7 WORKI'JG PAPERS

|eb and Intwemdonal Finanoe

Intemational Economics UspartmentThe World BankSeptember 1993

WPS 1186

Recent Estimatesof Capital Flight

Stijn Claessensand

David Naude

Estimates of capital flight calculated using several methodolo-gies do not differ widely. Capital flight is more widespread thancommonly assumed and, relative to GDP, evenly distributed.The capital flight-GDP Lorenz curve is close to the 45-degreeline.

lhePdicy Rueaach Wading Papas dsanite e findinu of wotk in progreu and enoagptheeichangeof ideas arnangBank sffand afl adm inunsued in d.voopmn t isues. Tbese pape.. disributad by dte Rsearch Adviscy Staff. cary the namn of the aShos,uCas1lythirvis.andshouwbldbeudedtecdaodingly.bThefndinp.intaptmati. ad andcmcau meaaudto own. TheyAdcod so be aflbuted to the Wadd BDan, iu Boed of Diectos, its managneat, o any of its nmber countnie.

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

_ * t ~~~~~~~ ~ ~~Polloy R !"Debt and Inllkdnal Fbwnoo

WPS 1186

This paper - a product of the Debt and International Finance Division, International EconomicsTepartment - is part of a larger effort in the department to study the integration of developing countriesin world financial markets. Copies of the paper are available free from the World Bank, 1818 H Street NW,Washington, DC 20433. Please contact Rose Vo, room S8-042, extension 31047. (Septemrber 1993, 27pages). For individual country data, available on a floppy disk, contact Shelley Fu, extension 33885 (fax477-)661).

Researchers and policymakers have in recent Claessens and Naudd discuss the data usedyears paid considerable attention to the phenom- for calculating capital flight and the adjustmentenon of capital flight. Researchers have focused that must be made. They present aggregateon four questions: What concept should be used capital flight figures using the various measuiresto measure capital flight? What figure for capital for o4 developing countries.flight will emerge, using this measure? Can theoccurrence and magnitude of c-apital flight be The figures show a pattem of increasingexplained by certain (economic) variables? What capital flight until 1988, followed by a return ofpolicy changes can be useful to reverse capital flight capital between 1989 and 1991.f-light?

Claessens and Naude present regionalClaessens and Naude focus strictly on aggregates of capital flight and rank countries

presenting estimates of capital flight using a and regions by the level of capital flight relativenumber of alternative methodologies. In their to GDP. They find that capital flight is morediscussion of these methodologies, they show widespread than commonly assumed and,that although the approaches to measuring relative to GDP, is rather evenly distributed. Thecapital flight differ, the identities used in balance capital flight-GDP Lorenz curve is above the 45-of payment data make thcm close in final degree line, indicating that countries with ameasurement. In particular, the so-called World smaller GDP have more capital flight than oneBank residual and Dooley methods - presented would expect if it were distributed proportionatein the past as very different approaches to to GDP.measuring capital flight - actually producesimilar measurements.

The Policy Research Working Paper Series disseminates the findings of work under way in the Bank. An objective of the seriesis to get these findings out quickly, even if presentations are less than fully polished. The findings, interpretations, andconclusionLs in these papers Jo not necessarily represent official Bank policy.

Produced by the Policy Research Dissemination Censer

RECENT EST[MATES OF CAPITAI FLIGHT

by

Stijn ClaessensWorld Bank

and

David NaudeWorld Bank

The opinions expressed do not necessarily represent those of the World Bank or itsBoard of Directors. The paper draws on joint work with Kevin Chang and Bob Cumby.We would like to thank Kevin Chang, Bob Cumby and Mike Dooley for usefulcomments, David Stewart and Cliff Papik for support with the data, and VinceMcCulough for editorial assistance.

Summary

The phenomnenon "capital flight" has over the years received considerableinterest from researchers as well as policy makers. Papers on capital flight deal withone or more of the four following questions: what is the right concept to use formeasuring capital flight; what is the figure for capital flight following from thismeasure; can the occurrence and mnagnituide of capital flight be explained by certain(economic) variables; and what policy implications follow (e.g., which policy changescan he useful in reversing capital flight).

The purpose of this paper is not to add any new insights on which (if any)capital flight measure is the right one to use or what causes capital flight. Neither willthe paper attempt to draw any policy iimplications. The sole purpose of this paper is topresent recent estimates of capital flight using a number of alternative, existingmethodologies.

The paper first discusses the various methodologies which have been used formeasuring capital flight. The paper draws attention to the fact that, while the alternativemethodologies may differ in their approach to meastiring capital flight, tne identitiesused in balance of payment data make these methodologies close in final measurement.In particular, the so called "World Bank Residual" and the "Dooley" methods, while inthe past presented as very different approaches to meastring capital flight, are actuallyvery close in measurement.

The paper discusses in detail the data used for calculating capital flight figuresand the various adjustments which need to be made. It then presents aggregate capitalflight figures using the various measures for a group of 84 developing countries.(Individual country data are available from the World Bank on a floppy disk, contactMs. Shelley Fu, tel. (202) 473-3885, fax. (202) 477-0661). I'he figures calculated bearout that the various capital flight figures are highly correlated and, with the exceptionof a few years, show the same pattern of (increasing) capital flight until 1988, followedby a reversal (return of flight capital) in the years 1989-1991.

The paper also presents various regional aggregates of capital flight, and rankscountries and regions by the lev'el of capital flight relative to GDP. The paper tinds thatcapital flight is a much wider spread phenomenon than commonly asserted. Whencompared to the respective countries' GDP, capital flight is quite even distributed. Thecapital flight-GDP Lorenz-curve is even above the 45-degree line, indicating thatcountries with smaller GDP have more capital flight than one would expect if it weredistributed over countries proportionately to GDP.

Table of Contents

I - Introduction p. 1

II - Different Measures of Capital Flight p. 2

1 - Residuial Methods of Capital Flight p. 4

2 - The Dooley Method P. 5

3 - Hot Money (Cuddington) Method p. 8

4 - Trade Misinvoicing p. 8

III - Data Issues p. 10

IV - Capital Flight: Some Figures p. 14

V - References p. 22

Annexes

I. Introduction

Over thle years, "capital flight" has received much attention from researchersand policy makers (see, for example, World Bank, 1985 and 1993). Using estimates ofcapital flight from developing countries, authors have drawn attention to the magnitudeof this phenomenon--particularly for Latin Amer 'a. It is generally agreed that theselarge private capital outflows have represented an important macroeconomic problemfor many developing countrics in the past two decades. The reversal of capital flight inthe early-1990's for some Latin American countries has added a new dimension to thisphencmenon. Yet, capital flight remains a little understood aspect of these countries'economies and it even remains difficult to ascertain the magnitude of capital flight.Consequently, a better understanding of the extent of past capital flight, as well as areliable measure of possible capital flight reversals in recent years, may be a usefulinput to a realistic assessment of current prospects for renewed invetrment and growthin developing countries.

Papers on capital flight try to answer a number of questions. What is the bestway to measure capital flight? Can the occurrence and magnitude of capital flight beexplained by (economic) variables? And what policy implications follow (for example,how to rev erse capital flight)? This paper is not intended to debate what causes capitalflight or the best measure; nor will it draw policy implications. Its sole purpose is toprovide a set of consistent figures on capital flight using existing measures for 84developing countries over a long period (1971-91). T'his should be useful forresearchers, since much detailed data used here originates from the World Bank andIMF and may not be generally available. '

Since capital flight -' calculated for many developing countries, no attempt wasmade to correct the figures in cases where data problems are known (or appear) toexist. Nor was any adjustment made for country-specific information or circumstances2.For individual countries or time periods, researchers may want to make adjustments tothe data calculated here.

Section II describes different concepts and methodologies used for measuringcapital flight. Section m describes the data and daa sources used. Some summaryfigures are provided in section IV on the various measures for 84 countries and forgroups of developing countries (regional) and compares the capital flight measures toother (macro-) variables. Section V lists recent papers on capital flight. Detailed capitalflight numbers ale found in Annex 1, while country coverage and notts on data are inAnnexes 2 and 3. Annex 4 provides the relative coverage fcr the 84 countries comparedto all developing countries.

Ilndividual country data are available from the World Bank on a floppy disk (contact: Ms. Shelly Fu, Tel.(202) 473-3885, Fax. 477-0661).2Eggerstedt et al. (1992), for example, draw attention to the fact that most capital flight measures do notcorrect for the foreign currency deposits held by state-owned companies.

