10 Capital Flight in Kenya

Mohsin Khan, and Simeon Ajayi
Published Date:
May 2000
  • ShareShare
Show Summary Details
N.K. Ng’eno

The withholding of quick-disbursing aid by donors in November 1991 contributed to one of the worst episodes of balance of payments crisis in Kenya’s history. The crisis brought to the fore the debate on capital flight, as it became evident that capital flight continued even as the foreign exchange crisis deepened. At the time capital flight was viewed as the major cause of slow economic growth, declining investments, and mounting external debt. The growth of GDP declined from 4.5 percent in 1990 to 2.1 percent in 1991, and declined further to 0.5 percent and 0.2 percent in 1992 and 1993, respectively. Long-term outstanding debt rose by 10 percent between 1990 and 1991, and the balance of payments deficit was 12 percent of GDP in 1990 (Ng’eno, 1997). Foreign direct investment declined from $19 million in 1991 to $6 million and $2 million in 1992 and 1993, respectively (World Bank, 1994).

The data relied on nonbank cross-border deposits, reported in the IMF’s International Financial Statistics, as a measure of capital flight. Total capital flight from this source was $2.53 billion at the end of 1992, equivalent to 37 percent of the country’s total external debt.

While accepting that capital flight was a problem, the government disputed its magnitude, but began nonetheless to take some steps to deal with it. The government on many occasions reminded Kenyans of the illegality of operating foreign accounts, and in 1991 it gave a one-month amnesty to holders of foreign assets. The most significant step, however, was to relax foreign exchange controls beginning in the early part of 1993, which led to the convertibility of the shilling in June 1994. Ng’eno (1997) provides a detailed narrative on economic reforms in Kenya since the 1970s.

The reforms were intended to restore confidence in the economy and thereby stem capital flight and promote foreign investment and increased exports, and, ultimately, a higher rate of economic growth. The reforms were so successful in attracting foreign exchange inflows that by September 1994 holders of foreign currency accounts were being encouraged by the central bank to retain their funds abroad to stabilize the exchange rate.

The capital flight debate in Kenya mirrors the common problem surrounding the issue in the literature. Studies on capital flight have not established a consensus on the right definition, method of measurement, or causes and effects of the problem. There is indeed no consensus as to whether capital flight is a problem or not. These issues need to be addressed, at least to some acceptable degree, if capital flight estimates are to be useful for policy purposes.

The extent of capital flight in Kenya, like that of many other sub-Saharan African countries, has not been given much attention in the literature. The available estimates of capital flight in Kenya (for example, Chang and Cumby, 1991) are sketchy and inconclusive. Ajayi (1996) provides estimates of Kenya’s capital flight in 1980–91.

This chapter aims to analyze the extent of the problem of capital flight in Kenya using different methods of estimation and to determine empirically the causes of capital flight, with emphasis on the role of macro-economic variables. The chapter first addresses the conceptual and practical problems related to the definition of capital flight and discusses the causes and effects of capital flight. It then reviews the different methodologies used for estimating capital flight and provides empirical results obtained by estimating a simple capital flight function. These results are used to discuss the policy implications of capital flight in Kenya.

Capital Flight: Definition, Causes, and Effects

Capital flight is simply the movement of capital from one country to another, motivated by differentials in returns to capital across countries. The movement of capital from developing to developed countries has generated much controversy, however.

The use of the word “flight” to describe the movement of capital from developing to developed countries connotes illegal or unsanctioned movement, but this cannot be true of all capital flows. This has led to attempts to distinguish normal or legal capital flows from “abnormal” or “illegal” flows. Assuming the existence of controls on the capital account, normal capital outflows would be those that occur with government approval, and capital flight would be movement of capital not sanctioned by the authorities. In practice, it is difficult to distinguish between normal and abnormal capital flows. Because of this problem, Kindleberger (1987) and Walter (1987) suggested that capital flight should simply be defined as all capital that “flees,” irrespective of the motive. Such a measure, while neutral, would be too general to be of use in assessing the problem of capital flight. An alternative definition used by the World Bank (1985), Erbe (1985), Morgan Guaranty Trust (1986), and Chang and Cumby (1991) is “the change in the private sector’s net foreign assets.” This definition is still too broad but is the most commonly used measure.

The other issue that arises in the analysis of capital flight is the existence of “cross-transfer” of capital between developed and developing countries. Theory predicts that capital-scarce developing countries would have higher returns to capital than capital-abundant developed countries. This should lead to net capital inflows to developing countries. However, simultaneous foreign borrowing and lending (private capital outflows) by developing country residents is common. Increased external indebtedness of developing countries has indeed been accompanied by high capital flight.

