Journal Issue
Share
Article

Estimating real interest rates in LDCs: What is the best approach?

Author(s):
International Monetary Fund. External Relations Dept.
Published Date:
June 1986
Share
  • ShareShare
Show Summary Details

Deena R. Khatkhate

Interest rate policy is one of the more important instruments of macroeconomic management. The rate of interest performs several functions in an economy. It is a reward for saving, affecting how an income earner allocates his income between present and future consumption. The structure of interest rates governs the distribution of savings, once accumulated, among different assets, financial and real. Interest rate policy has effects on the saving-investment processes and through them on the development and diversification of financial assets, the capital intensity of production, and the rate of growth of output.

When prices are stable, the nominal and the real rate of interest coincide. But when there is inflation, economic agents need to distinguish between these rates, generally estimating the real rate of interest by subtracting from the nominal one the rate of change in some measure of domestic inflation. Where interest rates are fixed administratively, policy makers may adjust the nominal rate of interest to try to keep the real rate of interest positive, so as to provide incentives for saving, to increase financial intermediation, and to promote the unification of financial markets.

There remains disagreement in the literature, however, on the actual effects of interest rate policy on saving, investment, and the rate of growth of output. Some empirical studies show that significantly negative real interest rates do not always have the adverse effects on economic growth predicted by much of conventional theory. Others have found no direct relationship between the real interest rate and the level of saving or between the real interest rate and the level of investment. As these empirical results depend partly on how real interest rates are estimated, it is helpful to understand the conceptual problems involved in their estimation.

This article, based on a longer study by the author, first describes the characteristics of the real interest rate and some of the complexities involved in its estimation on the basis of various domestic price indices. To emphasize the limitations of the methods commonly used, contrasts are drawn between series of real interest rates derived using different indices for a group of developing countries over a ten-year period. From this analysis it is obvious that the different price indices yield different guidelines as to the level at which the nominal rate should be set in order to achieve a desired level of the real rate. The article then offers an alternative method for guiding the selection of a nominal rate, whereby a representative foreign interest rate is adjusted for expected changes in the exchange rate.

Conceptual issues

What is the meaning of the real interest rate and how does this rate adjust to its equilibrium level in market economies? The economist Irving Fisher defined the real interest rate as simply the percentage premium paid on present goods over future goods of the same kind. Essentials of Fisher’s concept are (1) definite and assured payments; (2) definite and assured repayments; and (3) definite dates. Using this concept, the real interest rate can also be seen as the relative cost of current consumption in terms of forgone future consumption.

Fisher’s concept of the real interest rate has two important implications. First, present sacrifice is balanced by access to a stream of consumption goods and services at a future date. That stream is determined by the assets that are created through the investment that is financed by current savings. Second, the real interest rate determines the proportion of resources devoted to producing capital goods, as against consumer goods.

Both of these implications mean that savings and investment and their uses are critically influenced by perceptions of the real interest rate. To people making decisions on savings and investment, the long-term real interest rate is the most germane, because capital goods have a long time span. And, because these decisions depend on what people expect to happen in the future, it is not the historical or the current inflation rate that must be used as an adjustment factor for nominal interest rates, but the inflation rate expected in the long term.

Given the importance of the real long-term interest rate in the decisions made by economic agents, changes in its level have far-reaching effects. If its level rises, the present value of future claims will decline, with a consequent fall in the wealth of individuals. The opposite happens if there is a decline. The effect of movements in real interest rates on the value of any particular asset will vary according to the asset’s durability. When the long-term real interest rate rises, individuals with more durable assets will suffer greater loss in the present value of their assets than those with less durable assets, and they will gain more when it falls.

A change in the real long-term interest rate is the combined result of the change in the nominal interest rate and the change in the expected rate of inflation. The expected rate of inflation is usually measured with reference to a basket of goods, such as the consumer price index (CPI) or wholesale price index (WPI). There are, however, several grounds for objection to the use of these indices.

Using the CPI and WPI. It can be argued that the CPI gives disproportionate weight to the prices of services relative to those of goods. Unlike the consumption of services, goods are an alternative to holding money. Wealth owners hold either commodities, the rate of return on which is given by the rate of change in their prices, or financial assets, the rate of return on which is denoted by the nominal interest rate. It follows from this that expected change in the price of commodities is more relevant to the allocation of savings by individuals than those of services. In economies where labor productivity as well as real wages are rising, prices of services tend to rise in relation to the prices of commodities (an outcome that can be traced to relatively slow technical change in the production of services). Thus, the use of the CPI will tend to overestimate the rate of inflation in the price of goods, and thus bias downwards an estimate of the real interest rate that depends on expected changes in the CPI. For this reason, use of the wholesale price index, which is composed wholly of goods, is often suggested as an alternative for calculating the real interest rate.