2

H. Different Measures of Capital Flight

What is capital flight? Definitions are manv They can be broad, wovering allprivate outflows, or narrow, seeking to exclude some "normal" flows oi internationaltransactions. for each definition, moreover, there are several different measures. Andsome measures are complementary; for example, trade misinvoicing can be added toone rneasure to create another. Finally, data sources differ, which give conflicting finalfigtlres and make comparisons among the various measures difficult. For instance, Erbe(1985) uses the same methodology as the World Bank (1985) but uses OECD (notBank) debt data. Hardly surprising, then, that many measures of capital flight can beconstructed. The four most common approaches are the residual measure (used by theWorld Bank, Morgan Guaranty and Cline); measuring the stock of unreported foreigrnassets (Dooley's method);3 hot money measutl,s (Cuddington); and measuring trademisinvoicing. For each, there are variations which lead to (minor) differences. Thestarting point for all measures is balance-of-payment figures.

Balance of payments. Let us take a stylized balance-of-payments framework,using standard notaijon, but supplemented by World Bank debt data and based on theEMF's Balance of Payments Yearbook.4

3We will show below that in -ffect tne Dooley method is also a residual method.4The presentation of items below follows that in Ciimby and Levich (1987), Chang and Cumby (1991)and Chang, Claessens and Cumby (1993). The Balance of Payments Yearbook line item numbers arethose reported in the hard copy of the' BOP Year books. The corresponding codes are listed in Annex 3.

3

Table 1: Balance of Payments

A. Current Account (IMF BoPY* line 1 .44),which includesAl. Travel: credit (IMF BoPY line 9)A2. Reinvested earnings on direct investment abroad (IMF BoPY line

11)A3. Reinvested earnings on direct investment domestically (IMF

BoPY line 12)A4. Other investment income: credit (IMF BoPY line 19)

B. Net Equity Flowsof which:B1. Net foreign direct investmnent (IMF BoPY line 45..52)B2. Portfolio Investment: Corporate equities (IMF BoPY line 59..61)

C. Other short-term capital of other sectors: net (IMF BoPY line 93 .97)of r.'hich:Cl Other assets (IMF BoPY line 94)

D. Portfolio investment, other bonds (IMF BoPY line 56-58)

E. Change in deposit money banks' foreign assets (IFS line 7a.dzf)

F. R,..erves (IMF BoPY line 98.. 111)

G. Net errors and omissions (IMF BoPY line 112)

H Other long-term capital of resident official sector (IMF BoPY line 62-68)

or

H'. Change in external debt (World Debt Tables - see further section 4)

* IMF Balance of Payments Yearbook.

Since all capital flight measures attempt to estimate private capital flows, this wouldimply that capital flight can be simply measured as the sum of identified outflows (C +D + E). And, if all unidentified capital flows were private capital outflows, then neterrors and omissions (G) would also be included.

4

This seems straightforw;.id cnough. So, why so many different measures? Themain reason is doubts about tGe quality and accuracy of the balance-of-paymentsstatistics. As a result, analysts have pieferred to derive the annual piivate capital flowsby using the balance-of-payments ide:itity and proxying other balance-of-paymentsitems. How? The balance-of-payments identity implies that:

A + B + C + D + 8i + F + G -t H = Q.

01'

C + D + E + =- (A + i + F + H)

The second equation implies that p.ivate capital flows plus net errors andomissions (capital flight) is equal to (the negative of) the sum of the cuffent-accountdeficit, net equity flows (FDI and corporate equity), iprreases in reserves, and otherlong-term capital of the official resident sector. Measuring capital flight through eitherside of the equation give the same result. On the left hand side of the equation,however, analysts can use only balance-of-paymens for some statistics. On the righthand side, however, there is more (and presumably better) information availableelsewhere. So, tP- starting point for most capital flight methods is the right hand sideof the equationt.

Residual Method.

This measures the "residual" of the "sources of funds" over the "uses of funds".Sources of funds include all net official inflows (increases in net external indebtednessof the public sector) and the net flow of foreign direct investment.5 User of fundsinclude the currei'+-account deficit and additions to reserves. Outward capital flightexists when sources of funds exceed uses of funds, and vice-versa for inward capitalflight. In terms of balance-of-payments items, capital flight under the residual methodis thus the sum of: A + B + F + H. By the balance of payments identity, this is alsoequalto-(C + D + E + G).

The residual method, however, does not rely only on balance-of-payments data.For some items, other sources are used to get a better estimate of private capital flows.Most notable is item H (other long-term capital of resident offic;al sector or net officialexternal borrowing), for which the year-to-year change in external debt according toWorld Bank data (H') may be more accurate. Thus. the residual method measurescapital flight as A + B + F + H'; by the balance-of-payments identity, this is equal to- [C + D + E + G + (H-H')]. Capital flight is simply the sum of identified privatecapital outflows (C + D + E), the net errors and omissions from the balance-of-

51n contrast to much of the literature, we inclide net acquisition of corporate equities in our measure offoreign direct investment. Apart from establishing major ownership control, flows of corporate equitiestepresent a claim on a country's resources similar to foreign direct investment.

5

payments accounts (G), and the difference between reported net official capital and thechange in external debt acLording to World Bank deta (H-H').6

The World Bank, Morgan, and Cline methods use variations of the residualmethod.

World Bank: A + B + F + H'Morgan: A + B + E + F + H'Cline: A - (Al +A2+A3+A4) + B + E + F + H'

The Morgan mnethod differs slightly from the World Bank's. It includes E, thechange in banking system's foreign assets (which has a negative sign in BoPY and isthus subtracted). The Cline method is similar to Morgan's but substracts travel (credit),reinvested FDI Income (abroad and domestically), and other investmeni income(credit).

The Doolgy Method.

This seeks to measure the stock of privately held foreign assets that do notgenerate income reported to the domestic authorities. It does so by cumulating theidentified capital outflows in the balance-of-payments accounts arid making threeadjustments to capture unreported capital flows. The first is to add errors and oixiissions(G). The second is based on a comparison of the World Bank data on the stock ofexternal debt and the external borrowing flows reported in the balance-of-paymentsaccounts. Dooley adds the difference between each year's change in external debt(according to the World Bank) and the lows as officially recorded (that is, H' - H) tohis estimate of the increase in private-sector foreign assets. Dooley assumes thus thatthe entire difference is private-sector acquisition of foreign assets.

The third adjustment is to calculate the stock of external assets needed to givethe (balance-of-payments) investment income, by using an international market interestrate (for exaimple, the one-year US Treasury Bill rate). If investment income is under-reported, then the imputed stock of external assets will less than external assets usingbalance-of-payments figures (and after making the previous two adjustments). Thedifference between the two is the stock of flight capital; and difference from year-toyear is the measure of capital flight.7

61f both the balance of payments and the World 3ank report net official capital accurately, then thechange in the stock of debt reported by the World Bank will match the net borrowing flows reported inthe balance-of-payments accounts. This is often not the case. Exchange rate revaluation effects, debtreclassification, and "discoveries" of existing debt may cause estimates to diverge. If, however, adiscrepancy remains after corrections for these effects are made (see section 4), then the un:ecordelincrease in external liabilities must be due to an urnderestimation of balancing transactions, such as anunrecorded increase in external assets by the private sector (that is, capital flight).7As the Dooley method obtains sorne stock figures for external assets a. ' lainis by cumulating balanceof payments flows figures, flight ca.pital depends on the stock assumed ii the ini al year (i.ormally zero).

