Estimates and Causes of Capital Flight from Central and East European Countries
Josef C. Brada
W. P. Carey School of Business, Arizona State University
Tempe, AZ 85287-3806 USA
Ali M. Kutan
Southern Illinois University at Edwardsville
Edwardsville, IL 62026-1102 USA
Institute of Public Finance, Zagreb, Croatia
We estimate capital flight from twelve transition economies of Central and Eastern Europe (CEE) for the period 1995-2005 using the residual method. Capital flight from some of these transition economies, when adjusted for country size, is comparable to the more highly publicized capital outflows from Russia despite East Europe’s seemingly better transition and reform performance and greater political stability. We find that capital flight from CEE is mainly an economic phenomenon, driven by differences in interest rates and investors’ expectations about future macroeconomic conditions in their countries. Our empirical results are thus consistent with the mainstream explanations of capital flight and they mirror results obtained for other countries and time periods, suggesting that transition-related phenomena are not important factors in capital flight from CEE.
JEL Classification Numbers: E26, F31, F32, P33, P37
Key words: capital flight, external sector liberalization, money laundering, transition economies
A great deal of attention has been paid to measuring and explaining the causes of capital flight from Russia. For example, Abalkin and Whalley (1999) estimated capital flight from Russia at $56-70 billion for the period 1992-3 and at an annual rate of $17 billion from 1994 to 1997. Buiter and Szegvari (2002) offer somewhat higher estimates, as do Sarafanov (1995) and Loukine (1998) who show accelerating levels of capital flight, up to $50 billion per year, in the mid-2000 period. Tikhomirov (1997) estimated that the mispricing of Russian trade in the years 1990-95 alone resulted in capital flight that was six-fold the official Russian government estimates of $35-40 billion. Part of the reason for this widespread interest in the Russian experience with capital flight is the sheer magnitude of these flows and their implications for the Russian economy in terms of loss of government revenue and of foregone domestic investment. In part, the interest has also been driven by the perception that much of the capital flight is related to the process of economic transition in Russia. Many observers see it as being caused by Russian oligarchs and politicians who form an interlocking kleptocracy. Certainly Russia’s vast earnings from energy and mineral exports, the concentration of wealth and the uncertainty of the property rights of both Russian oligarchs and foreign investors provide incentives for moving assets offshore beyond the grasp of the Russian state.
In contrast, comparatively little effort beyond the early work of Sheets (1995) has gone into estimating capital flight from the transition economies of Central and Eastern Europe (CEE), including the Baltic States, which, for the purposes of this paper, we simply call East Europe. Estimates of recent capital flight from CEE countries based on a consistent methodology and data sources are not available, and the estimates that are available cover only some countries, often only for a few years, and use a variety of methodologies making cross-country comparisons difficult.
To address this lack of consistent data on capital flight from these transition economies, in this paper we present estimates of capital flight using the so-called World Bank or “residual method” for Albania, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia for the years 1995 to 2005. In order to gain a better understanding of capital flight from these countries, we also model the drivers of capital flight. In this regard the CEE countries are a particularly apt subject because in the course of transition they have significantly liberalized their international trade and financial regimes as well as their domestic economies, to the point where a number of them have become members of the European Union. This extensive and rapid opening up and liberalization provide a valuable opportunity to better study the effects of institutional change and economic development on capital flight than is possible using samples consisting of countries whose institutions and economic performance have not changed so rapidly over time
II. Capital Flight
Capital flight is a rather loose term for unregistered private capital flows (Walter, 1985) that encompasses a broad variety of activities ranging from the legal and economically beneficial to those that are illegal and harmful to the economy. As Buiter and Szegvari (2002) note, some of what is termed capital flight represents a rational reallocation of capital from the home country to other countries in response to more favourable risk-return opportunities abroad and to investors’ desire for portfolio diversification. Such portfolio decisions benefit both the investors undertaking them as well as the home and host countries. At the other end of the spectrum is money laundering, which means undertaking transactions that hide the illegal origin of the funds and convert them into legal income.1 Money laundering involves illegality in two ways. First, the money to be laundered is often earned though illegal activity such as prostitution, drug distribution, bribe taking, etc. Second, the money is then moved abroad, possibly in contravention of capital or currency controls, to hide its criminal origin and to evade payment of taxes in the country where it is earned. Falling somewhere in the middle in terms of legality are transactions that involve income that may be legally earned, but, because the home country has restrictions on capital outflows, investing such capital abroad effectively criminalizes these investments.
