonesia and Malaysia are defined as open using therestrictiveness measure from the early 1970s on, whileKorea is classified as closed through the end of the sam-ple in 1995.

strictiveness measure underestimates the de-gree of liberalization, the rapid rise in open-ness probably also reflects other factors, in-cluding the opening up of the Chineseeconomy, and the rapid development andgrowth—and associated capital require-ments of—the newly industrializedeconomies and members of the Associationof South East Asian Nations (ASEAN).While a substantial proportion of inflowstake the form of foreign direct investment,there was a very large increase in portfolioand bank flows in the early 1990s, which ul-timately proved unsustainable.

• For Africa, both the restriction measure andthe openness measure show little changeuntil the early 1990s. The substantial in-crease in openness in the 1990s is mainlydue to events in a limited number of coun-tries. In particular, this has involved largeportfolio flows to South Africa, which hasthe most developed financial markets in theregion, as well as large FDI flows to Lesothoand Nigeria, which reflected investments inresource production.

• In the Middle East and Europe the restrictionindex shows some liberalization in the early1970s, as oil producers such as Oman,Qatar, Saudi Arabia, and the United ArabEmirates lowered restrictions. The opennessmeasure shows a more gradual increase inliberalization. By contrast, gross capitalflows spiked with the two oil price hikes, buthave stagnated since then.

• For the Western Hemisphere, both measuresshow that, on average, the region was rela-tively open in the 1970s, with Argentina,Chile, and Mexico having liberalized. Theopenness of the region led to large bank-based inflows of oil surpluses in the mid- tolate 1970s, but many countries then im-posed controls in response to outflows dur-ing the 1980s debt crisis, sparked by theMexican crisis in 1982. These capital con-trols were relatively ineffective, and capitalflight continued through most of the “lostdecade” until longer-term institutional re-


150

1970 74 78 82 86 90 94 980.0

0.1

0.2

0.3

0.4

0.5Developing Countries

0.6

0.2

0.4

0.0

1.0

0.0

1970 74 78 82 86 90 94 980.0

0.3

0.6

0.9

1.2

1.5Advanced Economies

0.2

0.4

0.8

1.0

0.6

0.8

Figure 4.2. Summary of Measures of Capital Account Openness

The two measures of liberalization show similar overall patterns, but the openness measure points to greater progress in the 1990s.

Openness measure(right scale)

Restriction measure(left scale, inverted)

1

1970 74 78 82 86 90 94 980.0

0.1

0.2

0.3

0.4

0.5

1970 74 78 82 86 90 94 980.0

0.1

0.2

0.3

0.4

0.5Asia

0.8

0.2

0.6

0.0

1.0

0.0

1970 74 78 82 86 90 94 980.0

0.1

0.2

0.3

0.4

0.5Africa

0.2

0.4

0.8

1.0

0.6

0.4

1970 74 78 82 86 90 94 980.0

0.1

0.2

0.3

0.4

0.5Middle East and Europe

1.0

0.6

0.8

0.00.0 Western Hemisphere

0.2

0.6

1.0

0.8

0.2

0.4 0.4

Sources: IMF, Annual Report on Exchange Arrangements and Exchange Restrictions, various issues; International Financial Statistics; and IMF staff calculations. The restriction measure is calculated as the "average" value of the on/off measure for the country group. The openness measure is calculated as the average stock of accumulated capital flows (as percent of GDP) in a country group. For country coverage, see Table 4.2.

1

Developing Countries by Region 2

2

forms allowed the region to reenter interna-tional capital markets at the end of the1980s. Hence, one of the lessons from the1980s debt crisis is that imposing capitalcontrols on markets that have been closelylinked to the rest of the world may be oflimited value in stemming outflows.

• After the setback of much of the 1980s,Latin American liberalization has continuedthrough the late 1980s and early 1990s. Asin Asia, the openness measure shows amuch larger increase, fueled by portfolioflows and FDI. The U-shaped pattern ob-served in the indicators for the WesternHemisphere illustrate the endogeneity in-herent in these measures; if the economicperformance deteriorates, then a countrymight impose controls.

Finally, some countries have played a more im-portant role than others in global capital mar-kets over the last three decades. Among the in-dustrial countries, since 1970 the United Statesand the United Kingdom were both dominantsuppliers and users of capital. More generally,the Group of Five (France, Germany, Japan,United Kingdom, and United States) countrieswere the largest suppliers and users of capital,accounting for about two-thirds of all privatecapital inflows and outflows. For the developingcountries, the bulk of capital flows have been di-rected toward Asia and Latin America (Figure4.3). Asia (primarily China) received a higherportion of FDI than other regions, while coun-tries in Latin America received a higher portionof portfolio flows than other regions. These re-gional differences mirror partly the general his-tory of larger fiscal deficits in Latin Americacompared with Asia, leading to more floating ofgovernment bonds, as well as the earlier devel-opment of some capital markets in LatinAmerica. For example, most investments inChina are FDI flows, partly because local capitalmarkets are less open and developed. Moregenerally, the figures reflect that countries withrelatively liberalized capital accounts receivehigher capital flows whereas countries that aremore closed receive moderate amounts of capi-


151

Figure 4.3. Concentration of Capital Flows: Largest Developing Country Users(Average of 1970–2000)

The largest users of foreign direct investment (FDI) flows are generally in Asia, while portfolio flows are more concentrated in Latin America.

Largest Users of FDI

Largest Users of Portfolio Flows

Sources: IMF, International Financial Statistics; and IMF staff estimates.

Korea(13.8%)

South Africa(7.3%)

Argentina(13.8%)

Mexico(15.8%)

Brazil(19.7%)

All others(24.7%)

China(4.9%)

All others(29.9%)

China(33.2%)

Brazil(10.8%)

Mexico(9.9%)

Argentina(6.2%)

Singapore(6.0%)

Malaysia(4.0%)

tal flows, mainly in the form of foreign directinvestment.7

Overall, according to both the restriction andopenness measures, most regions of the worldtoday have been opening their capital accountsduring the 1980s and 1990s.

The Impact of International CapitalAccount Liberalization on Growth

The impact of liberalization on growth de-pends crucially on the initial conditions andpolicies in the country. From a theoretical per-spective, models of perfect markets and full in-formation suggest that liberalization, by allowinga better allocation of resources across countries,benefits both lenders and borrowers and raisesgrowth. In practice, however, this result dependscritically on a variety of preconditions, includinga supportive and consistent macroeconomic andinstitutional framework. One example is whenliberalization is associated with a governmentpledge to maintain a pegged exchange rate thatis unsustainable in the medium term. As a resultof the unsustainable and inconsistent macroeco-nomic policy, consumers and producers borrowexcessively in the short term from the rest of theworld to bring forward purchases of (temporar-ily cheap) foreign goods, creating a domesticboom. When the exchange rate peg eventuallycollapses, however, domestic demand will col-lapse and international capital flows will reverse,often accompanied by serious banking systempressures.8 A similar story could clearly be toldfor inadequate financial supervision. This, in anutshell, is the tension between the benefits andcosts of open liberalization. While an open capi-tal account can provide great benefits if the ap-propriate institutional requirements are inplace—notably adequate financial supervision

and consistent macroeconomic policies—it canbe destabilizing if they are not.

Financial instability may also affect poverty lev-els and other social conditions. On the onehand, recent evidence shows that growth is oneof the main factors contributing to the reductionof poverty and the improvement of social condi-tions, so that liberalization can help reducepoverty to the extent it increases the long-runimpact on growth. On the other hand, liberaliza-tion can be associated with an increase in macro-economic volatility and the occurrence of crises,which may entail social costs such as worseningincome distribution, poverty levels, and healthand education conditions. This is because thepoor and less educated have very limited accessto financial markets to hedge or diversify risk,and are thus hurt disproportionately more byeconomic contraction and by the frequently as-sociated reduction in public health and educa-tion spending. 9 This points to the value of com-prehensive social safety nets to help mitigate theimpact of structural changes—including that as-sociated with liberalization—on the populationin general and the poor in particular.

Reflecting these considerations, experiencewith liberalization is quite varied. Hence, identi-fying “the impact” of capital market opening ongrowth has been difficult. In many respects, amore fruitful approach is to examine the mainchannels through which liberalization affects theeconomy. On the positive side, liberalization willtend to raise investment by supplementing do-mestic savings, as well as by allowing better riskdiversification and greater consumption smooth-ing. In addition, foreign direct investment flowsin particular can provide technology spilloversvia the transfer of ideas. The impact on domesticfinancial intermediation, however, is mixed.Liberalization can improve the domestic finan-


152

7See Johnston and Ryan (1994); and Mody and Murshid (2001). The impact of liberalization on capital flows dependsalso on timing. When liberalization occurs in tranquil times, capital flows tend to increase, but when liberalization occursduring a crisis, capital flows tend either to not react or to decline—see Edison and Warnock (forthcoming).

8On these issues, see Eichengreen and others (1998); and Calvo and Végh (1999).9On these issues, see, for example, Chapter IV in the May 2000 World Economic Outlook; Dollar and Kraay (2001); and

Asian Development Bank (2001).

cial system over time by strengthening competi-tion, providing access to best practices elsewhere(particularly if foreign financial institutions en-ter the market), and reducing resources spenton circumventing capital controls. However, ex-cessive inflows facilitated by lax financial supervi-sion, macroeconomic policy inconsistencies, orexcessive zeal by foreign investors can over-whelm the ability of the domestic financial sys-tem to allocate funds efficiently, leading to fu-ture financial and social problems.

