Chapter

9 Impact of Taxation on International Capital Flows: Some Empirical Results

Editor(s):
Vito Tanzi
Published Date:
June 1984
Share
  • ShareShare
Show Summary Details
Author(s)

In recent years, only a few studies taking into account the impact of taxation have been added to the large body of literature on international finance. These studies have attempted to show analytically how the introduction of taxation can affect the interrelationships among interest rates, inflation rates, and exchange rates. This paper aims to demonstrate empirically the impact of tax factors on international capital flows.

The interrelationships among interest rate differentials, expected inflation differentials, and expected exchange rate movements, which are also known as the interest rate parity and the purchasing power parity, have been shown in theoretical papers (Aliber (1973) and Hod-jera(1973)) to hold simultaneously in equilibrium. However, empirical studies have found deviations between interest rate differentials and forward exchange rate premiums, on the one hand, and between these and expected inflation differentials, on the other. These deviations have been explained by transaction costs, degrees of political risks, government interventions, and varying degrees of exchange controls; tax factors have not been suggested as a possible explanation (Aliber (1975); Frenkel and Levich (1975)). Thus, it can be argued that these studies have implicitly assumed that tax factors do not affect these differentials or, alternatively, that tax factors affect all relevant variables proportionally so that the net effect is neutral.

The recent studies introducing tax considerations have demonstrated that differences in tax practices among countries may affect the relationships between interest rate parity and purchasing power parity (Ben-Zion and Weinblatt (1982) and Blejer).1 In addition, they have shown how taxes may affect the direction of capital flows (for example, capital can flow simultaneously in opposite directions (Levi (1977)) and how taxes introduce nonneutrality in that a change in expected inflation in one country affects the expected real rate of interest or the path of the real exchange rate in a two-country setting (Hartman (1979); Howard and Johnson (1982)).

The first section of this paper provides empirical evidence on the impact of tax factors on the relationship between interest rate differentials and expected inflation differentials for the United States and each of seven other industrial countries—Canada, France, the Federal Republic of Germany, Italy, Japan, the Netherlands, and the United Kingdom, based on quarterly time series data over the period 1972–80.

The second section focuses on the potential for “abnormal” simultaneous capital flows in opposite directions between the United States and each of the seven industrial countries, based on Levi’s methodology, and shows to what extent such flows are significant.

I. TAXATION AND INTEREST RATE DIFFERENTIALS

International capital flows arise from interest rate differentials across countries when investors choose securities yielding the highest return. In times of exchange rate uncertainty, investors in foreign-denominated securities can hedge against the risk of fluctuations in the value of foreign currencies by buying a contract to sell the foreign exchange-in the forward market upon redemption of the foreign-denominated securities. Consider a two-country case: Country A and Country B. Residents of Country A can buy domestic securities and receive a yield, iA, or they can buy foreign securities whose yield is composed of two components—the foreign interest, iB, and the expected foreign exchange gain (or loss), Se, which is defined as the expected percentage change of the spot exchange rates and is denominated in units of domestic currency per unit of foreign currency. From the point of view of the residents of Country A, portfolio equilibrium will exist when

Viewed differently, the expected change in the exchange rate is equivalent to the differential in expected inflation, or

where πe is the rate of expected inflation.

The first is known as the interest rate parity, while the second is the purchasing power parity. A third relationship can be established by applying the Fisher equation to a two-country setting.

where ie is the expected real rate of interest.

Thus, subtraction of equation (4) from equation (3) yields

Equation (2) differs from equation (5) in that it assumes that there is no difference in expected real rates of interest (actual real rates of interest can diverge because of differences in productivity of capital). In other words, equation (5) assumes that a Mundell effect may exist in a two-country case operating in a similar manner to the Mundell effect in a single-country case. Namely, a change in the differential of expected inflation between two countries may affect the differential of expected real interest rates.

Before introducing tax factors into the analysis, it would be useful to describe briefly how industrial countries tax income arising from international transactions. The treatment of taxation on interest income and capital gains arising from foreign exchange transactions varies across countries. As the tax laws in industrial countries generally do not contain explicit provisions about the treatment of foreign exchange gains, practices tend to reflect accounting and legal practices. For example, taxes on foreign exchange gains are payable on an accrual basis in Canada and France. Furthermore, foreign exchange gains are taxed at the capital gains tax rate rather than at the income tax rate in Canada, the United Kingdom, and the United States. With the exception of these countries, all other countries in the sample allow deductions for unrealized foreign exchange losses. Moreover, the eight industrial countries studied have tax treaties allowing for reduced tax rates on withholding taxes that apply to interest income of nonresidents. This does not necessarily mean that investors abroad enjoy lower tax rates; the general practice has been that the withholding tax paid on interest income abroad is deducted from total domestic income tax payments. Thus, investors continue to pay on the basis of domestic tax rates. In effect, tax treaties divide tax revenue between the capital-exporting and capital-importing countries.