There are other, practical, objections to the use of the CPI to estimate the real interest rate in developing countries. First, in many of these countries, the index does not reflect actual changes in equilibrium market prices, because a large proportion of the commodities featured in the index have official prices that are administered or controlled by government. In Egypt in 1977-82, for instance, 27 percent of the commodities represented in the CPI were subject to central government price controls; in Mali in 1976-80, the proportion was around 25 percent.

Second, in many developing countries the CPI records price changes in only a few cities, so that it may not measure changes in purchasing power in the country as a whole. Third, even when the CPI is calculated for a particular city or a region, the basket of goods and services it includes may be insufficiently representative of the consumption patterns of particular groups of the population.

In practice, the consumer price index and the wholesale price index generally move together in developing countries. In only 3 of 24 developing countries for which a wholesale price index (1970-80) was available was the correlation coefficient between changes in the CPI and WPI less than 0.75; in 19 of these countries the coefficient was above 0.9. That is to say, the year-to-year changes in both series were very similar in magnitude and direction. In contrast, in 7 out of 10 industrial countries for that period, the corresponding coefficient was below 0.8.

This pattern in developing countries is probably explained by two factors. First, expenditures on services—or at least those expenditures that are measured—are likely to be less important in the public’s consumption pattern because of the relative underdevelopment of such economies. Therefore, commodities enter into both the CPI and WPI with more or less the same weights. Second, the effects of controls on prices of commodities entering the CPI may ultimately be reflected in the WPI, in which case percentage changes in both would follow similar patterns.

Using the GDP deflator. Perhaps a more fundamental disadvantage of the CPI and WPI is that they both give too much weight to the prices of consumer goods and too little to those of capital goods and long-lived assets. This suggests that a better index to Use for estimating inflation is one that assigns weights to consumption goods, services, and long-lived goods that reflect an appropriate average of changes in their prices. For LDCs, the closest equivalent to such an index is the GDP deflator (which measures changes in the value of the total final output of the economy); the weights assigned to goods in these indices change as the pattern of expenditure shifts in response to movement in the relative prices of goods. Being more comprehensive in coverage than the CPI and WPI, the GDP deflator is a better indicator of the purchasing power of money.

It is worth noting the results obtained from correlating the GDP deflator with the CPI and the WPI for developing countries for which data were available. The coefficient of correlation between changes in the GDP deflator and changes in the CPI was above 0.75 in only 10 of 55 countries for which data were available for the period 1970-80; it was above 0.75 for only 15 out of 24 countries for which the WPI was available. That is, the different indices give quite different impressions of the rate of price change in the economy.

The GDP deflator has been used to construct a series of long-term real interest rates for the 70 LDCs for which data were available for the years 1971-80. To do this, nominal interest rates on bank deposits were adjusted by the percentage changes in the GDP deflator. In 70 countries, the resulting estimates of expected real interest rates are generally less positive than those estimated using the CPI.

Using asset prices. Obtaining estimates of real interest rates through using price indices requires assumptions about how past inflation rates are reflected in the expected inflation. An alternative means of inferring inflation expectations is to examine the behavior of asset prices that incorporate information about expected future yields.

Brown and Santoni suggest that inferences can be drawn about changes in the real interest rate from the behavior of prices of long-lived assets relative to the prices of short-lived assets (“Unreal Estimates of the Real Rate of Interest,” Federal Reserve Bank of St. Louis Review, January 1981). Their approach is a partial one, in that it assumes that the real interest rate changes only because of changes in the propensity to save, and ignores the effects of change in demand for investment. Nonetheless, their method can be applied to the available evidence for developing countries. Table 1 shows illustrative findings for Brazil, India, and the Republic of Korea on the relationship of prices of durable and nondurable goods, stocks, and commodities to the expected real rate of interest, as derived using the CPI.