The opening part of this section provides abroad assessment of the relationship betweencapital market openness and growth, examiningthe raw data, the existing literature, and the re-sults of a background study carried out by IMFstaff and a consultant. The analysis finds a weakpositive link between capital account liberaliza-tion and economic growth. A similar approachfocusing on the channels suggests that capitalmarket openness is linked over time with higherdomestic private investment in developing coun-tries, some positive spillovers from FDI, and aboost to domestic financial depth. However, ex-cessive capital inflows can lead to financial insta-bility, so that the net benefits will obviously de-pend, as already noted, on the strength of thedomestic macroeconomic policies and financialstructures.

Has Liberalization Led to Higher Growth?

Empirical work finds a weak positive relation-ship between international capital account liber-alization and growth. For example, when coun-tries are classified using the openness measuresinto open and closed, it appears that countriesthat are more open to international capital flowstend to grow faster than those countries classi-fied as less open in the 1980s and 1990s (Figure4.4). In this calculation open developing coun-tries are defined as those whose openness meas-ure exceeds the average value for the entire sam-ple period for both industrial and developingcountries, and the remainder are defined asclosed. Although such a correlation ignores thenumerous other factors that explain growth, and

THE IMPACT OF INTERNATIONAL CAPITAL ACCOUNT LIBERALIZATION ON GROWTH

153

(Percent)

More liberalized countries tended to grow faster in the 1980s and 1990s, according to the openness measure.

Figure 4.4. Per Capita Growth by Liberalization in Developing Countries

1970s 1980s 1990s0

1

2

3

4

Closed Open

Sources: IMF, International Financial Statistics; and IMF staff calculations. A country is defined as open when its openness measure exceeds the average value for the entire sample period for both industrial and developing countries. The remaining countries are defined as closed. See Table 4.2 for country classification.

1

1

assumes that liberalized markets affect growthimmediately, approaches using more sophisti-cated statistical models come to similarconclusions.

A growing number of academic studies haveexamined the relationship between capital ac-count liberalization and growth by adding ameasure of such liberalization to the conven-tional growth model (see Box 4.2). The resultsfrom these exercises have been mixed, withabout half identifying a significant positive im-pact and the other half failing to find such a re-lationship (Table 4.1).10 Overall, this suggeststhat liberalization is mildly beneficial forgrowth.

The wide divergence in results reflects a num-ber of differences across studies. First, the coun-try coverage is different, with some authors ana-lyzing industrial countries, others developingcountries, and others a mixture. Second, thereare differences in the sample period, which maybe particularly important for developing coun-tries given the recent nature of many capital ac-count liberalizations. Third, the applied method-ology (cross-sectional, time series, or panel) andthe estimation technique (ordinary least squares,instrumental variables, or generalized method ofmoments) differ across studies. In addition,there are some general drawbacks with the liter-ature that analyzes the relationship between lib-

eralization and growth. The restrictiveness meas-ure of financial controls used in the bulk ofthese studies is relatively crude. Finally, capitalcontrols are often treated as exogenous to thegrowth process, but in practice countries withparticular macroeconomic and financial charac-teristics are especially prone to adopt controls,implying the potential for reverse causality.11

To delve into these issues more deeply, IMFstaff and a consultant undertook a new study fo-cusing on 38 developing countries over the pe-riod 1980 to 1999, using both the restrictionmeasure of liberalization generally included inexisting studies and the new openness measureof liberalization discussed earlier (Table 4.2).Particular attention was paid to exploring the re-sults for developing countries using the open-ness measure, including how openness to portfo-lio flows and FDI affect the results, and howliberalization interacts with the channels dis-cussed above—domestic investment andspillover, financial depth, and institutionalarrangements. Details of the approach are con-tained in Box 4.2.12

On the central issue of the impact of liberal-ization on growth, the results are supportive of amildly beneficial impact. The results, reported inTable 4.3, and other regressions using alterna-tive sample periods, can be summarized asfollows:


154

Table 4.1. Summary Studies on Capital Account Liberalization and Growth

Study Number of Countries Years Covered Effect on Growth

Alesina, Grilli, and Milesi-Ferretti (1994) 20 1950–89 No effectGrilli and Milesi-Ferretti (1995) 61 1966–89 No effectQuinn (1997) 64 1975–89 PositiveKraay (1998) 117 1985–97 No effectRodrik (1998) 95 1975–89 No effectKlein and Olivei (2000) 92 1986–95 PositiveChanda (2000) 116 1976–95 PositiveArteta, Eichengreen, and Wyplosz (2001) 59 1973–92 MixedBekaert, Harvey, and Lundblad (2001) 30 1981–97 PositiveEdwards (2001) 62 1980s Positive

10For a comprehensive survey see Edison, Klein, Ricci, and Sløk (forthcoming).11For example, a country with weak economic performance might choose to adopt capital controls and there is a danger

in such a case to interpret incorrectly that the country’s low growth depends on those controls.12A comprehensive discussion of the results is contained in Edison, Levine, Ricci, and Sløk (forthcoming).

• Increased international financial integration isgenerally associated with an economically mean-ingful rise in growth in developing countries, al-though these effects are generally not statisticallysignificant. When a developing country liber-alizes, moving from a closed to open status,its per capita growth is estimated to rise onaverage by slightly over !/4 percent peryear.13 The lack of significance of most ofthe coefficients presumably reflects the widediversity of developing country experienceswith growth, including the financial instabil-ity associated with liberalization in manycases.

• The growth effects on developing countries comethrough both FDI and portfolio liberalization. Onfurther examination, there are some intu-

itive differences in results between the1980s and 1990s that imply these relation-ships may have altered over time. While thebenefits of FDI liberalization appear to havebeen higher in the 1990s than in the 1980s,that of portfolio liberalization appears tohave been lower, consistent with the finan-cial crises that affected many developingcountries with access to portfolio flows dur-ing the 1990s.

Channels Linking Capital Account Liberalizationwith Growth

The debate over the impact of capital accountliberalization on growth often reflects differ-ences in the assumed potency of the variouschannels through which capital account liberal-ization may occur, including the impact on in-vestment, the existence of technologicalspillovers, and the ability of the financial systemto cope with large inflows. Higher investment,technological spillovers, and deeper financialmarkets are all associated with higher growth


155

Table 4.2. Openness of the Capital Account(According to the openness measure)

Algeria India Papua New Guinea80,90

Argentina Indonesia PeruBangladesh Israel PhilippinesBotswana80 Jamaica70,80,90 Saudi ArabiaBrazil Jordan Singapore70,80,90

Chile90 Kenya South AfricaChina Korea Sri LankaCosta Rica90 Lesotho90 Syrian Arab RepublicColombia Malaysia80,90 ThailandDominican Republic Mauritius Tunisia90

Ecuador Mexico90 Trinidad and Tobago70,80,90

Egypt80 Morocco TurkeyEl Salvador Namibia90 United Arab EmiratesGabon70,80 Nepal UruguayGambia Nigeria80,90 Venezuela90

Guatemala Oman ZimbabweHaiti Pakistan

Note: The table lists the countries identified in the sample and in-dicates those that are classified as open (according to the opennessmeasure defined in the text) in the 1970s, 1980s, and 1990s. A su-perscript 70 denotes that the country was open in the 1970s, super-script 80 that it was open in the 1980s, and superscript 90 that itwas open in the 1990s. Developing countries in bold (plusSwaziland) were included in the regression sample. Tables andFigures in the text encompass four additional developing countries(Bhutan, Comoros, Cyprus, and Qatar) and 20 industrial countries(Australia, Austria, Canada, Denmark, Finland, France, Germany,Greece, Ireland, Italy, Japan, Netherlands, New Zealand, Norway,Portugal, Spain, Sweden, Switzerland, the United Kingdom, and theUnited States).

Table 4.3. Liberalization and Economic Growthfor Developing Countries(Percent change in per capita growth per year)

Restriction Opennessmoves from moves from

closed to open closed to open1

Overall liberalization 0.3 0.3FDI liberalization . . . 0.2Portfolio liberalization . . . 0.3

Note: Bold indicates significance at the 5 percent confidence level(using a one-tailed test), and italic indicates significance in regres-sions that account for endogeneity. Control variables are level of in-come in 1980; level of secondary schooling in 1980; inflation; andgovernment balance. For the details of the analysis see Edison,Levine, Ricci, and Sløk (forthcoming).

1Overall, FDI, and portfolio openness are assumed to increase by40, 30, and 10 percent of GDP, respectively. Each change is equiva-lent to the difference between the average of the correspondingstock for open and closed developing countries (defined as thoseabove and below the average for all countries in the sample).

13Liberalization is proxied by a move from one to zero when using the restriction indicator, or by the increase in the av-erage level of the openness measure as a country moves from closed to open. The calculation implies that the overall FDIand portfolio openness measures are assumed to increase by 40, 30, and 10 percent of GDP, respectively, in the move fromclosed to open. Table 4.2 contains details on the countries defined as open and closed using the openness measure for the1970s, 1980s, and 1990s.


156

The background study whose results are re-ported in this chapter, as with earlier studies,uses a standard growth model (Eichengreen,2001). This model, which has been widely usedto test numerous hypotheses about which factorsaffect economic growth, relates real per capitaGDP growth to initial per capita GDP (to cap-ture the convergence effect), and various otherconditioning variables.1 In the backgroundstudy, a relatively standard set of such variableswere included: the initial level of education at-tainment (proxying the level of human capital),a measure of the government surplus as a shareof GDP, and a measure of average inflation(both controlling for macroeconomic stability).2

Some studies also include real investment as anexplanatory variable of growth, but since invest-ment is a key channel through which capital ac-count liberalization might affect growth, it wasnot included.

