Article

Capital Controls and Interest Rate Parity: The Japanese Experience, 1978–81

Author(s):
International Monetary Fund. Research Dept.
Published Date:
January 1981
Share
  • ShareShare
Show Summary Details

Over the past several years, the exchange rate for the Japanese yen has fluctuated widely, as has the intensity of capital controls. At times when the yen rate was believed to have “overshot” some unspecified appropriate level, the Japanese Government introduced measures to influence capital flows in an attempt to narrow the amplitude of exchange rate fluctuations. For example, in early 1978, when an appreciation of the yen was widely expected, the Government introduced capital controls to reduce net inflows and to moderate the yen appreciation; in 1979 and early 1980, when the Government thought that the yen had weakened too rapidly, measures aimed at inducing net inflows were introduced. However, since December 1980, when the Government implemented a law embodying the principle that all transactions in foreign currency may be carried out freely unless expressly restricted, capital movements have, in principle, been virtually free of controls. These policy changes are frequently credited with having had an important impact on (a) measured deviations from interest rate parity, (b) the exchange rate, and (c) the efficiency of the foreign exchange market.

The main purpose of this paper is to examine the extent to which these capital controls or the absence of them have influenced deviations from interest rate parity during the period 1978–81. An attempt is also made to measure the impact of such capital controls on distortions in the foreign exchange-cum-securities market during that period. Questions regarding the effectiveness of capital controls on influencing the exchange rate and the impact of these controls on the market efficiency will not be dealt with in this paper.

The approach adopted builds upon work by Frenkel and Levich (1975, 1977).1 Utilizing basic price theory, it can be shown that the introduction of transaction costs into the Marshallian demand and supply apparatus has the effect of creating a band around the demand curve. Even though the realized price may deviate from the equilibrium price (which would prevail without transaction costs), the market would still be in equilibrium in the sense that there would be no unexploited profits.2 Similarly, the introduction of transaction costs into the foreign exchange and securities markets will create a band (neutral band in the Frenkel-Levich terminology) around the interest rate parity line. It can be shown that all covered interest differentials falling within this band are equilibrium points.3 The introduction of capital controls will increase the cost of transactions to market participants and consequently increase the width of the “neutral band.” The comparison of two such bands, where one reflects transactions in a market with free movements of capital while the other reflects controls on such movements, will provide a quantitative indicator of the effect of capital controls.

The imposition of controls may also have the effect of changing the expectations of market participants regarding possible capital controls in the future. This will determine, in part, whether the burden of adjustment of the forward premium (if any) owing to the imposition of controls falls on the spot exchange rate or the forward exchange rate or both. A comparison of the changes in the width of the band (owing to capital controls) with changes in interest rates (on domestic and foreign assets) and the forward premium shows the extent to which capital controls have been successful in affecting the exchange rates. This issue, although important, is not pursued in this paper.

The remainder of this paper is organized as follows. Section I presents an overview of the issues involved in discussing the relationships between capital controls and measured deviations from interest rate parity. Section II briefly describes the methodology that is to be used to quantify the influences exerted by capital controls in foreign exchange markets, Section III presents empirical results obtained from daily observation of exchange rates and interest rates in the Tokyo and London foreign exchange markets for the period 1978–81. Section IV summarizes the major findings and presents concluding remarks. Appendix I provides a technical note on selecting an appropriate data set of observations necessary for the estimation of the neutral band. Appendix II contains data descriptions, sources, and definitions.

I. An Overview of the Issues.

Consider two worlds, A and B. In each world, assets may be denominated in different currencies. The distinguishing characteristics of World A are that all assets are traded freely, and there are no institutional controls, taxes, tariffs, or quotas on transactions by residents of different regions. In contrast, in World B, the government of one region imposes controls on transactions between its residents and those of other regions. The exchange markets that we observe today are characterized by elements of both Worlds A and B. In addition, in each country, there is a unique combination of the taxes, tariffs, and quotas employed, and within each country, the tax/tariff/quota structure is applied differently to goods, factors, and securities.

Suppose that the portfolio of the economic agent is initially invested in domestic securities. The agent can either continue to hold assets denominated in domestic currency and earn income at the domestic interest rate (rd or transfer funds abroad to hold assets denominated in foreign currency and earn a return at the foreign rate of interest (rf). Given an interest rate differential, if the forward exchange rate (F) (number of domestic currency units per unit of foreign currency) is high relative to the spot rate (S)—that is, if there is a covered interest rate differential in favor of foreign assets—agents will purchase foreign exchange at the spot rate, invest in foreign securities, earn income at the foreign interest rate, and buy a forward contract to convert back to domestic currency at the forward rate. Since all agents possess the same information set, arbitrage implies that if one agent can make profits then all can. However, if all attempt to make profits simultaneously, then no one can. Something in the system must change. Either the forward premium will adjust or the foreign interest rate differential will change to bring the system into equilibrium.

Much of the theoretical literature formalizing the relationship between the asset and exchange markets under perfect competition has come under the heading of interest rate parity theory and posits that in World A the current money market interest differential on assets denominated in different currencies is uniquely related to differentials between spot and forward exchange rates. Formally, in the absence of any distortions, this relationship is expressed as

or

where

S=number of units of domestic currency per unit of foreign currency for immediate delivery (say, at time t0
F=number of units of domestic currency per unit of foreign currency for delivery at some time (t1) in the future (t0 <t1)
rf=foreign interest rate that can be secured for the period from t0 to t1
rd=domestic interest rate that can be secured for the period from t0 to t1

Thus, in World A, where no distortions exist (i.e., free capital mobility and flexible exchange rates in the context here), the interest rate differential between interest-bearing domestic and foreign assets will identically determine the differential between the forward and the spot exchange rates. Put differently, the forward premium or discount is completely determined by the current interest rate differential.