Table 1.Expected real interest rates, and prices of stocks and nondurable and durable goods in three countries, 1970–80
19701971197219731974197519761977197819791980
Brazil
Expected real interest rate6.28.49.08.80.5−1.2−5.62.78.3−16.5−4.4
Ratio of nondurable to durable goods prices0.790.880.910.941.001.001.081.081.141.231.32
Percentage change from previous year(—)(11.39)(3.40)(3.29)(6.38)(0.0)(8.00)(0.0)(5.55)(7.90)(7.40)
Stock price index, 1975 = 10033.3113.085.566.763.8100.0123.2144.9176.8184.1391.3
Ratio of nondurable goods to stock prices1.090.410.630.951.251.01.181.431.692.512.42
India
Expected real interest rate4.02.4−8.1−16.4−0.113.93.46.14.1−0.3
Ratio of nondurable to durable goods prices0.980.910.890.981.051.001.001.051.051.000.98
Percentage change from previous year(—)(−7.14)(−2.19)(10.11)(7.14)(−4.76)(0.0)(5.0)(0.0)(−4.76)(−2.0)
Stock price index, 1975 = 100103.7102.3102.3102.5100.8100.0100.6101.2101.3102.3100.6
Ratio of nondurable goods to stock prices0.550.570.620.781.031.001.011.091.081.231.46
Republic of Korea
Expected real interest rate6.5−1.01.91.4−1.30.50.54.83.96.2
Ratio of nondurable to durable goods prices0.660.720.830.850.861.01.2
Percentage change from previous year(—)(9.09)(15.28)(2.40)(1.17)(16.27)(2.00)
Stock price index, 1975 = 10091.590.0100.0122.7
Ratio of nondurable goods to stock prices0.660.851.000.94
Source: “Assessing the Level and Impact of Interest Rates in Less Developed countries,” by the author.—Indicates data not available.
Source: “Assessing the Level and Impact of Interest Rates in Less Developed countries,” by the author.—Indicates data not available.

The market for durable and nondurable goods is considered first. If the real interest rate rises, the prices of durable goods would be expected to decline in relation to the prices of nondurable goods, and vice versa if there is a decline in the expected real interest rate. However, the evidence for Brazil, India, and Korea shows no clear relationship between the changes in real interest rates as derived from the ratio of prices of nondurable goods to durables and those based on the consumer price index. Table 1 shows, for example, that in much of the period when this ratio rose in Brazil (1970-74 and 1975-80), implying an increase in the real interest rate, the expected real rate derived using the CPI actually decreased. Similarly, when this ratio fell in India during 1974-75, the expected real interest rate derived from the CPI rose. For Korea, the ratio rose continuously in 1970-76, yet the real interest rate derived from the CPI declined in 1971-72 and 1973-75.

The second measure used for gauging the change in long-term interest rates is movements in stock market prices. The most important stock market indicators in this connection are the earnings-price ratio, and the ratio of stock market prices to an index of nondurable goods prices. Since stock prices reflect the current value of expected future earnings, an increase in the price of shares relative to current earnings would indicate a decrease in the expected real rate of interest, provided the relationship between current earnings and expected future earnings remained unchanged. Similarly, a decrease in the expected real interest rate will be represented by a fall in the price of consumption goods relative to stock prices.

Unfortunately, information for the sample countries is only available on the ratio of prices of nondurable goods to those of stocks. Table 1 shows that for Brazil and India, the ratio of nondurable goods prices to stock prices did not always change in line with the expected real interest rate derived from the CPI. For Korea, no meaningful conclusion can be drawn, for lack of information.

The fact that different ways of calculating real interest rates can lead to significantly different results suggests that policy makers should exercise caution in using estimates of real interest rates as a guide in setting nominal interest rates. It also suggests that there is value in searching for alternative guides for policy.

An alternative

In view of the difficulties of estimating real interest rates, an alternative approach might be to set domestic interest rates by adjusting a representative foreign interest rate (say, the London Interbank Offered Rate) by the differential between the domestic and foreign inflation rates. This approach offers monetary authorities an alternative way to identify an appropriate level for the nominal interest rate. Its virtue rests on the belief that a fairly close relationship should be maintained between the domestic nominal interest rate and the adjusted foreign interest rate, since even in countries with trade and exchange controls, people substitute domestic currency and financial assets for foreign ones, legally or otherwise. If the domestic nominal rate is appreciably lower than the adjusted foreign rate, it should probably be adjusted upwards, to prevent capital flight. Of course, this method suffers from the disadvantage that it relies on estimates of current or past inflation to provide a proxy for the theoretically more desirable concept of expected future inflation. This is probably a less serious difficulty when setting short-term rates (in which case the last period’s inflation rate is probably a reasonably good guide to the next period’s) than it is for long-term interest rate.

Like the series based on the GDP deflator, the series based on foreign interest rates was constructed using the average annual rate of inflation over the past ten years. Table 2 illustrates the two series for the major developing countries. From a comparison between the two full series presented in the longer paper on which this article is based, it is clear that the countries which have negative real interest rates, as calculated using the GDP deflator, also tend to be the countries in which domestic nominal rates are lower than the foreign rate adjusted for expected exchange rate changes. Moreover, the countries where domestic nominal interest rates are higher than foreign rates adjusted for exchange rate changes are those where the real interest rates, as computed using the GDP deflator, are positive. This conclusion does not alter significantly when the countries are grouped by rapidity of inflation, degree of openness of their economies, or by region.