In the background study by IMF staff and aconsultant, the basic growth model was estimatedusing cross-sectional data, averaged over the pe-riod 1980–99 (as well as the 1980s and 1990s sep-arately) for a data set consisting of 57 industrialand developing countries.3 All models were esti-mated with ordinary least squares (OLS) and, toexamine causality, instrumental variables (IV).4

The Table shows that in the basic growth modelthe control variables are generally statisticallysignificant.

Building upon this model, five aspects of therelationship between capital account liberaliza-tion and economic performance were examinedusing a variety of dependent and independentvariables. The results presented in the text ta-bles are derived as follows:5

• The overall impact of capital account liberalizationon growth. The basic growth model was ex-tended by including either the IMF restrictionmeasure or the openness measure (see Box4.1). The end of the period was used foropenness, given that the stock of foreign as-sets as a ratio to GDP is a somewhat backward-looking measure of liberalization (average val-ues give very similar results). The twocomponents of the last measure—FDI open-ness and portfolio openness—were also testedseparately. For all countries and for all meas-ures the results were positive, but generallynot significant.

• The impact of capital account liberalization on pri-vate investment. In these regressions, private in-vestment as a ratio to GDP was introduced asthe dependent variable instead of real percapita growth. The control variables were kept

Box 4.2. The Impact of Capital Account Liberalization on Economic Performance

Basic Growth Regression

Ordinary InstrumentalLevel Squares Variables

Level of GDP in 1980 –0.0102 –0.0058(0.0045) (0.0037)

Level of schooling in 1980 0.0155 0.0099(0.0067) (0.0064)

Government balance 0.1220 0.0829(percent of GDP) (0.0660) (0.1220)

Inflation –0.0154 –0.0309(0.0094) (0.0145)

R2 0.2861 0.1617

Note: The coefficient is reported, with standard errors inparenthesis.

1As an illustration of its wide use, Sala-i-Martin (1997)in the paper entitled “I just ran two million regressions”attempts to map which of the numerous variables usedin studies variables are most important in determininggrowth. Other survey studies include Levine and Renelt(1992) and Barro and Sala-i-Martin (1995).

2In some cases the model is augmented for addi-tional factors such as political stability, governmentcorruption, black market exchange rate premium,and trade with the rest of the world.

3The country coverage is given in Table 4.2. Notethat, because of the long period covered by the study,three countries currently included in the advancedcountry group—Israel, Korea, and Singapore—wereincluded in the developing country group.

4The instruments are those commonly adopted inthe literature: the composition of religious beliefs(share of population embracing catholic, muslim, orother religion), the origin of the legal system (French,German, or English), the latitude, and an index of eth-nic diversity developed by Easterly and Levine (1997).

5For the details of the statistical analysis, see Edison,Levine, Ricci, and Sløk (forthcoming).

(King and Levine, 1993). Financial crises are as-sociated with generally temporary, but oftenlarge, losses in output (Barro, 2001).

Does Capital Account Liberalization PromoteDomestic Investment?

Greater access to foreign saving associatedwith opening the capital account generally leadsto greater capital inflows and—if these flows aremanaged appropriately—more investment andhigher growth. This is particularly true forpoorer countries and for “greenfield” FDI, inwhich a company is starting a new enterprisefrom scratch. By contrast, the impact on invest-ment in industrial countries is likely to be muchmore muted, as they have continuous access tocapital markets and most FDI flows involve thepurchase of an existing enterprise.

Simple comparisons of investment rates acrossdeveloping countries indicate that openness is in-deed generally associated with higher domestic

investment and hence somewhat higher growth.More liberalized economies have investment ra-tios that are higher than closed economies(Figure 4.5) and this behavior is associated withhigher FDI inflows (including many relativelypoor economies with open capital accounts,which typically rely primarily on FDI inflows—Box 4.3). Of course, some of this correlation mayreflect other characteristics—open countries alsotend to be richer and often have more stablemacroeconomic environments.

Recent academic studies that control for suchfactors confirm a positive association betweencapital inflows and investment, although some au-thors have found that this effect has declinedduring the 1990s. Some of these studies focusspecifically on FDI inflows, while others examinethe strength of this relationship for various cate-gories of inflows (see Bosworth and Collins,1999). The evidence across a wide range of de-veloping country groups indicates that all forms


157

the same, as the key source of growth in theSolow growth model that underlies the empir-ical specification is investment. Overall, the re-sults show that there are relatively strong posi-tive links between liberalization and privateinvestment.

• The impact of capital account liberalization on tech-nology transfers to developing countries. To testthe impact of FDI spillovers, three additionalvariables were added to the basic growthmodel—private investment (to eliminate thebenefits of FDI through this channel), FDIopenness (measured both as stocks and asflows), and the interaction of FDI with humancapital. The interaction term can be used tomeasure the impact of human capital on FDIspillovers that are not connected with privateinvestment. The results suggest that higherlevels of human capital raise the benefits fromopening up to FDI.

• The impact of capital account liberalization on fi-nancial development. In these regressions, meas-ures of financial development were substi-

tuted for growth in the specification. In oneregression, private sector credit and a ratio toGDP was used and, in another, stock marketturnover as a ratio to GDP. The regressionsgenerally support a significant link from liber-alization to financial development.

• The role of institutions when opening the capital ac-count. This was examined by successivelyadding into the basic growth regression vari-ous variables capturing the quality of the insti-tutional and policy environment, as well as in-teraction terms between such variables andcapital flows into the basic growth model. Theinstitutional and policy variables are: a meas-ure of law and order, the two measures of fi-nancial development discussed earlier, andfiscal balance as a ratio to GDP. The impactof the institutional environment on the bene-fits of liberalization can then be measuredthrough the interaction terms. As in the caseof the overall impact of liberalization ongrowth, the coefficients are generally correctlysigned but not statistically significant.

of capital inflows can increase investment, butthat this is particularly true of FDI flows (for arecent survey of this literature, see the WorldBank, 2001).

The impact of capital account liberalization ondomestic investment has received less attentionin the academic literature (see Henry, 2000).The results from the background study indicatethat liberalization is associated with higher do-mestic investment in developing countries, par-ticularly when the openness measure for FDI isused (Table 4.4).14 Increasing the opennessmeasure by the average gap between countriesdefined as closed and open is associated with arise in the private investment ratio of about2 percentage points of GDP, somewhat largerthan the effect implied by the restrictions meas-ure.15 These openness coefficients are in generalsignificant and, with the exception of the FDImeasure, the estimates using instrumental vari-ables indicate that the relationship appears to becausal, implying that liberalization increases in-vestment ratios. Using a standard coefficient be-


158


1970s 1980s 1990s0

1

2

3

4

5

6

Figure 4.5. Private Investment and Foreign Direct Investment (FDI) Inflows by Liberalizationin Developing Countries(Percent of GDP)

1970s 1980s 1990s0

5

10

15

20

25

30

Open economies tended to invest more, and attracted more FDI flows, according to the openness measure.

Open Closed

Private Investment

FDI Inflows

1

1

Table 4.4. Liberalization and Private Investmentfor Developing Countries(Investment as a ratio to GDP)



Overall liberalization 1.2 1.9FDI liberalization . . . 1.9Portfolio liberalization . . . 1.0

Note: Bold indicates significance at the 5 percent confidence level(using a one-tailed test), and italic indicates significance in regres-sions that account for endogeneity. Control variables are level of in-come in 1980; level of secondary schooling in 1980; inflation; andgovernment balance. For the details of the analysis, see Edison,Levine, Ricci, and Sløk (forthcoming).

1Overall, FDI, and portfolio openness are assumed to increase by40, 30, and 10 percent of GDP, respectively. Each change is equiva-lent to the difference between the average of the correspondingstock for open and closed developing countries (defined as thoseabove and below the average for all countries in the sample).

14The background study examined this issue using asimilar approach to that taken to examine the impact ongrowth, but with the ratio of private investment to GDPreplacing economic growth as the dependent variable.

15Results using total investment indicate a somewhatlarger impact, with similar results in terms of significanceand causality.


159

Foreign direct investment (FDI) flows provideone mechanism for helping to integrate poorercountries into the global economy. Such flowshave become increasingly important over thepast decade and offer considerable promise forboosting economic performance.

As shown in the Table, FDI flows to all devel-oping countries increased by more than six-foldbetween 1986–90 and 1996–99. FDI flows to the49 countries classified by the United Nations asleast developed countries, rose by a similar fac-tor—from an annual average of just $0.6 billionto $3.6 billion over the same period. In contrast,official development assistance, traditionally themajor source of capital financing for the leastdeveloped countries, has declined over the pastdecade. While FDI continues to flow mainly be-tween advanced economies, the share to poorercountries rose modestly in the 1990s.1 More im-portant, FDI inflows are often large relative tothe size of economies of the least developedcountries. During 1997–99, these inflows aver-aged 8 percent of gross fixed capital formationfor all least developed countries, and for six ofthese countries, exceeded 30 percent of fixedcapital formation.

Though still concentrated, FDI flows topoorer countries, have become increasingly dis-persed. During 1996–99, 10 countries received74 percent of the total flows to the 49 least de-veloped countries, compared with 84 percentduring 1991–95 and 92 percent during1986–1990.2 Countries such as Cambodia,Mozambique, and Uganda were able to join thegroup of top recipients, following substantialchanges in their economic and/or political envi-ronments. Moreover, while a significant share ofFDI remains in natural resources, investments inother sectors have also become more important.