Because it takes both legal and illegal forms, the measurement of capital flight is subject to considerable uncertainty even though central banks, treasuries and other government agencies as well as multilateral financial institutions have been increasingly concerned to improve their measures of capital flight and their understanding of what motivates it.2 While there is general agreement that it is the difference between the risks and returns available to investors at home and abroad that leads to capital flight, both theoretical and empirical work on this topic has tended to emphasize either the differences in returns or the differences in risks as being more salient to decisions to move funds off shore.
Those stressing differences in returns tend to take an explicitly portfolio-oriented approach where returns and risks are measured by macroeconomic conditions (Harrigan et al., 2002; Le and Zak, 2006). Such models view interest rate differentials and other measures of expected returns in the home country as important explanatory variables. Cuddington (1987a) also stresses the portfolio approach, building a model with domestic and foreign financial assets and consumer durables to model capital flight. Collier et al. (2004) model capital flight as the result of investors’ desires for portfolio diversification resulting from perceptions of relative returns and risks at home and abroad. They draw on a theoretical model of Sheets (1995), who constructs a demand function for domestic assets as a function of wealth, risk aversion and the differential between domestic and world interest rates. Hermes and Lensink (2001) model macroeconomic policy uncertainty explicitly, and they conclude that uncertainty about macroeconomic variables increases capital flight.
Other researchers tend to view capital flight as being more sensitive to what may be termed idiosyncratic country risk factors, meaning specific aspects of political, institutional and economic arrangements or policies in countries. Uncertainty about future fiscal policies due to changes of governments with different policy objectives is emphasized by Alesina and Tabellini (1989). Dooley and Kletzer (1994) focus on the tax treatment of residents and non-residents as does Bachattarya (1999). In a paper that is particularly suggestive for capital flight from transition economies, Kant (2002) argues that the short-term and long-term uncertainties and risks related to instability and lack of transparency in property rights, excessive taxation, corruption, lack of contract enforcement and the like are the key drivers of capital flight. Likewise, Khan and Ul Haque (1985) and Schineller (1993) emphasize risk of expropriation as one of the main incentives for capital flight. Khan and Ul Haque (1985) were among the first to observe that private capital flight can occur simultaneously with public foreign borrowing, and they explain the phenomenon on the grounds of asymmetric risk of expropriation, which is higher for domestic investors than for foreign ones. Other papers suggest that foreign direct investment inflows can play a role similar to that of borrowing abroad in stimulating capital flight. These and similar papers (Eaton 1987 and Boyce 1992) also show that capital flight can also take forms other than portfolio investment.
In the case of transition economies, both portfolio and country-specific risk factors would seem to be important a priori. Portfolio considerations rest in part on the fact that, prior to the transition, there had been very little capital outflow from these countries due to currency inconvertibility and the state’s almost total control over foreign trade and foreign exchange transactions. As a result, at the start of the transition, domestic agents had virtually no foreign assets, and thus basic portfolio theory would suggest that normal considerations of portfolio diversification would cause significant capital flight even in the face of restrictions on outward investment.3 Moreover, returns to capital were likely quite low during the onset of transition. Inflation was high due to price liberalization, output and profits were falling due to the so-called transition recession, and many firms were in financial distress and facing bankruptcy. Real deposit rates were generally low in the transition economies so as to prop up the banking system, which faced a daunting stock of bad debts. Thus foreign assets likely offered a much more attractive combination of risk and return, encouraging capital flight from transition economies. Consequently, if portfolio diversification is the dominant motive for capital flight from CEE, then variables related to expected returns at home and aboard should play a major role in explaining capital flight.
On the other hand, if idiosyncratic risk factors predominate, then capital flight from the transition economies can be seen as due to the weakness of the state and the resulting high levels of corruption and criminality. Criminal activity such as drug dealing, prostitution, fraud, bribery and various economic crimes generates large amounts of cash, and so do bribes paid to politicians, regulators and business executives. In order to safely use the money generated in these illegal ways, its recipients use the financial system to hold and move this money so as to hide its origins in illegal activity. Through this process of laundering, the criminals make the proceeds of their illegal activity assume the form of legitimate income so that the authorities are unable to identify its criminal origins. While much money laundering takes place in the country where it is earned, laundering too much money domestically is both risky and expensive because large cash transactions arouse the suspicions of the authorities and require an ever increasing number of accomplices.