6

The Dooley method can, thus, be described as:

Stock of Unreported Foreign Claims = A + B + C - D

where

A. Cumulative Recorded non-equity Balance of Payments Assets

= AI + A2 + A3 + A4 + A5 + So + A7 + A8

o/w Other Long-term Capital (Assets)Al ?ortfolio investment (IMF BoPY lines 53,36)A2 Resident official sector (IMF BoPY lines 62..64)A3 Depo0t money banks (IMF BoPY lines 69..71)A4 Other sectors (IMF BoPY lines 77..79)

o/w other Short-term Capital (Assets)AS Resident official sector (IMF BoPY lines 84.. 85)A6 Deposit money banks (IMF BoPY line 89)A7 Other sectors (IMF BoPY lines 93..94)A8 Reserves (IMF BoPY lines 98.. 109)

B. Cumulative Errors and Ormissions (IMF BoPY line 112)

C. Adjustmenit for Unrecorded Claims

- Stock of external debt as reported to the World Bank (see Section 4)Cumulative Recorded Balance of Payments Liabilities

where: Cumulative Recorded Balance of Paymients Liabilities

= C1 + C2 + C3 + C4 + C5 + C6 + C7 + C8

o/w Other Long-termn Capital (Liabilities)Cl Portfolio investr"!nt (IMF BoPY lines 54,55,57,58)C2 Resident officiai aector (IMF BoPY lines 65..68)C3 Deposit money banks (IMF BoPY lines 72..76)C4 Other sectors (IMF BoPY lines 80..83)

o/w Other Short-term Capital (Liabilities)C5 Resident official sector (IliF BoPY lines 86. .88)C6 Deposit money banks (IMF BoPY lines 90..92)C7 Other sectors (IMF BoPY lines 95..97)C8 Reserves: IMF credit (IMF BoPY lines 110.. 111)

Consequently, flow figures are more meaningful than stock figures because they do not depend on anyassumptior for the initial stock value.

D. Capitalized Reported non-FDI Inconme

= Flow of other investmlenit income: credit (IMF BoPY lines 15,17,19) /One-year US Treasury Bill rate (IFS line 60c..zf, US)

While the residual and Dooley methods differ greatly in conceptuial approach(one looks at flows, the other at stock figures), they u.se some of the sant'. measuires, orthose closely linked through the balance-of paymenits identity. Under the residualmethod, capital flight equals A + B + F + H'. By the balance-of-payments identity,this equals - (C +D + E + G+ (LI-H')), the negative of private capital flows (sholl-term, including banks and 'ionbanks, and long-term; C, D and E). errors andomissions, and the difference between World Bank and balance-of-payments reportedversions of offici.0 capital. These last four items are those that Dooley uses to cal2ulatethe total reported and unreported) assets held abroad. These annual changes in total(reported and unreported) assets held abroad, according to tne Dooley method, aresimply annual capital-flight estimates according to the residual method.

These similarities also imply that the Dooly estimate can be made more easilyby taking the World Bank residual flow measure and the subtracting the flow of capitalcorresponding to the series for the imputed stocks of reported assets. It is not possiblehowever, to equah. the year-to-year difference in the imputed stock of reported assetswith a flow, since they include both new flows froin the country, as well as interestearned abroad but not repatriated.8 In principle, new flows and reinvested earningsshould be reflected in the capital account, just as retained earnings on FDI enter thecapital account. In Dooley, the new flows reported in the capital account are deemedunreliable. Why otherw:se impuite the stock of reported assets from the cu.rent-accountand not measure it directly from the capital account'? And. in practice, earnings on non-FDI inivestments which are reinvested abroad are seldom reported &cujrately in thecapital accoullt of developing coLlntries.

This implies that the year-to-year change in the irnputed stock of reported assetscannot be divided between a new flow from the country and earnings reinvested. Anarbitrary assumption has to be made that none of the reinvested interest enters thecapital accouint, that all earned interest is reinvested and, thus. that the year-to-yeardifference in stocks includes interest accrued.9 The flow on reported assets is thereforecalculated as the difference between the stock this year miniuis the stock last yeargrossed up by one plus the interest rate. The Dooley measuire of the flow of capitalflight is then obtained by subtracting this flow from thie World Bank residual measLureof the annual flow of capital flight.

8New flows from the countryn means new flows beyond those occurring through reinvestment of interestearnings.9For some individual countries, better adjustine)its canl perhaps be made. See for example Eggerstedt etal. (1993) for Mexico.

8

Hot Money Method.

This calculates private capital flows directly by taking the (negative of) errorsancd omissions and private short-term capital figtires from the balance-of-payments. Themeasure of private short-tenm capital varies (from country to country) in sorne of thestudies employing this method and three different hot money measures has thusdeveloped.

Hot Money 1 = - (G + C1)Hot Money 2 = - (G + C)Hot Money 3 = - (G + C + D1 + D2)

Where

G Net errors and omissions (IMF BoPY line 112)

C Other short-term capital of other sectors (IMF BoPY line 93 ..97)of which:C1. Other assets (IMF BoPY line 94)

DI Portfolio investment: other bonds (IMF BoPY line 56.. 58)

D2 Portfolio investment: corporate equities (IMF BoPY line 59. .61)

Trade Misinvoicing

Export underinvoicinig and import overinvoicing can hide capital flight, anddifferences in statisLics of the reporting country and its trading partners can helpidentify it. '° Here, we adopt the industrial countries' reported trade figures as thestandarc6 of reference from which to calculate invoicing biases and use the IMFDirection of Trade (IMF-DOT) data. To put imports as reported by the country andimports as reported by the world (exports by the country) on a comparable basis, bothare adjusted from a CIF (cost, insurance, freight) basis to a FOB (free-on-board) basis.This implies that reported imports, nonnally expressed on a CIF basis, are adjusteddownward bv a couintry-specific CIF/FOB ratio so that exports and imports can becompared on a consistent FOB basis.

Export misinvoicing = (Xw/CIFFOB factor) - Xc

Import misinvoicing = (MC/CIFFOB factor) - Mw

10 Difference in trade statistics can he due not only to capital flight (under-invoicing exports,over-invoicing imports), but also tax evasion (under-invoicing for both exports and imports), inconsistentreporting methods, and bad reporting.

9

where

Xw: Imports from that country as reported by the world CIF (IMFDOT-imports of world with that country as partner, lineIMP_cif)

Xc: Exports as reported by the country FOB (IMF DOT-exports ofcountry with world as partner, line EXP_fob)

Mc: Imports as reported by the country CIF (IMF DOT-imports ofcountry with world as partner, line IMP cif)

Mw: Exports to that country as reported by the world FOB (IMFDOT-imports of world with that country as partner, lineEXP fob)

CIFFOB factor: CIF to FOB ratio (IFS line ..v..z.)

Imports are adjusted downward by a country-specific CIF/FOB ratio. In thispaper, a positive sign signifies capital flight (overinvoicing of imports orunderinvoicing of exports) and a negative sign signifies reverse capital flight(underinvoicing of imports and overinvoicing of exports). Since both exportunderinvoicing and import overinvoicing add to capital flight, the two should be addedfor the net effect of trade misinvoicing on capital flight.

111. Data Issues

Sources: This paper relies on World Bank and IMF data, and ptiblished data. (The onlyunpuiblislhed data is the currency-adjuistment factor.) The IMF Balance of PaymentsYearbook (BoPY) is the source for almost all BOP cuirrent and capital-accounlit itenms.Data from thie BoPY were taken fromii the 13 March 1993 update of the IMF BoPY-tape. The World Bank Debtor Reporting System (DRS) is the source for all externaldebt statistics. The external debt figures comile froml thie DRS-data base (as frozeni on 29January 1993) and used for the putblished World Debt Tables 1992/93. The IMFDirection Of Trade (DOT) statistics provide the trade data. Data froin DOT were takenfrom the 16 March 1993 update of the tape. The one-year US-Treasury interest rateinformation came from the IMF International Financial Statistics (IFS) data set. IFS-data were also used for the change in the deposit money banks foreign assets. Datafrom the IFS were taken from the 12 March 1993 update of the tape.

As data are constantly revised, variations can arise becauise of different vintagesof data and differences between the published versions of data and tapes (and othercomputerized data sources). The effect of such variations on calculations, however,should be small.

IMF Balance of Payments Yearbook (BoPY) Data

Although no adjustments were made to any of the BoPY-figures, some conventions areworth mentioning:

a) Signs: Items in the IMF BoPY are prefixed by signs: + for debits (for example,exports), - for credits (for example, imports). In the capital account, assets (lending)and liabilities (borrowing) are stocks which increase (-/+) or decline (+/-). Net figures(flows) and net balances (stocks) are calculated not hy subtractinig debits from credits,but by adding them, for example, exports + imports = the trade balance. Thus, thecapital account, reserves, and errors and oimissionls are added to (rather than subtractedfrom) the currenit accounlt to arrive at the overall balance.

b) Aggregated versus Detailed Presentation: The Aggregated Presentation in the BoPYprovides capital account and reserves figures net of counterpart items (valuationchanges in reserves, {de}monetization of gold, and SDR's), exceptional financing, andliabilities constituting foreign authorities reserves, which are slhown separately. As aresult, the current account. capital account, Limange in reserves, and errors andomissions in the aggregated accouints do not add to zero. They do, however, balance inthe Detailed Presentation, which incluides counterpart items. Throughout, this paperuses the Detailed Presentation.