As will be evident to the reader, it is not surprising that conditions for interest rate parity may be found not to hold with strict equality in any empirical exercise designed to verify that proposition. In an illuminating survey, Officer and Willett (1970) identified two causes for deviations from interest rate parity. One is transaction costs and the other is a positively sloped (i.e., a less than perfectly elastic) “medium”-run or “long”-run supply curve of funds for arbitrage. Transaction costs depend on the differential costs of trading across different currencies and include not only brokerage costs but also information and time costs. One would expect the brokerage cost to vary with the volume of transactions, while the information cost would depend on the number of outlets searched and the marginal benefit from searching. The time cost will vary with the wage/shadow-wage rate or the marginal disutility of work.

Another significant contribution pertaining to the empirical verification of the interest rate parity theory was made by Aliber (1973). He argued that assets denominated in different currencies and issued in different countries carry with them two additional risks. First, different currencies imply that an economic agent, when formulating his portfolio, will have to incorporate the possibility of a change in the exchange rate. Second, assets issued in different countries carry with them different political risks. Differences in political risk can be thought of as arising out of different tax/tariff structures or capital controls. Political risk will not only reflect the existing structure of controls but also the markets’ perception of future government actions based on past and current information. Aliber concluded that, for noncomparable assets (i.e., assets issued in different political jurisdictions), political risk accounts for much of the observed differential.

The above discussion suggests that the relationship presented by equation (1) would not hold exactly in a world where transaction costs exist. Typically, for every transaction, there is a positive transaction cost associated with it, whether it is a transaction between securities in the same currency or across different currencies. Furthermore, the presence of controls will increase transaction costs.4

The role of transaction costs in the foreign exchange and securities markets can be illustrated by the following simple example. Let us examine the sequence of transactions that an individual in Japan will need to consider in order to determine the portfolio that he should hold. Suppose the individual borrows yen-denominated funds amounting to x at the rate of interest rd. Let the per unit transaction cost associated with borrowing the funds be td. For this individual, the total cost of borrowing the yen-denominated funds, including the transaction costs, is x(1 + rd)(1 + td).5 This individual now sells yen at the prevailing spot exchange rate, 5, and pays in yen the per unit transaction cost, ts, for the sale in the spot market. Through such an exchange, the total amount of foreign currency (U.S. dollars) that the individual can obtain is x/S(1 + ts), which is equal to, say, y. These y dollars will now be invested in the dollar-denominated assets, which will yield a return of rf. Investment in the dollar-denominated assets carries with it the per unit transaction cost of tf, which must be paid in dollars. At the time of investing in the dollar-denominated assets, the individual also makes a forward contract to exchange y(1 + rf)/(1 + tf) dollars at the prevailing forward exchange rate, F, and pays the associated transaction cost of tF in dollars.6 Thus, the amount of yen the individual expects to get by investing in the dollar-denominated securities is y(1 + rf)F/(1 + tf)(1 + tF, or x.F(1 + rf)/S(1 + ts)(1 + tf)(1 + tF).7 The individual will continue to borrow the yen denominated funds and go through the sequence of the transactions outlined above as long as equation (3) is satisfied.

or

Under the circumstances governed by equation (4), capital moves out of the home country and into the foreign country.

Analogously, capital moves into the home country from abroad if equation (5) is satisfied.

Since each expression for costs associated with transactions in the foreign exchange and securities markets is a very small fraction, total costs (TC) can be approximated by (td + tf + ts + tF). The expression for capital flows seeking unexploited profits can be rewritten as

and

If no unexploited profits exist, the following condition must be satisfied:

where

or

Actual deviations from interest rate parity, therefore, are expected to be scattered around zero within a “neutral band” limited by the total transaction costs (TC), if no unexploited profits exist.

II. Methodology

To empirically study the effect of capital controls on transaction costs, or the deviations from the interest rate parity line, two empirical counterparts to World A and World B have been chosen.

World A is represented by London (Euro-currency market), where transactions are free of controls, taxes, and quotas so that capital is perfectly mobile. Two assets have been chosen: one denominated in Japanese yen (Euro-yen deposits) and another denominated in U. S. dollars (Euro-dollar deposits).8 It can be hypothesized that any observed deviation from the theoretical (and deterministic) relationship in equation (1), defined as the “gap in London” (GLON), is randomly distributed9 and has the expected value of zero; that is

where

Glon(1+rey)SLON(1+red)FLON1
red=interest rate on three-month Euro-dollar deposits
SLON=Japanese yen/U. S. dollar spot exchange rate in London
FLON=Japanese yen/U. S. dollar three-month forward exchange rate in London

World B is perceived as consisting of two locations—Tokyo and London. Since there has been a variety of controls on capital movements between the two cities, any observed deviation from equation (1), defined as the “gap in Tokyo” (GTKY) will not be randomly distributed around a mean of zero; that is,

where

GTKY=(1+rgk)STKY(1+red)FTKY1
rgk=interest rate on tnree-month Gensaki 10 bonds
STKY=Japanese yen/U. S. dollar spot exchange rate in Tokyo
FTKY=Japanese yen/U. S. dollar three-month forward exchange rate in Tokyo