Of the eight countries, only Canada, the United Kingdom, and the United States apply capital gains tax rates that are lower than the tax rates on interest income or company income. And only in Canada and the United Kingdom do these capital gains tax rates apply to short-term gains. In the United States, the capital gains tax is applied only on gains realized after 12 months; short-term gains are taxed as normal income.

The introduction of taxation into the Fisher equation in the single-country case results in a modified Fisher effect, as has been demonstrated theoretically by Darby (1975) and Tanzi (1976), and empirically for eight industrial countries.2

Consider the introduction of tax factors into the analysis. Suppose a tax rate, τA, is applied to interest income of residents of Country A earned in any of the countries. If the tax applies to interest income while capital gains are exempt, the relationship in equation (1) will be modified as follows:

or

If the tax applied to capital gains, θA is equal to the tax rate on interest income, then equation (6) reverts to equation (1). However, if there is no capital gains tax (θA = 0), then equation (7) remains intact.

If a reduced capital gains tax, θA < τA, applies, then equation (6) becomes

and similar considerations apply to residents of Country B.

When the expected rate of change of the exchange rate is replaced by the differential of expected inflation rates, the introduction of tax factors viewed in terms of the portfolio equilibrium requirements of residents of Country A would require equating the net-of-tax interest rate differential to the differential of expected inflation. If capital gains tax rates are equal to interest income tax rates, then equation (2) will not be affected by the introduction of tax factors:

If, however, foreign exchange gains are not subject to any tax or if foreign exchange gains tax docs not apply to unrealized gains, then equation (2) becomes

Alternatively, if foreign exchange gains are subject to a reduced capital gains tax, θ < τ, then equation (10) becomes

As the United States applies a capital gains tax of 30 percent on foreign exchange gains to individuals realized after at least 12 months, it would be expected that this would affect the relationship between the differentials for interest rates and expected inflation rates. In order to verify this hypothesis, the following tests were conducted:

where superscripts s and L represent short-term (3 months) and long-term (12 months) maturities, respectively. In order to account for a possible two-country Mundell effect, an alternative formulation to equation (12) is considered. This alternative formulation constrains the constants, αs and αL, to the actual ex post differential of real interest rates, (rArB).

The time series for expected inflation were derived from the term structure of interest rates using the Frankel (1982) method3

To the extent that the differential tax treatment of foreign exchange gains on short-term and long-term flows on the part of the United States affects the link between interest rate and expected inflation differentials, this can be demonstrated in coefficients of equations (12) and (13). If short-term foreign exchange gains are not realized, for tax purposes, it would be expected that ßS in both formulations—(12) and (13)—would be greater than ßL, as the first reflects 1/(1 – τA), while the latter reflects (1 – θA)/(1 – τA)(>1 if θA < τA). If, however, short-term foreign exchange gains are realized and taxed as income, ßS could be smaller than ßL.

The results of the estimations as summarized in Table 1 demonstrate that ßS is significantly greater than ßL for all differentials between the United States and each of the seven countries. These indicate that the relationship between interest rate differentials and expected inflation differentials is affected by tax considerations. The results also demonstrate that the two alternative formulations do not result in substantially different coefficients.

Table 1.United States and Seven Industrial Countries: Short-Term and Long-Term Interest Rate Differentials and Expected Inflation Differentials, 1972–80
ßSßLßSßL
CountryWithout adjustment for

real interest rates1
With adjustment for

real interest rates2
Canada1.5171.0291.5151.024
France1.3331.0751.2481.069
Germany, Fed. Rep.1.1841.0221.1421.023
Italy0.2310.1230.2310.121
Japan31.3491.0121.2651.005
Netherlands1.4131.0781.4211.090
United Kingdom1.2131.0171.0381.003

Based on Tables 3 and 4 in the Appendix.

Based on Tables 5 and 6 in the Appendix.

Data for the second quarter of 1977 through the fourth quarter of 1980.

Based on Tables 3 and 4 in the Appendix.

Based on Tables 5 and 6 in the Appendix.

Data for the second quarter of 1977 through the fourth quarter of 1980.

II. TAXATION AND DIRECTION OF CAPITAL FLOWS

Section I empirically demonstrates how tax factors affect the relationship between interest rate differentials and expected inflation differentials. Section II studies to what extent tax factors may lead to simultaneous movements of capital in opposite directions between two countries.