World Bank Debtor Reporting System

External debt data from the World Bank Debtor Reporting Systeim (DRS),employed for the residuial and Dooley measurLs, cannot be used directly. The mieasuresrequire putting together a few separate components to obtain the annual net increase inexternal liabilities. The stalling point is the year-to-year change in the dollar-mlieasuire(ddebt stock. This can inaccurately reflect net borrowings (new disbursements minuisprincipal repayments), since it may incilide tht effects of, say, cross-ctirrencyexchange-rate flucttiation.s, debt found (reschieduiled or converted), and debt forgiveni orreduced. To correct for these, the change in debt stock (inclusive of short-terim, IMF,and exclusive of private, non-guaranteed debt) is reduced by the change in debt stocksdue to cross-currency exchange-rate fluctuiations, while forgiven or reduced debt anddebt service is added back to the annual change.

a) Cross-Currency Exchange-Rate Fluctuations: One problem in using the change indollar-measured debt stock is that non-dollar denomninated debt will fluctuiate in dollarterms from year to year (in addition to any net flows) due to cross-currency exchange-rate fluctuations, Excludinig that part of the change due to such fluctuiations, however,can only be done for the public and publicly guaranteed portion of external (debt (PPG)since the currency mix is known only for these loans. (For World Bank loans, it isdone only after 1983.) The currency adjuistment on SDR denomilinated debt (forexample, IMF credits) has not been calculated.

b) Debt "found": Debt stocks miay increase eacih year because of debt found (forinstance, during a rescheduling), debt which was incurred earlier. Since thlis"discovery" is often not reflected in (past) net flows. the change in the stock of debt ispreferable to net flow figures, but may not reflect when the debt was incurred. Wedidn't correct the yearly change for any debt foLuind, but incurred earlier.

c) Debt converted: When private-sector extenial debt is converted to public-sector debt(in, say, a rcscheduling agreement). the private sector substituites a liability to thegovernment for a foreign liability. Although there has been no new net flow of lunds.the private sector will increase its net foreign assets, which may increase the measuireof capital flight. The change in the stock of debt yields better estimates of increases ilthe private sector's net foreign assets (and thlius of capital flight) than the flow of netborrowings.

d) Capitalizations: The annual stock of debt may change because of capitalization otarrears, reschedulinig of principal or interest payments due. and capitalization of(interest) penalties imposed, all to the extent that are not fully reflected in net flows otborrowings. Whether to include rescheduled interest (and other capitalizations) or not?In principle, balance-of-payments data show interest paymentis due. rathier thiall inade.As a result, when interest is rescheduled withinl the year, the currenlt-accoLiunt deficit

will be overstated on an actual cash flow basis and thus the resicidial estimiiates of privatecapital flows understated. Incliding rescheduled interest duie as an ofticial flow in the

12

capital account will correct this. If, however, interest paid is measured in the currentaccouInt (and no new flow for the rescheduled interest payments is recorded in thecapital accounit) using the adjiusted change in the stock of debt, including rescheduledinterest, will overstate thie extent to which the private sector has increased its claimsab)road." This paper uses current-accouint figtures from the balance of payments andloes not correct thie change in debt stock for the possibility that interest paid (instead ofinterest dIlue) is recorded in the balance of payments.

e) Adhistment for Forgiveness: Forgiveness of debt, principal, and interest and otherdebt and debt service re(ductions (throuigh market-based transactions) will decrease thedebt stock in that year. This would imply a lower annual change in the stock of debt,but with no correspontding net flow (or decrease in capital flight). Therefore, the debtstock figures are adjusted for the net amounts forgiven or reduced each year. '

f) Private Non-G(uaranteed Debt (PNG): Including or excluding private non-guaranteeddebt in capital flight depends on whether gross or net private sector foreign assets are tobe measuired. If net private external claims are to be measured, the relevant debt figureis the change in public and publicly guarantee(d external debt. If it is gross private-sector foreigin assets, then the change in private, non-guaranteed debt should beincltlded, and the residual measuire of capital flight will be higher. However, sinceprivate external indebtedness represents an actual liability of the private sector(expected to be serviced and repaid by the private sector), arguably the simultaneousacquisition of a foreign asset and foreign liability shouild not be considered "capitalfliglht". In calculating the World Bank residual measures net acquisition of foreignassets by the private sector is chosen as the relevant measure and PNG-debt is thusexcluded'. However, a measuire of capital flight is also provided which includes PNG-debt (and the necessary adtjustimients for debt reduction. forgiveness, etc.).

h) Short-Termi Debt: The same argument for excluding PNG-debt can be made forshort-tenr debt-niamilely, that it sihould be included when measuring gross capitaloutflows, while only public and publicly guaranteed short-term debt should be used fornet capital outtflows. Untfortuniately. World Bank data on short-term external debt arenot clisaggregated between public and private flows. Anecdotal evidence for Sub-

'"l1Ihe same reasoning suge,ests that interest arrears ought to be added to the stock of debt if interest duehut unpaid is included in the current-account bal-snce-of-pavnments data.'"For market-hased debt reductions transactions. for example, debt huyhacks. the change in debt stock isonly1 corrected bN the discount in the transaction. The repaNyment part. say the cash used in a buNback(the price times the face value of the debt). represents a prepayment bN the borrower, and acorresponding reductioll in net liabilities. This is normally reported in the W'orld [)ebt Tables under tileline Amortizations. Once corrected for the discount in the debt reduction transactions, the year to yearchanige in the debt stockls Vould thus corre..pond to the net borroAings (abstracting from possible otherdi tetrences).

In examining net private sector acquisition of foreign assets. we differ from much preNious literature.Dooley et. aL. ( 1985). the World Bank ( 1985), and Miorgan Guaranty, ( 1986) all use private, non-guaranteed debt in addition to public and publicly guaranteed debt in computing residual estimates ofprivate sector acquisition of foreign assets.

13

Saharan Africa suggest that much short-term debt is publicly guaranteed, but this isprobably not so for Latin America and Asia with their freer private sector. Includingshort-term debt can thus be seen as calculating a upper bound of capital flight asmeasured by this method, and excluding the short term as a lower bound. For thepurposes of this paper, short-term debt is included.14

Data Convention: Interest Compounding

Given the assumption made for the Dooley measure of flows on the imputedstocks of reported foreign assets (that is, no reinvested earnings are reported in thecapital account) and the corresponding calculation of the flow figures, all flowsreported here are on the same basis and are not compounded for interest earned abroad.All flows can be grossed ip using an international interest rate for a flight capitalmeasuire which includes the cumulative returns on the assets held abroad.

14Erbe (1985) uses OECD data on medium and long-term gross external indebtedness with estimates ofshort-term debt for some countries. Cumby and Levich (1987) use public and publicly guaranteed long-tern debt and all short-term debt. Morgan Guaranty (1986) excludes the increase in short-term foreignassets of the banking system from the increase in total private sector claims. Acquisition of foreignassets by nonbank agents, however, continues to be considered capital flight by Morgan. Since Morganoffers no explanation for treating the banking system differently from other firms and individuals. we donot pursue its distinction here.

14

IV. Capital Flight: Some Figures

We calculated capital flight usirng eight of the different measures--World BankResidual, idem plus Private Non-Guaianteed debt, Dooley, Morgan, Cline, and threeHot Money Methods--as well as measures of trade misinvoicing (see Annex 1). Thesemeasures require many data items from various sources over a long period, starting in1971. For many developing countries, consistent data are not always available. Here,we study a group of developing countries for which data were available at least since1975: we did not exclude any countries for which the most recent year was missing.These selection criteria led to a sample of 84 countries (see Annex 2). The 1990 GDPof these represent about 75 percent of all (151) developing countries' GDP. We haveverified that there is a close relationship between capital flight in the 84 countries andall countries (Annex 4).

Annual flows in 1971-90 for the 84 countries using the World Bank Residualand Dooley methods, follow a similar pattern (Figure 1). This is hardly surprisinggiven the similarities in approach. The exceptions are 1990 and 1991, when Dooleyshows much less capital flight than the World Bank Residual measure. Why thedifference?