One way of overcoming the difficulty and risk of laundering large sums of money domestically is to launder it through foreign financial institutions and companies. To do this, the money is deposited into the financial system of the home country, often through a shell corporation. The money is then moved off shore to disguise its origins. Businesses such as restaurants and retail establishments are particularly attractive because their cash deposits are less likely to attract attention. Firms engaged in international trade and financing are also attractive because moving the money to a foreign location, especially one where banking regulation and privacy laws are favourable to money laundering, and then bringing it back to the country of origin further disguises its criminal origins.
A common way of moving money offshore, even in the presence of capital controls, is through under- and over-invoicing. Companies that engage in international trade can disguise the movement of money being laundered by manipulating the prices of the goods and services they buy or sell. Research suggests that the amount of money moved thorough under- and over-invoicing is large. For example, De Boyrie et al. (2005) estimated that over- and under-invoicing accounted for the movement of $1.01 to $4.85 billion per year between Russia and the United States during 1995-1999.
Consequently, if capital flight from the CEE countries largely represents flows of illicitly earned money that is being laundered, then explanatory variables related to differences in expected returns, etc. are likely to have less explanatory power because the money being laundered is directed not to places where returns are the highest but rather to countries that are known as money laundering centers where illicit transactions are sheltered from the view of the authorities.
III. Estimates of Capital Flight
In this paper, we take capital flight to be the net unrecorded private capital outflows from the home country. Because these flows are unrecorded, we estimate them as the difference between the recorded sources and uses of funds for the country. This definition was developed by the World Bank (1985), and it is often referred to as the World Bank or the “residual method”.4 We choose to use the residual method for several reasons. We want to obtain estimates for all the CEE countries, and the residual method lends itself to this effort because it is relatively straightforward to implement. Second, because it is also a widely used measure, international comparisons are easier to make. In doing this we accept two drawbacks of the residual method. The first is that this measure of capital flight is somewhat similar to the errors and omissions component of the balance of payments. Nevertheless, given the magnitude of the estimates of capital flight from CEE produced by this method, it would be hard to argue that errors and omissions of this size could all be caused mainly by recording errors rather than by capital flight. A second problem is that we do not have the currency composition of the external debt of the CEE countries, and, since some of this debt is denominated in US dollars and some in Euros, movements in the $/Euro exchange rate will affect our estimates of capital flight. We deal with this problem by econometric means described below.
The residual method estimates capital flight indirectly, using balance of payments and international asset data. It weighs the country’s sources of funds, as given by the net increase in external debt and the net inflow of foreign investment against the uses of these funds as given by the current account deficit and the change in foreign reserves. If the recorded sources are greater than the recorded uses then there is capital flight from the country. Thus
Capital Flight = ΔED + NFI – CA – ΔR Eq. 1 where ΔED is the change in the stock of gross external debt, NFI is the net foreign investment inflow, CA is the current account deficit and ΔR is the change in the stock of official foreign reserves.5
Two methodological caveats are necessary in the case of our application of the residual method in the case of the East European countries. One of these is how to treat cash holdings of the countries’ residents and the dollarization or euroization of these economies. In the beginning of the transition process, agents in the transition economies had little experience with, and few realistic possibilities for, portfolio optimization through foreign investments. Therefore they may have chosen the easiest form of portfolio diversification into foreign assets by allocating part of their wealth to foreign currency holdings, with the currency often held inside the East European country. In the spirit of the model of capital flight discussed above, such cash holdings should be counted as part of capital flight. However, some of the East European economies also had high levels of dollarization or euroization, and whether the foreign currencies that went into transactions demand in those countries should count as capital flight is more open to discussion. Since we cannot estimate the euros, dollars or other foreign currencies going into transactions demand, we assume all such foreign cash holdings in these countries do represent capital flight in the logic of Equation 1. 6
Table 1 reports our estimates of capital flight for the 12 East European countries in our sample. For each country we provide estimates of capital flight in US dollars for the 1995-2005 period, and, to aid in interpreting the results, we normalize the capital flight estimates by current GDP, also in US dollars. The use of current US dollar GDP adds volatility to the ratios of capital flight to GDP for some countries because of exchange rate fluctuations, but, since capital flight is also measured in current US$, the normalization provides a degree of market-based comparability for the two variables. It also tends to overstate the ratio of capital flight to GDP because of the systematic undervaluation of these counties’ currencies. We therefore also report the ratio of capital flight to GDP measured in current international purchasing power dollars to eliminate the effects of undervaluation. For purposes of comparison in Table 2 we provide estimates of capital flight from Russia for the same period, calculated by the same method as our estimates for the East European economies and also normalized by Russian GDP.