Several explanations have been given for this phenomenon. One is the existence of abnormal risks, or what Khan and Ul Haque (1987) call “expropriation risks.” The underlying assumption behind this explanation is that developing country residents invest in developed countries as a means of diversifying their portfolios. Investing in the domestic economy is perceived to be risky because of poor economic performance, inefficient capital markets, and/or fragile political systems. Moreover, the risks of expropriation of assets are higher in developing countries than in developed countries. This argument does not explain why foreign residents would want to invest in developing countries, but this problem of cross-transfer of capital could be explained by the existence of asymmetric risks between residents and nonresidents (Dooley, 1986, 1988). The asymmetric risks may arise from different tax treatment and credit guarantees offered to domestic and foreign residents. A tax system that discriminates in favor of foreign investment may induce increased foreign capital inflow and an outflow of domestic capital trying to avoid taxes and other risks. This is quite common in developing countries, where governments try to attract foreign capital inflows and promote exports by offering substantial tax incentives to foreign investors. Government guarantee of foreign debt can also induce capital inflows, and when domestic investments do not enjoy similar guarantees, capital is more likely to flee to safer havens. At the extreme, public and publicly guaranteed external debt could be contracted by local investors who at the same time acquire foreign assets. This is the most likely scenario in many developing countries facing high capital flight. These explanations led Khan and Ul Haque (1987) and Dooley (1986, 1988) to define capital flight as the unreported stock of foreign claims that generate income that is not recorded in the balance of payments.

The choice of a measure of capital flight depends on the issues of importance to the country under consideration. Chang and Cumby (1991) argue that if capital flight reduces domestic investment, then the outflow of capital, whether reported or not, should be considered a measure of capital flight. However, if the issue of interest is the effect of capital flight on the domestic tax base, then the unreported capital outflow is the appropriate definition. The two problems coexist in Kenya. This chapter will therefore not limit itself to particular measures.

Capital flight results from macroeconomic and political instability. A poor investment climate, characterized by high poverty, low savings and investment rates, a narrow technological base, and poor performance of the economy, also contributes to capital flight. The macroeconomic variables that play a major role in capital flight are inflation, domestic and foreign interest rates, and exchange rates. Inflation causes capital flight by reducing the value of domestic assets such as money and savings. Similarly, an overvalued currency implies a high probability of devaluation, which would reduce the value of domestic assets. Low or negative interest rates are common in the financially repressed economies of developing countries. The large difference between domestic and foreign interest rates makes it attractive to acquire foreign assets. However, it is possible that even if domestic interest rates were higher than foreign interest rates, political instability, which leads to high expropriation risks, would make it profitable to invest (“save”) abroad.1 Thus, in the face of expropriation risks, higher foreign interest rates may not necessarily explain capital flight. The performance of the economy can also serve as an indicator of the profitability of investments. Poor economic growth can encourage capital flight in search of better returns.

Capital flight has several negative effects on the economy. First, it reduces domestic investment and therefore depresses economic growth. Second, it reduces the domestic tax base, which may force the government to resort to distortionary taxes that introduce inefficiencies in the market, thus reducing economic growth. Third, by increasing foreign exchange, which could be siphoned out, capital inflows can contribute to capital flight. This is especially true where corruption results in the flight of public and publicly guaranteed debt. Last, capital flight creates a need for higher foreign borrowing and therefore leads to higher indebtedness, and also slows the solution of the debt problem. Creditors would be unwilling to supply new money on favorable terms when most of it is likely to be repatriated to developed countries.

Measurement of Capital Flight

The following methods have been most commonly used to estimate capital flight: the residual method, the Dooley method, the “hot money” method, the nonbank cross-border deposits method, and the trade misinvoicing method.

Residual Method

This method estimates capital flight as a residual based on the balance of payments identity. The residual is measured as the difference between capital inflows and official allocation of funds as recorded in the balance of payments statistics. This method was developed by Dooley and others (1983) and has been applied by the World Bank (1985), Erbe (1985), Morgan Guaranty Trust (1986), and Cumby and Levich (1987). The variants of this model are discussed by Chang and Cumby (1991) and Gajdeczka (1990).

The most basic form of this method can be specified as follows:


CF=capital flight
dD=change in external debt
NFI=net foreign direct investment flow
CAD=current account deficit
dR=change in official reserves.