Figure 1: World Bank Residual and Dooley Measure of Capital Flight (Annual Flows)8p[,o,0s of USS

190

100

80

71 72 73 74 76 79 77 78 79 80 8l 8. 8 84 85 86 87 88 8Y 90 9

-50

.100

World Bank Residual Dooley

Unlike other measures, the Dooley method attempts to distinguish betweenlegitimate and illegitimate foreign assets by excluding the stock of imputed legitimateforeign exchange holdings (obtained by dividing reported foreign interest earnings byan international interest rate, one-year U.S. Treasury bill rate). Since all measurescalculate the stock of total foreign holdings in a similar way, exclusion of legitimateforeign assets must account for any major divergence.

15

The U.S. Treastiry bill rate used in these calculations was 5.4 percent in 1991,down from 8.1 percent in 1989. All else equal, lower interest rates will increase theimputed stock of legitimate foreign assets, leading to lower capital flight under theDooley measure. Reported interest earnings were $24 billion in 1991, up from $19billion in 1989. These two effects reinforced each other, resulting in substantiallyhigher and increasing estimated stocks of legitimate foreign assets for 1990 and 1991--up from $235 billion in 1989 to $436 billion in 1991, and thus lower capital flight forthese years under Dooley (see table in Anniex 1). In general, the Dooley measure ismore variable than the World Bank residual measure because it depends on a ratio oftwo variables (reported earnings and the international interest rate), which may havelittle contemporaneous correlation.

The (broadest) Hot Money and the World Bank Residual measure are ,,lotted inFigure 2. The variation between Hot Money measures and the World Bank Residualmeasure arises from differences between balance-of-payments figures on the flow ofpublic and publicly guaranteed debt and World Bank data on the change in the stock ofpublic-sector external debt (adjusted for currency-movements, resch dulings, etc.).Even so, the two measures broadly reflect the same trends, save for 1977, and 1986-89, when the Hot Money measure shows lower capital flight. This suggests that largeincreases in public-sector debt went unrecorded in t'ie balance-of-payments accounts inthose years.

Figure 2: World Bank Residual and Hot Money (Annual Flows)Bl11ons of USS

120

t 00

80

e0

40

20 e

0 -f--'..._ ... .

71 72 73 74 76 78 77 78 79 80 81 82 83 84 86 86 87 88 8a 90 91

.20

World Bank Residual Hot Money 1

The three misinvoicing measures are exports underinvoicing, importsoverinvoicing and the sum of the two (Figure 3); positive numbers for exports indicateunderinvoicing and for imports overinvoicing, while negative numbers for exportsdenote overinvoicing and for imports underinvoicing. Underinvoicing of imports isclear after 1981. Exports are overinvoiced up to 1985 and undervoiced thereafter.

16

While the underinvoicing of exports is consistent with capital flight, thesystematic underinvoicing of imports after 1981 and the overinvoicing of exports before1985 are not. Underinvoicing of imports may be because many developing countrieshave heavy import taxes and other value-based irnport restrictions, counteractingcapital-flight incentives to overinvoice imports. Trade co.itrols, especially on imports,give incentives not to record goods. Reasons for overinvoicing exports before 1985may be related to export subsidies and other incentives to generate legitimate foreignexchange eamings (see Eggerstedt et al. (1993). Taking the two misinvoicing measurestogether, there is negative capital flight until 1989 (a lower trade surplus reported bythe trading partners than reported by the countries), implying unrecorded capitalinflows to these 84 countries.

Assuming that these trade misinvoicing estimates are a useful approximation foractual unrecorded imports and exports, we adjust our capital flight based estimatesaccordingly, since funds spent on imported goods would no longer constitute assetscontrolled by the country.

Figure 3: Capital Flight Due to Trade Misinvoicing (Annual Flows)8llon of US$

30 T

1 6 -.-

7Z .72 73 74 76 7e 77 78 79 sO 81 82 83 84 85 8e 87 es 89 90 91

-16 -r

-30 -

-46 -

Trade Misinvoicing, - Trade Misinvoicing, Trade Misinvoicing, TotalExports Imports

What happens when misinvoicing is added to the World Bank residual andDooley measures? They both show that outward capital flight flows peaked in 1988 atabout $100 billion and $130 billion respectively for each measure (see Figure 4). It isalso seen that capital flight was reversed in 1989 and 1990. According to anecdotalevidence, this has continued in 1991 and 1992. In the past four years, much flightcapital has reportedly returned particularly to three Latin American countries--Argentina, Brazil, and Mexico."5

150ur numbers show for Mexico a return of flight capital of about $20 billion total in 1989. 1990 and1991. For Argentina and Brazil. our data show lower levels of capital flight reversal during the sameperiod.

17

Figure 4: World Bank and Dooley Combined with Trade Misinvoicing (Annual Flows)BSllhone of USt

160 T

100

50 X\

° --0 ---- - ------/71 72 73 74 78- -1_ 77 78 79 80 81 82 83 84 86 8e 87 88 89 90 91

World Bank Residual + Trade Misinvoicin - Dooley + Trade Misinvoicing

Over a long period (in this case, 20 years) stock figures of flight capital aremore interesting that annual flow figures. According to the World Bank Residualmethod, at the end of 1991, the accumulated stock of flight capital was $500 billion, orabout 44 percent of their stock of external debt. The Dooley measure shows capital-flight stock of about $140 billion (12 percent of debt stock). The difference between thetwo (about $360 billion) is equal to the imputed stock of reported assets held abroad.

Figure 5: Stocks of Flight Capital (with and without interest compounding)

600

4C0

300

200

100

0 t : -* . - . ._. . ..71 72 73 -74A, 75 76 77 78 79 90 8 82 83 84 85 86 87 88 89 90i 6

-100

- - - -World Bank Dooley + Trade World Bank Dooley + TradeResidual + Misinvoicing Residual + Misinvoicing (noTrade (with interes:) Trade interest)Misinvoicing Misinvoicing (no(with interest) interest)

lx

Most analytical attention has focused on Latin America's capital flight. Inabsolute amount, this accounts for a large share of the total capital flight--by ourmneasure about 44 percent in 1985 and 37 percent in 1991 (see Figure 6).

Figure 6: Top Ten Countries for World Bank Residual MeasureAverage Annual Flows 1981-1991

M,Ilons of US$

4500

4000

3500

3000

2500

2000

1 500

1000U

0 4-'- -- - ~-- --

China Argentina Mexico Venezuela Poland Brazil Nigeria Egypt, Turkey IndonesiaArab Rep.

Capital flight, however, is more widespread than previously thought, and ismore important for many developing countries, not just Latin America. Relative toGDP, exports, and external debt, for instance, capital flight of Latin Americancountries is much less than most others. For Sub Saharan Africa, t': stock of capitalflight at the end of 1991 (excluding misinvoicing) represented more than 85 percent ofthe region's GDP. By this measure, capital flight was, relatively, the largest for theMiddle-East and North Africa region, with a ratio of a stock of flight capital to GDP of118 percent. For all other regions, the ratio was 30 percent or less at the end of 1991.Least capital flight was in South-East Asia, only 15 percent of 1991 GDP.

19

Figure 7: Ratio of Flight Capital to GDP by RegionWorld Bank Residual, Stocks 1991

Per,entge5

120

100

80

40

20 ioE U.*.._+.-,_ ..iE. t Asia Europe and Latin North South Asia Sub- Alland the Central America Africa and Saharan DevelopingPacific Asia and the the Middle Africa Countries

Caribbean East

Ranking the 84 countries by the stock of flight capital to GDP shows that onlythree are Latin American (Nicaragua, Bolivia and Venezuela), four are African(Gabon, Zambia, Sudan and Uganda) and three are from the Middle-East and NorthAfrica (Egypt, Syria and Jordan). For Gabon, the stock of capital flight is almost tripleits 1991 GDP (Figure 8). Much the same picture emerges when the stock of capitalflight is compared to external debt, adjusted to exclude the effect of cross-currenicymovements on the dollar-measured debt stock (Figure 9).

Figure 8: Top Ten Countries for the Ratio of Flight Capital to GDPWorld Bank Residual. Stocks - 1991

Percentoage

300 r

2001

1100

50 IiSudan Egypt, Panama Gabon Bulgaria Nigeria Venezuela Syrian Jordan Uganda

Arab Rep. Arab Rep.