For example, most of the inward FDI stock inCambodia and Uganda is in manufacturing,while most of the stock in Cape Verde andNepal is in services.

Growing FDI inflows promise a variety of po-tential benefits to poor country recipients. FDIflows could provide a relatively stable and grow-ing source of finance for poorer countries, andthey have tended to be considerably less volatilethan other types of capital inflows for middle-and high-income countries. Notably, FDI flowsstagnated, but did not collapse during the re-cent financial crises.3

How do FDI flows affect recipient countries?• FDI inflows tend to raise domestic investment,

including in low-income and sub-SaharanAfrican countries. While the link between cap-ital inflows and investment may have weak-ened during the 1990s, this is probably due toshifts within FDI, toward mergers and acquisi-tions instead of “greenfield” investments (i.e.,a firm started from scratch), and thus is lessrelevant for poorer countries where mergersand acquisitions account for less than 10 per-cent of direct investments.4

• Multinational corporations tend to pay higherwages than domestic enterprises and can offervaluable training opportunities to workers.

Box 4.3. Foreign Direct Investment and the Poorer Countries

FDI and ODA Flows (Billions of U.S. dollars, annual averages)

1986–90 1991–95 1996–99

Foreign direct investmentWorld 160.9 229.1 641.8Developed 133.0 149.8 459.7All developing 27.9 79.3 182.2Least developed 0.6 1.8 3.6

Official development assistance Least developed 13.9 16.6 12.7

Source: UNCTAD, FDI in Least Developed Countries at aGlance (April 2001), Figure 2, pp..2 and Table 3, pp. 8.

1This share rose from just 0.4 percent in the late1980s to 0.8 percent in the early 1990s. It declinedsomewhat to 0.6 percent during 1996–99 as flows toadvanced economies surged.

2The least developed countries with the largest FDIinflows during 1996–99 were Angola, Cambodia,Ethiopia, Lesotho, Mozambique, Myanmar, Sudan,Tanzania, Uganda, and Zambia.

3See for example, Mody and Murshid (2001); andWei (2001).

4See for example, Bosworth and Collins (1999);Mody and Murshid (2001); and United Nations Confer-ence on Trade and Development, or UNCTAD (2000).

tween investment and growth would indicatethat this increase in the investment ratio wouldraise growth by 0.3 percent a year.

Does Capital Account Liberalization PromoteTechnology Spillovers?

A second channel through which capital ac-count liberalization can have a positive impact is

through technology spillovers. These spilloversare most clear in the case of foreign direct in-vestment, especially through foreign firms in-corporating new technologies in their sub-sidiaries. As new technologies are generallydeveloped and adapted by firms in industrialcountries, foreign direct investment may be themost efficient way for developing economies to


160

• Multinational corporations can also promotethe transfer of technology, with possiblespillovers to domestic firms. However, empiri-cal evidence on such transfers or spillovers ismixed (for a recent review, see Hanson, 2001).

• Evidence on economic growth is also mixed.Some studies find a positive relationship onlyfor countries above a minimum “threshold” ofabsorptive capacity—measured by educationof the workforce, capital infrastructure orother development indicators.5 By promotinginvestment in poorer countries, FDI can helpmove them closer to this threshold.Recent analyses find that those developing

countries with stronger policy environments at-tract a larger share of the total FDI flow to de-veloping countries, while higher levels of cor-ruption act as a deterrent. These factors arelikely to be particularly important for countrieshoping to attract flows outside of the natural re-source sector, given the greater range of alterna-tive locations. Many countries compete to attractFDI flows using a range of financial incentives.However, such incentives can significantly re-duce the benefits from FDI flows accruing to thedomestic economy—for example, through fore-gone tax revenues and increased distortions.Their role in attracting multinational corpora-tions is unclear.6

Some examples of poorer countries that haverecently attracted increased FDI inflows, basedon improved structural and macroeconomicperformance are:

• Uganda, where FDI inflows began to grow in1993, averaging $182 million per year during1997–99, or nearly 20 percent of gross fixedcapital formation. The sectoral distribution ofFDI has been quite diverse, with 52 percent ofthe 1998 stock allocated to manufacturing, 35percent to services (including transport andtelecommunications) and 13 percent to theprimary sector. Uganda was one of the firstcountries to benefit from the Initiative for theHeavily Indebted Poor Countries (or HIPCInitiative) reflecting its commitment to soundmacroeconomic policy since 1987, includingprivate-sector led development and povertyreduction.

• Bolivia, which was also an early HIPC partici-pant, saw its FDI flows rise after a successfulprogram to eliminate a hyperinflation in 1985,followed by ongoing structural reform. Fromminimal levels in the late 1980s, FDI flowsgrew to an average of $818 million a year dur-ing 1996–99. At the same time domestic grossfixed capital formation increased considerably.FDI inflows remained high in 1999 despiteshort-term macroeconomic difficulties.

• Bangladesh, where FDI inflows surged fromminimal levels before 1995 to an annual aver-age of $170 million during 1996–99 (about 4percent of domestic investment). This growthfollowed a program of macroeconomic policyreforms undertaken since the early 1990s, andmoves more recently to encourage (domesticand foreign) private investment. The FDI flowshave been allocated to the energy sector (espe-cially gas and power), to manufacturing (tex-tiles, garments and electronics), and to trans-portation and telecommunications services.

Box 4.3 (concluded)

5For example, see Borensztein, De Gregorio, andLee (1998); and Eichengreen (2001).

6See Hanson (2001); Wei (2001); and UNCTAD(1996).

gain access to them. In addition, this knowledgemay become more widely available in the coun-try over time, as employees with experience inthe techniques used in foreign companiesswitch to other firms. Finally, foreign investmentcould increase competition in the host-countryindustry, and hence force local firms to becomemore productive by adopting more efficientmethods or by investing in human and/or physi-cal capital.

Recent analyses using macroeconomic datasuggest that FDI can have a positive impact ongrowth, particularly when the receiving countryhas a highly educated workforce, allowing it toexploit FDI spillovers (Borensztein, De Gregorio,and Lee, 1998). In a similar vein, other studieshave found that FDI spillovers are greatest inricher countries, while in poor countries thetechnologies being used are often less attuned tothe needs of the economy, limiting the benefitsfrom technological spillovers (the mining indus-try is often a good example of this type of effect;see Blomström, Lipsey, and Zejan, 1994). Theevidence on spillovers between foreign-ownedand domestic-owned firms is less clear-cut. Whilestudies find that sectors with a higher degree offoreign ownership exhibit faster productivitygrowth, firm-level data provide little evidence ofspillovers (Aitken and Harrison, 1999; andBlomström, 1986).

The background study confirmed existingresults that FDI spillovers appear to dependon human capital. The results in Table 4.5indicate that higher levels of human capitalraise the benefits from FDI liberalization andflows. For a country with a high level of humancapital, such as Korea, increasing the opennessmeasure by the average gap between closedand open economies can raise growth by asmuch as a quarter of a percent a year. Further,in both cases this relationship appears to becausal, which is important given the highly en-dogenous nature of FDI openness and humancapital. Slightly more than two-thirds of thecountries in the sample have sufficient humancapital to gain (to varying degrees) from suchspillovers.

Capital Account Liberalization and DomesticFinancial Development

The interaction between capital account lib-eralization and financial development is a dou-ble-edged sword, holding out the promise ofmedium-term benefits to growth throughdeeper domestic capital markets, but also carry-ing the risk of severe financial difficulties if theappropriate institutional framework is not inplace. On the positive side, liberalization associ-ated with steady inflows of capital can graduallydeepen domestic financial systems, particularlywhen control on foreign ownership of banksis relaxed (see Chapter 5 in IMF, 2000). Forexample, in Latin America, foreign involvementin banking systems increased dramatically inthe 1990s, particularly in Argentina and Mexico,which resulted in higher levels of financialdevelopment. In addition, greater foreign own-ership of the banking sector may actually helpreduce a country’s vulnerability to financialcontagion. International banks with their largecapital base and geographically diversifiedoperations may be less susceptible to a dome-stic bank run or financial panic and may serveas a safe haven for local depositors. There isalso ample evidence at the firm, sector, and


161

Table 4.5. Liberalization and FDI Spillovers forDeveloping Countries(Percent change in per capita growth per year)

FDI openness rises Gross FDI flows rise by 30 percent of GDP by 3 percent of GDP

FDI term –1.87 –1.88

Interaction between FDI and human capital 0.50 0.48

Percentage of countries benefiting from FDI spillover 71 68

Note: Bold indicates significance at the 5 percent confidence level (usinga one-tailed test), and italic indicates significance in regressions that ac-count for endogeneity. Control variables are level of income in 1980; level ofsecondary schooling in 1980; inflation; and government balance. Three addi-tional variables were added to the basic growth specification—the private in-vestment ratio, FDI openness (or actual FDI flows), and the interaction ofFDI with human capital. Adding private investment (which is highly signifi-cant) eliminates the direct impact of FDI through capital formation, so thatthe coefficients of FDI and its interaction with human capital measure thespillover effects of FDI on growth. For details of the analysis see Edison,Levine, Ricci, and Sløk (forthcoming).

macroeconomic level that a deeper and moreefficient financial system can, in turn, raisegrowth through more efficiently allocating re-sources among competing projects.16

Does Liberalization Benefit Domestic FinancialDepth in the Medium Term?

