Abstract
In recent years, an extensive literature has emerged on the relationship between interest rates on financial assets denominated in different currencies. One approach to this topic is embodied in the covered interest parity theorem, which applies when there are organized spot and forward exchange markets that operate without administrative restrictions. The theorem implies that short-term capital movements will result in the equalization, when expressed in terms of the same currency and after covering the operation in the forward exchange market, of the returns on assets that are identical in all respects except for the currency of denomination. The empirical validity of the proposition that covered interest parity holds in such markets has been studied extensively, and the nature of the departures from interest parity have been modeled in detail. Frenkel and Levich (1975, 1977) have shown that observed deviations from parity in Euromarkets are consistent with the level of existing transactions costs. Others have claimed that, in fact, observed departures from interest parity arise because assets issued in different countries are unlike in more ways than currency of denomination. Political and sovereign risks, capital and exchange controls, taxation, and other practices differ across countries, and they tend to introduce a wedge between covered nominal returns.1 Differences in exchange risk obviously do not matter in this context, since covered interest parity deals with observed forward exchange rates and therefore there is no exchange rate uncertainty.2
In recent years, an extensive literature has emerged on the relationship between interest rates on financial assets denominated in different currencies. One approach to this topic is embodied in the covered interest parity theorem, which applies when there are organized spot and forward exchange markets that operate without administrative restrictions. The theorem implies that short-term capital movements will result in the equalization, when expressed in terms of the same currency and after covering the operation in the forward exchange market, of the returns on assets that are identical in all respects except for the currency of denomination. The empirical validity of the proposition that covered interest parity holds in such markets has been studied extensively, and the nature of the departures from interest parity have been modeled in detail. Frenkel and Levich (1975, 1977) have shown that observed deviations from parity in Euromarkets are consistent with the level of existing transactions costs. Others have claimed that, in fact, observed departures from interest parity arise because assets issued in different countries are unlike in more ways than currency of denomination. Political and sovereign risks, capital and exchange controls, taxation, and other practices differ across countries, and they tend to introduce a wedge between covered nominal returns.1 Differences in exchange risk obviously do not matter in this context, since covered interest parity deals with observed forward exchange rates and therefore there is no exchange rate uncertainty.2
An alternative approach to analyzing the relationship between nominal interest rates between countries is provided by the Fisher hypothesis. Applied to the open economy, this hypothesis maintains that, for instruments that are identical except for their currency denomination, interest rate differentials should be explained by the expected exchange rate change between the two currencies. The Fisher proposition and the interest parity theorem coincide only when there is no risk premium in the forward market, that is, when the forward rate and the expected future exchange rate are one and the same.3 Clearly, departures from Fisher parity respond to the same set of variables as departures from covered arbitrage (such as transactions costs, political risk, and differential taxation4). In addition, if expectations about the future course of the exchange rate are not held with perfect certainty, exchange risk will also be part of the observed differential.5
Only the Fisher “open” hypothesis is a testable proposition for currencies for which there are no organized forward exchange markets. One test (Cumby and Obstfeld (1981)) showed that, in the case of six industrial countries, the relationship does not hold, that is, assets denominated in different currencies cannot be taken as perfect substitutes.
The purpose of this study is to test the Fisher “open” hypothesis for the case of Argentina, a country that has undergone a large measure of financial market liberalization since 1976, with considerable movements of capital having taken place in the past four years, but a country in which an organized forward market has not yet developed.6
I. The Argentine Experience
The various stabilization plans implemented in Argentina since 1976 have involved increased reliance on market forces in several areas of economic activity. The financial reform completed by mid-1977 entailed the full liberalization of interest rates and the elimination of the highly centralized banking arrangements, which were preventing the growth of domestic financial intermediation.7 Exchange controls were eliminated, barriers to free capital movements were largely reduced,8 and exchange rate policies changed over the period. Until May 1978, a crawling-peg system was followed, and the exchange rate for the peso was adjusted by the central bank at relatively short intervals in line with the differential between domestic and foreign inflation. After May 1978, the exchange rate was allowed to float, with a small degree of intervention. In December 1978, a new system was implemented. As part of a complete package of measures, the authorities preannounced the future path of the exchange rate by publishing a schedule of daily rates for the subsequent eight months. The schedule was extended for additional periods until its abandonment in June 1981. Except for some slight departures from the announcement in late 1980 and two large devaluations in February and March 1981, the authorities followed closely the previously announced policy.9 In June 1981, the foreign exchange market was split into a commercial market and a financial market. The commercial market was reserved for trade operations and the exchange rate was determined by the central bank, while in the financial market the exchange rate was, in principle, to be determined by the free play of supply and demand.