This method was used by the World Bank (1985) and Erbe (1985). It assumes that the residual represents the net foreign claims by the private sector, so that if CF > 0, then capital flight exists, and CF < 0 represents capital inflows.

Morgan Guaranty (1986) modified the World Bank and Erbe method by excluding from the estimation of capital flight the acquisition of short-term foreign assets by the banking sector. However, foreign claims by other private sector agents are considered capital flight. This method can be specified as follows:

where dFCB is the increase in net foreign claims by commercial banks.

Cline’s method (Claessens and Naude, 1993, and Cumby and Levich, 1987) in turn modified the Morgan Guaranty method by excluding the following items from the current account: travel (credit), reinvested foreign investment income (abroad and domestically), and other investment income (credit).

Cline’s rationale for the exclusion of these items in method is that they are beyond the control of the foreign exchange authorities and therefore should not be treated as capital flight. The problem in Kenya’s case is that earnings from tourism form a large part of the current account. The exclusion of these earnings would overestimate the extent of capital flight. Cumby and Levich (1987) suggest that low estimates from Cline’s method would indicate that controls on the capital account have been successful in reducing capital flight.

Dooley Method

Dooley (1986) defines capital flight as a stock of privately held foreign assets that earn income but is not reported to domestic authorities and is therefore not recorded in the balance of payments accounts. The stock of private sector external claims is derived from adjusted capital flows in the balance of payments accounts and the stock of external debt as reported by the World Bank.

The Dooley estimate, however, can be obtained more easily by subtracting the imputed stocks of reported assets from the World Bank residual measure (Claessens and Naude, 1993). The problem with this is that the imputed stock of reported assets is made up of new capital flows and interest earned abroad but not repatriated, which is a stock. The imputed stock of reported assets would not be a pure measure of flows and therefore distorts the Dooley measure of capital flight.

“Hot Money” Method

The “hot money” method, associated with Cuddington (1986), is the simplest measure of capital flight, comprising short-term capital plus errors and omissions. Errors and omissions are used as proxies for unreported short-term capital outflows. This measure reflects speculative private capital movements that can easily move in and out of the country depending on economic conditions; hence the term “hot money.”

This measure is too restrictive because it excludes long-term assets. Cumby and Levich (1987) and Eggerstedt and others (1993) argue that unreported short-term capital is not the only item in the errors and omissions account of the balance of payments. The other items in the account would include measurement errors, unrecorded imports, and lagged registration. Eggerstedt and others argue further that long-term capital cannot be easily distinguished from short-term capital. Moreover, there is no particular reason why agents cannot acquire long-term foreign assets in highly developed foreign capital markets as a hedge against unfavorable economic conditions in their countries.

Despite these shortcomings, estimates obtained from this method are useful and may be treated as lower-bound estimates of capital flight. The estimates can be used as benchmarks for evaluating estimates from other methods.

Nonbank Cross-Border Deposits Method

This method measures capital flight as the change in deposits held by a country’s residents in foreign banks. It is based on the assumption that foreign assets held by residents abroad are mainly in the form of bank deposits. Khan and Ul Haque (1987) argue that this measure is more reliable because capital flight is directly recorded, not estimated as in the other methods. Vos (1990), however, notes that nonbank private deposits held abroad may not be accurately recorded, because some of the funds are held outside reporting institutions and the nationalities of depositors are not always known or easy to establish. Foreign assets are also not always held in the form of bank deposits but may be stocks, bonds, treasury bills, or real estate.

Similar to estimates obtained from the hot money method, nonbank cross-border bank deposits may be taken as lower-bound estimates of assets that can be repatriated easily. The estimates can also be used as easy measures of the extent of capital flight.

Trade Misinvoicing

Trade misinvoicing is one way of effecting capital flight. Bhagwati (1964), Bhagwati and others (1974), Gulati (1987), and Vos (1990) have shown that overinvoicing of imports and underinvoicing of exports are important means of capital flight in developing countries. Misinvoicing occurs as a result of high trade taxes, quantitative restrictions, and exchange controls, all of which are common regimes in developing countries.

This chapter analyzes the role of trade misinvoicing in capital flight using the (Bhagwati 1964) methodology. The methodology uses partner country trade statistics to determine the level of misinvoicing and hence capital flight. The analysis starts with the assumption that for purposes of repatriating capital, exporters will underinvoice exports while importers will overinvoice imports.

