Illicit Financial Flows From Developing Countries: 2001-2010

Dev Kar and Sarah Freitas
December 2012

Illicit Financial Flows From Developing Countries: 2001-2010
Dev Kar and Sarah Freitas1
December 2012

Global Financial Integrity Wishes to Thank The Ford Foundation for Supporting this Project

1

Dev Kar, formerly a Senior Economist at the International Monetary Fund (IMF), is Lead Economist at Global Financial Integrity (GFI) and Sarah Freitas is an Economist at GFI. The authors would like to thank Simón Ramírez Amaya, an intern at GFI, for assistance with data research as well as Raymond Baker and other staff at GFI for helpful comments. Any errors that remain are the authors’ responsibility.

We are pleased to present here our analysis of Illicit Financial Flows From Developing Countries: 2001-2010. In our previous annual reports we have utilized the World Bank Residual model adjusted for trade mispricing, presented in both gross non-normalized and in filtered normalized calculations. In this year’s report we are adding a second form of analysis, the Hot Money Narrow model adjusted for trade mispricing, again presented in non-normalized and normalized calculations. The results for 2010 are summarized as follows: World Bank Residual Plus Trade Mispricing, Non-Normalized World Bank Residual Plus Trade Mispricing, Normalized Hot Money Narrow Plus Trade Mispricing, Non-Normalized Hot Money Narrow Plus Trade Mispricing, Normalized US$ 1,138 billion US$ US$ US$ 892 billion 859 billion 783 billion

The consideration which led us to include a second type of measure of illicit flows has to do with the potential for some level of licit financial flows to appear in the gap between the source of funds and use of funds. This will bear further examination, as we continue to augment our analytical methodologies. What is perhaps most important to appreciate is that none of our estimates include several major components of illicit flows, such as smuggling, cross-border movements of cash, trade mispricing that occurs in the same invoice exchanged between importers and exporters, and the mispricing of all services and intangibles which are not covered in IMF Direction of Trade Statistics. If we had reliable figures or estimates on these exclusions, without question our estimates of illicit flows from emerging market and developing countries would be much higher. Our preceding 2009 analysis utilizing the World Bank Residual model produced a range of estimates of illicit flows from US$775 billion to US$903 billion for the year. The 2010 estimates summarized above in four calculations and depicted in charts in the text indicate a growing order of magnitude, suggesting that the slightly improving global economy afforded rising levels of unrecorded flows. Whatever strengthened financial regulations may be in place or may be contemplated cannot yet be seen to have an effect on the continued passage of funds out of poorer countries, through the global shadow financial system, and ultimately into richer western economies. The somewhat more conservative analysis produced by the Hot Money Narrow methodology suggests that trade mispricing is rising in importance in the shift of illicit funds abroad.

Illicit Financial Flows from Developing Countries: 2001-2010

c

We have added to this year’s report a special section on Sovereign Wealth Funds (SWF), recognizing their growing importance. The lack of clarity and consistency with which SWF-related transactions are handled in balance of payments compilations of some countries calls for remedial action by regulatory and statistical agencies with such large assets under management. Libya’s SWF appears to have been used for political as much as investment purposes, whereas Norway’s SWF has made every citizen of the country a comfortable kroner millionaire. With now more than 60 SWFs around the globe an enormous pool of capital exists, and standards of accounting for such funds need to be regularized through the auspices of the International Monetary Fund. Six years ago when Global Financial Integrity was formed, the term “illicit financial flows” was nonexistent or insignificant in the global political-economy lexicon. Today this term and its surrounding concepts are used and addressed by the G20, UN, World Bank, IMF, OECD, European Union, and national governments across the planet. A UN official recently commented that GFI’s job is to “unpack the opaque.” And this will continue to be our role in years to come. We thank Dev Kar and Sarah Freitas for their excellent work in producing this analysis. The ongoing support of the Ford Foundation is most gratefully acknowledged and appreciated. Raymond W. Baker Director, Global Financial Integrity December 12, 2012

d

Global Financial Integrity

Contents
Abstract ......................................................................g Executive Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .i I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 a. Normalization through use of filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 III. Trends in Illicit Financial Flows from Developing Countries and Regions . . . . . . . . . . . . . . . . 9 IV. Special Issues: Sovereign Wealth Funds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 V. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

II. Coverage of flows in the World Bank Residual method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Appendix Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Charts and Tables within Report
Table A. Chart 1. Table B. Chart 2. Chart 3. Table C. Table D. Chart 4. Chart 5. Chart 6. Chart 7. Table E. Table F. Chart 8. Chart 9. The United States and China: Balance of Payments Components, 1991 . . . . . . . . . . . . . 4 Volume of Illicit Financial Flows in Nominal Terms from All Developing Countries 2001-2010 . . . . . . . . . . . . . . . . . . . . . . . . . 5 Four Estimates of Capital Flight, All Developing Countries, 2001-2010 . . . . . . . . . . . . . . . 6 Illicit Financial Flows by Region, 2001-2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Normalized vs. Non-normalized GER, 2001-2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Illicit Financial Flows Broken Down by Region in Nominal Terms . . . . . . . . . . . . . . . . . . 10 Illicit Financial Flows Broken Down by Region in Real Terms . . . . . . . . . . . . . . . . . . . . . 12 Illicit Flows in Real Terms 2001-2010; Regional Shares in Developing World Total . . . . . 14 Real Rates of Growth of IFFs from 2001-2010 by Region . . . . . . . . . . . . . . . . . . . . . . . . 15 Regional Illicit Flows in Nominal Terms 2001-2010; Shares Related to HMN and GER Components. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Top 20 Countries’ Cumulative Illicit Flows, Nominal HMN+GER Non-normalized, 2001-2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Total Illicit Financial Flows from the Top Ten Developing Countries . . . . . . . . . . . . . . . . 17 Changes in Cumulative Non-Normalized Illicit Outflow Rankings in Nominal Terms . . . 18 Top 10 Countries of 2010 Tracking Nominal Illicit Financial Flows . . . . . . . . . . . . . . . . . 19 Oil Prices and Illicit Flows Out of Five Major Countries . . . . . . . . . . . . . . . . . . . . . . . . . . 20 from Developing Countries, 2001-2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Chart 10. Heat Map of Cumulative Illicit Financial Flows