The results in Table 1 show that there are considerable differences in the experience of individual East European countries with this phenomenon. For example the first country in Table 1, Albania, with the exception of 1998, appears not to have experienced any unrecorded capital outflows. Indeed, it consistently has unrecorded capital inflows. This is not surprising given that between 20 and 25 percent of the Albanian labor force has migrated overseas either on a semi-permanent or seasonal basis to seek work. These migrant workers send home remittances, but they may also be investing in Albania, either by financing the business investments of family members or by purchasing real estate that they intend to occupy when they return to Albania.
The Central European countries that joined the EU in the first eastward expansion, the Czech Republic, Hungary, Poland, Slovakia and Slovenia generally have low levels of capital flight until 2002 or 2003. In those years, as we shall see, we likely overestimate capital flight because of large $/Euro exchange rate fluctuations that influence our estimates of foreign borrowing. Relative to PPP GDP, capital flight from these countries is around 1 percent of GDP per annum, with some exceptions, such as the effects of contagion from the Ruble crisis in 1998, and there are also years in which there are small unrecorded net inflows of capital. This is consistent with the belief that these countries have relatively low levels of risk of expropriation, improving political and macroeconomic stability, and improving returns on domestic assets. When measured against official GDP, the ratio of capital flight is higher, in the neighbourhood of 3 to 5 percent on average due to the undervaluation of these countries’ currencies. In view of the fact that domestic resources are diverted from investment in the home economy and converted to foreign assets at the official exchange rate, the use of GDP at the official exchange rate as the basis for measuring the resource costs of capital flight may be a more accurate measure of the true costs of capital flight.
The Baltic Republics joined the EU at the same time as the Central European countries. Lithuania had quite low levels of capital flight comparable to the Central European countries, but Estonia and Latvia had ratios of capital flight to PPP GDP in the range of 3-4 percent and much higher at the end of the sample period. Indeed, comparing the results for these two countries relative to the calculations for Russia reported in Table 2 shows that capital flight in these counties was as much, if not more, of a problem that it was in Russia. It may be possible that such high levels of unrecorded capital outflows reflect not the behaviour of local investors, but rather the possibility that the banking systems of these countries serve as a conduit for capital flight from nearby countries such as Russia and Belarus.
Finally there are three European countries that did not join the EU during our sample period, Bulgaria, Croatia and Romania. The first two have rather erratic unrecorded capital flows, negative in some years, positive in others, but the magnitude of the flows appears to be somewhat bigger than they are for the Central European EU member countries. Croatia in particular has some years during which capital flight seems relatively high. In contrast, Romania, despite a rather erratic transition experience and political instability has comparatively low levels of capital flight.
Overall, while the levels of capital flight from the East European countries are somewhat lower than those from Russia when scaled by GDP in PPP, they are comparable to the magnitude of Russian capital outflows when scaled for GDP at official exchange rates in some of these countries. This is somewhat surprising given the different perceptions of returns and risks encountered by domestic investors in Russia and East Europe. Nevertheless, some reflection suggests that levels of capital flight from East Europe comparable to those from Russia are not unreasonable. Russian investors may face greater macroeconomic instability as well as higher risks of expropriation and government predation than do those in East Europe, but they also face greater restrictions on their ability to move money out of the country because of capital controls as well as the concentration of trading activities in a few large firms. On the other hand, the East European economies are much more permeable to unrecorded money flows because of their higher trade volumes as well as greater openness to the international movement of capital and people and greater integration into global capital markets. Although returns on foreign direct investments in East Europe are high, domestic capital markets are weak and under-regulated (Glaeser et al. 2001). The difference between capital flight from East Europe and Russia may thus not be quantitative, but rather qualitative in that Russian capital flight is driven by political and macroeconomic uncertainty and the fear of expropriation while capital flight from East Europe is more the consequence of a rational portfolio diversification on the part of investors.7 To examine this hypothesis more closely, we now turn to an examination of the determinants of capital flight from the region.