It is worthwhile to note that the imports and exports used in determining overinvoicing are valued c.i.f. and f.o.b., respectively. This makes it difficult for the data to be compared on a consistent basis. This problem is corrected in this chapter by adjusting Kenya’s imports from the partner countries and imports of the partner countries from Kenya (Kenya’s exports) by an f.o.b.-c.i.f. ratio obtained from the IMF’s International Financial Statistics (1996). Trade data are obtained from the IMF’s Direction of Trade Statistics.

Trade misinvoicing can therefore be expressed as


IM=import misinvoicing
EX=export misinvoicing
Mk=imports by Kenya from partner country as reported by Kenya
Mp=exports to Kenya as reported by partner country p
Xp=imports from Kenya as reported by partner p country
Xk=exports to partner country as reported by Kenya.

Positive signs for Equations (3) and (4) indicate misinvoicing, that is, overinvoicing of imports and underinvoicing of exports, implying capital flight. Negative signs indicate reverse capital flight or underinvoicing of imports and overinvoicing of exports. The net effect of misinvoicing on capital flight can be obtained by adding import misinvoicing (IM) and export misinvoicing (EM).

Data from Kenya’s major trading partners—the United Kingdom, Germany, France, Italy, the United States, the Netherlands, Japan, India, the United Arab Emirates, Tanzania, and Uganda—are used to estimate the extent of trade misinvoicing. These countries account for 60 percent of Kenya’s imports and exports. The regional diversity of the countries will allow us to determine whether the source or the destination of trade affects the level of misinvoicing. Yeats (1991) suggests that it does.

Estimates of Capital Flight

Estimates of capital flight in sub-Saharan Africa are scanty. Empirical studies on capital flight are predominantly confined to the Latin American countries. Chang and Cumby (1991) estimate capital flight for 36 sub-Saharan African countries, including Kenya. Ajayi (1996) estimates capital flight for 25 severely indebted countries of sub-Saharan Africa, also including Kenya. Ajayi (1991) provides capital flight estimates for Nigeria using a variety of methods, and Olopoenia (1995) provides estimates for Uganda. Using the residual method and the cross-border bank deposits method, Chang and Cumby (1991) estimate Kenya’s stock of capital flight between 1967 and 1987 to have been $0.6 billion and $1.2 billion, respectively. The study also found that there was misinvoicing of both imports and exports. Yeats (1991) found that Kenya’s exports to sub-Saharan Africa were underinvoiced by 12.3 percent in 1982/83 and that exports of manufactured goods were underinvoiced by 32.8 percent. The same study found that Kenya’s exports to the European Community, the United States, and member countries of the European Free Trade Association were underinvoiced by 24.3 percent, 27.1 percent, and 68.2 percent, respectively. Ajayi (1996) established that there were instances of overinvoicing of exports and underinvoicing of imports in Kenya. While indicating the possible causes of capital flight, these results also highlight the inconsistent outcomes from the use of different methodologies and even data types to estimate capital flight.

Estimates of Kenya’s capital flight are derived using the World Bank, Morgan Guaranty Trust, Cline, and Cuddington methods for the period 1971–95. Data on cross-border bank deposits could only be obtained from 1981 onward. The notations used by Cumby and Levich (1987) and Claessens and Naude (1993) are shown in Table 1.

Table 1.Components of Capital Flight
A.Current account balance (August 27, 1998)
A1: Travel: Credit
A2: Reinvested earnings on direct investment abroad
A3: Reinvested earnings on direct investment domestically
A4: Other investment income: credit
B.Net direct foreign investment
C.Other short-term capital of other sectors: net
C1: Other assets
D.Portfolio investment, other bonds
E.Change in deposit money bank’s foreign assets
F.Change in reserves
H.Change in external debt
G.Net errors and omissions
World Bank method:(H + B + A + F)
Morgan Guaranty method:(H + B + A + E + F)
Cline method:(H + B) + (A –(A1 + A2 + A3 + A4) + E + F
Cuddington method I:–(G + C)
Cuddington method II:–(G + C1)
Sources: Data for A, B, C, D, F, and G, International Monetary Fund, Balance of Payments Yearbook; data for E and cross-border bank deposits, International Monetary Fund, International Financial Statistics; and data for H, World Bank World Debt Tables.
Sources: Data for A, B, C, D, F, and G, International Monetary Fund, Balance of Payments Yearbook; data for E and cross-border bank deposits, International Monetary Fund, International Financial Statistics; and data for H, World Bank World Debt Tables.