Table 2.Interest rates and the productivity of investment in selected LDCs, 1971–80
Nominal

interest

rate1
Real

interest

rate2
Adjusted

foreign interest

rate3
Incremental

output/capital

ratio4
Marginal

rate of return

on capital
Africa
Ghana5.29−41.7033.76− 2.000.23
Cote d’lvoire3.97−12.9012.4523.5010.78
Kenya5.37−4.9713.9726.1013.33
Nigeria5.54−10.809.2832.1022.11
Tanzania3.79−7.3112.4530.8013.41
Zambia4.76−2.9513.57−2.000.68
Asia
Bangladesh8.145.3820.5018.10
India6.79−2.2411.2516.9017.46
Indonesia11.06−7.8617.3041.8037.53
Korea, Rep. of15.46−2.6716.6131.7016.18
Malaysia7.611.207.1630.305.92
Pakistan8.39−0.4718.2128.4028.82
Philippines9.45−3.2713.2022.0918.73
Thailand9.660.169.7428.2016.34
Latin America
Argentina58.95−71.34154.5312.40
Bolivia13.66−21.0145.0318.109.80
Brazil31.07−13.2559.0731.0029.14
Chile60.31−118.42158.2512.002.64
Colombia20.981.2320.5425.9013.47
Mexico15.40−3.6531.9230.8016.04
Peru22.96− 15.1444.3720.2010.42
Uruguay44.10−18.5855.9525.109.46
Other
Morocco6.00−2.5512.08
Tunisia4.00−4.1111.3030.7026.25
Yugoslavia9.91−11.7326.5117.00
Source: See Table 1.—Indicates data not available.

Observed rate on bank deposits.

Nominal rate adjusted for percentage change in GDP deflator (see text).

Representative foreign interest rate adjusted for expected change in exchange rate(see text).

Growth rate of real GDP divided by the ratio of investment to income.

Source: See Table 1.—Indicates data not available.

Observed rate on bank deposits.

Nominal rate adjusted for percentage change in GDP deflator (see text).

Representative foreign interest rate adjusted for expected change in exchange rate(see text).

Growth rate of real GDP divided by the ratio of investment to income.

Whether the use of a foreign reference rate is an appropriate method for setting the nominal interest rate depends to a large extent on country characteristics. Given these characteristics, policy makers will have their own views on what weight to give foreign influences, relative to domestic ones. If the rate of return to capital is much higher or lower than the adjusted foreign interest rate, authorities may wish to give relatively more weight to domestic factors.

What of the returns to capital investment? Calculations were made for 1971-80 for 63 developing countries on which data were available. (Table 2 shows results for the major developing countries.) They show that in 48 of 63 countries, the adjusted foreign interest rate is lower than the incremental output/capital ratio—that is, the rate of growth of real GDP divided by the ratio of investment to income. (Among the countries showing the opposite pattern are Argentina, Brazil, Chile, Peru, Turkey, and Uruguay. These may be seen as a special category in which the prevalence of low nominal interest rates in the face of high and rising inflation tended to depress the expected rate of return on capital.) The evidence on the marginal rate of return on capital, also illustrated in Table 2, is more ambiguous: the rate of return was higher than the adjusted foreign interest rate in only 18 of 50 countries. (Here the same group of high inflation countries forms an exception.)

There is also another reason, however, for assigning a high weight to foreign factors in assessing the real interest rate. From domestic investors’ point of view, the differential between the expected rate of interest and the expected rate of return on investment is crucial. Suppose the domestic nominal interest rate is appreciably lower than the foreign interest rate, adjusted for expected changes in the exchange rate. Capital flows out of the country, while a failure to attract foreign capital may delay the country’s investment program, possibly depressing the expected rate of return. If this happens, the differential between the expected rate of interest and the expected rate of return will narrow, leading to lower investment. If, on the other hand, domestic interest rates remain in rough parity with the relevant foreign interest rate, the country’s general economic environment is likely to be more attractive, raising the return on present and future investment.

To sum up, it seems desirable for those setting nominal interest rates in LDCs at least to supplement their knowledge by looking at the foreign interest rate adjusted for expected changes in the exchange rate, without being much concerned with the measurement problems involved in approximating the real interest rate on the basis of domestic price movements.

Other Resources Citing This Publication