To investigate the effect of controls, it is important to distinguish between (a) disparities owing to pure transaction costs and those owing to controls and (b) changes in disparity owing to the effect of a specific control promulgated by the government as opposed to a change in disparity reflecting a change in international economic conditions. Fortunately, GLON is, in the aggregate, a joint indicator of both pure transaction costs and the effects of the world economic situation. Thus, the difference between GTKY and GLON captures the net effect of capital controls. If the hypotheses advanced by equations (10) and (11) are supported by empirical data, then the following relationship can be used to quantify the effect of controls on covered interest rate differentials between Tokyo and London.

where

CTL=measure of the effect of controls on transactions between Tokyo and London
E=expectations operator

Since most of the deviations from interest rate parity can be explained by “transaction costs,” as defined earlier, CTL represents opportunity costs that market participants have to incur if they are to evade capital controls in order to make a profit from covered interest rate differentials.

III. Empirical Results

This section presents the empirical results of testing the hypotheses represented in equations (10) and (11) and quantifies the extent of capital controls represented by equation (12). The period under study runs from January 1978 to March 1981. Daily observations are used. The statistical results are followed by a brief summary of changes in capital controls that can explain the value of CTL.

HYPOTHESES TESTING

That the gap in London, as represented by equation (10), is randomly distributed narrowly around a mean of zero implies that there are negligible transaction costs in the London market and that if market participants were to trade on every non-zero value of the gap in London they would, on the average, make zero profits. Similarly, the hypothesis that the gap in Tokyo, as represented by equation (11), is not randomly distributed around a non-zero mean implies that if market participants were to trade on every non-zero value of the gap in Tokyo they may end up with non-zero profits. The hypothesis also implies that transaction costs (which account for the gap) are not negligible. However, the persistence of a significant non-zero gap would, a priori, indicate that the market is not free of controls. Thus, a reliable check of government intervention would be to examine basic statistics on gaps to determine whether or not they are randomly dispersed around a mean of zero.

Table 1 summarizes the means and t-statistics of the measured deviations from interest rate parity in the London and Tokyo markets. The results clearly support the hypothesis that the measured disparities are, by and large, randomly distributed around the mean of zero in London for the entire sample period as well as for most of the subperiods.11 The means in Tokyo were positive during the first five three-month periods ended March 1979, while they were negative during the following three-month periods ended March 1981. All of the estimated values of the means for Tokyo were significantly different from zero, except for the first quarter of 1981 (Charts 1 and 2).

Table 1.Japan: Mean, Standard Deviation, andt-Statistics of Deviations from Interest Rate Parity1
Deviations in LondonDeviations in Tokyo
PeriodMean2Standard deviation2t-statisticsMean2Standard deviation2t-statistics
1978
First quarter−0.0940.179−3.8530.8370.25623.053
Second quarter0.0130.1390.680.5740.21619.543
Third quarter−0.0050.208−0.170.6410.12438.683
Fourth quarter0.0020.2090.080.6150.22419.763
1979
First quarter−0.0290.185−1.190.3560.2938.593
Second quarter0.0560.2012.03−0.0740.073−7.403
Third quarter0.0060.1550.28−0.0600.109−4.093
Fourth quarter0.0670.1792.77−0.1840.093−14.193
1980
First quarter−0.0150.148−0.78−0.2680.271−6.923
Second quarter0.0150.2220.50−0.1360.185−5.403
Third quarter−0.0400.124−2.37−0.2860.113−18.283
Fourth quarter0.0460.1602.11−0.0820.097−5.943
1981
First quarter0.0540.1532.61−0.0050.094−0.40

Chart 1.Japan: Deviations from Interest Rate Parity in London, 1978–811

(In per cent)

1 Daily data; 100 GLON (defined as the “gap in London”).

The zone bounded by ± 0.146 represents the average width of the neutral band for the sample period as a whole.

Chart 2.Japan: Deviations from Interest Rate Parity in Tokyo, 1978–811

(In per cent)

1 Daily data; 100 GTKY (defined as the “gap in Tokyo”).

The zone bounded by ± 0.339 represents the average width of the neutral band for the sample period as a whole.

QUANTIFICATION OF CAPITAL CONTROLS

In the previous section, it was established that deviations from the interest rate parity in the London market are, for all practical purposes, randomly distributed around the mean of zero. Thus, it can be said that, for the period 1978–81, the London market closely resembled World A, as described earlier. As mentioned above, by comparing the expected absolute value of the covered interest rate differentials in the two markets, one should be able to capture distortions present in the Tokyo market relative to those in the London market.

A reasonable estimate of the expected absolute value of the covered interest rate differentials, as represented by equation (9), can be obtained in two steps. First, one must obtain a data set of uniformly distributed observations.12 This is necessitated by the strong possibility that some of the observed deviations from the interest rate parities seem to have extremely large or small values relative to most observations and to reflect measurement errors of large magnitudes. The empirical analysis suggests that the data set mentioned above can be obtained by eliminating the observations in any given quarter that belong to the set of the extreme 13 per cent of their absolute values when they are arranged in ascending order. Second, one must obtain the mean absolute value of the uniformly distributed sample in the data set for each market. The results are summarized in Table 2; these means are the estimated values of the transaction costs (or opportunity costs) defined earlier and represent the outer bound of the neutral band.