The estimates of capital flight using cross-border bank deposits are presented in Table 2. The stock of capital flight peaked at $2.6 billion in 1990 and remained above $2 billion in 1989. The stock of capital flight is substantial, accounting for an average of over 30 percent of external debt since 1982 and over 20 percent of GDP since 1985. The estimates decline, however, when measured as the year-to-year change in deposits. The measure indicates that there was reverse capital flight between 1991 and 1993 and that subsequently capital flight was low. It would seem puzzling that there were substantial returns of capital during 1991–93, given that this was also a period of serious political turmoil and economic decline. It is possible, however, that the capital was repatriated to finance the elections of 1992 and as a clearly positive response to foreign exchange and exchange control liberalization at the time.

Table 2.Kenya: Cross-Border Deposits of Nonbanks by Residence of Depositor, 1981–95
Total Deposits (In millions of U.S. dollars)Change in Deposits (In percent)Ratio of Total Deposits to External Debt (In percent)Ratio of Total Deposits to GNP (In percent)
Source: International Monetary Fund, International Financial Statistics, various issues.
Source: International Monetary Fund, International Financial Statistics, various issues.

The residual and Cuddington estimates are shown in Table 3; it is evident from these measures that capital flight peaked in 1979, 1987, and 1990. With the exception of 1987 these were also years of balance of payments crisis. This implies that the flight of capital was used to hedge against the poor economic conditions. Capital flight in 1987 must have resulted from the availability of large foreign exchange reserves owing to the mini-boom in tea in 1986. The periods in which capital flight bottomed out (1984, 1992, and 1994) were also marked by poor economic performance, which reduced the availability of savings to be expatriated. Decline in capital inflows, especially after 1990, also contributed to lower capital flight.

Table 3.Kenya: Estimates of Capital Flight, 1971–95(In millions of U.S. dollars)
World BankMorganClineCuddington ICuddington II
Sources: Author’s calculations.
Sources: Author’s calculations.

The results also indicate that estimates from the residual method have similar trends. These trends are similar to those obtained by Cumby and Levich (1987) and Vos (1990). There is little difference between the estimates obtained from the World Bank and Morgan Guaranty methods. This is not surprising since the only difference between the two is the addition of net foreign claims by commercial banks to the Morgan Guaranty method. The estimates obtained using Cline’s method are higher than those of the World Bank and Morgan Guaranty methods. This is also to be expected because Cline’s method excludes, among other items, tourism earnings, a major element of Kenya’s current account.

Cuddington’s method generates the lowest estimates of capital flight. The peaks of the Cuddington estimates follow a pattern similar to those obtained with the residual method.

Table 4 presents misinvoicing of exports by country. Misinvoicing of exports is high for exports to the United Kingdom, the United States, and Germany, and to a lesser extent for exports to Italy and France. The incidence of misinvoicing is low for exports to India, the United Arab Emirates, Tanzania, and Uganda. This indicates that exports to industrial countries have been used as conduits for capital flight. It could also suggest, as argued by Gulati (1987), that data from these countries are more accurately recorded, thus showing up as high capital flight.

Table 4.Kenya: Export Misinvoicing, 1974–95(In millions of U.S. dollars)
United KingdomGermanyFranceItalyUnited. StatesNetherlandsJapanIndiaUnited Arab EmiratesTanzaniaUganda
Source: Author’s calculations.
Source: Author’s calculations.

It is also evident from Table 4 that misinvoicing has mainly been in the form of underinvoicing of exports and that this has been a major problem for exports to the United Kingdom, Germany, France, Italy, the United States, and Japan. The problem has been consistent for exports to the United Kingdom throughout 1974–95 but rose in Germany, Italy, and the United States after 1985. This result confirms the finding that exports to industrial countries have been used for capital flight. Exports to Tanzania, Uganda, and the Netherlands are overinvoiced, implying that they are used to repatriate capital. Exports to the Netherlands, however, began to be underinvoiced after 1991. The direction of misinvoicing for exports to India and the United Arab Emirates is not clear.

The extent of misinvoicing of imports is shown in Table 5. Imports from all the countries for almost all the years were underinvoiced, implying that imports were used to bring capital back through unofficial channels. This situation was caused by the restrictive trade regime, especially the high import taxes and quantitative restrictions that were prevalent in Kenya before 1993. The underinvoicing of imports would also have been used to obtain foreign exchange outside the foreign exchange control system under the import licensing regime.

Table 5.Kenya: Import Misinvoicing, 1974–95(In millions of U.S. dollars)
United KingdomGermanyFranceItalyUnited. StatesNetherlandsJapanIndiaUnited Arab EmiratesTanzaniaUganda
Source: Author’s calculations.
Source: Author’s calculations.