The following investigation is an extension of Levi’s (1977) framework (which relates to flows between the United States and Canada) to flows that occur between the United States and each of six additional industrial countries.

Levi’s demonstration of abnormal capital flows is based on a differential tax treatment of interest income and capital gains.

Suppose there is a pretax advantage of investing in Country A

where S and F are the spot and forward exchange rates denominated in units of domestic currency per unit of foreign currency.

Residents of Country A are subject to interest income tax, τA, and capital gains tax, θA, where θA < τA; residents of Country B are subject to income tax, τB and no capital gains tax. Residents of Country A will prefer to buy securities of Country B when the foreign currency is at a premium because the exchange gain component of their earnings is taxed at a lower rate. Residents of Country B will prefer to buy securities of Country A because of their higher yield.

This situation arises when

or

where

Such a situation may result in capital flows in both directions as residents of both countries buy each other’s securities. Because only Canada and the United Kingdom apply differential tax rates on interest income and foreign exchange gains on short-term securities having maturities of less than 12 months, the frequency of such potential occurrences for these two countries vis-à-vis other countries can be measured. Canada applies a corporate income tax rate of τA = 0.46 and a capital gains tax of θA = 0.23. Let Canada be Country A and the United States Country B. Canadians will buy short-term U.S. securities and U.S. residents will buy Canadian securities simultaneously when

During 1972–80, using weekly data on three-month treasury bills and three-month forward exchange rates with annualized premiums, such a potential existed during 3.2 percent of the weekly observations (14 out of 436). The frequency of abnormal capital flows arrived at by Levi is considerably higher. This may be due to the fact that he does not use annualized three-month forward exchange rate premia. When three-month Eurocurrency deposit rates were used, no such observations were recorded (Table 2). Similarly, with three-month treasury bills, a potential for simultaneous flows between the United Kingdom and the United States existed in 2 out of 437 weekly observations.

Table 2.United States and Seven Industrial Countries: Frequency of Abnormal Capital Flows, 1972–80
Country and RateNumber of

Observations

When U.S. Dollar

at Forward

Premium
Number of

Observations

When U.S. Dollar

at Forward

Discount
Total

Number of

Observations
Canada
3-month treasury bill rate141436
3-month Eurocurrency deposit rate06423
12-month Eurocurrency deposit rate151406
France
12-month Eurocurrency deposit rate132431
Germany, Fed. Rep.
12-month Eurocurrency deposit rate00436
Italy
12-month Eurocurrency deposit rate120390
Japan
12-month Eurocurrency deposit rate01262
Netherlands
12-month Eurocurrency deposit rate31435
United Kingdom
3-month treasury bill rate21437
3-month Eurocurrency deposit rate00433
12-month Eurocurrency deposit rate828430

Because the United States applies lower foreign exchange gains tax rates for realized gains of at least 12 months, such an exercise can be conducted on differentials between the United States and other countries using 12-month Eurocurrency deposit rates. Consider the U.S. tax treatment of interest income and capital gains arising from foreign exchange gains. U.S. corporations are subject to a 48 percent tax rate that applies to interest income (τB = 0.48) and to a 30 percent capital gains tax (θB = 0.30) that applies to foreign exchange gains realized after at least 12 months. Denote the foreign country as Country A and the United States as Country B. Suppose there is a pretax advantage to investing in U.S. securities:

where iA and iB are 12-month Eurocurrency deposit rates, and SFF, which is derived from 1/F1/S1/S, is the 12-month forward premium of the foreign currency vis-à-vis the U.S. dollar. Without taxes. U.S. residents will dearly prefer to buy securities denominated in U.S. dollars, as will residents of Country A. With the introduction of taxes on interest and foreign exchange gains on U.S. residents, the above relationship becomes

Under certain conditions, the pretax advantage of buying U.S. securities can be reversed and U.S. residents will prefer to buy foreign securities on which the exchange gain component of their earnings is taxed at a lower rate. Algebraically,

or

As far as residents of Country A are concerned and as long as there is no distinction between interest income tax and foreign exchange gains tax, they will prefer U.S. securities when equation (17) obtains. Thus, if equation (22) obtains, U.S. residents will prefer buying foreign securities and residents of France, the Federal Republic of Cer-many, and Japan, inter alia, will prefer buying U.S. securities, thus giving rise to simultaneous capital flows in opposite directions. The frequencies of a potential for occurrences such as those in column 2 of Table 2 are very low for all of the above countries. Such a situation may or may not arise regarding Canada and the United Kingdom, which apply differential tax treatment to foreign exchange gains. Because Canada and the United Kingdom also apply lower foreign exchange gains tax rates, the existence of equation (22) will result in capital flows to the country with the lower pretax yields.