20

Figure 9: Top Ten Countries for the Ratio of Flight Capitai to Debt SockWorld Bank Alsidual, Stocks 1991

Pwcarn,ge

900

coo_

700

400

300

100IiiliEE0

Romania Vanuatu Oman Gabon Venezuela Egypt, Trinidad Panama Lesotho MalaysiaArab Rep. and

Tobago

That capital flight is widespread in relative terms is clearly demonstrated bysome Lorenz-curves, when capital flight in dollars is ranked from smallest to largest(Figure 10). By this measure, 80 percent of the countries represent less than 20 percentof all capital fight stock. For about 65 percent (or 38 countries), there is negativecap.tal flight. The remaining 20 percent of the countries represent more than 80 percentof all capital flight stock.

Figure 10: Lorenz Curve of Flight Capital - World Bank Residual, Stocks - 1991

.. I

0.8 t*

0.6

Share of Stock of I.,Total Flight Capital

0.2 -

O -+- e-- -- S--, -

o 0.1 0.2 b 03 0.4 0.6 0oe 0. 0.8 0.9

-0.2

Deciles

21

Figure 11: Lorenz Curve of Flight Capital weighted by GDPWorld Bank Residual, Stocks - 1991

0,8 -

0 a

Shtia of Stock ofTotal Flight Capital

0.4

0 0.1 0.2 0.3 0.4 05 0F . 0.7 0.8 0.9 1

GDP Deciles

When deciles of GDP are ranked against deciles of share of capital flight stock,the picture changes drastically (Figure I1). The curve is close to (but above) the 45degree-line, indicating that countries with smaller GDP have a higher share of capitalflight. Capital flight is thus not just widespread but is a substantial drain of scarceexternal resources for many developing countries.

22

V. Recent Papers on Capital Flight

Cuddington, John T., 1986, "Capital flight: Estimates, issues, and explanations,"Studies in International Finance, no. 58, Princeton University Department ofEconomics, Princeton, NJ.

Chang, P. H. Kevin and Robert E. Cutinby, 1991, "Capital Flight in Sub-SaharanAfrican Countries," in Ishrat Husain and John Underwood, eds, Africaii ExternalFinance in the 1990s, Washington, DC, World Bank, 162-184.

Chang, P. H. Kevin, Stijn Claessens and Robert E. Cumby, 1993, "Conceptual andMethodological Issues in the Measurement of Capital Flight," forthcoming in a volumeedited by Benu Varman-Schneider, Christian-Albraechts-Universitat, Kiel.

Cumby, Robert E. and Richard M. Levich, 1987, "On the Definition and Magnitude ofRecent Capital Flight," in Donald R. Lessard and John Williamson, eds, Capital Flightand Third World Debt, Washington, DC, Institute for International Economics, 27-67.

Dooley, Michael P., William Helkie, Ralph Tryon, and John Underwood, 1986, "Ananalysis of external debt positions of eight developing countries through 1990," Journalof Development Economics, Vol 21, no. 2, 283-318.

Dooley, Michael P., 1986, "Country-specific risk premiums, capital flight and netinvestment income payments in selected developing countries," International MonetaryFund, mimeo.

Dooley, Michael P., 1987, "Comment," in Donald R. Lessard and John Williamson,eds, Capital Flight and Third World Debt, Washington, DC, Institute for InternationalEconomics, 79-81.

Dooley, Michael P., 1988, "Capital flight: A response to differences in financialrisks,' International Monetary Fund Staff Papers, 35, September, 422-436.

Eggerstedt, Harald, Rebecca Brideau Hall, and Sweder van Wijnbergen, 1993,"Measuring Capital Flight: A Case Study of Mexico," Working Paper, No. 1121, TheWorld Bank.

Erbe, Suzanne, 1985, "The flight of capital from developing countries,"Intereconomic.s, November/December, 268-275.

Gajdeczka, Przemyslaw, 1989, "Financial Flows to Developing Countries," OuarterlyReview, World Bank, March.

23

Gajdeczka, Przemyslaw and Daniel Oks, 1989, "Domestic deficits, debt overhang, andcapital outflows in developing countries," in Finance and the Intemational Economy,R. O'Brien and I. Iversen, eds., (Oxford University Press), 103-120.

Gulati, Sunil, (1987), "A Note on Trade Misinvoicing," in Donald R. Lessard andJohn Williamson, eds, Capital Flight and Third World Debt, Washington, DC, Institutefor International Economics, 68-78.

International Monetary Fund, 1987, Final Report of the Working Party of theStatistical Discrepancy in World Current Account Balances, Washington, D.C.:International Monetary Fund.

Khan, Mohsin S., and Nadeem Ul Haque, 1987, "Capital flight from developingcountries," Finance and Development, March.

Kindleberger, Charles P., 1987, "A historical perspective," in Donald R. Lessard andJohn Williamson, eds, Capital Flight and Third World Debt, Washington, DC, Institutefor Intemational Economics, 7-26.

Lessard, Donald R. and John Williamson, 1987, "Introduction" and "The problem andpolicy responses," in Donald R. Lessard and John Williamson, eds, Capital Flight andThird World Debt, Washington, DC, Institute for Intemational Economics, 1-5 and201-254.

Morgan Guaranty Trust Company, 1986, "LDC capital flight," World FinancialMarkets, March.

World Bank, 1985, World Development Report, Washington, DC, World Bank.