Table 2.Japan: Estimated Transaction Costs—Means of Absolute Deviations from Interest Rate Parity, 1978–81(In per cent)
Transaction CostsEffects of Capital Controls on Transaction Costs in Tokyo (B–A)
PeriodLondon (A)Tokyo (B)
1978
First quarter0.1610.8370.676
Second quarter0.1140.5740.460
Third quarter0.1640.6410.477
Fourth quarter0.1710.6150.444
1979
First quarter0.1440.3940.250
Second quarter0.1710.0930.078
Third quarter0.1240.1030.021
Fourth quarter0.1550.1850.030
1980
First quarter0.1200.2890.169
Second quarter0.1860.1990.013
Third quarter0.1130.2860.173
Fourth quarter0.1370.107−0.030
1981
First quarter0.1350.081−0.054
Average0.1460.3390.208

Charts 1, 2, and 3 and Tables 1 and 2 point to several interesting observations.

Chart 3.Japan: Measure of Distortions in Tokyo Market, 1978–811

(In per cent)

1 Daily data. These observations are constructed by the expression (GTKYGLON)100.

(1) The movements in the deviations show no trends in the London market but definite trends in the Tokyo market.

(2) The deviations in the Tokyo market were the largest in the first quarter of 1978, and their excess over the deviations in the London market was also the largest during that quarter. The difference in the mean value of absolute deviations is estimated to be about 0.7 percentage point in that quarter.13

(3) The deviations in the Tokyo market were significantly lower in the last three quarters of 1979 than during 1978 and the first quarter of 1979, and the gap between the London market and the Tokyo market is virtually nonexistent in that later period. This indicates that the Tokyo market was as free from capital controls in the April-December period of 1979 as was the London market.

(4) In the first quarter of 1980, the mean value of the deviation from the interest rate parity in Tokyo increased sharply from the level reached during the fourth quarter of 1979, suggesting a significant increase in distortion created by capital controls. (Note that the deviations in the London market remained virtually unchanged in the first quarter of 1980 compared with the previous quarter.)

(5) During the second quarter of 1980, the deviations in the London market increased considerably. This may have been due to increased uncertainty in the capital market following a change in U.S. monetary policy late in the first quarter, that is, the introduction of controls over credit extended to consumers and businesses. On the other hand, the deviations in the Tokyo market declined sharply in that quarter, implying that distortions were reduced significantly, even though both the Tokyo market and the London market were subject to the same uncertainty owing to the shift in U.S. monetary policy.

(6) In the third quarter of 1980, the mean value of the deviation from interest rate parity again increased considerably in Tokyo, indicating an increase in transaction costs. The increase reflected partly an introduction of capital controls; that is, deposit institutions were asked to exercise restraint in accepting foreign currency deposits resulting from sales of foreign currency assets.

(7) The month of December 1980 marks the implementation of the new Foreign Exchange and Foreign Trade Control Law. The new law provided that transactions in both the current and the capital account could be made freely unless expressly prohibited. Partly because of both the expectation that the new law would come into effect on December 1, 1980 and the actual implementation of the law, transaction costs in the Tokyo capital market declined significantly in the fourth quarter of 1980. These declined further in the first quarter of 1981. In both of these quarters, transaction costs were estimated to be lower in Tokyo than in London.

In short, it can be inferred that significant distortions were present in the Tokyo market for the five-quarter period ended March 1979 (Phase I). During the period April–December 1979 (Phase II), distortions were reduced considerably. During the first 11 months of 1980, temporary increases and decreases in distortions were observed. Distortions increased in the first and third quarters but declined in the subsequent quarters (Phase III). Since December 1980, distortions have remained low in Tokyo (Phase IV).

The implications drawn from this analysis closely correspond to changes in capital controls during the period under study.

Phase I (first quarter 1978–first quarter 1979). During this period, expectations of a yen appreciation were very strong, and a large amount of incipient capital was induced into the Tokyo market. Mindful of a further appreciation of the yen, the authorities introduced various measures to reduce capital inflows and to encourage outflows, particularly in the first quarter of 1978. As a result, the market condition for internationally traded yen-denominated assets became tight in Tokyo relative to London. Reflecting these market conditions, the Gensaki rate continued to be above the Euro-yen rate in London (Chart 4), even though there was a gradual easing of the restrictive measures on capital inflows from the second quarter of 1978.

Chart 4.Japan: Interest Rates on Yen-Denominated Assets, 1978–811

(In per cent per annum)

1 Daily data.

Since early November 1978, when the U.S. Government introduced a package of dollar defense measures, the yen started to depreciate rapidly in both the spot and forward exchange markets.

Phase II (second quarter-fourth quarter 1979). The measures in effect in Phase I that attempted to reduce capital inflows were removed and certain types of outflows were discouraged. As a result, capital moved more freely than before, and the interest rates on yen-denominated assets in Tokyo and London moved very closely. A deterioration in the external current account balance and increased uncertainty following the second oil shock caused a continued depreciation of the yen.

Phase III (first quarter–November 1980). A package of yen defense measures was introduced in early March 1980. These measures included a provision that allowed Japanese banks abroad to raise funds in London and transfer them to their head offices in Tokyo. In addition, the ceiling on the interest rate on the nonresident free yen account was removed. Owing to the effect of these measures, capital flowed into Japan. The interest rate on the Euro-yen deposits increased significantly beyond the Gensaki rate,14 reflecting capital inflows induced in part by the moral suasion exercised by the authorities in Tokyo. At the same time, the cost of transactions increased significantly, as mentioned earlier, owing mainly to increased uncertainty caused by changes in capital controls.