There does appear to be a positive long-termcorrelation between financial depth and capitalaccount openness, presumably reflecting, at leastin part, the fact that domestic and internationalfinancial market deregulation are often carriedout in tandem (Williamson and Mahar, 1998).Figure 4.6 illustrates the relationship betweenopenness and two measures of financial develop-ment that have been linked to higher growth inpast analysis—the value of credit to the privatesector as a ratio of GDP and stock marketturnover (as the ratio of total stock marketturnover to GDP). For both measures there is astrong positive correlation, partly reflecting jointcharacteristics, such as richer countries tend tohave more open international financial marketsand deeper domestic financial markets.

The existing academic literature, which takesaccount of other factors, such as income percapita, confirms that capital account liberaliza-tion is associated with financial development inthe longer term. For example, one recent studyfound that countries with open capital accountsenjoy a significantly greater increase in financialdepth than countries that maintain capital ac-count restrictions, although this link appears tobe closest for industrial countries (see Klein andOlivei, 2000; and Bailliu, 2000). Other work, fo-cusing on large emerging market economies,finds that stock markets become larger andmore liquid after the capital account is opened(see Levine and Zervos, 1998; and Henry,2000).

Further evidence of a positive link was foundin the background study. The results suggest


162

Figure 4.6. Financial Development by Liberalization in Developing Countries(Percent of GDP)

Open economies generally have greater domestic financial development, according to the openness measure.

0

12

24

36

Closed Open

Private credit (left scale) Stock market (right scale)

9

6

3

0


1

1

16For firm level evidence, see Demirgüc-Kunt andMaksimovic (1998); for industry-level evidence, Rajan andZingales (1998); for cross-country evidence, Beck, Levine,and Loayza (2000).

that liberalization has large and (in the case ofthe openness measure) significant effects onboth measures of domestic financial depth.For example, increasing the openness measureby the average gap between a closed and openeconomy raises private credit as a ratio to GDPby 20 percent and almost doubles stock marketturnover (Table 4.6). Based on standard esti-mates of the link between financial deepeningand growth, this could raise growth by a quarterof a percent. Turning to the results for port-folio and FDI openness, as expected, the impacton financial deepening of liberalization ofportfolio flows appears to be both significant,and, in the case of stock market turnover,causal. Liberalization of foreign direct invest-ment also appears to have a positive impact, al-though it is not significant. In short, there issuggestive evidence that over longer periodssuccessful liberalization, particularly on theportfolio side, is linked to deeper domestic fi-nancial markets.

Financial Instability: The Cost of Liberalization

These potential long-term benefits, however,need to be set against the danger that open in-ternational financial markets can also create fi-nancial problems, including financial crises, withlarge output costs. The problems are generallyassociated both with excessive inflows and out-flows and, more generally, the volatility of netcapital flows. Figure 4.7 shows that volatility ofnet capital inflows has increased substantiallyover time, particularly for countries that experi-enced more extensive capital account liberaliza-tion. The increase in volatility has been morepronounced for portfolio flows than FDI flows,reflecting the more long-term relationship im-plicit in FDI. Volatility has been particularly highin the 1990s, when many developing countrieshad recently liberalized. In this decade, highervolatility of capital flows has also been associatedwith somewhat lower growth (Figure 4.8).

As noted above, the 1990s have been charac-terized by numerous exchange rate and financialcrises, often associated with a drastic contractionof economic activity.17 In most cases, the crisesare associated with large inflows of capital—espe-cially portfolio inflows—which are not efficientlychanneled to the most productive investmentopportunities, leading to a progressive deteriora-tion in the balance sheets of the domestic finan-cial sector.18 This reflects the limited depth of fi-nancial markets in many developing countries,as well as the reduced incentive of lending agen-cies to screen for good projects when morecredit is easily available, and the maturity mis-match in trying to finance long-term projectswith short-term money. Asymmetric informationbetween foreign investors and domestic borrow-ers may allow inflows to continue even as bal-ance sheets deteriorate (see Eichengreen andothers, 1998). When international investors de-cide that repayment difficulties and default risksmay arise, however, large net inflows are quicklyreplaced by outflows, which can degenerate intoexchange rate and financial crises. These effects


163

Table 4.6. Liberalization and Domestic FinancialDevelopment for Developing Countries(Percentage increase)



Private credit2Overall liberalization 35.8 19FDI liberalization . . . 15Portfolio liberalization . . . 12

Stock market turnover2

Overall liberalization 54.7 90FDI liberalization . . . 86Portfolio liberalization . . . 40

Note: Bold indicates significance at the 5 percent confidence level(using a one-tailed test), and italic indicates significance in regres-sions that account for endogeneity. Control variables are level of in-come in 1980; level of secondary schooling in 1980; inflation; andgovernment balance. For the details of the analysis see Edison,Levine, Ricci, and Sløk (forthcoming).

1Overall, FDI, and portfolio openness are assumed to increaseby 40, 30, and 10 percent of GDP, respectively. Each change isequivalent to the difference between the average of the corre-sponding stock for open and closed developing countries (definedas those above and below the average for all countries in thesample).

2As a ratio to GDP.

17For a discussion of financial crises, see World Economic Outlook, May 1998 and May 1999.18See Calvo and Reinhart (2000) and the discussion in Chapter IV of the May 2001 World Economic Outlook.

can rapidly spread through global financial mar-kets, particularly at the regional level, affectingfinancing in other countries with open capitalmarkets.

In addition to sudden reversals of capitalflows, there are other possible costs. As with anypolicy change that provides access to greater re-sources, liberalization can exacerbate existingdistortions within an economy. For example, ifhigh levels of protection are supporting an inef-ficient industry, providing access to foreign sav-ing may allow more capital to flow to this sector,thereby accentuating the underlying problem.This reflects the general issue of the law of thesecond best, namely that in the presence ofother distortions, removing one particular dis-tortion does not necessarily improve the overalloutcome. That said, the major costs of liberal-ization appear to have been focused on exces-sive capital inflows followed by suddenreversals.

Policy responses to a reversal of inflows canattempt to deal with the short-term problem ofoutflows directly through reimposing capitalcontrols and/or deal with the longer term issueof improving financial supervision and strength-ening macroeconomic policies. In response tocrises in the 1990s, most countries have focusedon supervision and policy, tightening monetarypolicies during the crisis to regain investorconfidence, moving away from exchange ratepegs, and instituting financial and (where nec-essary) corporate structural reforms. The mostnotable countries to have reimposed capitalcontrols after a crisis are Malaysia and Russia. Inboth cases, controls appear to have providedsome breathing space in which to implementmore fundamental policy reforms, but at thecost of weakening the confidence of interna-tional investors, thereby increasing the cost offunding from abroad, weakening FDI flowssomewhat, and producing large administrativecosts. In Malaysia, the effectiveness of capitalcontrols was enhanced by macroeconomic andstructural adjustments. The authorities’ strongenforcement capacity and favorable exchangerate developments also played an important


164

Figure 4.7. Volatility of Net Capital Flows and Liberalization

According to the openness measure, portfolio and bank flows have been very volatile, particularly in the 1990s. Foreign direct investment (FDI) flows have instead been steadier.

(Billions of U.S. dollars)

Open Closed

1975 79 83 87 91 95 99-200

-100

0

100

200

300

400

1975 79 83 87 91 95 99-200

-100

0

100

200

300

400 Net Private Capital Flows Net FDI Flows

1975 79 83 87 91 95 99-200

-100

0

100

200

300

400

1975 79 83 87 91 95 99-200

-100

0

100

200

300

400 Net Bank FlowsNet Portfolio Flows


1

1

role.19 In Russia, in the absence of rapid struc-tural reforms, controls have done little to stemunderlying capital flight—a similar experienceto that of the 1980s debt crisis. There is alsoChile, which imposed capital controls on inflowsin the face of an increase in such capital. Theaim was to discourage short-term inflows butnot long-term ones and to increase the potencyof monetary policy. In Chile’s case, the breath-ing space provided was used to implement sig-nificant structural reforms.

Empirical evidence on the connection be-tween institutional arrangements and perform-ance is difficult to come by, largely reflecting thedifficulties of measuring institutional quality.Earlier work has found that, while higher levelsof gross capital flows increased growth over the1990s, the volatility of capital flows loweredgrowth, which is suggestive of a significant rolefor institutional factors (Mody and Murshid,2001). The background study also finds some ev-idence that stronger institutions increase thebenefits from capital inflows. Data limitations fordeveloping countries restricted the proxy vari-ables to measures of financial depth and fiscalpolicies, as appropriate data on such importantissues as financial supervision or corporate gov-ernance were not available. The results suggestthat three indicators of domestic financial mar-ket efficiency (private credit as a ratio to GDP,stock market turnover as a ratio to GDP, and ameasure of the underlying degree of law and or-der) tend to boost the growth benefits of capitalflows, although the coefficients are generallysmall and statistically insignificant (Table 4.7).20

On the macroeconomic side, a stronger fiscalposition also appears to enhance the growthbenefits from international capital flows.

These results suggest that domestic institu-tional failures, such as weak financial supervisionand regulation, can lead to crises and output


165

(Percent)

In the 1990s, higher volatility has been associated with somewhat lower growth.

Figure 4.8. Growth by Volatility of Gross Portfolio Flows in Countries with Open Capital Markets

Low volatility High volatility

Sources: IMF, International Financial Statistics; and IMF staff calculations. Volatility is measured by the standard deviation of gross portfolio flows over 1970–99. Countries are classified as high volatility if their standard deviation is above the average.