Illicit Financial Flows from Developing Countries: 2001-2010

e

f

Global Financial Integrity

Abstract
This 2012 report on illicit financial flows (IFFs) from developing countries and regions updates estimates provided in Illicit Financial Flows from Developing Countries Over the Decade Ending 2009 published by Global Financial Integrity in December 2011. The report presents an additional method of estimating flows based on the Hot Money Narrow measure adjusted for trade misinvoicing. The measure results in estimates of capital flows that are more likely to be illicit by nature. These conservative estimates of illicit flows are then compared against the previous estimates based on the World Bank Residual method adjusted for trade misinvoicing (the CED+GER method). The gap, representing flows of “licit” capital, has narrowed since the onset of the global economic crisis in 2008. We conclude by pointing out that estimates of illicit financial flows from certain countries with large sovereign wealth funds (SWFs) may be subject to significant margins of error due to incomplete or incorrect recording of SWF-related transactions.

Illicit Financial Flows from Developing Countries: 2001-2010

g

h

Global Financial Integrity

Executive Summary
Two main issues, which arose in the past year, encouraged us to supplement our standard methodology used to estimate illicit flows based on the World Bank Residual method adjusted for trade misinvoicing. First, we investigated the net measurement of inward from outward capital flight traditionally used by economists in academic journals. We reaffirm our commitment to a gross outflow approach, rather than a net approach, because only a return of licit capital that is recorded can offset loss of capital. The return of unrecorded and illicit capital cannot be used for productive purposes. In other words, the gross/net issue is linked to the nature of the capital. Second, we explored the effect of the global financial crisis on both illicit and licit flows, determining that the residual method of estimating illicit flows adjusted for trade misinvoicing may include some licit capital as well as illicit. Moreover, if the CED+GER method includes licit capital, the support for a gross outflows approach is strengthened, as one cannot be sure whether the inward capital flight is licit or illicit in nature. Therefore, we present estimates of illicit flows using both the CED+GER method and the conservatively focused Hot Money Narrow method adjusted for trade misinvoicing (HMN+GER). A firm judgment as to which method provides a more accurate method for estimating illicit flows is somewhat premature at this stage. While the HMN+GER method provides more conservative estimates of illicit outflows, it may exclude certain illicit transactions such as round-tripped FDI which could be erroneously recorded as private sector flows. We invite readers to comment on the appropriateness of the two methodologies for estimating illicit flows including reasons why one should be preferred over the other. Using robust (non-normalized) estimates for both measures, we found that in 2010 developing countries lost between US$858.8 billion to US$1,138 billion, implying that as much as US$279 billion of the higher figure could be licit capital flows of the private sector—outflows that took place as a result of “normal” portfolio maximizing considerations. While the two estimates were quite close in the early 2000s, capital market liberalization in many large emerging markets may have encouraged more licit or “normal” capital flight over the years. The gap between the HMN+GER and CED+GER estimates widened, reaching a peak in 2008 at the onset of the global economic crisis. In the following year, outflows of legal capital flight dropped more sharply than illicit outflows. The latter showed a steady upward trend for all developing countries more or less immune to macroeconomic shocks and adjustments.

Illicit Financial Flows from Developing Countries: 2001-2010

i

We then further analyzed the gap between the two non-normalized (or robust) estimates in order to shed light on possible legal capital flight from the various regions of the developing world during the 10-year period studied. We observed that in the case of developing Europe, the MENA region, and Western Hemisphere, the gap tends to widen over time, reaching a peak in 2008 although it has closed in the following two years. The widening gap is perhaps the result of more normal capital flight due to a relaxation of capital controls. In all three regions, licit outflows plunged in 2009 due to the effects of the crisis on domestic and foreign capital markets noted above. In the case of Asia, the gap, which was almost nonexistent in the early 2000s, began to widen in 2005 and reached a peak in 2008 at the onset of the crisis. But the gap closed almost completely in 2009 as both licit and illicit outflows from Asia fell in tandem. A finding that is worrisome is that the HMN+GER measure of illicit flows increased at a faster pace than the CED+GER measure (13.3 percent vs. 12.6 percent). The adverse implication is that increasing illicit flows are likely to result from a worsening of governance-related drivers given the scant evidence of a systematic increase in measurement errors. In order to avoid overlap and to focus more sharply on flows that are likely to be purely illicit, we analyze trends, shares, and country rankings based on the HMN+GER method. According to this measure, illicit flows from developing countries in the robust calculation increased by over US$500 billion since 2001 implying a real growth rate of 8.6 percent per annum on average, which exceeded their average rate of economic growth (6.3 percent per annum). We established that about 80 percent of illicit outflows were channeled through the deliberate misinvoicing of trade, although the shares of outflows from trade misinvoicing and the balance of payments have fluctuated. We found that Asia, accounting for 61.2 percent of cumulative outflows, was still the main driver of such flows from developing countries. Indeed, five of the ten countries with the largest illicit outflows (China, Malaysia, the Philippines, India, and Indonesia) are in Asia. The Western Hemisphere, led by Mexico, follows at 15.6 percent, with the Middle East and North Africa (MENA) at 9.9 percent. Developing Europe follows MENA in share size, making up 7.0 percent of illicit flows, with the balance flowing out of Africa (6.3 percent). MENA had the highest growth rate of illicit capital in real terms (26.3 percent per annum on average), followed by Africa (23.8 percent), Asia (7.8 percent), Europe (3.6 percent), and Western Hemisphere (2.7 percent). The rapid growth of outflows from the MENA region was due mainly to the increase in crude oil prices, which drove the region’s current account surplus. It seems that rising oil prices provide more incentive for unrecorded flows. The finding is consistent with Almounsor (2005) who also found a significant positive link between illicit outflows and crude oil prices. Trade misinvoicing continued to be the preferred method of transferring illicit capital from all regions except the MENA region where it only accounted for 37 percent of total outflows over the