Generally, the highest underinvoiced imports are those originating in the United Kingdom, Germany, France, Italy, the United States, the Netherlands, and Japan. These are followed by imports from India and the United Arab Emirates, but those from Tanzania and Uganda are the least underinvoiced. This confirms again that trade with industrial countries is the major source of misinvoicing.

The net effect of total misinvoicing, that is, the sum of export and import misinvoicing results, indicates that underinvoicing of imports outweighs the underinvoicing of exports (Table 6).

Table 6.Kenya: Total Misinvoicing, 1974–95(In millions of U.S. dollars)
United KingdomGermanyFranceItalyUnited. StatesNetherlandsJapanIndiaUnited Arab EmiratesTanzaniaUganda
Source: Author’s calculations.
Source: Author’s calculations.

This suggests that trade policy plays a major role in the misinvoicing of trade. This is confirmed by the dramatic decline in misinvoicing from 1992, following the dismantling of the import licensing and exchange control regimes in 1993.

It is evident from the overall results that the level of capital flight is sensitive to the method of estimation. It is also true that the cross-border bank deposit and Cuddington estimates are too narrow and therefore form lower-bound estimates of capital flight, while the World Bank and Morgan Guaranty estimates are the upper bound. The trade misinvoicing estimates indicate how policy distortions can lead to undesirable effects with wide-ranging macroeconomic consequences.

A comparison of the estimates of the residual method and trade misinvoicing clearly shows that the residual method generates the best estimates of capital flight. The estimates of this method are more reliable because they are based on coherent and well-tested data. Estimates from Cline’s method are problematic, because they overestimate capital flight. The World Bank and Morgan Guaranty methods therefore provide the best estimates of capital flight for Kenya.

Empirical Analysis of Capital Flight

Empirical studies on capital flight have been limited by data availability. Pastor (1990) reviews the available empirical literature, which mainly covers studies of Latin American countries. The study also specifies a model that estimates capital flight as a function of changes in inflation, interest rates, degree of currency overvaluation, net long-term capital flows as a percentage of GDP, difference between domestic and U.S. growth rates, tax revenue as a percentage of GDP, and value of labor share in income. The study used pooled data (eight countries over 14 years) to estimate the model. All the coefficients were found to have the right sign (positive, except the difference between the domestic and U.S. growth rates) and were significant, except tax revenue as percentage of GDP.

Cuddington (1986, 1987) uses individual country data to estimate the determinants of capital flight. The 1986 study modeled capital flight as a function of domestic and foreign interest rates, inflation, and the depreciation of currency. The model was estimated for Argentina, Brazil, Chile, Korea, Mexico, Peru, Uruguay, and Venezuela. The study estimated capital flight as a function of domestic and foreign interest rates, inflation, real effective exchange rates, and foreign aid disbursements using data from Argentina, Mexico, and Venezuela. The variables were found to have the correct signs (positive) and were significant in all the countries. The Cuddington studies are limited by the fact that they are restricted to fewer than 12 observations.

Following Cuddington (1986, 1987), we specify an asset portfolio adjustment model that assumes that domestic agents diversify their portfolio by holding domestic and foreign assets. The study uses quarterly data from 1981:Q4 to 1995:Q1. This gives a reasonable number of observations, which allows for dynamic specification of the model. The data are obtained from the IMF’s International Financial Statistics, but GDP data were obtained by interpolation. In this study nonbank cross-border deposits are used as the dependent variable and assumed to be a function of GDP, the real effective exchange rate, and relative returns, measured as the difference between the real foreign interest rate and the domestic interest rate.

The general functional form of the model is as follows:


R=domestic interest rate
RW=foreign interest rate
RD=relative returns (RW – R)
RER=real effective exchange rate
Y=real GDP.

It is postulated that RER is positively correlated and Y negatively correlated with capital flight. The coefficient of RD cannot be determined a priori but is expected to be positively correlated to CF if RW>R and negatively correlated otherwise.

The model was tested for the order of integration of the variables, and it was found that the variables were integrated of order one, implying stationarity (Table 7). Cointegration was tested using the Engle and Granger (1987) two-stage procedure. Since cointegration was established, the variables were entered into the equation in first differences. The model was reduced starting with six lags for each variable.

Table 7.Unit Root Tests
Note: CF, capital flight; RD, difference between foreign and domestic interest rates; RER, real effective exchange rate; Y, real income. Figures in parentheses are the probability values for the null.
Note: CF, capital flight; RD, difference between foreign and domestic interest rates; RER, real effective exchange rate; Y, real income. Figures in parentheses are the probability values for the null.