IV. Drivers of Capital Flight from Transition Economies
The literature on capital flight we have surveyed above provides relatively consistent ways of estimating the drivers of capital flight. In order to account for differences in country size, we scaled capital flight by current GDP at official exchange rates, and, consequently, we scaled a variety of potential explanatory variables by current GDP also. The explanatory variables can be grouped into three broad categories. The first set of variables, which we call the economic variables, reflects the portfolio motive, meaning mainly economic variables that measure the benefits of relatively higher returns abroad, more efficient intermediation, and less risk of losing the value of liquid assets to inflation, currency depreciation and high taxes in the home country. The real interest rate differential, the growth of GDP, the ratio of private credit to GDP, inward FDI, the current account deficit and the government deficit are all related to this general portfolio approach to capital flight.8
We expect that the sign for the interest rate differential to be negative, since the higher the domestic interest rate, ceteris paribus, the greater the opportunity cost of holding foreign assets, while a higher domestic inflation rate makes foreign assets more attractive. The government deficit is a sign of future inflation as well as of the likelihood of higher taxes in the future, and both of these should encourage capital flight. The sign for the growth of GDP is unclear ex ante; falling GDP can be a signal of economic difficulties that encourage capital flight, but higher GDP also means higher incomes and profits, increasing the funds that investors can send off shore. The ratio of private credit to GDP is a measure of the availability of money to borrowers in the home country. To the extent that borrowing costs are less than the expected returns on money moved off shore, the provision of credit can be seen as a facilitator of capital flight. Inward FDI is another source of funds, and privatizations that are financed through FDI may be arranged so that the payment to local residents remains outside the country. Finally, we the country’s currency is subject to devaluation??, and thus it affects investors’ perceptions of risk from holding domestic assets. This risk of future devaluation is captured by the size of the current account deficit.
The political factors affecting capital flight are captured by the index of index of economic freedom, as compiled by the Heritage Foundation, and the country’s polity score. The coefficient of economic freedom variable cannot be signed a priori. Greater economic freedom should encourage domestic investment and increase protection against predation and expropriation, but, at the same time, it reduces the legal and bureaucratic obstacles to capital flight by promoting greater capital mobility, the freedom to travel abroad, etc. The second measure of the political environment we use is the polity score, which is the difference between a country’s rating on a ten point scale for democracy minus its ranking on a ten point scale for autocracy (Marshall and Jaggers, 2007); higher values thus reflect a more democratic regime, lower values a more autocratic one. The coefficient of economic freedom variable cannot be signed a priori. A more democratic regime provides investors with protection through the rule of law and limits on predation, thus reducing capital flight, but it also means that one political party can easily succeed another, resulting in changes in economic policy. On the other hand, autocratic regimes are likely to be replaced in unpredictable fashion by revolutions or mass protests, which themselves can cause considerable economic disruption. Consequently this variable cannot be signed a priori either.
A third set of variables measures the ease with which agents in the economy can move capital out of the country in ways that are neither seen by the authorities nor captured by the balance of payments accounts. We call these variables the facilitating variables, as they either make capital flight easier or more difficult. In the preceding section we have identified some of the means for unreported transfers of money overseas, such as under- or over-invoicing and “fictitious” outward foreign direct investment. The amount of money that can be moved abroad through miss-invoicing should be proportional to the volume of trade. If the volume of trade is low relative to GDP, or if it is strictly controlled by the government, then engaging in capital flight will be more difficult. Alternatively, the removal of restrictions on foreign trade and foreign exchange transactions has an ambiguous effect on capital flight. On the one hand, such liberalization makes it possible to move money abroad more easily, thus promoting capital flight. On the other hand, such liberalization allows individuals to move money aboard legally and in ways that are recorded in the balance of payments accounts. Moreover it serves to integrate domestic and foreign capital markets, thus reducing differences between domestic and foreign returns to investors, thus reducing incentives for capital flight (Lensink et al., 1998). Consequently such liberalization may reduce estimates of illegal or unrecorded capital flight, replacing it with measured capital outflows. We capture these variables by means of a measure of openness, the ratio of trade to GDP and by the average of indexes of trade and foreign exchange liberalization compiled by the European Bank for Reconstruction and Development (EBRD).