Given the swings in exchange rate policies and the uncertainties related to the changes in the characteristics of the various stabilization programs adopted, it would not be surprising if the Fisher “open” hypothesis failed to hold in Argentina. As in Cumby and Obstfeld (1981), the test of this hypothesis is conducted as follows. Let it be the domestic nominal interest rate for the period starting at t and it* the equivalent foreign rate. If xt is the logarithm of the rate of exchange at time t and
If the realized exchange rate at t + 1 is expressed as the sum of its expected value at t plus the prediction error (i.e.,
where Dxt+1 is the actual rate of change in the spot rate over the maturity period, that is, Dxt+1 = xt+1 - xt. A test of the Fisher hypothesis can be conducted by calculating equation (2) and analyzing the time-series properties of δt,. If the foreign exchange market is weakly efficient10 and expectations are therefore formed on the basis of all the available information, including past forecast errors, and if there are no significant anticipations of a radical change in exchange rate policies, then δ should be white noise, that is, it should not display any pattern of serial correlation.11 If, however, the Fisher hypothesis does not hold in its simple form because domestic and foreign financial instruments are not perfect substitutes, the difference between interest rate differentials and the realized change in the exchange rate will include, in addition to the forecast error δt, a risk premium θt, to account for exchange, sovereign, and political risk. Equation (2) then becomes
where kt:, = δt, + θt. Moreover, additional elements reflecting transactions costs, regulations, and so forth would be included in kt. If these elements, as well as θt, display a high degree of persistence, kt may be serially correlated despite the informational efficiency of the foreign exchange market.
The value of kt for Argentina is calculated for two alternative time periods: (a) June 1977 to August 198112 and (b) December 1978 to January 1981, the period for which the preannounced exchange rate system was in effect.13 The domestic interest rate, it, is the 30-day deposit rate, and i* is the 30-day Eurodollar rate. The exchange rate used is the closing selling rate. All the interest and exchange rates are as of the last trading day of the month.14 An alternative series of kt is calculated by adjusting it for reserve requirements to allow for a more appropriate comparison of the cost of funds to financial intermediaries.15
Using the Box-Jenkins method for analysis of univariate time series, the autocorrelation function is estimated for the two alternative series of kt. The Ljung-Box adjusted version of the Box-Pierce test for the smallness of the whole set of the sample auto correlation of lags 1 through n was computed.16 The adjusted Q-statistics reported below were computed with 12 lags for the complete sample, and therefore distributed as χ2 with 12 degrees of freedom, and with 6 lags for the preannounced period. The obtained results are the following (significance level is in parentheses)17
| i | i (adjusted) | |
| June 1977-August 1981 | 4.35 | 4.44 |
| (0.95) | (0.95) | |
| December 1978-January 1981 | 2.11 | 2.02 |
| (0.90) | (0.90) |
| i | i (adjusted) | |
| June 1977-August 1981 | 4.35 | 4.44 |
| (0.95) | (0.95) | |
| December 1978-January 1981 | 2.11 | 2.02 |
| (0.90) | (0.90) |
The autocorrelation functions are reported in Table 1. For both sample lengths, the means of k and their standard deviations are (in percentage terms)
Autocorrelation Function of Interest Differentials
Autocorrelation Function of Interest Differentials
| Lag: | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| June 1977-August 1981 | ||||||||||||||
| i | Coefficient | −0.01 | 0.03 | −0.07 | −0.07 | −0.03 | −0.02 | −0.03 | −0.02 | −0.02 | −0.03 | −0.02 | −0.00 | |
| Standard error | 0.14 | 0.14 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | ||
| i’ | Coefficient | −0.01 | 0.25 | −0.08 | 0.07 | −0.03 | −0.02 | −0.03 | −0.02 | −0.02 | −0.03 | −0.02 | −0.00 | |
| Standard error | 0.14 | 0.14 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | ||
| December 1978-January 1981 | ||||||||||||||
| i | Coefficient | 0.11 | 0.03 | −0.03 | −0.05 | −0.08 | −0.24 | |||||||