j

Global Financial Integrity

decade ending 2010. At one extreme, Asia preferred trade misinvoicing over balance of payments leakages by 94 percent to 6 percent. Trade misinvoicing was also the dominant channel of illicit outflows from the Western Hemisphere (84 percent), Africa (65 percent), and developing Europe (53 percent). According to the HMN+GER method, the ten countries with the largest outflows of illicit capital (in declining order of magnitude) were China, Mexico, Malaysia, Saudi Arabia, the Russian Federation, the Philippines, Nigeria, India, Indonesia, and the United Arab Emirates. Total outflows from China over the decade ending 2010 (US$2,742 billion) exceeded total cumulative outflows from all other nine countries on the list (US$1,728 billion). The new rankings imply that illicit flows impact more people more adversely than what the previous IFF reports indicated. This is because the CED+GER rankings included Kuwait, Venezuela, Qatar, and Poland among the top ten countries with the largest outflows. However, these countries have relatively much higher income and fewer people living on less than US$2 a day, compared to the Philippines, Nigeria, India, and Indonesia which are ranked among the top ten countries under the HMN+GER methodology. Hence, the revised rankings do a much better job of reflecting the adverse impact of illicit flows on poverty compared to the CED+GER method. Finally, we explored the significant statistical issues related to the recording of sovereign wealth funds (SWFs) in the balance of payments and how incomplete or incorrect recording of SWFrelated transactions can lead to errors in estimating illicit flows (due to errors in recording balance of payments variables). If, for instance, there is a drawdown of reserve assets to invest in SWFs and the drawdown is fully recorded, while an SWF-related drawdown to pay off external debt is not recorded then the increased use of funds is not offset by a decline in external debt which would be reflected in an increase in unrecorded capital outflow. Had the subsequent debt repayment been correctly recorded, there would have been no change in unrecorded outflows. Errors could also be introduced in the appropriate recording of reserves due to SWF-related deposits. We conclude that the criteria as to whether specific SWF funds are to be considered part of reserve assets should not be based on mechanical rules but should be based on judgments regarding encumbrance, control, and ease of availability. We looked at the net errors and omissions (NEO) in the balance of payments for a group of ten countries with the largest SWFs. While NEOs are driven by many factors, the purpose was to see whether there is a simple casual link between SWFs and NEOs given the statistical capacity of the SWF country. Normally we would expect countries with strong statistical systems to do a better job of capturing SWF transactions. In general, we found that there is little correlation between the balance of payments of certain countries with large SWFs and the relative strength or weakness of their statistical systems. This led us to believe that SWF transactions do not seem to adversely impact the NEO, although there are a few notable exceptions. The finding that the NEO in the balance of payments data reported by United Arab Emirates, Saudi Arabia, and Qatar to the IMF

Illicit Financial Flows from Developing Countries: 2001-2010

k

are relatively high imply that estimates of illicit flows from these countries must be interpreted with caution due to the risk of significant measurement errors.

l

Global Financial Integrity

I. Introduction
1. Studies at Global Financial Integrity (GFI) on illicit financial flows from developing countries have been based on the World Bank Residual method (using the change in external debt or CED version) adjusted for trade misinvoicing. Economists such as Claessens and Naudé (1993), Cumby and Levich (1989), Epstein (2005), Gunter (2004), Ndikumana and Boyce (2008), Schneider (1997), and others have used this method for many years to estimate the volume of capital flight from developing countries and entire regions. The methodology used in GFI studies has been consistent with this overall approach, except for the fact that the “traditional” approach netted out flows in both directions, while GFI’s methodology is based on gross outflows. In this report, we revisit our methodology, reaffirming the “gross outflow” approach and fine-tuning our balance of payments estimates to provide the reader with alternative estimates of illicit financial flows. 2. The need to broaden the methodology was based on two reasons. First, we looked more closely at the rationale for preferring the gross outflow approach in contrast to the traditional net approach. Some economists, such as Fuest and Riedel (2012) and Nitsch (2012), imply that our gross approach may significantly overstate the problem of capital flight.2 However, the rationale for netting capital flows rests on the premise that net inflows of legitimate capital (i.e. reversal of capital flight) represent a benefit to a country. Legitimate inflows need to offset the original loss of capital through other channels either within the same year or across previous years in order to arrive at a net cumulative position over a given period. However, if we are concerned with estimating illicit financial flows or illegal capital flight, the netting out procedure makes little sense. This is because there is no such concept as net crime— flows in both directions are illicit. Hence, illicit inflows which cannot be used productively and are much more likely to end up in the underground economy provide little or no benefit to governments. The rationale of netting flows is reasonable in analyses of legal or “normal” capital flight. We will show that the method traditionally used by economists may well capture both “normal,” or legal, and “abnormal,” or illegal, capital flight. The gross versus net issue is therefore linked to the nature of capital (i.e. whether it is licit or illicit) which required us to examine, more closely, the types of capital included in the traditional versus GFI methodologies. 3. Second, during the course of our study on illicit flows in connection with the report Illicit Financial Flows from Developing Countries Over the Decade Ending 2009 (henceforth the 2011 IFF Report), we noticed a sharp decline in total outflows of illicit capital from developing countries and regions in 2009. However, the 2011 IFF Report found no evidence that major
2

See, for example, Tax Evasion and Tax Avoidance: The Role of International Profit Shifting, Clemens Fuest and Nadine Riedel and Trade Mispricing and Illicit Flows, Volker Nitsch, in Draining Development? Controlling Flows of Illicit Funds from Developing Countries, edited by Peter Reuter, The World Bank, 2012, Washington DC.