1

1980s 1990s0.0

0.5

1.0

1.5

2.0

2.5

1

19Malaysia has subsequently reversed all controls onportfolio capital flows that were imposed in September1998.

20Arteta, Eichengreen, and Wyplosz (2001) also investi-gate the relationship with law and order.

losses. Therefore, adequate financial supervisionand regulation is also essential to ensure that thedomestic lending agencies do not undertake ex-cessive risk due to the presence of limited liability(see Bhattacharya and Thakor, 1993). This is par-ticularly true in the presence of explicit or im-plicit government guarantees of loans, which re-duce the incentives of both domestic financialintermediaries and international lenders to con-trol risk. More generally, weak corporate gover-nance, poor accounting practices, and the ab-sence of timely diffusion of accurate informationon the macroeconomic situation and the finan-cial sector raise the general level of uncertaintyamong international investors, worsening theaforementioned problems of asymmetric infor-mation between domestic and internationalinvestors.

Strong domestic fundamentals or institutions,however, do not rule out the possibility of crises,as shifts in portfolio preferences by external in-vestors can also be a source of sudden changes innet capital flows. When the Hong Kong dollarcame under pressure in October 1997, despite itssolid fundamentals, investor sentiment towardAsia shifted sharply, and countries that had beenawash with foreign money suddenly found it diffi-cult or impossible to obtain new financing, oreven experienced large net capital outflows.

These shifts in sentiment may reflect opportuni-ties in other markets, as saving and investmentopportunities fluctuate in the industrialcountries—partly due to changes in interest rates(this is discussed further in Chapter II). However,the rapidity of the apparent shift in investor sen-timent has also led to discussion of financial con-tagion and models of “self-fulfilling crisis,” inwhich the vagaries of international investors de-termine whether a crisis occurs or not.21 There isa growing empirical literature on contagion, fo-cusing largely on the crises that occurred in thesecond half of the 1990s, that suggests that prob-lems in one country did cause lenders to reducetheir exposures to other countries in the region.

The international financial community canhelp countries that undertake capital accountliberalization to reduce the associated costs.Promoting standards for improved transparencyand wider dissemination of information—including the IMF’s initiatives on data dissemina-tion and financial and fiscal transparency—canreduce the asymmetry of information betweendomestic borrowers and international investors.Initiatives aimed at increasing private sector in-volvement in resolving financial crises are alsoaimed at inducing international investors to takefull account of available information on borrow-ers’ default risks. Nonetheless, the major effortto reduce the costs of capital account liberaliza-tion need to be undertaken by the countriesthemselves, by implementing appropriate poli-cies and creating favorable domestic conditions.Sound macroeconomic policies, efficient super-vision and regulation of the financial system, re-duced government guarantees (e.g., though lim-ited deposit insurance schemes), effectiveaccounting and legal systems, and enhancedtransparency improve the effectiveness of the fi-nancial system in channeling foreign resourcesto domestic productive activities, limit the sys-temic risks of this intermediation process, andhelp reduce the moral hazard of both domesticand international agents.


166

Table 4.7. The Benefits of Capital Flows fromStronger Institutions in the 1990s(Increase in per capita growth per year)

Private credit doubles 0.15Stock market turnover doubles 0.07Law and order improves from worst to best 0.43Fiscal balance rises by 10 percent of GDP 0.13

Note: Bold indicates significance at the 5 percent confidence level(using a one-tailed test), and italic indicates significance in regres-sions that account for endogeneity. The estimates are derived fromthe coefficient of the interaction (product) term of gross capitalflows and the respective measure of institution; they are evaluatedat the average level of gross capital flows over the sample. Controlvariables are level of income in 1990; level of secondary schoolingin 1990; inflation; government balance; gross capital flows; and therespective measure of institution. For the details of the analysis seeEdison, Levine, Ricci, and Sløk (forthcoming).

21For surveys on self-fulfilling crises, see Obstfeld (1998) and Jeanne (2000). For evidence on financial contagion, seeCaramazza, Ricci, Salgado (2000) and Van Rijckeghem and Weder (2000).

It is often not possible—and not always desir-able—to implement such a wide range of re-forms at the same time. Some of the risks arisingfrom capital account liberalization can be re-duced by appropriately sequencing the stages ofliberalization and coordinating them with com-plementary policies (including sound macroeco-nomic policy and reform). The next section ex-plores how reforms related to liberalization canbe sequenced.

Sequencing Capital AccountLiberalization

Successful international capital account liber-alization often requires careful sequencing ofpolicies, requiring a detailed assessment of coun-try-specific circumstances. In particular, it is im-portant that reforms take place in an institu-tional environment that can support theopening of the capital account and avoid ex-change rate or financial crises. In particular, fo-cus is needed on the ordering of liberalizationof types of flows, such as FDI, and long- andshort-term financial instruments, as well as thespeed at which the often numerous restrictionson such flows are lifted.

Two basic approaches have been put forwardas being the best method to achieve financialintegration:

• The conventional view emphasizes the pre-conditions for liberalization. In particular,this approach suggests that liberalizationshould come after macroeconomic stabilityhas been achieved; financial reform hasbeen implemented, and trade liberalizationundertaken.22 Based on this view, liberaliza-tion should occur gradually and late in theprocess of economic reform.

• The political economy view stresses the con-straints imposed by political factors on re-forms and the limited capacity of countriesto reform themselves in the absence of ex-ternal pressures. This view recognizes that

the approach to reform is often dictated asmuch by political feasibility as technical re-quirements and hence suggests that liberal-ization should come early in the process asa “big bang,” serving as a catalyst for furthereconomic reforms and helping to overcomeopposition to reform.

The experiences of countries that have en-gaged in international financial market liberal-ization indicate that both views have merit (Box4.4). However, the pace and timing of reformsare often not as important as the consistency ofthe reforms and policies that are followed ateach particular point in time. For capital ac-count liberalization, the coordination of specificreforms and policies in the domestic and exter-nal sectors, especially domestic financial sectorreforms and the implementation of a consistentmonetary and exchange rate policy mix, areconsidered the most important (Johnston,Darbar, and Echeverria, 1999). In Asia, for ex-ample, problems have generally occurred in situ-ations where financial supervision was not suffi-ciently rigorous, while in Latin Americaproblems have more often come through themacroeconomic policy mix.

In practice, liberalization in industrial coun-tries has tended to follow the gradual andphased approach to economic reforms.23

Developing countries, on the other hand, havefollowed both “big bang” and gradual ap-proaches to liberalization of their capital ac-counts. The differences in experiences reflectedexisting conditions as much as the speed of re-form and other factors.

While country experiences illustrate there isno simple rule for the sequencing and coordina-tion of capital account liberalization with otherpolicies, the following general three principlescan help guide liberalization in any particularcase (see Ishii and others, 2001):

• Sound and sustainable macroeconomic policiesare an important pre-condition for liberalization.Macroeconomic instability can exacerbate

SEQUENCING CAPITAL ACCOUNT LIBERALIZATION

167

22See for example McKinnon (1973), Shaw (1973), and Hanson (1995).23In those countries, capital account liberalization followed trade reforms and domestic financial reforms.


168

Country experiences illustrate that there is nounique correct approach to sequencing and co-ordinating capital account liberalization withother policies.1 This box examines the experi-ences of eight advanced and emerging marketeconomies that exemplify many of the issues thatcountries have encountered in the course of cap-ital account liberalization. Four of the countries(Austria, Hungary, South Africa, and the UnitedKingdom) were able to fully or very substantiallyliberalize the capital account without suffering asystemic financial crisis, and no or only mild bal-ance of payments problems. The four othercountries (Korea, Mexico, Sweden, and Turkey)suffered severe financial system and external

crises, even though at least one (Korea) stillmaintained a broad range of capital controls.

The country experiences are summarized inthe Table. Four important conclusions emergefrom the analysis of these experiences:• The pace of capital account liberalization had

no systematic effect on the likelihood of a cri-sis. Of the four countries that experiencedmajor financial sector disruptions followingcapital account liberalization, Korea, Mexico,and Sweden took a gradual approach to liber-alization. Conversely, the United Kingdomand Hungary avoided crises even though theyrapidly liberalized capital flows.

• Sequencing capital account liberalization in aparticular way is not by itself sufficient to pro-tect a country from crisis. Most of the coun-

Box 4.4. Country Experiences with Sequencing Capital Account Liberalization

Summary of Country Experiences with Sequencing

Pace and Sequencing ofLiberalization Financial Sector Policies Macroeconomic Policies

1. Countries that avoided a crisis

Austria Gradual. Long-term flows Sound and well-supervised Stable macroeconomic liberalized before short-term flows. financial sector. environment.

Hungary Rapid. FDI and other long-term Rapid financial sector reforms. Macroeconomic stabilization flows liberalized before short-term Foreign bank participation following 1995 crisis.flows. encouraged early.

South Africa Gradual. Restrictions on non- Well-capitalized banks. Steps to Sound macroeconomic policies.residents’ capital flows liberalized strengthen prudential regulation first. and supervision.

United Kingdom Rapid. Strong market discipline and Generally sound policies, despiteprudential policies. 1992 ERM exchange rate crisis.

2. Countries that experienced a crisis

Korea Gradual and partial. Financial Weaknesses in the financial sector. Sound macroeconomic policies,institutions not subject to significant Poor corporate governance and with low inflation and stable restrictions on short-term external high leverage. public finances.borrowing, but limits on long-term external borrowing.