Finally, we include two dummy variables and two variables that account for some of the data obstacles we faced in estimating capital flight. The first dummy variable accounts for the effects on these countries of the Russian ruble crisis in 1998, which had serious, though temporary, effects on the real economies of CEE as well as on CEE stock markets and investors’ perceptions of the financial stability of the transition economies and likely returns on domestic assets. The second dummy variable accounts for the effects of EU accession in the behavior of the residents of the countries who joined the EU in 2004. Specifically, we set the dummy equal to one for 2003 and 2004 to account for this effect on investor behavior of the pre-accession year as well as of the first year of membership. The first measurement correction variable captures the effect of changes in the $/euro exchange rate on our measure of capital flight to account for the fact that we do not have data for all countries and all years on the currency composition of their external debt and thus our estimates in Table 1 include these revaluation effects as unrecorded capital flight. The second measurement error we noted was the possibility that a part of unrecorded capital flight was “internal” to the CEE countries and represented a buildup of foreign exchange holdings, both to meet transaction needs and as components of household portfolios. To capture the domestic accumulation of foreign currency asserts, we use change in foreign currency deposits on the assumption that such deposits should parallel the evolution of foreign exchange held outside the domestic banking system.
The most general specification is:
CFi,t = α + Σ βm (Economic variable m, i, t) + Σ δk (Political variable k, i, t,) + Σ γj (Facilitating variable j, i, t) + Σ εn (Ruble Crisis and EU Accession dummy and Measurement Corrections Variables n, i, t) + ui,t Eq. 2
where the dependent variable is:
CFi,t – Capital flight/ nominal GDP in country i in year t as calculated in Table 1.
The economic variables are:9
RIRDi,t– the real ex post interest rate in country i minus LIBOR in year t
CABi,t – current account/GDP in country i in year t
GGBi,t – government balance/GDP in country i in year t
GDPGi,t – real GDP growth of country i in year t.
IFDIi, t – Inward foreign direct investment / GDP in country i in year t
DPCi, t – Change in the ratio of private credit to GDP in country i in year t.
The political variables are:
IEFi,t – Index of economic freedom in country i in year t, scaled from 0 to 100 where 100 represents the greatest freedom
PSi, t – Polity score in country i in year t, ranging from +10 for most democratic political institutions to -10 for most autocratic political institutions.
The facilitating variables are:
OPi,t – ratio of exports plus imports to GDP in current US $ in country i in year t
TFORi, t – average of EBRD indexes of trade and foreign exchange liberalization in country i in year t
The Dummy and measurement correction variables are:
D98i, t – The Russian crisis dummy takes a value of 1 for all countries when t =1998, equals zero otherwise
DEUi, t – The EU membership dummy, which is equal 1 when i = Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovak Republic or Slovenia and t = 2003 or 2004, equals zero otherwise. We include the year of 2003 to capture the expectations of the accession.
XR t –Euro(ECU)/USD exchange rate in year t
DFX i, t - change in foreign exchange deposits in country i in year t
Our estimation of the model’s parameters over 1995-2005 is by means of an OLS panel regression with 120 observations. To account for unobserved country differences in home-country institutions and policies, we use a country fixed-effects specification (Arellano and Bond, 1991). The regression results are reported in Table 3. Because a number of explanatory variables in the economic and facilitation categories were highly correlated, parameter estimates of the full specification as set out in Equation 2 were plagued by high standard errors. We thus estimated more parsimonious specifications, eliminating variables that duplicated the information provided by others.
Turning first to the economic drivers of capital flight, we find that, as in many other studies of capital flight, economic factors have considerable explanatory power. Specification 1 focuses primarily on the economic drivers of capital flight, with the inclusion of trade openness as a facilitating variable that has an economic character as well. The desire to obtain higher real returns by home country residents appears to be a very important motivator for capital flight. The real interest rate differential, RIRD, is significant at the 1 percent level in this and every other specification and its coefficient is very stable across specifications. Policies to increase domestic returns to investors thus have the potential to reduce unrecorded capital flight; financial repression designed to hold deposit rates down in order to boost bank profits is thus costly in terms of the consequent capital flight.