It is possible to conceive of situations in which U.S. residents will prefer buying U.S. securities while foreign residents will prefer buying their own securities. Such situations may arise when the U.S. dollar is at a forward premium and equation (22) obtains. Suppose there is a pretax advantage to buying foreign securities. Under such conditions, U.S. residents will borrow in foreign currency in order to invest in securities denominated in U.S. dollars. Because the U.S. dollar is at a forward premium, the foreign exchange gain is treated as a capita) gain that will more than compensate for any higher foreign interest cost. Every dollar lost through the high foreign interest cost will represent a net lax loss of $0.52 but every exchange gain will represent a net tax gain of $0.70. The frequencies of a potential for such occurrences, which are shown in column 1 of Table 2, are low, with the exception of the United Kingdom. Therefore, it can be argued that, while the concept of abnormal capital flows is appealing because it can explain simultaneous flows in opposite directions, the fact is that during 1972–80 such flows rarely took place.

III. CONCLUSIONS

This paper empirically examines two ways in which tax factors may affect the flow of international capital.

The first method is designed to test whether tax factors affect the relationship between the differentials of interest rates and expected inflation rates. The empirical tests show that the coefficients of the regression of long-term interest rate differentials on expected inflation differentials are smaller than those for short-term interest rates. These findings suggest that if the difference in the coefficients is wholly attributable to tax factors, then short-term foreign exchange gains are effectively taxed at lower rates than long-term gains. This is a surprising conclusion, which would bear further analysis.

The second, which is an application of the Levi (1977) method, demonstrates how differential tax treatment by the United States of interest income and foreign exchange gains realized after 12 months can lead to simultaneous capital flows in opposite directions. Under certain conditions and when the U.S. dollar is at a forward discount, U.S. residents will he inclined to buy foreign securities while residents of the foreign country will be inclined to buy U.S. securities; in the opposite case, and when the U.S. dollar is at a forward premium, residents of both countries will be inclined to buy their own securities. In practice, however, the frequency of a potential for such flows is shown to be very low.

Appendix
Table 3.United States and Seven Industrial Countries: Short-Term Interest Rate Differentials and Expected Inflation Differentials, 1972–801

iAtsiBts=αs+βs(πAteπBte)+ut

CountryαsβsR¯2D-W
Canada0.3931.5170.9441.78
(0.134)(0.091)
France0.8561.3330.9011.96
(0.356)(0.102)
Germany, Fed. Rep.–0.008*1.1840.8621.88
(0.341)(0.094)
Italy–5.0210.2320.8492.05
(1.871)(0.022)
Japan2–2.5711.3490.8861.89
(0.405)(0.141)
Netherlands0.9711.4130.6771.77
(0.336)(0.186)
United Kingdom3.8511.2130.9101.96
(0.532)(0.080)

Short-term interest rates are three-month Eurocurrency deposit rates; * represents insignificant coefficients; standard errors are in parentheses.

Data for the second quarter of 1977 through the fourth quarter of 1980.

Short-term interest rates are three-month Eurocurrency deposit rates; * represents insignificant coefficients; standard errors are in parentheses.

Data for the second quarter of 1977 through the fourth quarter of 1980.

Table 4.United States and Seven Industrial Countries: Long-Term Interest Rate Differentials and Expected Inflation Differentials, 1972–801

iAtLiBtL=αL+βL(πAteπBte)+ut

CountryαLβLR¯2D-W
Canada0.1941.0290.9892.12
(0.039)(0.025)
France0.210*1.0750.9902.01
(0.121)(0.288)
Germany, Fed. Rep.–0.123*1.0220.9951.84
(0.065)(0.016)
Italy–5.7070.1230.7611.98
(1.908)(0.023)
Japan20.3331.0120.9991.88
(0.025)(0.054)
Netherlands0.8461.0780.9351.78
(0.121)(0.058)
United Kingdom2.8291.0170.9992.06
(0.042)(0.006)

Long-term interest rates are 12-month Eurocurrency deposit rates; standard errors are in parentheses; * represents insignificant coefficients.

Data from the second quarter of 1977 through the fourth quarter of 1980.

Long-term interest rates are 12-month Eurocurrency deposit rates; standard errors are in parentheses; * represents insignificant coefficients.

Data from the second quarter of 1977 through the fourth quarter of 1980.