Phase IV (December 1980–first quarter 1981). In December 1980, the Government implemented the new Foreign Exchange and Foreign Trade Control Law. This law provided that many types of capital transaction would be free under normal circumstances; others would require prior notice. For a few types of transaction, advance approval is still required. On balance, capital controls have been reduced considerably, and the reduction in these controls has in turn reduced the transaction costs of short-term securities.

IV. Concluding Remarks

The main purpose of this paper has been to examine how capital controls affected deviations from interest rate parity during the period 1978–81. An attempt was also made to devise a quantitative indicator that can measure the extent to which capital controls create distortions in the foreign exchange market.

The indicator of distortions caused by capital controls shows that, since the beginning of the period under study (the first quarter of 1978), distortions in the market were on a generally declining trend. By mid-1979, the Tokyo market was as free from distortions as was the London market. However, in the first quarter of 1980, the distortions increased significantly, reflecting the moral suasion exercised by the authorities to encourage Japanese residents to bring in funds from abroad. However, the distortions declined considerably thereafter, possibly because moral suasion ceased to be applied, and has remained low since then, except for a brief period in the third quarter of 1980. Data also indicate that, in recent months, deviations from interest rate parity have been reduced to levels observed in the second half of 1979. The relaxation of capital controls following the implementation of the new Foreign Exchange and Foreign Trade Control Law in December 1980 contributed further to reducing transaction costs in the short-term securities market in Tokyo.

APPENDICES
I. Obtaining Uniformly Distributed Observations: An Example

The purpose of this appendix is to describe how to distinguish observations that are uniformly distributed from those that are not and to obtain a set of uniformly distributed observations that are useful in estimating the expected absolute value of the covered interest rate differentials (Chart 5, top panel). For this purpose, we have used, as an example, a set of observations in the London market, for the fourth quarter of 1979, and arranged data points (i.e., the absolute value of covered interest rate differentials (in per cent)) in ascending order (Chart 5, middle panel).

Chart 5.Japan: Deviations from Interest Rate Parity, Distribution and Mean (An Example)

(In London market, fourth quarter 1979)

1 The mean of the observation points represented by the segment B.

The graph in the middle panel in general exhibits the following pattern:

(a) A relatively few points in the lower end of the range (i.e., 0.0–0.05) of values exhibit relatively small differences between their magnitude. These data points are represented by a segment, A, in the middle panel.

(b) A few points in the high end of the range (i.e., 0.25–0.8) of values exhibit relatively large differences between their magnitude. These data points are represented by a segment, C, in the middle panel.

(c) The majority of points (about 70 per cent) lie in the range 0.05–0.35, which is significantly smaller than the range for the entire set of observation points (i.e., from 0.001–0.8), and the magnitude of these observations in the range between 0.05 and 0.35, by and large, increases linearly. These points are represented by a segment, B, and thus these data points are approximately uniformly distributed.

From these observations, one can note that some data values are clearly separated from the majority, and the extreme observation points must be eliminated so as to obtain a uniformly distributed observation data set. For this example, the 12 observations represented by the segments A and C are removed; as a result, the remaining data points constitute a set of uniformly distributed observations. The mean of the observation points represented by the segment B is represented by D in the bottom panel. It is also observed that, as the extreme points are eliminated successively, the decline in the mean values is initially large, but after a certain number of iterations, the decline becomes minimal, and one observes some indication of a kink in the mean value. Thus, by observing this kink, one has additional information indicating the approximate number of points to be eliminated for each period. The results obtained by applying these methods—finding kinks both in the density function and the curve representing the mean absolute value of deviations—to each quarterly sample period are shown in Table 3.

Table 3.Number and Percentage Share of Observations Eliminated, 1978–81
Observations Eliminated
London marketTokyo market
PeriodTotal numberPercentage shareTotal numberPercentage share
1978
First quarter46
Second quarter4623
Third quarter81346
Fourth quarter813
1979
First quarter81347
Second quarter101647
Third quarter81346
Fourth quarter1220
1980
First quarter610
Second quarter101623
Third quarter813
Fourth quarter610814
1981
First quarter610611

In the London market, the approximate location of the kink in the curve depicting the mean absolute value of the deviations from the interest rate parity was observed in the first quarter of 1978, when four extreme observations, or 6 per cent of the observations, were eliminated. In the second quarter of 1979 and 1980, the location of such a kink was obtained when ten extreme observations, or 16 per cent of the observations, were eliminated. In the Tokyo market, such a kink was observed when only a few observations were removed during most of the sample periods. Therefore, the percentage of the observations that are to be eliminated differs from one quarter to another, but these criteria are by no means clear cut. Because of both an element of uncertainty concerning the definitiveness of the criteria and for the sake of computational ease, without impairing the quality of the uniform distribution of the data set, it was decided to use a single value to decide the cutoff points for both series for all periods. It was decided that 13 per cent of the highest and the lowest observations were to be eliminated from each series. A higher-than-average value was chosen for the cutoff point because the change in mean owing to eliminating extra points in a series with less extreme points is smaller than eliminating too few points in a series with many extreme values.

The results of the mean absolute values after eliminating 13 per cent of the extreme observations are reported in Table 2 in the text.