Illicit Financial Flows from Developing Countries: 2001-2010

1

developing countries adopted macroeconomic, structural, or governance-related policy measures which could account for this decline. We attributed the sharp fall in illicit flows to the slowdown in recorded source of funds (such as new loans and foreign direct investment) relative to use of funds. This can also be thought of as an increase in the latter relative to the former. Hence, the need to explain the fall in illicit outflows as a result of the global economic crisis became apparent. The question was if illicit flows reacted so strongly to an economic crisis, what is the response of licit or “normal” capital flight? 4. This report is organized as follows. Section II discusses the rationale for adding a second methodology to focus more sharply on illicit flows and minimize the risk of including legitimate capital flows. We will compare estimates of illicit flows using the new approach against the previous method based on change in external debt (CED) adjusted for trade misinvoicing based on the gross excluding reversals (GER) method. To maintain a sharp focus, section III presents our analysis of the trends in illicit outflows using the new non-normalized methodology from developing countries and regions over the period 2001-2010. Section IV discusses the impact of sovereign wealth funds (SWFs) on the reliability of estimates of illicit flows from developing countries that maintain large SWFs. The final section will draw the main conclusions of this study.

2

Global Financial Integrity

II. Coverage of Flows in the World Bank Residual Method
5. The World Bank Residual method estimates the gap between recorded source of funds and use of funds. The equation is quite straightforward. If a country’s source of funds total US$100 million and its use of funds only amounts to US$75 million, then the Residual method indicates that US$25 million in unrecorded capital must have leaked out of the balance of payments. The approach we adopted thus far in our studies assumes that if flows are unrecorded then they must be illicit, because there is no logical reason why legitimate capital transactions should go unrecorded. 6. In economic literature, the World Bank Residual measure is typically used in isolation without consideration of the balance of payments identity from which it is derived, as shown by Claessens and Naudé (1993). The conclusion that the gap between recorded flows is unrecorded (and therefore illicit) follows from this isolation. However, full balance of payments accounting reveals that the gap between the source of funds and use of funds may include some licit as well as illicit flows. The following analysis shows why licit flows may be included. 7. As Claessens and Naudé (1993) demonstrate, the equation for the World Bank Residual method can be derived directly from the balance of payments identity. Using their nomenclature, let A be the current account balance, B represent net equity flows (including net foreign direct investment and portfolio investment), C the other short-term capital of other sectors, D the portfolio investments involving other bonds, E the change in deposit-moneybanks’ foreign assets, F the change in reserves of the central bank, G the net errors and omissions (NEO), and H the change in external debt. Then, equation (1) demonstrates the balance of payments identity: A + B + C+ D + E + F + G + H = 0 Or, C + D + E + G = - (A + B + F + H) (1) (2)

Equation (2) implies that recorded (and therefore legal) private capital flows (C + D + E) plus net errors and omissions (G) must equal the negative of the sum of the current account balance (A), net equity flows (B), change in reserves (F), and the change in external debt (H). The right hand side of the above equation is the World Bank Residual equation.

Illicit Financial Flows from Developing Countries: 2001-2010

3

Table A. The United States and China: Balance of Payments Components, 1991 1/
(in U.S. dollars)
Balance of Payments Components Scale A. Current Account Capital Account (B+C+D+E+F+H) B. Net Equity Flows FDI abroad FDI in the country Portfolio invest. (corporate equities) C. Other short-term capital Other sectors Resident official sector D. Portfolio investment Other bonds E. Change in DMB foreign assets Short-term capital Long-term capital F. Reserves H. Other long-term capital Resident official sector Other sectors G. Net errors and omissions Memoranda Items Balance of payments check 2/ Private sector flows +NEO (C+D+E+G) World Bank Residual (A+B+F+H) 0.00 28.12 -28.12 0.00 -6,151.00 6,151.00 31.64 -8.80 -15.50 6.70 5.76 6.45 6.45 0.00 -1.12 -8,143.00 1,655.00 558.00 1,097.00 -14,537.00 2,885.00 2,236.00 649.00 533.00 United States (US$ billions) -3.69 4.81 -36.64 -27.15 11.50 -20.99 6.40 -6.17 12.57 China (US$ millions) 13,765.00 -14,298.00 4,038.00 -913.00 4,366.00 585.00 -196.00 -196.00 0.00

1/ Corresponds to the format and figures published in the Balance of Payments Yearbook, Part 1, 1992, IMF. The position of items H and G are not in the order that they appear in Recent Estimates of Capital Flight, Stijn Claessens and David Naude (1993). As item H is classified under the capital account, the order was switched with item G. 2/ The balance of payments check consists of the fact that the current account plus the capital account and the net errors and omissions must sum to zero. 3/ According to the BOP identity, as pointed out by Claessens and Naude, the BOP equation implies that C+D+E+G = -(A+B+F+H). The last two line items verify this for the United States and China.

8.

One could estimate capital flight using either the left- or right-hand side of the above equation— the result will be equivalent. Table A demonstrates that the World Bank Residual estimates of capital flight can be derived using the 1991 balance of payments data reported by China and the United States to the IMF. Because the classification of the balance of payments items is consistent with those used by Claessens and Naudé in accordance with the Balance of Payments Manual in effect at the time, we had to use the data published in the 1992 Balance of Payments Yearbook. The reported data show that, in fact, the right hand side of the equation (private sector capital flows plus the NEO) is equal to the World Bank Residual estimate based on change in external debt (with sign reversed). This implies that the illicit component of the CED method (i.e., the NEO) is simply the difference between the CED estimates and private sector licit flows. Hence, more conservative estimates of illicit flows are based on the illicit component of the CED plus trade misinvoicing based on the gross excluding reversals (GER) method.

4

Global Financial Integrity

9.

According to the above formulation, a narrower version of illicit flows can be derived simply by adding the NEO to the GER estimates of trade misinvoicing. The NEO has been traditionally used by economists as the Hot Money Narrow (HMN) method. As we pointed out in our 2008 study, there are three versions of the Hot Money method, starting with the Narrow version and progressively including more types of private sector flows.3 The broader Hot Money measures yield larger estimates of capital flight. However, the broader Hot Money method suffers from the same drawback as the World Bank Residual method: both methodologies produce estimates that could include both licit and illicit flows. This goes against the purpose of GFI studies, which is to look solely at illicit flows.