The estimates for Equation (5) are shown in Table 8. The coefficients of all the variables are significant, but that of Y has an unexpected sign. The positive effect of currency appreciation on capital flight is confirmed. Real income was used as a proxy for the investment climate; that is, higher income is expected to raise the credibility of policy reforms and therefore reduce capital flight. But the results indicate the opposite effect. It is plausible, however, that where policies are not credible, an increase in income provides an opportunity for increased outflow of capital. The positive coefficient of Y can therefore be rationalized in the face of noncredible economic reform. The negative coefficient of relative interest rates indicates that the positive effects of foreign interest rates on capital flight have been outweighed by the negative effects of domestic interest rates. In other words, real domestic interest rates are higher than foreign interest rates. This phenomenon has been common in the 1990s with interest rate liberalization and relaxation of exchange controls.

Table 8.Solved Static Long-Run Equation for Capital Flight
VariableCoefficientStandard Error
Step dummy (91)0.9910.229
Note: The null hypothesis that coefficients are zero is rejected. Wald test chi2(7) = 86.63 [0.0000]**.
Note: The null hypothesis that coefficients are zero is rejected. Wald test chi2(7) = 86.63 [0.0000]**.

The coefficient for ECM is low and insignificant. It is also evident that there were seasonal effects. Two dummies, D91 and D93(3), introduced to account for interest rate liberalization and capital flight in 1991 and the third quarter of 1993, respectively, were found to be significant.

The model confirms that real exchange rate appreciation encourages capital flight. It also suggests that if reforms are not credible, economic growth would lead to increased capital flight. The increases in incomes then simply provide the means for greater accumulation of foreign assets.


This chapter has derived estimates of capital flight using the residual method, the cross-border bank deposits methods, and the trade misinvoicing method. The determinants of capital flight were also estimated.

The chapter establishes that estimates of capital flight depend on the method used, but estimates derived from the residual method have similar trends. Estimates from cross-border deposits and the Cuddington method should be treated as a lower bound, while the World Bank, Morgan Guaranty, and Cline estimates are the upper bound. There is no objective rule for determining the best estimates for Kenya, but the World Bank and Morgan Guaranty estimates are more reliable since they are derived from consistent and well-tested data.

Results from trade misinvoicing indicate that underinvoicing has been a major problem and that it has been prevalent with exports to developed countries. Exports to Tanzania and Uganda have been over-invoiced. These results suggest that exports to industrial countries have been used for capital flight, while those to developing countries have been used to repatriate capital. Imports from all countries were underinvoiced throughout the period, suggesting the use of imports to transfer capital held abroad into the country without going through official channels. The net effect of total misinvoicing also indicates underinvoicing of imports, suggesting that trade policy plays a major role in trade misinvoicing. This conclusion is supported by the decline in trade misinvoicing from 1992, a period in which trade and exchange control regimes were relaxed.

The estimated model of capital flight indicates that real exchange rate appreciation encourages capital flight. The model also suggests that economic growth in the face of reforms that are not credible would result in increased capital flight.

In general, the chapter establishes that sound macroeconomic policies are the right way to deal with the problem of capital flight. It is evident that Kenya’s move in this direction has contributed to the marked reduction in capital flight. This is shown by the low trade misinvoicing in 1994 and 1995 and the net inflows of capital as shown by the Cuddington and cross-border deposits methods. Indeed, the positive change in cross-border deposits in 1994 and 1995 reflects the legitimization of ownership of foreign assets through the total removal of exchange controls. The trends of capital flight derived from the other residual methods also point to the reduction of the problem since serious economic reforms began.


    AjayiS.I.1991“Capital Flight and External Debt in Nigeria,”paper presented at the African Economic Research Consortium WorkshopNairobiMay2630.

    AjayiS.I.1996“An Analysis of External Debt and Capital Flight in the Severely Indebted Low-Income Countries in Sub-Saharan Africa,”paper presented at the IMF-World Bank Conference on External Financing for Low Income CountriesWashingtonDecember 9–12.

    BhagwatiJ.N.1964“On the Underinvoicing of Imports,”Bulletin of Oxford University Institute of Economics and StatisticsVol. 26.

    BhagwatiJ.N.A.Craggier and C.Webulswasdi1974“Capital Flight in LDCs: A Statistical Analysis,” in Illegal Transactions in International Tradeed. by J.N.Bhagwati (Amsterdam: North-Holland).