II. Description of Data

No single source was available from which data on all relevant variables could be retrieved. In fact, the sources were so diverse that simultaneous observation of all variables was not achieved. Fully aware of McCormick’s (1979) criticism of Frenkel and Levich and of the effects that nonsimultaneous observations have on the estimation of transaction costs, we have assembled a highly reliable data set on the Japanese economy. Researchers in this field are warned against using much of the published data on cross exchange rates for many countries (including Japan), because one exchange rate is observed and the cross exchange rates are calculated through “conversion” rather than through direct observation. Under such circumstances, interest rate parity is identically “satisfied,” the Frenkel-Levich transaction costs are by definition identically zero, and the capital market is efficient in a trivially perverse sense. Researchers are thus advised to investigate the method by which data are collected.

Spot exchange rates

(SLON) refers to the yen/dollar spot exchange rate. This series is the midpoint of the extreme values observed during the London trading day. In effect, they correspond to the midpoint of the lowest bid and the highest “ask” values during the day. The extreme values were collected from various issues of Montagu Monthly Review (London).

(STKY) refers to the yen/dollar spot exchange rate, observed at the close of the day in the Tokyo foreign exchange market. This series was compiled from various issues of Nihon Keizai Shimbun (Tokyo). This series was crosschecked for accuracy with the daily issues of the Fund’s “Morning Press.”

Forward exchange rates

(FTKY) refers to the three-month yen/dollar forward exchange rate, observed at the close of business in the Tokyo foreign exchange market. This series, which is the midpoint of the ask and bid quotations, was obtained from various issues of Nihon Keizai Shimbun (Tokyo).

(FLON) refers to the three-month yen/dollar forward exchange rate. The ask and bid quotations were reported by Data Resources Incorporated. Their source was the Bank of America (San Francisco), and their quotations reflect a consensus of several interbank dealers at the opening of the New York foreign exchange market. The values reported in this series are the midpoints of the ask and bid quotations.

Interest rates

red refers to the interest rate on three-month Euro-dollar deposits. This series is the midpoint of the bid and ask quotations reported by Data Resources Incorporated. The ask and bid quotations reflect the prevailing market levels in the London market as perceived by Reuters at the New York opening time.

rey refers to the interest rate on three-month Euro-yen deposits. No single source was available for this data set. For 1978, data were collected from the Foreign Exchange Yearbook, 1979 Edition, ed. by Trevor Underwood (New York and Cambridge, England, 1979). For the period January 1, 1979 to September 23, 1979, the Treasurer’s Department of the Fund collected data on the bid rate. This series was adjusted by adding the mean of the difference between the bid-offered rate from September 24, 1979 to March 30, 1981 to obtain a proxy for the Euro-yen rate during that period. For September 24, 1979 to March 30, 1981, Data Resources Incorporated published the ask and bid quotations, and the subsequent values correspond to their midpoints.

rgk refers to the three-month interest rates on Gensaki in the Tokyo securities market. The series reported here reflects the average of the rates quoted during a day and has been obtained from various issues of Nihon Keizai Shimbun (Tokyo).

REFERENCES

SUMMARIES

A General-Equilibrium Approach to the Analysis of Monetary and Fiscal Policiesandrew feltenstein (pages 653–81)

A major task confronting policymakers in recent years has been to predict the effects of changes in the various parameters under their control. When these changes are small—for instance, a slight shift in a tax rate—partial-equilibrium models have normally been used for prediction, on the assumption that the macroeconomic effect of the proposed changes are small enough to be disregarded. Often, however, there are parameter changes that are intended specifically to bring about structural shifts in the economy—for example, a trade and exchange liberalization under an extended Fund facility program. In such cases, the assumption of insignificant aggregate effects becomes invalid, and a general-equilibrium rather than a partial-equilibrium model must be used.

This paper constructs a general-equilibrium model of an open economy and develops a computational technique for deriving a market-clearing solution to the model. The model, which incorporates an input/output framework, allows for disaggregated commodities, taxes, and tariffs, so that the individual parameter changes often considered by a government can be examined. The government in this model is treated as a producer of public goods that pays for its purchases of factor inputs from the tariffs and taxes (income, profits, and sales) that it collects. Since these taxes will not necessarily be sufficient to finance the cost of the public goods, the government issues a combination of money and bonds to finance whatever deficit it incurs, so that the rates of change of both the nominal money supply and the level of public indebtedness are endogenous. This endogenicity is an advance over current macroeconomic models, which have provisions for disaggregated commodities but also require the money supply to be exogenous, thus precluding any investigation of the impact of government behavior on the financial sector. The model also contains foreign consumers and foreign bonds and can therefore generate endogenous capital flows.

The Divergence Indicator: A Technical Notejoanne salop (pages 682–97)

This paper investigates the divergence indicator used in the European Monetary System (EMS) to see how its actual specification affects the identification of divergent currencies under various circumstances. Two principal questions are examined: (1) Does the design of the divergence indicator lead to unbiased treatment of all member currencies participating in the exchange rate mechanism of the EMS? (2) How does the design of the divergence indicator further the EMS objective of greater “economic alignment” among participating countries? To address these questions, a measure is constructed that can be used to make intercountry comparisons of the degree of constraint embodied in the indicator. The idea behind the measure is to posit a specified percentage deviation from parity between two member currencies and to calculate, for each of these currencies, how large a departure from parity vis-à-vis third currencies would be required for its divergence indicator to be triggered.