Chart 1. Volume of Illicit Financial Flows in Nominal Terms from All Developing Countries 2001-2010 1/ !"#$%&'(&)*+,-.&*/&0++121%&314#421#+&3+*56&14&7*-14#+&8.$-6&/$*-&9++& (in millions of U.S. dollars)

:.;.+*??'@>?'?&'A&
B14&-1++1*46&*/&C(D(&E*++#$6F&

($&!!$!!!" ($%!!$!!!" ($#!!$!!!" ($!!!$!!!" '!!$!!!" &!!$!!!" %!!$!!!" #!!$!!!" !" #!!(" #!!#" #!!)"

#!!%"

#!!*"

#!!&"

#!!+"

#!!'"

#!!,"

#!(!"

-./0/.12345678"9::;"?@A>BC" -.12345678"9::;"?@A>BC"

B7D5;78"9::;"B"-.//.12C" B7D5;78"9::;"B"-.12C"

10.

Licit capital flight can be simply estimated as the difference between the World Bank Residual estimates and the HMN. Chart 1 plots normalized and non-normalized estimates based on the CED+GER and the HMN+GER methods. The green and the purple lines represent non-normalized CED+GER and HMN+GER estimates which are also presented in Table B.4 We can see that in the early 2000s, the two lines were quite close. The gap between the two lines, representing licit capital flight, increased as capital controls were eased in many large emerging markets such as in Brazil, China, India, Mexico, and Russia. The gap is widest in 2008, at the onset of the global economic crisis. Then in the following year, as outflows of legal capital flight dropped much more sharply than did illicit outflows, the gap
3

4

Reference, Illicit Financial Flows from Developing Countries: 2002-2006, Dev Kar and Devon Cartwright-Smith, Global Financial Integrity, December 2008, pp. 4-5. See Kar, Dev, and Sarah Freitas, Illicit Financial Flows from Developing Countries Over the Decade Ending 2009, Global Financial Integrity, 2011 for details on the process of normalization.

Illicit Financial Flows from Developing Countries: 2001-2010

5

narrowed followed by some widening in 2010 as licit outflows picked up along with the pace of economic activity. Outflows of licit capital fell in 2009 because economic agents retained more capital domestically due to the financial squeeze resulting from the crisis and the fact that major capital markets in the United States and Europe were in turmoil.

Table B. Four Estimates of Capital Flight, All Developing Countries, 2001-2010
(in billions of U.S. dollars)
Non-normalized IFFs (CED+GER) 1/ Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries 2001 24.1 225.5 72.3 55.8 99.5 477.1 2002 25.0 216.1 64.4 37.8 98.7 441.9 2003 33.2 273.4 104.1 89.6 116.7 617.0 2004 41.6 345.6 128.7 133.0 120.0 768.9 2005 37.4 425.0 104.9 157.7 128.7 2006 53.7 497.6 141.0 218.3 135.7 2007 85.3 535.6 253.0 283.4 189.6 2008 101.9 608.9 329.2 305.9 213.8 2009 76.3 423.1 111.7 186.6 138.5 2010 86.1 584.0 126.3 178.4 163.1 Cumulative 564.6 4,134.9 1,435.6 1,646.4 1,404.2 9,185.7 Average 56.5 413.5 143.6 164.6 140.4 918.6 Logarithmic 18.08 12.03 10.83 20.21 7.07 12.61

853.7 1,046.2 1,346.9 1,559.8

936.1 1,138.0

Normalized IFFs (CED+GER) 1/ Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries 2001 9.6 221.4 67.3 49.9 81.1 429.3 2002 16.0 192.5 56.0 32.6 93.9 391.0 2003 28.1 253.7 92.3 84.7 108.7 567.5 2004 33.3 331.5 109.5 128.6 97.3 700.3 2005 31.7 395.0 91.0 151.5 110.8 780.0 2006 48.3 383.5 134.4 210.1 125.1 2007 77.5 424.4 242.2 218.1 154.8 2008 93.9 513.6 314.5 288.8 149.9 2009 72.3 388.8 80.4 175.0 127.1 843.6 2010 63.0 490.6 43.3 158.0 137.0 891.9 Cumulative 473.6 3,595.0 1,230.9 1,497.3 1,185.7 7,982.5 Average 47.4 359.5 123.1 149.7 118.6 798.3 Logarithmic 24.73 10.40 4.63 20.07 6.22 11.45

901.3 1,117.1 1,360.7

Revised IFFs (HMN+GER Non-normalized) 2/ Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries 2001 11.1 177.8 41.1 33.2 67.2 330.5 2002 8.8 193.0 23.5 8.0 66.4 299.8 2003 9.9 244.2 32.4 7.3 65.3 359.0 2004 17.9 332.9 39.4 22.1 77.6 490.0 2005 35.5 387.9 29.4 63.7 98.5 615.1 2006 46.5 384.6 19.4 55.6 82.7 588.7 2007 59.1 418.8 44.2 41.1 106.7 669.9 2008 74.4 478.3 56.7 140.7 121.1 871.3 2009 70.5 415.5 46.5 141.3 102.1 776.0 2010 51.1 535.7 73.7 89.2 109.3 858.8 Cumulative 384.8 3,568.8 406.3 602.3 896.9 5,859.2 Average 38.5 356.9 40.6 60.2 89.7 585.9 Logarithmic 29.15 12.43 8.03 31.74 7.06 13.28

Revised IFFs (HMN+GER Normalized) 2/ Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries 2001 7.5 175.2 39.2 28.1 62.3 312.3 2002 7.1 189.9 18.9 6.7 62.0 284.5 2003 9.3 241.9 25.7 5.5 62.4 344.7 2004 11.7 328.8 23.5 20.4 71.0 455.5 2005 30.0 379.1 27.1 61.1 91.6 588.9 2006 44.0 368.0 18.5 53.0 78.8 562.3 2007 51.9 409.2 36.6 38.5 95.9 632.0 2008 70.1 461.9 49.7 129.5 97.7 809.0 2009 69.2 398.2 28.6 134.6 91.6 722.1 2010 41.5 523.1 30.2 82.2 106.1 783.2 Cumulative 342.3 3,475.4 297.8 559.6 819.4 5,494.6 Average 34.2 347.5 29.8 56.0 81.9 549.5 Logarithmic 32.38 12.13 2.94 34.01 6.59 12.88

1/ Estimates include both licit as well as illicit financial flows. Estimates updated from 2011 IFF Update. 2/ Estimates pertain to illicit financial flows.