Table 5.United States and Seven Industrial Countries: Short-Term Interest Rate Differentials and Expected Inflation Differentials, 1972–801

(iAtsiBts)(r¯Ar¯B)=βs(πAteπBte)+ut

CountryβsR¯2D-W
Canada1.5150.9411.83
(0.100)
France1.2480.8912.02
(0.101)
Germany, Fed. Rep.1.1420.8651.86
(0.053)
Italy0.2310.8382.11
(0.022)
Japan21.2650.8881.83
(0.110)
Netherlands1.4210.6871.77
(0.170)
United Kingdom1.0380.9011.85
(0.029)

Short-term interest rates are three-month Eurocurrency deposit rates; standard errors are in parentheses; r¯Aandr¯B represent real ex post interest rates.

Data for the second quarter of 1977 through the fourth quarter of 1980.

Short-term interest rates are three-month Eurocurrency deposit rates; standard errors are in parentheses; r¯Aandr¯B represent real ex post interest rates.

Data for the second quarter of 1977 through the fourth quarter of 1980.

Table 6.United States and Seven Industrial Countries: Long-Term Interest Rate Differentials and Expected Inflation Differentials, 1972–801

(iAtLiBtL)(r¯Ar¯B)=βL(πAteπBte)+ut

CountryβLR¯2D-W
Canada1.0240.9872.12
(0.026)
France1.0690.9902.10
(0.029)
Germany, Fed. Rep.1.0230.9961.84
(0.010)
Italy0.1210.7392.06
(0.022)
Japan21.0050.9991.90
(0.003)
Netherlands1.0900.9371.77
(0.053)
United Kingdom1.0030.9991.86
(0.002)

Long-term interest rates are 12-month Eurocurrency deposit rates; standard errors are in parentheses; r¯Aandr¯B represent real ex post interest rates.

Data for the second quarter of 1977 through the fourth quarter of 1980.

Long-term interest rates are 12-month Eurocurrency deposit rates; standard errors are in parentheses; r¯Aandr¯B represent real ex post interest rates.

Data for the second quarter of 1977 through the fourth quarter of 1980.

References

    AliberRobert Z.“The Interest Parity Theorem: A Reinterpretation,”Journal of Political Economy (Chicago) Vol. 81 (November/December1973) pp. 145159.

    AliberRobert Z.“Exchange Risk, Political Risk, and Investor Demand for External Currency Deposits,”Journal of Money Credit and Banking (ColumbusOhio) Vol. 7 (May1975) pp. 16180.

    Ben-ZionUri and JamesWeinblatt“Purchasing Power Interest Rate Parity and the Modified Fisher Effect in the Presence of Tax Agreements” (unpublishedAugust1982).

    DarbyMichael R.“The Financial and Tax Effects of Monetary Policy on Interest Rates,”Economic Inquiry (Long BeachCalifornia) Vol. 13 (June1975) pp. 26674.

    FrankelJeffrey A.“A Technique for Extracting a Measure of Expected Inflation from the Interest Rate Term Structure,”Review of Economics and Statistics (Cambridge, Massachusetts) Vol. 64 (February1982) pp. 13542.

    FrenkelJacob A. and Richard M.Levich“Covered Interest Arbitrage: Unexploited Profits?”Journal of Political Economy (Chicago) Vol. 83 (April1975) pp. 32538.

    HartmanDavid G.“Taxation and the Effects of Inflation on the Real Capital Stock in an Open Economy,”International Economic Review (Osaka, Japan) Vol. 20 (June1979) pp. 41725.

    HodjeraZoran“International Short-Term Capital Movements: A Survey of Theory and Empirical Analysis,”Staff PapersInternational Monetary Fund (Washington) Vol. 20 (November1973) pp. 683740.

    HowardDavid H. and Karen H.Johnson“Interest Rates, Inflation and Taxes: The Foreign Connection,”Economics Letters (Amsterdam)Vol. 9No. 2 (1982) pp. 18184.

    LeviMaurice D.“Taxation and ‘Abnormal’ International Capital Flows,”Journal of Political Economy (Chicago) Vol. 85 (June1977) pp. 63446.

    ModiJitendra R.“Survey of Tax Treatment of Investment Income and Payments in Selected Industrial Countries” (unpublishedFiscal Affairs Department, International Monetary FundMay271983).

    TanziVito“Inflation Indexation and Interest Income Taxation” Quarterly ReviewBanca Nazionale del Lavoro (Rome) No. 116 (March1976) pp. 6476.

See Chapter 8 (paper by Blejer).

See Chapter 7 (paper by Katz).

An application of this method to single-country Fisher effects appears in Chapter 7 (paper by Katz).

    Other Resources Citing This Publication