| Standard error | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | ||||||||
| i’ | Coefficient | 0.10 | 0.02 | −0.04 | −0.05 | −0.05 | −0.24 | |||||||
| Standard error | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | ||||||||
Autocorrelation Function of Interest Differentials
| Lag: | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | ||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| June 1977-August 1981 | ||||||||||||||
| i | Coefficient | −0.01 | 0.03 | −0.07 | −0.07 | −0.03 | −0.02 | −0.03 | −0.02 | −0.02 | −0.03 | −0.02 | −0.00 | |
| Standard error | 0.14 | 0.14 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | ||
| i’ | Coefficient | −0.01 | 0.25 | −0.08 | 0.07 | −0.03 | −0.02 | −0.03 | −0.02 | −0.02 | −0.03 | −0.02 | −0.00 | |
| Standard error | 0.14 | 0.14 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | 0.15 | ||
| December 1978-January 1981 | ||||||||||||||
| i | Coefficient | 0.11 | 0.03 | −0.03 | −0.05 | −0.08 | −0.24 | |||||||
| Standard error | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | ||||||||
| i’ | Coefficient | 0.10 | 0.02 | −0.04 | −0.05 | −0.05 | −0.24 | |||||||
| Standard error | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | 0.20 | ||||||||
| i | i (adjusted) | |
| June 1977-August 1981 | 0.093 | 0.128 |
| (9.666) | (9.729) | |
| December 1978-January 1981 | 1.604 | 1.566 |
| (1.800) | (1.808) |
| i | i (adjusted) | |
| June 1977-August 1981 | 0.093 | 0.128 |
| (9.666) | (9.729) | |
| December 1978-January 1981 | 1.604 | 1.566 |
| (1.800) | (1.808) |
The results clearly indicate that the hypothesis that the observed interest rate differentials are white noise cannot be rejected at the 5 per cent level for the complete period and at the 90 per cent level for the preannouncement period.
These results are quite surprising, since they do not allow rejection of the joint hypothesis of efficiency and perfect substitutability of assets.18 It is an intuitive tenet, however, that, in addition to institutional and organizational constraints specific to the Argentine financial structure, a substantial risk premium has been an important component of interest rate differentials following the opening of the capital market. Exchange risk was certainly not eliminated by the policy of exchange rate preannouncements, since the time horizons of the various schedules were not always well defined and the central bank refused to sell forward contracts based on the announced future rates. Given the increasing overvaluation of the Argentine currency, there was always a positive, nontrivial, probability that a large discrete devaluation would take place despite the preannounced schedule. This was the well-known “peso problem.”19
Since it would not be realistic to rule out the existence of a risk premium and postulate perfect substitutability, a possible rationalization for the result is that the Argentine market is indeed informationally efficient and that the risk premium appears to be time varying and serially uncorrelated. An implication is that interest rates have responded quickly to “news,” that is, new information that appears randomly in the market.20
II. Summary
Considering the Argentine financial market after its liberalization, it has been shown that observed nominal interest differentials, adjusted for exchange rate changes, do not display any pattern of serial correlation, and therefore behave as forecast errors in an efficient market. A sensible interpretation of this result is that, although domestic and foreign assets are not likely to be perfect substitutes, the Argentine financial market is informationally efficient and has been characterized by an uncorrelated, time-varying risk premium on domestic assets.
REFERENCES
Aliber, Robert Z., “The Interest Rate Parity Theory: A Reinterpretation,” Journal of Political Economy, Vol. 81 (November/December 1973), pp. 1451–59.
Aliber, Robert Z., “Monetary Independence Under Floating Exchange Rates,” Journal of Finance, Vol. 30 (May 1975), pp. 365–76.
Blejer, Mario I., and Mohsin S. Khan, “Foreign Exchange Market Regularities in a Developing Economy,” Economics Letters, Vol. 6, No. 3 (1980), pp. 279–86.
Blejer, Mario I., and Donald J. Mathieson, “The Preannouncement of Exchange Rate Changes as a Stabilization Instrument,” Staff Papers, Vol. 28 (December 1981), pp. 760–92.