According to the analysis, relatively more disparate third-currency movement is tolerated for large countries than for small. Specifically, there exist configurations of deviations from central rates that would trigger the divergence indicator of the small country if the deviations were vis-à-vis its currency, and that would not trigger the divergence indicator of the large country if percentage deviations of the same magnitude prevailed vis-à-vis its currency. Nevertheless, there are situations—if all member currencies deviate from parity with the currency by the same critical percentage—under which the divergence indicator would be triggered for a country regardless of its size. Furthermore, the design of the divergence indicator is less biased in favor of large countries than it might have been, with the remaining bias being a natural consequence of the differences in the weights of EMS participants.

The analysis also established the existence of a preference for coalitions of currencies in the following sense: For two or more currencies moving in tandem vis-à-vis another currency or group of currencies, greater third-currency deviation from parity is tolerated than if the coalition were a single country whose weight equaled the sum of the weights of the coalition members. Moreover, if the divergence indicator is to be triggered for any coalition member, it will be for that of the largest country. However, once this country has taken such adjustment measures as are required and its currency value moves closer to parity with the other coalition, the divergence indicator of one or more of its former allies may ring.

Real Responses Associated with Exchange Rate Action in Selected Upper Credit Tranche Stabilization Programsdonal j. donovan (pages 698–727)

The paper examines certain aspects of a selected set of 12 upper credit tranche stand-by arrangements approved by the Fund during the period 1970–76 in which a once-and-for-all currency depreciation played a major role. Following a brief review of the relevant theoretical and empirical considerations, the article discusses the problems facing the 12 countries and the intended effects of the exchange rate action. It then examines the outcome of the economic stabilization programs from the perspective of the actual performance of the volume of exports and imports, real gross domestic product, and inflation. The results for these variables are analyzed in relation to the average performance experienced by similar groups of countries in both the short run and the long run.

The principal conclusions of the study are as follows: (i) The majority of programs cited the emergence of significant balance of payments pressures (reflected in many instances in an intensification of restrictions) as evidence of the inappropriateness of the then current exchange rate, (ii) On average, export performance improved sharply in the postdepreciation period, especially over the longer run; however, the improvement was not universal, (iii) The behavior of imports differed between the import-restraint programs, where real import growth declined noticeably—reflecting the intended “expenditure reducing” effect of the depreciation—and the liberalization programs, where the volume of imports rose sharply, partly as a result of measures envisaged in the program to liberalize the exchange and trade system. For the latter group, however, real export growth also increased markedly, (iv) On average, the differential between program countries’ inflation rates and world inflation rates remained about the same; abstracting from the once-and-for-all inflationary effect of the depreciation, however, some improvement was recorded relative to world inflation rates, (v) For the import-restraint programs, average economic growth declined sharply, in both the short run and the long run; the converse was true for the liberalization programs.

The Fund Agreement in the Courts—XVIIjoseph gold (pages 728–59)

This article discusses two cases involving the much-litigated Article VIII, Section 2(b) of the Fund’s Articles of Agreement, under which, if an exchange contract involving the currency of a member of the Fund is contrary to exchange control regulations of the member that are consistent with the Articles, the courts of other members must forebear from enforcing the contract, or awarding damages for nonperformance, if a suit is brought to compel performance or obtain damages. The meaning of almost every aspect of the provision has been litigated in numerous countries. In England, the Court of Appeal has interpreted exchange contracts restrictively to mean contracts for the sale of one currency against another, but has made a reservation that includes within the concept other contracts if they are monetary transactions in disguise. In a case decided recently, the Court of Appeal found that there was a monetary transaction in disguise because of overinvoicing in the currency of payment, but the main importance of the case is that it decides for the first time, probably in any country, the relationship between Article VIII, Section 2(b) and irrevocable letters of credit. In the interests of commerce, the primary principle has been that a letter of credit is distinct from the underlying transaction and is enforceable against the confirming bank even if there are disputes between the parties to the underlying transaction. This principle has been subject to the single exception of fraud in the underlying transaction, so that only honest commerce will be facilitated. The question in the case under discussion was whether there is another exception if the conditions of Article VIII, Section 2(b) are met and the established exception does not apply. The Court of Appeal held that there is such a second exception.

The other case discussed in the article, decided by the Supreme Court of the Netherlands, raised the unusual question of the application of Article VIII, Section 2(b) by courts in a region of the Kingdom to a contract contrary to the regulations of another region that had become independent after the contract was made but before the action was brought.

The Preannouncement of Exchange Rate Changes as a Stabilization Instrumentmario i. blejer and donald j. mathieson (pages 760–92)

In recent years, some countries have adopted stabilization programs that have included the preannouncement of the future path of the exchange rate. This paper considers the circumstances under which the preannouncement of exchange rate changes helps to generate an adjustment process that results in a sustained decline in inflation without creating an overvalued currency. The favorable effects of this type of policy are shown to depend on its ability to strengthen the linkages between domestic goods and financial markets and the corresponding international markets. The private sector’s willingness to arbitrage in any interest rate and price differentials between comparable domestic and foreign goods and financial assets is an especially important element in determining whether a successful anti-inflation program will be generated. This is most likely to take place when the preannounced exchange rate path is credible because it is consistent with the size of the government fiscal deficit. The potentially most disruptive element is that capital inflows induced by greater exchange rate certainty may lead to growth of domestic monetary aggregates that is inconsistent with a declining rate of inflation. In that situation, a growing overvaluation of the currency is possible.