6

Global Financial Integrity

11.

Charts 2a-2e show the gap between two estimates representing possible legal capital flight from the various regions of the developing world. We observe that in the case of developing Europe, the MENA region, and Western Hemisphere (charts 2c -2e)—the gap tends to widen over time, reaching a peak in 2008. In all three regions, licit outflows plunged in 2009 due to the effects of the crisis in domestic and foreign capital markets noted above. In contrast, Chart 2b indicates that the gap in Asia, which was almost nonexistent in the early 2000s,

Chart 2. Illicit Financial Flows by Region, 2001-2010 1/ (in millions of U.S. dollars)
!"#$%&'#(&)*+,-.&*/&0++121%&314#421#+&3+*56&14& in Chart 2a. Volume of Illicit Financial Flows 7*-14#+&8.$-6&/$*-&9/$12#:&';;=& Terms from Western Hemisphere, 2001-2010
@14&.1,,1+46&+/&A)B)&C+,,#$6D& (in millions of U.S. dollars) &#!$!!!" &!!$!!!" %#!$!!!" %!!$!!!" #!$!!!" !" &!!%" &!!&" &!!'" &!!(" &!!#" &!!)" &!!*" &!!+" &!!," &!%!" -./01.2"345654789:;" ?@301.2"345654789:;"

1/ Estimates of GER in the CED+GER and the HMN+GER lines are non-normalized. All tables and charts in section III of this report and Tables 1, 2, and 14 of the Appendix use non-normalized estimates, as discussed in the following section on normalization.

Illicit Financial Flows from Developing Countries: 2001-2010

7

began to widen in 2005 and reached a peak in 2008 at the onset of the crisis, closing almost completely in 2009 as both licit and illicit outflows fell in tandem. In the last year, both types of flows recovered along with increase in global economic activity. China, Malaysia, the Philippines, and India led Asia as the major drivers of licit and illicit flows. The relaxation of capital controls by these countries over the years perhaps encouraged more legal or “normal” capital flight accounting for the widening regional gap between the CED+GER and HMN+GER measures through 2008. Further research is needed in order to analyze the factors driving licit capital flows from the various regions. For instance, legal capital flight seems to be driving the widening gap between the two measures in the case of developing Europe. In fact, the Central Bank of Russia as well as the IMF corroborates the existence of large capital flight from the country, which are predominantly recorded private sector outflows.

a. Normalization through use of filters
12. As Chart 1 showed, the HMN+GER approach to estimating illicit flows is more conservative than the CED+GER approach, which may include some legitimate private sector capital flows.

Chart 3. Normalized vs. Non-normalized GER, 2001-2010 (in millions of U.S. dollars) !"#$%&'(&)*$+#,-./0&12(&)*343*$+#,-./0&5678&9::;49:;:&
(?(&0*,,#$2@& *!!$!!!" )!!$!!!" (!!$!!!" '!!$!!!" &!!$!!!" %!!$!!!" #!!$!!!" !"

%!!#"

%!!%"

%!!&"

%!!'"

%!!("

%!!)"

%!!*"

%!!+"

%!!,"

%!#!"

-./0/.12345678"9:;"

-.12345678"9:;"

Moreover, Chart 3 shows that the conservative or normalized GER estimates are so close to the non-normalized estimates that not much would be gained by generating a range. Therefore, non-normalized GER estimates will be used throughout the remainder of the report in generating both the HMN+GER and CED+GER estimates.

8

Global Financial Integrity

III. Trends in Illicit Financial Flows from Developing Countries and Regions
13. Table B shows estimates of illicit financial flows from developing countries based on the HMN+GER and CED+GER methods. Using non-normalized estimates, the data indicate that on average, developing countries lost between US$585.9 billion to US$918.6 billion per annum over the period 2001-2010. In 2010, they lost a minimum of US$858.8 billion and as much as US$1,138.0 billion. 14. We also observe that the HMN+GER measure has increased at a faster pace than the CED+GER measure (13.3 percent vs. 12.6 percent). The relatively faster rate of increase in purely illicit outflows is worrisome given that there is no reason for human statistical errors (included in both the CED and the HMN measures) to have increased in a systematic manner throughout the decade. In fact, with the increasing availability and adoption of new technologies, and the provision of technical assistance by the IMF to developing countries in order to build their statistical capacities, one would expect the proportion of statistical errors to decline over the past decade.5 The implication is that the significant increase in illicit flows is likely to result from a worsening of governance-related drivers. 15. It is somewhat premature to make a definitive judgment as to which method provides a more accurate method for estimating illicit flows. While the HMN+GER method is the most conservative measure, it may exclude some illicit transactions (such as round-tripped FDI) which show up as recorded private sector flows. We invite our readers to provide comments on the two alternative methodologies and the reasons why one of them should be preferred over the other. 16. For the sake of brevity and sharper focus on illicit flows, we shall henceforth confine the discussion of trends, shares, and country rankings in terms of the HMN+GER estimates. Going by that measure, the increase in illicit flows of over US$500 billion since 2001 implies a nominal growth rate of 13.3 percent per annum (Table C). In inflation-adjusted or real terms, illicit flows grew by 8.6 percent per annum on average (Table D), which exceeded their average rate of economic growth (6.3 percent per annum). About 20.0 percent of total outflows were channeled through balance of payments leakages while the bulk (approximately 80.0 percent) was transferred through the deliberate misinvoicing of external trade. Over the decade, the shares of outflows from trade misinvoicing and the balance of payments have fluctuated. In 2004, trade misinvoicing reached a peak of 86.1 percent of total IFFs, dropping to a low of 62.3 percent in 2009. However, in 2010 outflows through trade misinvoicing picked up again to reach 64.2 percent of the total (Table C).
5

There is no evidence that net errors and omissions have a clear increasing pattern to them; reference IMF Committee on Balance of Payments Statistics, Annual Report 2011, IMF, Table 1, pp. 17-18.