Buiter, Willem H., “Implications for the Adjustment Process of International Asset Risks: Exchange Controls, Intervention and Policy Risk, and Sovereign Risk,” Working Paper No. 516, National Bureau of Economic Research (Cambridge, Mass., July 1980).
Cumby, Robert E., and Maurice Obstfeld, “A Note on Exchange-Rate Expectations and Nominal Interest Differentials: A Test of the Fisher Hypothesis,” Journal of Finance, Vol. 36 (June 1981), pp. 697–703.
Cumby, Robert E., and Maurice Obstfeld, “International Interest-Rate and Price-Level Linkages under Flexible Exchange Rates: A Review of Recent Evidence,” in Exchange Rates: Theory and Practice, ed. by John F. O. Bilson and Richard C. Marston (University of Chicago Press, forthcoming in 1983).
Dooley, Michael P., and Peter Isard, “Capital Controls, Political Risk, and Deviations from Interest-Rate Parity,” Journal of Political Economy, Vol. 88 (April 1980), pp. 370–84.
Fama, Eugene F., “Efficient Capital Markets: A Review of Theory and Empirical Work,” Journal of Finance, Vol. 35 (May 1970), pp. 383–417.
Fernandez, Roque B., “Consideraciones ex-post sobre el Plan Economico de Martinez de Hoz,” Centro de Estudios Macroeconomicos de Argentina (Buenos Aires, 1981).
Frankel, Jeffrey A., “The Diversifiability of Exchange Risk,” Journal of International Economics, Vol. 9 (August 1979), pp. 379–93.
Frenkel, Jacob A., and Assaf Razin, “Stochastic Prices and Tests of Efficiency of Foreign Exchange Markets,” Economics Letters, Vol. 6, No. 2 (1980), pp. 165–70.
Frenkel, Jacob A., and Richard M. Levich, “Covered Interest Arbitrage: Un-exploited Profits?” Journal of Political Economy, Vol. 83 (April 1975), pp. 325–38.
Frenkel, Jacob A., and Richard M. Levich, “Transaction Costs and Interest Arbitrage: Tranquil versus Turbulent Periods,” Journal of Political Economy, Vol. 85 (December 1977), pp. 1209–26.
Grauer, Frederick L. A., Robert H. Litzenberger, and Richard E. Stehle, “Sharing Rules and Equilibrium in an International Capital Market Under Uncertainty,” Journal of Financial Economics, Vol. 3 (June 1976), pp. 233–56.
Hansen, Lars Peter, and Robert J. Hodrick, “Forward Exchange Rates as Optimal Predictors of Future Spot Rates: An Econometric Analysis,” Journal of Political Economy, Vol. 88 (October 1980), pp. 829–53.
Krasker, William S., “The ‘Peso Problem’ in Testing Efficiency of Forward Exchange Markets,” Journal of Monetary Economics, Vol. 6 (April 1980), pp. 269–76.
Levi, Maurice D., “Taxation and ‘Abnormal’ International Capital Flows,” Journal of Political Economy, Vol. 85 (June 1977), pp. 635–46.
Lizondo, José S., “Interest Differential and Covered Arbitrage,” Conference Paper No. 137, National Bureau of Economic Research (Cambridge, Mass., 1981).
Ljung, G. M., and G. E. P. Box, “On a Measure of Lack of Fit in Time Series Models,” Biometrika, Vol. 65 (August 1978), pp. 297–303.
Mutén, Leif, “Some Topical Issues Concerning International Double Taxation” (unpublished, International Monetary Fund, January 13, 1982).
Tanzi, Vito, and Mario I. Blejer, “Inflation, Interest Rate Policy and Currency Substitution in Open Economics: A Discussion of Some Major Issues.” It is scheduled to be published in World Development in 1982.
Mr. Blejer, economist in the Financial Studies Division of the Research Department, is a graduate of the Hebrew University of Jerusalem and of the University of Chicago. He has been on the faculty of the Hebrew University and of Boston University.
See Aliber (1973, 1975), Dooley and Isard (1980), and Levi (1977). A useful survey of the issues related to the absence of perfect substitutability of assets across countries and a careful analysis of its implications is presented in Buiter (1980). For additional complications arising in developing countries, see Tanzi and Blejer (1982). While most of the studies on covered interest arbitrage involve the major industrial countries, there have been studies for Singapore and Mexico; see Blejer and Khan (1980) and Lizondo (1981), respectively.