    ChangK.P.H. and R.E.Cumby1991“Capital Flight in Sub-Saharan African Countries,” in African External Finance in the 1990sed. by I.Husain and J.Underwood (Washington: World Bank).

    ClaessensS. and D.Naude1993Recent Estimates of Capital Flight Policy Research Papers (Washington: World Bank).

    CuddingtonJ.T.1986“Capital Flight: Estimates, Issues and Explanations,”Princeton Studies in International Finance No. 58 (Princeton, New Jersey: Princeton University Press).

    CuddingtonJ.T.1987“Macroeconomic Determinants of Capital Flight: An Econometric Investigation,” in Capital Flight and Third World Debted. by D.R.Lessard and J.Williamson (Washington: Institute for International Economics).

    CumbyR.E. and R.M.Levich1987“On the Definition and Magnitude of Recent Capital Flight,” in Capital Flight and Third World Debted. by D.R.Lessard and J.Williamson (Washington: Institute for International Economics).

    DooleyM.P.1986“Country-Specific Risk Premiums, Capital Flight and Net Investment Income Payments in Selected Developing Countries” (unpublished; Washington: International Monetary Fund).

    DooleyM.P.1987“Comment,” in Capital Flight and Third World Debted. by D.R.Lessard and J.Williamson (Washington: Institute for International Economics).

    DooleyM.P.1988“Capital Flight: A Response to Differences in Financial Risks,”Staff PapersInternational Monetary FundVol. 35 (September).

    DooleyM.P.W.HelkieR.Tryon and J.Underwood1983“An Analysis of External Debt Positions in Eight Developing Countries Through 1990,”Journal of Development EconomicsVol. 21No. 2.

    EggerstedtH.R.B.Hall and S.van Wijnbergen1993“Measuring Capital Flight: A Case Study of Mexico,”Working Paper No. 859 (Washington: World Bank).

    EngleR.F. and C.W.J.Granger1987“Co-integration and Error Correction: Representation, Estimation and Testing,”EconometricaVol. 55No. 2.

    ErbeS.1985“The Flight of Capital from Developing Countries,”IntereconomicsNovember/December.

    GajdeczkaP.1990“The Measurement of Capital Right and Developing Country Foreign Assets,”paper presented at the Kiel Institute Workshop.

    GulatiS.N.1987“A Note on Trade Misinvoicing,” in Capital Flight and Third World Debted. by D.R.Lessard and J.Williamson (Washington: Institute for International Economics).

    International Monetary FundInternational Financial Statistics various issues (Washington).

    KhanM.S. and N.Ul Haque1987“Capital Flight from Developing Countries,”Finance & DevelopmentMarch.

    KindlebergerC.P.1987“A Historical Perspective,” in Capital Flight and Third World Debted. by D.R.Lessard and J.Williamson (Washington: Institute for International Economics).

    Morgan Guaranty Trust Company1986“LDC Capital Flight,”World Financial MarketsMarch.

    Ng’enoN.K.1991“Kenya’s Export Performance,” in Trade and Development in Sub-Saharan Africaed. by J. H.Frimpong-AnsahS. M. RaviKanbur and P.Svedberg (Manchester: Manchester University Press).

    Ng’enoN.K.1997“The External Debt Problem of Kenya,”a revised version of the final report presented at the African Economic Research Consortium WorkshopNairobiMay24281992.

    OlopoeniaR.A.1995“Capital Flight from Uganda,”draft final report presented at the African Economic Research Consortium WorkshopNairobi.

    PastorM.Jr.1990“Capital Flight from Latin America,”World DevelopmentVol. 18No. 1.

    VosR.1990On the Measurement of Private Accumulation of Foreign Assets: The Case of the Philippines (The Hague: Institute of Social Studies).

    WalterI.1987“The Mechanism of Capital Flight,” in Capital Flight and Third World Debted. by D.R.Lessard and J.Williamson (Washington: Institute for International Economics).

    World Bank1985World Development Report 1985 (New York: Oxford University Press).

    World Bank1994World Debt TablesVol. 2 (Washington: World Bank).

    YeatsA.1991“On the Accuracy of Economic Observations: Do Sub-Saharan Africa Trade Statistics Mean Anything?”World Bank Economic ReviewVol. 4No. 2.

The first case in Kenya of capital flight induced by political events was at independence in 1964, when white settlers transferred their capital abroad (Ng’eno, 1991).

    Other Resources Citing This Publication