Illicit Financial Flows from Developing Countries: 2001-2010

9

Table C. Illicit Financial Flows by Region in Nominal Terms 1/
(millions of U.S. dollars, unless otherwise indicated)
HMN (Hot Money Narrow, Balance of Payments component) Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries 2001 2,597.64 11,425.56 12,932.53 12,844.44 13,725.55 53,525.72 2002 3,221.96 6,430.32 11,479.23 4,248.36 13,030.16 38,410.03 2003 4,147.09 6,696.46 15,647.52 4,243.83 10,447.70 41,182.60 2004 1,705.94 6,656.05 10,397.39 2,841.25 12,129.15 33,729.78 2005 20,296.96 15,306.33 16,150.18 51,940.31 22,021.10 125,714.88 2006 17,871.45 16,869.06 9,754.15 44,103.30 7,101.79 95,699.75 2007 17,825.18 17,679.33 27,928.90 31,300.12 8,805.86 103,539.39

GER (Gross Excluding Reversals, Trade Mispricing component) Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries Total HMN + GER Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries HMN Percent of Total GER Percent of Total 2001 11,098.22 177,834.24 41,129.31 33,229.92 67,185.39 330,477.08 16.2 78.3 2002 8,837.76 192,980.03 23,509.90 8,044.87 66,441.95 299,814.50 12.8 82.1 2003 9,899.14 244,157.77 32,359.23 7,269.11 65,289.21 358,974.46 11.5 84.6 2004 17,929.52 332,933.72 39,424.00 22,119.82 77,592.67 489,999.73 6.9 86.1 2005 35,519.97 387,942.18 29,448.18 63,715.70 98,480.53 615,106.56 20.4 75.3 2006 46,484.80 384,623.66 19,360.65 55,570.33 82,684.82 588,724.27 16.3 79.2 2007 59,065.06 418,841.21 44,173.92 41,141.34 106,727.73 669,949.26 15.5 78.9 2001 8,500.58 166,408.68 28,196.78 20,385.49 53,459.84 258,803.00 2002 5,615.80 186,549.71 12,030.67 3,796.51 53,411.79 246,137.54 2003 5,752.05 237,461.31 16,711.71 3,025.28 54,841.52 303,518.10 2004 16,223.58 326,277.67 29,026.61 19,278.57 65,463.52 421,816.06 2005 15,223.01 372,635.85 13,298.00 11,775.39 76,459.43 463,176.89 2006 28,613.35 367,754.60 9,606.50 11,467.04 75,583.04 466,563.29 2007 41,239.88 401,161.88 16,245.02 9,841.22 97,921.87 528,479.34

Source: Staff estimates, Global Financial Integrity, based on official balance of payments and trade data reported to the IMF by member countries. 1/ Based on cumulative outflows from the region in total outflows from developing countries over the period 2001-2010

10

Global Financial Integrity

2008 23,488.48 22,225.26 43,109.26 114,750.53 1,804.47 205,378.00

2009 30,941.93 66,647.09 18,469.56 100,655.75 22,249.87 238,964.21

2010 20,755.05 91,726.77 23,145.48 67,021.29 28,820.58 231,469.16

Total 142,851.68 261,662.22 189,014.21 433,949.18 140,136.22 1,167,613.50

Share of Region in Total (in %) 1/ 12.23 22.41 16.19 37.17 12.00 100.00

Percent Change 2009-2010 -49.08 27.34 20.20 -50.18 22.80 -3.24

Logarithmic Growth 2001-2010 35.50 30.67 10.25 44.33 -0.27 25.21

2008 50,915.87 456,103.67 13,624.66 25,978.67 119,301.74 603,580.03

2009 39,584.24 348,866.43 28,039.81 40,684.36 79,868.07 483,156.50

2010 30,301.94 443,945.48 50,525.16 22,167.25 80,433.72 551,710.80

Totals 241,970.30 3,307,165.27 217,304.91 168,399.77 756,744.54 4,691,584.79

Share of Region in Total (in %) 1/ 5.16 70.49 4.63 3.59 16.13 100

Percent Change 2009-2010 -30.63 21.42 44.50 -83.53 0.70 12.43

Logarithmic Growth 2001-2010 27.00 10.91 5.02 17.11 7.27 9.95

2008 74,404.35 478,328.93 56,733.92 140,729.20 121,106.21 871,302.61 23.6 69.3

2009 70,526.16 415,513.52 46,509.37 141,340.11 102,117.94 776,007.11 30.8 62.3

2010 51,056.99 535,672.25 73,670.64 89,188.53 109,254.30 858,842.70 27.0 64.2

Totals 384,821.97 3,568,827.49 406,319.12 602,348.95 896,880.76 5,859,198.29 19.9 80.1

Share of Region in Total (in %) 1/ 6.57 60.91 6.93 10.28 15.31 100.00

Percent Change 2009-2010 -38.13 22.43 36.87 -58.47 6.53 9.65

Logarithmic Growth 2001-2010 29.15 12.43 8.03 31.74 7.06 13.28 18.1 76.0

Ave. HMN % (2001-2010) Ave. GER % (2001-2010)

Illicit Financial Flows from Developing Countries: 2001-2010

11

Table D. Illicit Financial Flows by Region in Real Terms 1/
(millions of 2005 U.S. dollars, unless otherwise indicated)
HMN (Hot Money Narrow, Balance of Payments component) Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries 2001 3,046.90 13,401.60 15,169.20 15,065.87 16,099.37 62,782.94 2002 3,867.83 7,719.32 13,780.33 5,099.98 15,642.15 46,109.61 2003 4,725.81 7,630.93 17,831.09 4,836.05 11,905.65 46,929.53 2004 1,830.78 7,143.15 11,158.29 3,049.18 13,016.79 36,198.18 2005 20,296.96 15,306.33 16,150.18 51,940.31 22,021.10 125,714.88 2006 17,074.15 16,116.48 9,318.99 42,135.72 6,784.96 91,430.29 2007 16,249.92 16,116.95 25,460.74 28,534.03 8,027.66 94,389.30