In fact, in the presence of the factors discussed above, similar inequalities in nominal interest rates will arise in a world of fixed exchange rates that are expected to remain constant or, in the absence of forward coverage, when exchange risk is fully diversified (see Frankel (1979)).
On the issue of risk premium in the forward exchange market, see Grauer, Litzenberger, and Stehle (1976) and Hansen and Hodrick (1980). Frenkel and Razin (1980) have shown that the neglect of risk premium considerations in efficiency tests of the foreign exchange markets is of little empirical significance.
In most countries, accrued capital (including exchange) gains, or even realized ones, are not taxed, or the taxes are very easily evaded. On the other hand, nominal incomes arising from interest payments are generally taxable if generated domestically but are often de facto exempt when generated outside the country. The differential taxation issue arises from the conflicting rules applied by different countries: the rule of source (claims to tax all income generated from sources located in the country) and the rule of residence (claims to tax all income of residents of the country regardless of the location of the source of income). On issues concerning international taxation practices, see Mutén (1982).
In other words, in the presence of exchange risk, assets that are identical except for their currency denomination will not be considered perfect substitutes.
A forward exchange market was in operation in Argentina during a limited period in the mid-1960s, but its activities were discontinued until July 6, 1978, when the central bank again authorized its operations. It never, however, developed to any significant size, owing to the restrictions imposed on its operations (forward transactions were limited to the hedging of dollar-denominated imports by privately owned companies). In 1979, the scope was widened to include a few financial operations, such as permitting firms to hedge balance-sheet exposure and loans. Although about US$100 million was transacted in that market in 1978, virtually no operations took place in 1980. In December 1981, the central bank initiated a process directed toward the reactivation of the market, and some operations were registered that year.
Before the Financial Entities Law was adopted in June 1977, a system of centralized deposits was in effect based on the imposition of 100 per cent reserve requirements. The new law enabled financial intermediaries to receive deposits on their own account. Reserve requirements were set at 45 per cent for all deposit liabilities. Financial intermediaries were allowed to allocate credit freely among economic sectors and to set interest rates in accordance with market conditions.
The restrictions to free capital mobility varied over the period. During a short period in 1978 there was a compulsory 15 per cent deposit requirement on foreign borrowing, and for most of the period some minimum terms for foreign borrowing were in effect. However, the limitations were easily avoidable, and they did not effectively prevent the integration of domestic and foreign capital markets.
For an analysis of the characteristics and implications of the system of preannounced exchange rates, see Blejer and Mathieson (1981).
Weak efficiency is used here in the sense of Fama (1970) and implies that past information on the variable analyzed cannot be used to improve its forecast.
The white noise test is therefore a joint test of informational efficiency and of the Fisher proposition.
Following the implementation, in June 1977, of the new Financial Entities Law, which assured free market determination of domestic interest rates.
Testing the hypothesis that the nature of exchange risk may have changed following the announcement of the exchange rate schedule.
The data sources are as follows: for it, Central Bank of Argentina; for it* Data Resources Incorporated, Databank; and for Dxt+1 International Monetary Fund, Datafund.
The alternative series substitute
The distribution of the Box-Pierce Q-statistic can deviate from χ2 in small samples. Ljung and Box (1978) propose a correction to account for the discrepancies. The adjusted statistic,
The critical values of the χ2(12) are 5.23 (0.95) and 6.30 (0.90); for χ2(6), the critical values are 1.64 (0.95) and 2.20 (0.90).
The result is particularly surprising in view of the results obtained by Cumby and Obstfeld (1981) for industrial countries.
On the “peso problem” and its empirical implications, see Krasker (1980) and Cumby and Obstfeld (1983). The “peso problem” tends to introduce a mean in the observed differential and may also introduce positive autocorrelation in the forecast errors when the discrete devaluation does not actually take place. This is not a necessary result if the perceived probability of devaluation fluctuates randomly, a conclusion that may be hinted at by the results obtained here.
Some evidence on the rapid reaction of interest rates to current events (including unexpected changes in the exchange rate schedules) is provided by Fernandez (1981).