_________ 1993, Global Economic Prospects. Washington, DC, World Bank.

Annex I1- Summary Table for the 84-Country SaMple16 - Billions of US$1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1-980 1961 1982 1983 19t4 9t5 !956 1987 19J5 1989 199fl 1991WorldBank Residual 0 -0 -I 2 2 3 -4 56 10 2 1 37 IS 27 33 21 32 64 98 106 -16 -3 IS+C-hange in Adjustod Debt Stock 0 6 9 12 la 26 29 89 65 73 95 79 70 56 39 46 S2 106 105 -I 20 S0f Adjuwted Debt Stock 37 43 52 63 81 107 136 226 291 365 464 543 613 668 708 754 835 942 1047 1046 1066 1145t Debt OusstaM. h Diab ...... . .. All 52 61 72 BB III 139 173 272 349 423 529 618 697 759 791 S79 962 10S0 1066 1073 1144 1202- Debi Outstanl. i& Disb. PNG is 17 20 23 29 33 37 43 51 59 69 SS 93 101 97 W9 So 73 61 54 63 74+ Debt Reduction, PPG 0 0 0 0 0 0 a 0 0 0 0 0 0 0 0 9 1 4 1 1 1 1 21 21+ Debt Reduction, Short-Tema 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0- Exc-hange Rate Impact 0 0 0 1 1 -2 0 4 7 -0 -3 -12 -9 -10 -14 36 47 69 -31 -16 37 4+ Foreign Direct IrBvcstiment 2 2 2 3 3 6 4 5 7 9 S 12 10 IL S 9 S 1`2 la 21 22 29+ Current Account -6 -8 -6 -4 -9 -29 -27 -31 -44 -42 -52 -83 -76 -35 -14 -23 -34 -3 -4 -12 -I -33+ Inerease in Rescrves I -0 -5 -9 -10 2 -11 -8 -9 -16 -11 12 23 5 -13 -0 7 -20 -14 -27 -46 -52t Corporxte Equities 0 -0 0 0 0 0 0 -0 0 0 0 0 0 0 0 0 0 2 2 3 3 0WorkJ Ban Residual t PNG 0 0 1 4 7 7 -0 62 IS 30 46 37 32 41 17 24 55 91 94 -23 6 2S+ World Bank Residual 0 -0 -I 2 2 3 -4 56 10 21 37 18 27 33 21 32 64 98 106 -16 -3 IS+ Ct nge in DOD. PNG 0 2 3 3 6 4 4 6 S 9 9 19 5 Fi 4 -S -9 -7 -12 -7 9 1 1• Debt Reduction, PNG (change) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ' 0aogn 0 -I1 -2 -I -4 2 -8 51 1 13 27 1 1 23 30 22 29 61 100 106 -24 -16 5+ World B nk Residmal 0 0 1 2 2 3 4 56 10 21 37 IS 27 33 21 32 64 9S 106 -16 -3 I- Change in Deposit Money Bans 0 0 1 3 6 2 3 5 9 9 10 7 4 3 -2 3 3 -2 0 a 13 12f Deposit Money Banks 3 4 5 S 14 17 19 25 34 44 54 62 67 69 68 71 73 71 72 S0 95 100Cline 0 -4 -6 7 -11 -6 -16 41 -12 -5 3 -17 -5 5 -6 0 31 64 64 -72 -74 47+ Morgw 0 I1 -2 -I 4 2 S9 51 1 13 27 If 23 30 22 29 61 10t) 106 -24 -16 5-Travel h Tourism 3 3 4 5 6 7 S 10 12 1 5 17 IS IS 17 19 20 24 30 36 40 47 42-Reinvested Famrnigo, Credit 0 -0 -0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0-Reinve led E mings. Dcbit -0 -0 -0 -0 - I -I I -2 -2 -2 -3 -3 -3 -2 -1 -2 -2 -2 -2 -2 -2 -2Other Investment Incomc: Private 0 0 0 1 2 '2 2 2 3 5 9 13 13 10 I11 10 9 8 a la 13 1 1Doolcy 0 -4 -5 0 -17 -12 6 52 9 19 4 24 30 57 30 IS5 56 I19 140 -9 -82 -9241 Chbange in Adjtistod Dcbt Sto,k 0 6 9 12 IS 26 29 S9 65 73 95 79 70 56 39 46 S2 lobm !05 -1 20 go- tCapital Account 6 S 5 4 10 33 30 31 46 41 56 96 93 38 20 27 36 4 7 S 1 32- Net Errors & Omissions -0 -0 0 -0 -2 -4 -5 -I -2 1 -4 -14 -17 -3 -5 4 2 -I -3 4 0 1+ Increae in Reacres -I -0 -5 -9 -10 2 -11 -S -9 -16 -11 12 23 5 -13 -0 7 -20 -14 -27 -46 -52+ Foireign Dircect Investment 2 2 2 3 3 6 4 5 7 9 a 12 10 a S 9 S I: la 21 22 29+ Compor te Equities 0 -0 0 0 0 0 0 -0 0 0 0 0 0 0 0 0 0 2 2 3 3 0-Change r Stock ofRepoiedAssts O 3 3 1 19 IS5 0 4 0 3 41 -6 -3 -24 -10 17 a -21 -34 -7 79 110f Sloct of Repontod Assets I11 1 5 2 1 24 45 64 71 79 S4 92 145 IS9 1S9 1S6 192 227 252 246 227 235 333 436+ Invcstmnen lInconie 0 0 0 2 4 4 4 4 6 9 I17 22 20 1 6 IS 1 7 1 5 14 1 5 4 25 24/ US Treasury Bill Rate 6.4 4.3 4.1 7.0 7.9 5.S 5.0 5.3 7.2 10.0 11.6 14.1 10.7 8.6 9.6 7.5 6.0 5.S 6.7 S.l 7.5 5.4Hi tMoncy I 0 0 -0 1 3 5 6 3 3 3 7 20 24 7 12 1 1 0 7 7 -I 3 2-Net Effrso h Omiwions -0 -0 0 _0 _2 _4 _5 -I _2 1 -4 -14 -17 -3 -5 4 -2 -I -3 4 0 1-Othr Shon-Tean CVital. Debit -0 -0 -0 -0 -0 -0 -2 -2 -2 -4 -4 -6 -S -4 -7 -7 2 -6 -3 -3 -3 -3Hot Mome 2 -I -0 -0 I 1 3 5 0 5 -3 3 22 30 19 15 13 -3 3 5 3 -10 -10-NelEro h. Omi o io -0 -0 0 -0 -2 -4 -5 -I1 -2 1 -4 -14 -17 -3 -5 4 -2 -I1 -3 4 0 1_-Other Sboni-Tcnn Capital, Net I 0 -0 -0 0 1 -0 0 -3 2 0 -9 -14 -16 -10 -Y 5 -2 -2 -7 10__Hot Money 3 -I -0 -0 I 1 3 5 0 5 -3 4 22 29 19 15 1 2 -5 0 3 - I -15 -19-Hot Money 2 -I -0 -0 I 1 3 5 0 5 -3 3 22 30 19 15 1 3 -3 3 5 3 -10 -10- nve tmenl in Bornds 0 0 -0 -0 -0 -0 -0 -0 -0 0 -0 0 0 -0 -0 0 0 0 0 0 2 10-Corporate Equitics 0 -0 0 0 0 0 0 -0 0 0 0 0 0 0 0 0 0 _ 2 3 3 0Tnhde Misinv.oicing, Exporb -5 4 -7 -9 1 2 -16 -19 -19 -23 -2S 44 -16 -12 -15 -9 -2 10 I10 1 3 24 25 20+ Expoon (world vrww adj to FOB) 37 39 46 611 10S 106 124 14S 160 214 273 307 2S6 285 31S 320 309 363 423 469 522 561+ Expons (wofld view, CIF) 42 45 53 iS 123 IIS 13S 164 178 238 304 339 315 313 349 351 339 397 463 512 571 612/ FOB-CiFConvcesionFactor 115 115 115 114 113 III III Ili III III III 110 110 110 ]10 110 110 109 109 109 109 109- Exponsu(country vicw, FOB) 40 38 50 73 115 113 133 155 17S 236 293 324 299 301 329 324 301 355 413 447 499 543TAde Misinvoicing. Impons 2 3 3 2 3 1 1 10 13 16 1 1 19 -2 -S 0 -11 -9 -12 -16 -19 -15 -26 -10+ ImpoiU (country view, adj toFOBR) 39 43 50 69 112 133 141 166 202 239 307 342 29S 277 281 289 291 32S 380 417 474 542+ Imporu (country view, CIF) 46 50 57 79 12S 146 154 182 221 262 335 376 32S 305 309 318 321 361 418 45S 519 591/FOI-C[F Convcrmon Fn--tor 115 115 115 114 113 III III III III III III 110 110 110 110 110 110 109 109 109 109 109- mporb (world vicw. FOB) 42 47 54 77 123 135 143 166 193 237 305 347 309 279 295 301 307 348 405 437 503 556Trade Muinvocn Total -2 -I 4 -7 -9 -6 -9 -6 -7 -17 -25 -IS -20 -14 -20 -11 -2 -6 -7 9 -I 10i Tnadc MisLnoicing, ExpoiU -5 4 -7 -9 -12 -16 -19 -19 -23 -28 -44 -16 -12 -15 -9 -2 10 10 13 24 25 20t Trade Misinvoicing Imporst 2 3 3 2 3 1 1 10 13 16 1 1 19 -2 -S 0 11i 9 -12 -16 -19 -15 -26 -10

16This table was derived through simple aggregation of individual series; this implies that items may sometimes not add to the total since wheii imnportant items were nmissing thetotal was set to zero whiile other ilems miay lhave been available, this is miostly true for early years. Also the CIF-FOB factor has been recalculated as ain weighted average.

Annex 2 - The Sample of 84 CountriesEast Asia and the Pacific Europe and Central Asia Latin America and the North Africa and the South Asia Sub-Saharan Africa

Caribbean Middle EastChina Bulgaria Argentina Algeria Bangladesh BeninFiji Cyprus Barbados Egypt, Arab Rep. Nepal BotswanaIndonesia Czechoslovakia Bolivia Jordan Pakistan Burkina FasoKorea, Rep. Hungary Brazil Morocco S-r Lanka ChadMalaysia Malta Chile Oman ComorosPhilippines Poland Colombia Syrian Arab Rep. Co igoSolomon Islands Portugal Costa Rica Tunisia Cote d'lvoireThailand Romania Dominica EthiopiaVanuatu Turkey Dominican Rep. GabonWestern Samoa Yugoslavia Ecuador Gambia, The

El Salvador GhanaG renada KenyaGuatemala LesothoHaiti MadagascarHonduras MaliJamaica NigerMexico NigeriaNicaragua RwandaPanama SenegalParaguay SeychellesPeru SudanSt. Kitts and Nevis SwazilandSt. Lucia TanzaniaSt. Vincent TogoTrinidad and Tobago UgandaUruguay ZaireVenezuela