Mexico Gradual. FDI liberalized first. Capital Poorly supervised and managed Growing macroeconomic imbalancesaccount substantially liberalized on financial sector that relied heavily inconsistent with the tightly the eve of the 1994 crisis. on short-term foreign borrowing. managed exchange rate regime.

Sweden Gradual, but accelerated in late Extensive domestic financial Expansionary macroeconomic 1980s. Long-term flows generally liberalization, but with inadequate policies leading to an unsustainable liberalized before short-term flows. supervision. credit and asset price boom.

Turkey Rapid. Most capital controls Weak banks, poor supervision, Growing macroeconomic removed between 1988 and 1991. government ownership. imbalances and uncertain policy Direct investment liberalized slightly setting marked by high and earlier than portfolio investment. variable inflation and interest rates.

1For a detailed discussion see Ishii and others (2001).

financial sector weaknesses; and financialand capital account liberalization in suchcircumstances can accentuate such instabil-ity. Sound macroeconomic policies are es-sential to defusing this two-way linkage.

• A series of financial sector reforms should be im-plemented during liberalization, if they do not al-ready exist. In particular, market-based mone-tary arrangements and associated centralbanking reforms should be implementedearly to foster domestic financial liberaliza-tion. Prudential regulation and supervisionand financial restructuring policies shouldbe phased in to complement other reformsaimed at enhancing competitive efficiencyand market development to help managerisks in liberalization and to foster financialsector stability.

• The pace, timing, and sequencing of liberaliza-tion need to take account of social and regionalconsiderations. In particular, account needsto be taken of the authorities’ commitmentto and ownership of a reform strategy, aswell as other considerations, such as mem-

bership in regional groups. Also, the opera-tional and institutional arrangements forpolicy transparency and data disclosure—in-cluding monetary and financial policy trans-parency—need to be adapted to supportcapital account opening.

These principles are consistent with variousspeeds of capital account liberalization; and theydo not imply that liberalization should be un-duly delayed. Countries should therefore feelable to proceed with both capital account liber-alization and financial sector development andreform in accord with these principles, and, asquickly as they can, develop the ability to effec-tively manage the risks associated with interna-tional capital flows.

Policy ConsiderationsThe remarkable growth of cross-border capital

flows has been in part associated with the reduc-tion in impediments to capital movements andin part associated with the general trends inglobalization. Recent experience has made it

POLICY CONSIDERATIONS

169

tries studied here liberalized FDI and otherlong-term flows before short-term flows. Evenso, some of them (for example Mexico andSweden) experienced serious crises. This doesnot mean that the order in which flows areliberalized is irrelevant. Korea liberalizedshort-term flows before long-term flows, whichencouraged excessive short-term foreign bor-rowing and left the economy vulnerable to ex-ternal shocks.

• Financial sector stability is of paramount im-portance. All of the countries that avoided acrisis following capital account liberalizationpaid careful attention to the soundness of thefinancial sector and put in place strong pru-dential policies. The financial system in theUnited Kingdom, for example, was able towithstand recession and withdrawal from theERM owing, among other things, to effective

supervision and strong market discipline. Bycontrast, there were major weaknesses in thefinancial sector in all of the countries that ex-perienced a crisis.

• Stable macroeconomic policies are importantto an orderly capital account liberalization. Allof the countries that avoided crises had soundmacroeconomic policies in place. For exam-ple, Austria consistently geared its macroeco-nomic policies to maintaining the peg withthe deutsche mark. By contrast, in Mexico,Sweden, and Turkey, a fixed or tightly man-aged exchange rate was maintained for toolong in the face of expansionary policy set-tings. A counter-example is provided byKorea, where sound macroeconomic policiesoffered no protection from vulnerabilitiesstemming from deep-seated structural prob-lems in the financial and corporate sectors.

clear that international financial market liberal-ization can have both favorable and adverse ef-fects. On the positive side, over time liberaliza-tion can significantly raise domestic investment,create spillovers to the rest of the economy fromtechnological transfer (particularly for FDIflows) and deepen domestic financial markets(particularly for portfolio flows). Estimates re-ported in this chapter provide evidence that, fora “typical” liberalization, these benefits are asso-ciated with an increase in growth of a half of apercent a year or more (a quarter percent fromhigher investment, a quarter percent fromgreater domestic financial development, and upto a quarter percent from FDI spillovers). As ex-perience has shown, however, liberalization en-tails significant risks. In particular, weak financialsupervision and inconsistent macroeconomicpolicies can be associated with excessive capitalinflows that are allocated inefficiently and leadto rapid capital outflows. As a result, strong over-all growth benefits from liberalization are diffi-cult to identify.

The challenge for emerging market countries,therefore, is to maximize net benefits from liber-alization. For those countries that already havesignificant involvement in international capitalmarkets, the key requirement is to create institu-tions that strengthen the positive aspects of fi-nancial integration. In this respect, the lessonsof the financial crises of the 1990s underscorethe importance of strong macroeconomic poli-cies and sound financial systems, which enablecountries to protect themselves against adverseswings in investor sentiment. The experience ofthe debt crisis in the 1980s and of the Asian andRussian crises more recently also suggests thatimposing capital controls during crises on finan-cial systems that have been liberalized interna-tionally has only a limited impact on outflowsand may distract governments from their primetask of strengthening the financial and macro-economic environment.

For those countries that are not involved—oronly partially involved—in global capital mar-kets, capital account liberalization should re-main the ultimate goal, but the pace at which it

can be achieved will vary significantly. Countryexperiences suggest that successful capital ac-count liberalization requires careful sequencingof policies that may help to reduce the likeli-hood of external or financial sector instabilities.These experiences illustrate that there is no sim-ple rule for sequencing either the speed or theorder of liberalization of capital flows. Rather,reforms and policies need to be implementedconsistently at each point. Two of the most im-portant areas of reform relate to the supportand reinforcement of sound macroeconomicpolicy and financial sector reform. If these twofactors are not in place it may be best for a coun-try to adopt a slow and gradual approach toopening the capital account.

Finally, turning to the poorest developingcountries, on the positive side, FDI flows canplay an important role. However, a top priorityshould be to implement policies that make theircountry more attractive for domestic savings andinvestment so as to encourage foreign capital.Improvements in financial development, espe-cially if associated with stronger financial andpolicy institutions, can also help cushion thenegative impact of liberalization on social condi-tions by reducing the poor’s exposure to macro-economic volatility and financial crises. It wouldbe a mistake for these countries to think that in-volvement with global capital markets offers amagic, near-term fix for their problems.Creating the conditions and institutions thatmake savings and investment attractive will, how-ever, over time offer poorer countries opportuni-ties for higher living standards, through, amongother sources, capital inflows from abroad.

ReferencesAitken, Brian, and Ann Harrison, 1999, “Do Domestic

Firms Benefit from Direct Foreign Investment?”American Economic Review, Vol. 89 (June), pp. 605–18.

Alesina, Alberto, Vittorio Grilli, and Gian Maria Milesi-Ferretti, 1994, “The Political Economy of CapitalControls,” in Capital Mobility: The Impact onConsumption, Investment, and Growth, ed. byLeonardo Leiderman and Assaf Razin (Cambridge:Cambridge University Press for CEPR).


170

Arteta, Carlos, Barry Eichengreen, and CharlesWyplosz, 2001, “On the Growth Effects of CapitalAccount Liberalization” (unpublished; Berkeley:University of California).

Asian Development Bank, 2001, Asian DevelopmentOutlook (Oxford: Oxford University Press).

Bailliu, Jeannine, 2000, “Private Capital Flows,Financial Development, and Economic Growth inDeveloping Countries,” Bank of Canada WorkingPaper No. 2000–15 (Ontario, Canada: Bank ofCanada).

Barro, Robert J., 2001, “Economic Growth in East AsiaBefore and After the Financial Crisis,” NBERWorking Paper No. 8330 (Cambridge, Mass.:National Bureau of Economic Research).

——— and Xavier Sala-i Martin, 1995, Economic Growth(Cambridge, Mass.: MIT Press).

Bekaert, Geert, and Campbell Harvey, 2000, “ForeignSpeculators and Emerging Equity Markets,” TheJournal of Finance, Vol. 55 (April), pp. 565–613.

Bekaert, Geert, Campbell R. Harvey, and ChristianLundblad, 2001, “Does Financial LiberalizationSpur Growth?” NBER Working Paper No. 8245(Cambridge, Mass.: National Bureau of EconomicResearch).

Beck, Thorsten, Ross Levine, and Norman Loayza,2000, “Finance and the Sources of Growth,” Journalof Financial Economics, Vol. 58 (Oct.–Nov.), pp.261–300.

Bhattacharya, Sudipto, and Anjan V. Thakor, 1993,“Contemporary Banking Theory,” Journal ofFinancial Intermediation, Vol. 3 (October), pp. 2–50.

Blomström, Magnus, 1986, “Foreign Investment andProductive Efficiency,” Journal of IndustrialEconomics, Vol. 35 (September), pp. 97–110.

———, Robert Lipsey, and Mario Zejan, 1994, “WhatExplains Developing Country Growth?” NBERWorking Paper No. 4132 (Cambridge: NationalBureau of Economic Research).

Bordo, Michael, Barry Eichengreen, DanielaKlingebiel, and María Soledad Martínez-Peria,2001, “Is the Crisis Problem Growing MoreSevere?,” Economic Policy, Vol. 16 (April), pp. 51–82.