GER (Gross Excluding Reversals, Trade Mispricing component) Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries Total HMN + GER Region/Year Africa Asia Developing Europe MENA Western Hemisphere All Developing Countries HMN Percent of Total GER Percent of Total 2001 13,017.64 208,590.52 48,242.59 38,977.01 78,805.05 387,632.81 16.2 83.8 2002 10,609.36 231,664.35 28,222.64 9,657.52 79,760.74 359,914.60 12.8 87.2 2003 11,280.54 278,229.38 36,874.88 8,283.50 74,400.16 409,068.46 11.5 88.5 2004 19,241.63 357,298.41 42,309.12 23,738.59 83,271.04 525,858.80 6.9 93.1 2005 35,519.97 387,942.18 29,448.18 63,715.70 98,480.53 615,106.56 20.4 79.6 2006 44,410.97 367,464.42 18,496.92 53,091.17 78,996.00 562,459.49 16.3 83.7 2007 53,845.31 381,826.96 40,270.14 37,505.56 97,295.88 610,743.84 15.5 84.5 2001 9,970.75 195,188.91 33,073.39 23,911.14 62,705.67 324,849.86 2002 6,741.53 223,945.02 14,442.30 4,557.54 64,118.59 313,804.99 2003 6,554.74 270,598.45 19,043.78 3,447.45 62,494.51 362,138.93 2004 17,410.85 350,155.26 31,150.83 20,689.41 70,254.26 489,660.62 2005 15,223.01 372,635.85 13,298.00 11,775.39 76,459.43 489,391.68 2006 27,336.82 351,347.94 9,177.93 10,955.46 72,211.05 471,029.19 2007 37,595.39 365,710.00 14,809.40 8,971.52 89,268.22 516,354.53

Source: Staff estimates, Global Financial Integrity, based on official balance of payments and trade data reported to the IMF by member countries. 1/ Current dollar estimates are deflated by the U.S. Producer Price Index base 2005 (from IMF IFS online database). 2/ Based on cumulative outflows from the region in total outflows from developing countries over the period 2001-2010.

12

Global Financial Integrity

2008 19,500.26 18,451.53 35,789.54 95,266.49 1,498.08 170,505.90

2009 28,166.71 60,669.44 16,813.01 91,627.83 20,254.26 217,531.25

2010 17,684.04 78,154.46 19,720.77 57,104.52 24,556.16 197,219.94

Totals 132,443.35 240,710.20 181,192.13 394,659.98 139,806.16 1,088,811.83

Share of Region in Total (in %) 2/ 12.16 22.11 16.64 36.25 12.84 100.00

Percent Change 2009-2010 -59.28 22.37 14.74 -60.46 17.52 -10.30

Logarithmic Growth 2001-2010 29.92 25.30 5.71 38.39 -4.38 20.06

2008 42,270.63 378,659.67 11,311.26 21,567.63 99,044.93 552,854.12

2009 36,033.88 317,576.23 25,524.89 37,035.34 72,704.62 488,874.95

2010 25,818.43 378,258.85 43,049.40 18,887.36 68,532.67 534,546.71

Totals 224,956.03 3,204,076.19 214,881.19 161,798.24 737,793.96 4,543,505.60

Share of Region in Total (in %) 2/ 4.95 70.52 4.73 3.56 16.24 100.00

Percent Change 2009-2010 -39.57 16.04 40.71 -96.09 -6.09 8.54

Logarithmic Growth 2001-2010 21.78 6.35 0.70 12.29 2.85 6.13

2008 61,770.88 397,111.19 47,100.80 116,834.12 100,543.02 723,360.02 23.6 76.4

2009 64,200.60 378,245.67 42,337.90 128,663.17 92,958.88 706,406.21 30.8 69.2

2010 43,502.47 456,413.32 62,770.17 75,991.87 93,088.82 731,766.65 27.0 73.0

Totals 357,399.38 3,444,786.39 396,073.32 556,458.22 877,600.12 5,632,317.42 19.3 80.7

Share of Region in Total (in %) 2/ 6.35 61.16 7.03 9.88 15.58 100.00

Percent Change 2009-2010 -47.58 17.13 32.55 -69.31 0.14 3.47

Logarithmic Growth 2001-2010 23.83 7.81 3.59 26.32 2.65 8.62 18.1 81.9

Ave. HMN % (2001-2010) Ave. GER % (2001-2010)

Illicit Financial Flows from Developing Countries: 2001-2010

13

17.

Adjusting the outflow estimates for inflation or for private sector flows marginally alters the regional shares found previously. Chart 4 shows the shares of cumulative illicit outflows from the various regions of the developing world over the decade ending 2010 under the HMN+GER method. We note that Asia still remains the main driver of illicit outflows from developing countries regardless of the method of estimation. On a cumulative basis, the region accounted for 61.2 percent of total outflows, mostly due to massive outflows from mainland China and India. The Western Hemisphere follows at 15.6 percent, with the Middle East and North Africa (MENA) at 9.9 percent. The MENA region has a smaller share than the Western Hemisphere in this study compared to the 2011 Update due to the fact that Algeria, Iran, and Iraq have not fully reported the balance of payments data necessary for the HMN+GER method. Developing Europe follows MENA in share size, making up 7.0 percent of illicit flows, with the balance flowing out of Africa (6.3 percent). The relative increase in outflows from Africa can be mainly attributed to a larger number of countries for which we were able to collect basic data on the balance of payments and bilateral trade flows. The relative shares are subject to the caveat that restricted data availability on important countries of certain regions may understate the regional shares and overstate others.

Chart 4. Illicit Flows in Real Terms 2001-2010; Regional Shares in Developing World Total 1/ ,-./0%1"%2334540%63789%4:%;