Annex 3 - Description of Input Series!'Output Name Database Label Field Database Name Last UpdateTravel & Tourism BOP ID.A D4Q Travel: credit (BOP line 9) 13 March 1993Reinvested Earnings, Credit BOP IEIA DAQ Reinvested earnings on direct investment abroad (BOP line 11) 13 March 1993Reinvested Earnings, Debit BOP I El B D4Q Reinvested earnings on direct investment in the country (line 12) 13 March 1993Investment Income BOP IG.A D4Q Other investment income: credit (BOP lines 15,17 and 19) 13 March 1993Other Investment Income: Private BOP IG3A D4Q Other invetsment income: credit (BOP line 19) 13 March 1993Investment in Bonds BOP 6N IX D4Q Portfolio investment: other bonds (BOP lines 56 to 58) 13 March 1993Corporate Equities BOP 6PIX D4Q Portfolio investment: corporate equities (BOP lines 59 to 61) 13 March 1993Other Short-Term Capital, Net BOP 8.2X D4Q Other short-term capital of other sectors (BOP lines 93 to 97) 13 March 1993Other Short-Term Capital, Debit BOP 8K2X D4Q Other short-term capital of other sectors: other assets (line 94) 13 March 1993Increase in Reserves BOP 2.. X D4Q Reserves (BOP lines 98 to 111) 13 March 1993Foreign Direct Investment BOP 3. .X D4Q Direct investment: net 13 March 1993Capital Account BOP ... X D4Q Capital account 13 March 1993Current Account BOP A.. C D4Q Current account 13 March 1993Net Errors & Omissions BOP .A.X D4Q Net errors and omissions 13 March 1993US Treasury Bill Rate IFS 60C . ZF USA US Treasury Bill Rate 12 March 1993Deposit Money Banks IFS .7A.DZF Deposit Money Banks: Assets 12 March 1993FOB-CIF Conversion Factor IFS .. V.. Z. FOB-CIF Conversion Factor 12 March 1993Debt Outstand. & Disb., All DRS SLT_ALL PLUSIMF DOD Total external liabilities (short and long term + IMF credit) 29 January 1993Debt Outstand. & Disb., PNG DRS PNG_ALL TOTAL DOD Private nonguaranteed, all creditors, total 29 January 1993Exports (country view. FOB) DOT EXP_FOB CRY Exports of goods & services, FOB, as reported by the country 16 March 1993Imports (country view, CIF) DOT IMP_CIF CRY Imports of goods & services, CIF, as reported by the country 16 March 1993Imports (world view, FOB) DOT EXP_FOB WLD Exports of goods & services, FOB, as reported by the world 16 March 1993Exports (world view, CIF) DOT IMP CIF WLD Imports of goods & services, CIF, as reported by the world 16 March 1993Debt Reduction, PPG DRS DRP Debt Reduction, PPG 29 January 1993Debt Reduction, Short-Term DRS DRS Debt Reduction, Short-Term 29 January 1993Debt Reduction, PNG DRS DRN Debt Reduction, PNG 29 January 1993Exchanige Rate, Impact DRS XR1 Exchange Rate Impact 29 January 1993

l7BOP. IFS and DOT are all I'F databases, respectivelv: Balance of Pavments, International Financial Statistics and Direction of Trade; thev reside on the

BESD svstem. DRS is the World Bank. IECDI Debt Reporting System database

Annex 4 - Relative Importance of the 84-Country Sample in Developing World'8

GDP of the 84-Country Sample as a share of GDP of GEP World

10oo0% I- -.-.-.-..-..... _-.

90%

90% _

70%

80%

60%

40%

30%

20% -

10%-

0% --- I1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1986 198e 1987 1988 1989 1990 1991

7 __ _______ _ _ __IE 84 Developing Countries C i Complement for GEP world

Stock of Flight Capital of the 84-Country Sample as a Share ofStock of Flight Capital of GEP World

100% - --

90% _

/0%

70%

D0%

D0%

40%

30%

20%

10%

0%

1977 1978 1979 1980 1981 1982 1983 1994 1995 1988 1987 1988 1989 1990 1991

Z 84 Developing Countries Complement for GEP world

18GEP world refers to 151 countries as used in Global Economic Prospects, 1993. The World Bank, The GEPworld figure of capital flight does not include data for all countries (outside the 84) for all years.

Policy Research Working Paper Series

ContactTitlb Author Date for paper

WPS1 164 Power, Distortions, Revolt, and Hans P. Binswanger July 1993 H. BinswangerReform in Agricultural Land Relations Klaus Deininger 31871

Gershon Fader

WPS1 165 Social Costs of the Transition to Branko Milanovic August 1993 R. MartinCapitalism: Poland, 1990-91 39026

WPSI 166 The Behavior of Russian Firms in Simon Commander August 1993 0. del Cid1992: Evidence from a Survey Leonid Liberman 35195

Cecilia UgazRu. an Yemtsov

WPS1 167 Unemployment and Labor Market Simon Commander August 1993 0. del CldDynamics in Russia Leonid Liberman 35195

Ruslan Yemtsov

WPS1 168 How Macroeconomic Projections Rashid Faruqee August 1993 N. Tannanin Policy Framework Papers for the 34581Africa Region Compare with Outcomes

WPS1 169 Costs and Benefits of Debt and Eduardo Fernandez-Arias August 1993 R. VoDebt Service Reduction 33722

WPS1 170 Job Search by Employed Workers: Avner Bar-Ilan August 1993 D. BallantyneThe Effects of Restrictions Anat Levy 37947

WPS117'i Finance and Its Reform: Beyond Gerard Caprio, Jr. August 1993 P. Sintim-Laissez-Faire Lawrence H. Summers Aboagye

38526

WPS1 172 Liberalizing Indian Agriculture: Garry Pursell September 1993 D. BallantyneAn Agenda for Reform Ashok Gulati 37947

WPS1 173 Morocco's Free Trade Agreement with Thomas F. Rutherford September 1993 N. Artisthe European Community: E. E. Rutstr6m 38010A Quantitative Assessment David Tarr

WPS1 174 Asian Trade Barriers Against Primary Raed Safadi September 1993 J. Jacobsonand Processed Commodities Alexander Yeats 33710

WPS1 175 OECD Trade Barriers Faced by the Bartlomiej Kaminski September 1993 J. JacobsonSuccessor States of the Soviet Alexander Yeats 33710Union

WPS1 176 Cash Social Transfers, Direct Taxes, Branko Milanovbc September 1993 R. Martinand Income Distribution in Late 39065Socialism

WPS1 177 Environmental Taxes and Policies Neil Bruce September 1993 C. Jonesfor Developing Countries Gregory M. Ellis 37699

Policy Resarch Working Paper Series

ContactTitle Author Date for paper

WPS1 178 Productivity of Public Spending, John Baffes September 1993 C. JonesSectoral Allocation Choices, and Anwar Shah 37699Eoonomic Growth

WPS1 179 How the Market Transition Affected Bartiomibj Kaminski September 1993 P. KokilaExport Performance in the Central 33716European Economies

WPS1 180 The Financing and Taxation of U.S. Harry Huizinga September 1993 R. VoDirect Investment Abroad 31047

WPS1 181 Reforming Health Care: A Case for Zeljko Bogetic September 1993 F. SmithStay-Well Health Insurance Dennis Heffley 36072

WPS1 182 Corporate Governance in Central Cheryl W. Gray September 1993 M. Bergand Eastern Europe: Lessons from Rebecca J. Hanson 31450Advanced Market Economies

WPS1 183 Who Would Vote for Inflation in Cheikh Kane September 1993 T. HollestellbBrazil? An Integrated Framework Jacques Morisett 30968Approach to Inflation and IncomeDistribution

WPS1 184 Providing Social Benefits in Russia: Simon Commander September 1993 0. del CidRedefining the Roles of Firms and Richard Jackman 35195and Government

WPS1 185 Reforming Hungarian Agricultural Morris E. Morkre September 1993 N. ArtisTrade Policy: A Quantitative David G. Tarr 37947Evaluation

WPS1 186 Recent Estimates of Capital Flight Stijn Claessens September 1993 R. VoDavid Naud6 31047

WPS1 187 How Should Sovereign Debtors Andrew Warner September 1993 J. QueenRestructure Their Debts? Fixed 33740Interest Rates, Flexible InterestRates, or Inflation-indexed

WPS1 188 Developmentalism, Socialism, and Mario Marcel September 1993 S. FlorezFree Market Reform: Three Decades Andres Solimano 39075of Income Distribution in Chile

WPS1 189 Can Communist Economies Alan Gelb September 1993 PRDTMTransform Incrementally? China's Gary Jefferson 37471Experience Inderjit Singh

WPS1 190 The Government's Role in Japanese Yoon Je Cho September 1993 T. Ishibeand Korean Credit Markets: A New Thomas Hellmann 37665Institutional Economics Perspective