The coefficient for the ratio of the current account to GDP is not significantly different from zero in this or any other specification. This probably reflects the ambiguous signal that a current account deficit sends to investors. On the one hand, it can be seen as a harbinger of austerity measures or depreciation of the currency, but for countries experiencing large capital inflows, a current account deficit is to be expected and reflects the country’s appeal to foreign investors. The coefficient of GGB, the ratio of the government balance to GDP, is negative and significant across all specifications as well. An increase in the government surplus to GDP of one percentage point will reduce the ratio of capital flight to GDP by about 0.5 percentage points. This result is consistent with the theories of capital flight that view government deficits as fueling investors’ fears of domestic inflation and higher future taxes. Real GDP growth, GDPG, has a positive and significant effect on capital flight, not only in this specification but across all specifications, suggesting that greater incomes and the accumulation of wealth by households translates into demand for foreign assets. Of course, some of this additional income may be directed toward the purchase of illegal drugs, prostitution, gambling, etc. that then has to be laundered abroad through unrecorded means. The volume of foreign direct investment also influences capital flight positively. This many be due to efforts of domestic owners of firms to divert to foreign assets the proceeds of sales of firms to foreigners, placing them beyond the reach of local authorities, or a further aspect of the restructuring of portfolios in favor of foreign assets. Finally, the increases in the amount of private credit also increase capital flight. This may reflect a form of intermediation if investors believe that domestic borrowing rates are less than foreign returns. Both this and the foreign direct investment are significant and suggest that about 30% of FDI inflows are turned into capital flight as is somewhat less than 30% of domestic credits created.
The facilitating variable we use in Model 1 is the trade to GDP ratio because of the importance of over- and under-invoicing in facilitating unrecorded capital flight. However, the variable is not significant in this or any other specification. The dummy variable for the Ruble Crisis is positive suggesting that the effects of contagion form the crisis may have driven CEE residents to seek the shelter of more secure assets in foreign countries. However, we note that the coefficient for this dummy is not significant in all Models. By way of contrast, the EU dummy is highly significant and large. EU accession sharply increased unrecorded capital outflows. Perhaps this is partly due to the increase in euro holdings for transactions purposes by residents of these countries in anticipation of accession, or it may reflect increased unrecorded capital outflows in anticipation of more efficient anti-money-laundering measures that would accompany EU accession. The XR variable (i.e. Euro(ECU)/USD exchange rate) is also highly significant across specifications, and its positive sign means that the appreciation of the Euro against the dollar increases measured capital flight. This suggests that significant improvements in estimates of capital flight from CEE could be obtained form additional data on the composition of CEE debt.
Finally, as can be seen from a comparison of the R2 for Model 1 relative to the other models, the economic variables alone explain a good deal of the variance in the dependent variable despite the significance of other explanatory variables in the other specifications. Moreover, our parameter estimates of the economic variables are quite stable across specifications and consistent with the theories of capital flight that we have surveyed above as well as with the findings of studies of capital flight in other countries around the world.
In Model 2 we add our second facilitating variable, the liberalization of trade and foreign exchange controls. The variable is significant and negative. This is likely due to the fact that liberalization enables home-country investors to move money abroad through legal, and thus recorded, ways, and as a result legal and recorded capital outflows are substituted for some illegal and unmeasured capital flight.
In Model 3 we add the change in foreign currency deposits as an explanatory variable, but it proves not to be significant; evidently changes in domestic holdings of foreign currencies are not an important factor in the estimation of unrecorded capital flight, except in the case of the new Member States, whose response is captured by the EU accession dummy.
In Models 4 and 5 we examine the role of political risk. We chose not to use dummy variables for short-term political events such as elections, domestic disturbances, regional tensions, etc. because, given the length of our sample and the turbulent nature of the region, such dummies would use up too many degrees of freedom. Moreover, the decisions on which events rate a dummy and which ones do not can become excessively subjective and data driven. Consequently, we focused on longer-term measures of the evolution of the political system. The index of economic freedom, IEF, compiled by the Heritage Foundation is widely used in studies such as this because it combines political characteristics with those related to secure property rights, protection of entrepreneurs, etc. which bear directly on the causes of capital flight. Surprisingly the index did not have a significant coefficient in Model 5, and, given its rather poor performance there, we did not use it in other specifications. The polity variable, PS, on the other hand, was significant in Model 4. Because this variable is quite sensitive to even nuanced changes in the balance between democracy and autocracy in the political regime of a country, it may convey information more relevant than the index of economic freedom. In any case, the coefficient suggests that increases in democracy tend to increase capital flight, a result that may seem paradoxical except for the fact that a strengthening of democracy may be seen by old elites in these countries as a danger to the wealth that they may have acquired early in the transition through corrupt privatizations, etc. Although, while there is some room in the results for political drivers of capital flight, our overall results suggest that capital flight from the transition economies of East Europe is largely an economic phenomenon driven by considerations of risk and return on domestic and foreign assets.