Capital Controls and Interest Rate Parity: The Japanese Experience, 1978–81ichiro otani and siddharth tiwari (pages 793–815)

Over the past several years, the exchange rate for the Japanese yen has fluctuated widely, as has the intensity of capital controls. In eary 1978, when an appreciation of the yen was widely expected, the Japanese Government introduced capital controls to reduce net inflows; in 1979 and early 1980, when the Government thought that the yen had weakened too rapidly and would depreciate further, measures aimed at inducing net inflows were introduced. However, since December 1980, when the Government implemented a law embodying the principle that all transactions in foreign currency may be carried out freely unless expressly restricted, capital movements have, in principle, been virtually free of controls. The main purpose of this paper is to examine the extent to which the capital controls or the lack of them have distorted interest rate parity during the period 1978–81.

The deviations from interest rate parity show that, since the beginning of 1978, the distortions were on a declining trend. By mid-1979, the Tokyo market was as free from distortions as was the London market. However, in the first quarter of 1980, the distortions increased significantly, reflecting the moral suasion exercised by the authorities to encourage Japanese residents to bring in funds from abroad. Since then, the distortions have declined considerably, possibly because moral suasion ceased, and have remained low except for a brief period in the third quarter of 1980. Data also indicate that, in recent months, the market distortions have been reduced to levels observed in the second half of 1979. The relaxation of restrictive capital controls following the implementation of the new Foreign Exchange and Foreign Trade Control Law in December 1980 contributed further to reducing the market distortions and the cost of transactions in short-term securities and foreign exchange in Tokyo.

RESUMES

RESUMENES

In statistical matter (except in the résumés and resúmenes) throughout this issue,

Dots (…) indicate that data are not available;

A dash (—) indicates that the figure is zero or less than half the final digit shown, or that the item does not exist;

A single dot (.) indicates decimals;

A comma (,) separates thousands and millions;

“Billion” means a thousand million;

A short dash (–) is used between years or months (e.g., 1977–79 or January–October) to indicate a total of the years or months inclusive of the beginning and ending years or months;

A stroke (/) is used between years (e.g., 1978/79) to indicate a fiscal year or a crop year;

Components of tables may not add to totals shown because of rounding.

International Monetary Fund, Washington, D.C. 20431 U.S.A.

Telephone number: 202 477 7000

Cable address: Interfund

Mr. Otani, Senior Economist in the Asian Department, is a graduate of the University of California at Berkeley and the University of Minnesota. Currently, he is on leave from the Fund and serves as a visiting scholar at the Program on U.S.-Japan Relations, Harvard University.

Mr. Tiwari, a doctoral candidate in economics at the University of Chicago, is a graduate of St. Stephen’s College, University of Delhi, and the London School of Economics. This paper was begun when Mr. Tiwari was a summer intern in the Fund in 1980.

The authors have benefited from discussions with Prabhat Goyal and Michael Mussa of the University of Chicago, John Makin of the University of Washington, and colleagues in the Fund.

Dooley and Isard (1980) constructed a model of portfolio behavior to study effects of German capital controls on deviations from interest rate parity.

Throughout this paper, the term “equilibrium” will be interpreted as a state where all profits have been exploited.

If assets are not denominated in the currency of the same political jurisdiction, this ban will reflect both transaction costs and political risk (see Aliber (1973)).

Transaction costs defined in this paper include not only brokerage fees but also costs in terms of taxes, search time, political risks, and ways in which market participants need to maneuver around any capital controls.

It is assumed that transaction costs are paid to lenders at the time of loan maturity.

It is assumed that the total transaction costs associated with buying and selling securities are paid at the time of selling the securities.

Given the nature of previous work in this area (e.g., Frenkel and Levich (1975, 1977)), there is need to briefly elaborate on the particular specification used here. Frenkel and Levich assumed that costs were paid in the currency bought and thus the effect of the introduction of transaction costs was a lower realized rate of return. However, the mirror image of this specification has been adopted here. Following Deardorff (1979), it has been assumed that costs are paid in the currency that is being sold. This specification shifts the focus from a lower realized rate of return to an increased cost (in current value terms) of undertaking the specified transaction. It is useful to note that for small transaction costs the two specifications are virtually the same.

The interest rates on these assets are freely determined by market forces; in other words, they are not institutionally fixed.

Randomness of the variables is due mainly to measurement problems—that is, “white noise”—which can be caused by recorded data that may not strictly correspond to the market rate. The differences between the recorded and the true market rate may be attributable to, say, rounding errors, differences in timing of sampling, or pure mistakes.

10

Bonds sold with the repurchase agreement. Interest rates on the Gensaki bonds are determined essentially by market forces but may be subject to distortions owing to imperfections in the market.

However, the mean was significantly different from zero in the first quarter of 1978.

The uniformly distributed observations are defined as those that have values increasing evenly when these values are arranged in ascending order. Therefore, these observations are linearly distributed with respect to their values. For a detailed discussion, see Appendix I.

In other words, the capital control measures had an impact of raising the transaction costs in Tokyo to 0.7 percentage point above those in London.

Despite the substantial covered interest differential in favor of London, short-term capital flows increased sharply, particularly in April, when they amounted to $4.4 billion, compared with $0.2 billion in March. This trend continued in May and June.

Other Resources Citing This Publication