Borensztein, Eduardo, José De Gregorio, and Jong-Wha Lee, 1998, “How Does Foreign DirectInvestment Affect Growth?” Journal of InternationalEconomics, Vol. 45 (June), pp. 115–35.

Bosworth, Barry, and Susan Collins, 1999, “CapitalFlows to Developing Economies: Implications forSaving and Investment,” Brookings Papers on EconomicActivity:1 (Washington: Brookings Institution).

Calvo, Guillermo, and Carlos Végh, 1999, “InflationStabilization and BOP Crises in DevelopingCountries,” in Handbook of Macroeconomics, ed. byJohn Taylor and Michael Woodford (New York:North-Holland), pp. 1531–1614.

Calvo, Guillermo, and Carmen Reinhart, 2000, “WhenCapital Inflows Come to a Sudden Stop:Consequences and Policy Options,” in Reforming theInternational Monetary and Financial Systems, ed. byPeter Kenen and Alexander Swoboda (Washington:International Monetary Fund), pp. 175–201.

Caramazza, Francesco, Luca Ricci, and Ranil Salgado,2000, “Trade and Financial Contagion in CurrencyCrises,” IMF Working Paper 00/55 (Washington:International Monetary Fund).

Chanda, Areendam, 2000, “The Influence of CapitalControls on Long Run Growth: Where and HowMuch?” (unpublished; Providence, Rhode Island:Brown University, Department of Economics).

Demirgüç-Kunt, Asli, and Vojislav Maksimovic, 1999,“Institutions, Financial Markets, and Firm DebtMaturity,” Journal of Financial Economics, Vol. 54(December), pp. 295–336.

Diaz-Alejandro, Carlos, 1985, “Good-Bye FinancialRepression, Hello Financial Crash, Journal ofDevelopment Economics, Vol. 19 (Sept.–Oct.), pp.1–24.

Dollar, David, and Aart Kraay, 2001, “Growth is Goodfor the Poor,” World Bank Policy Research WorkingPaper No. 2587 (Washington: World Bank).

Easterly, William, and Ross Levine, 1997, “AfricaGrowth Tragedy: Policies and Ethnic Divisions,”Quarterly Journal of Economics, Vol. 112 (November),pp. 1203–50.

Edison, Hali, and Frank Warnock, 2001, “A SimpleMeasure of the Intensity of Capital Controls,”International Finance Discussion Paper No. 705(Washington: Board of Governors of the FederalReserve, August).

———, forthcoming, “Cross-Border Listings, CapitalControls, and Equity Flows to Emerging Markets,”(Washington: Board of Governors of the FederalReserve).

Edison, Hali, Michael Klein, Luca Ricci, and TorstenSløk, forthcoming, “Capital Account Liberalizationand Economic Performance: A Review of theLiterature,” IMF Working Paper (Washington:International Monetary Fund).

Edison, Hali, Ross Levine, Luca Ricci, and TorstenSløk, forthcoming, “The Relationship BetweenCapital Account Liberalization and Growth

REFERENCES

171

Revisited,” IMF Working Paper (Washington:International Monetary Fund).

Edwards, Sebastian, 2001, “Capital Mobility andEconomic Performance: Are Emerging EconomiesDifferent?” NBER Working Paper No. 8076(Cambridge, Mass.: National Bureau of EconomicResearch).

Eichengreen, Barry, 2001, “Capital Account Liberaliza-tion: What do Cross-Country Studies Tell Us?” (un-published; Berkeley: University of California).

———, Michael Mussa, Giovanni Dell’Ariccia, EnricaDetragiache, Gian Maria Milesi-Ferretti, andAndrew Tweedie, 1998, Occasional Paper No. 172(Washington: International Monetary Fund).

Frankel, Jeffrey, and Alan MacArthur, 1988, “Politicalvs. Currency Premia in International Real InterestDifferentials: A Study of Forward Rates for 24Countries,” NBER Working Paper No. 2309(Cambridge, Mass.: National Bureau of EconomicResearch).

Grilli, Vittorio, and Gian Maria Milesi-Ferreti, 1995,“Economic Effects and Structural Determinants ofCapital Controls,” Staff Papers, InternationalMonetary Fund, Vol. 42 (September), pp. 517–51.

Hanson, Gordon, 2001, “Should Countries PromoteForeign Direct Investment?” G-24 Discussion PaperNo. 9 (February).

Hanson, James, 1995, “Opening the Capital Account:Costs, Benefits and Sequencing,” in Capital Controls,Exchange Rates, and Monetary Policy in the WorldEconomy,” ed. by Sebastian Edwards (Cambridge:Cambridge University Press).

Henry, Peter, 2000, “Stock Market Liberalization,Economic Reform, and Emerging Market EquityPrices,” Journal of Finance, Vol. 55 (April), pp.529–64.

International Monetary Fund, 1998, World EconomicOutlook, May 1998 (Washington: IMF).

———, 1999, World Economic Outlook, May 1999(Washington: IMF)..

———, 2000, World Economic Outlook, May 2000(Washington: IMF).

———, 2000, World Economic Outlook, October 2000(Washington: IMF).

———, 2001, World Economic Outlook, May 2001(Washington: IMF).

Ishii, Shogo, Karl Habermeier, Bernard Laurens, JohnLeimone, Judit Vadasz, and Jorge Ivan Canales-Kriljenko, 2001, “Capital Account Liberalizationand Financial Sector Stability,” (unpublished;Washington, D.C.: International Monetary Fund).

Jeanne, Olivier, 2000, “Currency Crises: A Perspectiveon Recent Theoretical Developments,” Special Papersin International Economics No. 20 (Princeton:Princeton University, Economics Department,International Economies Section).

Johnston, R. Barry, and Chris Ryan, 1994, “TheImpact of Controls on Capital Movements on thePrivate Capital Account of Countries Balance ofPayments,” IMF Working Paper No. 94/78,(Washington: International Monetary Fund).

Johnston, R. Barry, Salid Darbar, and ClaudiaEchevarria, 1999, “Sequencing Capital AccountLiberalization: Lessons from the Experiences inChile, Indonesia, Korea, and Thailand,” in Sequenc-ing Financial Sector Reforms: Country Experiences andIssues, ed. by R. Barry Johnston and V. Sundararajan(Washington: International Monetary Fund).

King, Robert, and Ross Levine, 1993, “Finance andGrowth: Schumpeter Might Be Right,” QuarterlyJournal of Economics, Vol. 108 (August), pp. 717–37.

Klein, Michael, and Giovanni Olivei, 2000, “CapitalAccount Liberalization, Financial Depth, andEconomic Growth” (unpublished; Medford, Mass.:Tufts University).

Kraay, Aart, 1998, “In Search of the MacroeconomicEffects of Capital Account Liberalization,” (unpub-lished; Washington: World Bank).

Lane, Philip, and Gian Maria Milesi-Ferretti, forth-coming, “The External Wealth of Nations,” Journalof International Economics.

Levine, Ross, and Sara Zervos, 1998, “Stock Markets,Banks, and Economic Growth,” American EconomicReview, Vol. 88 (June), pp. 537–58.

Levine, Ross, and David Renelt, 1992, “A SensitivityAnalysis of Cross-Country Growth Regressions,”American Economic Review, Vol. 82 (September), pp.942–63.

McKinnon, Ronald, 1973, Money and Capital inEconomic Development (New York: Oxford UniversityPress).

Mody, Ashoka, and Antu Panini Murshid, 2001,“Growing Up With Capital Flows” (unpublished;Washington: International Monetary Fund).

Obstfeld, Maurice, 1998, “The Global Capital Market:Benefactor or Menace?” Journal of EconomicPerspectives, Vol. 12 (Fall), pp. 9–30.

O’Donnell, Barry, 2001, “Financial Openness andEconomic Performance” (unpublished; Dublin:Trinity College).

O’Rourke, Kevin, 2001, “Globalization and Inequality:Historical Trends,” CEPR Discussion Paper No.


172

2865 (London: Centre for Economic PolicyResearch).

Quinn, Dennis, 1997, “The Correlates of Change inInternational Financial Regulation,” AmericanPolitical Science Review, Vol. 91 (September), pp.531–51.

Rajan, Raghuram, and Luigi Zingales, 1998, “FinancialDependence and Growth,” American EconomicReview, Vol. 88 (June), pp. 559–86.

Rodrik, Dani, 1998, “Who Needs Capital-AccountConvertibility” (unpublished; Cambridge, Mass.:Harvard University).

Sala-i-Martin, Xavier, 1997, “I Just Ran Two MillionRegressions,” American Economic Review, Vol. 87, pp.178–83.

Shaw, Edward, 1973, Financial Deepening in EconomicDevelopment (New York: Oxford University Press).

UNCTAD, 1996, Incentives and Foreign Direct Investment(New York).

———, 2000, World Investment Report 2000.Van Rijckeghem, Caroline, and Beatrice Weder, 2000,

“Spillover Through Banking Centers—A Panel DataAnalysis,” IMF Working Paper No. 00/88(Washington: International Monetary Fund).

Wei, Shang-Jin, 2001, “Domestic Crony Capitalism andInternational Fickle Capital: Is There a Connec-tion?” (unpublished; Washington: BrookingsInstitution).

Williamson, John, and Molly Mahar, 1998, “A Surveyof Financial Liberalization,” Essays in InternationalFinance, No. 211 (Princeton, New Jersey, PrincetonUniversity Press).

World Bank, 2001, Global Development Finance(Washington).

REFERENCES

173

international financial integration and

Documents