Economic Review - September/October 1990
Economic Review - September/October 1990
Inflation and the Dollar Index
Karen R. Hunter
At the time this article was written, the author was an economic analyst in the macropolicy section of the Atlanta Fed's research department. She is grateful to Mary Rosenbaum, Sheila Tschinkel, Frank King, and Steve Feinstein for helpful comments and to Mike Chriszt and Jeff Watson for research assistance.
n 1986 the Atlanta Fed first published its dollar index to provide an updated summary measure of the dollar's global value. Recently, a few of the eighteen countries within the Atlanta Fed trade-weighted dollar index have been experiencing inflation that is higher than the U.S. rate, prompting concern that the Atlanta Fed dollar index may have become a less accurate measure of the dollar's real value. The Atlanta Fed dollar index, as well as a host of other new and existing indexes, was specified as an aggregate measure, a summary statistic of the dollar's average value against the currencies of the world. The recent proliferation of currency indexes may be a product of the emergence of large bilateral trade imbalances since 1982.1 Because trade imbalances are often attributed to over- or undervalued currencies, a single statistic capable of tracking the dollar's value could be helpful in exploring trade issues.
The Atlanta Fed dollar index is constructed using nominal exchange rates weighted by a country's share of U.S. trade. Because each component country's exchange rate is not adjusted for inflation, one of the important criteria when considering whether to include a country in the index is that its inflation track U.S. inflation fairly closely. Otherwise, changes in exchange rates would reflect relative price movements between countries in addition to changes in real, or inflation-adjusted, value. If all the countries in an index are nearly price-stable or are inflating at similar rates, then a nominal index such as the Atlanta Fed's index behaves like a real, or price-deflated, index. If inflation rates diverge among countries in an index, then a nominal index may be distorted or biased in its representation of changes in real value among currencies. The purpose of this article is to examine whether or not the nominal Atlanta dollar index remains a reasonable proxy for the dollar's average real value, given changes in the rate of price change in several countries included in the index. This study is also an examination of the impact of inflation differentials on various regional subindexes into which the Atlanta Fed index is divided.
Since the 1973 conversion to a floating exchange rate system, currencies have not only fluctuated widely in value but also have moved by disparate margins against the dollar. For example, the yen depreciated roughly 14.7 percent and the Deutschemark declined 43.8 percent in value versus the dollar from 1982 to 1985, while the Canadian dollar depreciated 16.1 percent against the dollar over the same period. Clearly, any one bilateral exchange rate is an unsuitable measure of the dollar's overall international value. It is, however, possible to combine, or average, bilateral exchange rates to get an idea of the dollar's average movement.
Giving each currency within this average equal weight would be imprudent since individual fluctuations in each exchange rate are not of equal importance to the U.S. economy. Within the Atlanta Fed index the currencies are weighted by their respective proportion of trade (exports plus imports) with the United States in 1984. For example, Canada, the United States' largest trading partner, has the highest weight, 0.288, of the currencies in the index. Japan, the nation's second largest trading partner, has a weight of 0.213. When the index shows that the dollar appreciated 28.8 percent against the eighteen currencies in the Atlanta Fed dollar index from 1982 to 1986, it is understood that the dollar did not rise uniformly against all currencies but that, on average, it appreciated by 28.8 percent.
The Atlanta Fed dollar index is also divided into regional subindexes since countries within a particular region may have similar trading patterns with a third country and their currencies often move together. Subindexes for Europe, Canada, Asia, and Asia excluding Japan have been defined. See Table 1 for a list of the eighteen countries, weights, and the composition of the subindexes. Charts 1-5 show the movement of the nominal Atlanta Fed index and subindexes since 1981. It should be noted that the scale of the charts is not uniform.
From the beginning, the countries included in the nominal Atlanta Fed trade-weighted index were chosen to cover the largest amount of U.S. trade possible (roughly 80.0 percent) while limiting the number of currencies according to certain criteria.2 Although an index might ideally encompass all U.S. trading partners, inclusion of either countries with high inflation relative to that of the United States or developing countries that employ multiple exchange rates would introduce a bias.3 A country such as Mexico, the United States' third-largest trading partner, has been excluded from the index because of its historically high inflation rates and the tiered exchange rate in effect during much of the 1980s. Creating the Atlanta Fed index so that the average inflation rate of the countries in the index approximated that of the United States generated an index that could function as a reasonable proxy for a real aggregate dollar index, with the added advantage of timeliness. Because nominal exchange rates, which are available daily and immediately, are employed in the index, it is possible to generate a daily Atlanta Fed dollar index. Timely, high-frequency price data necessary for a deflated, or real dollar, index are less available.
Several countries, primarily those included in the Asian subindex, have experienced accelerating inflation in the late 1980s. Although comparatively high, these inflation rates, with the exception of China's, have been generally in line with those found in other industrialized countries. For example, in 1989, prices increased 7.6 percent in Australia, 9.7 percent in Hong Kong, and 7.8 percent in the United Kingdom, compared with an inflation rate of around 4.5 percent for most industrialized countries. In China prices advanced 16.3 percent last year.
To measure inflation's distortion of the nominal dollar index, a real index and subindexes are constructed. More specifically, the nominal dollar index is deflated by the relative rate of consumer price increases, a calculation that generates a real dollar index.4 The various exchange rates that constitute the trade-weighted index and subindexes are deflated by the inflation differential between the United States and the country in the bilateral comparison. These measures have produced both a nominal and real index and subindex for each period covered by the original nominal index and subindexes.
Atlanta Fed Trade-Weighted Currency Index and Subindexes*
Country Exports Imports Sum Percent
Canada 46,524 66,911 113,435 28.768 Japan 23,575 60,371 83,946 21.289 United Kingdom 12,210 15,044 27,254 6.912 West Germany 9,084 17,810 26,894 6.820 Taiwan 4,775 14,772 19,547 4.957 Korea 5,983 10,027 16,010 4.060 France 6,037 8,516 14,553 3.691 Italy 4,375 8,504 12,879 3.266 Hong Kong 3,062 8,899 11,961 3.033 Netherlands 7,554 4,329 11,883 3.014 Saudi Arabia 5,564 4,009 9,573 2.428 Belgium 5,301 3,287 8,588 2.178 Singapore 3,675 4,121 7,796 1.977 Australia 4,793 2,899 7,692 1.951 China 3,004 3,381 6,385 1.619 Switzerland 2,563 3,199 5,762 1.461 Spain 2,561 2,628 5,189 1.316 Sweden 1,542 3,427 4,969 1.260 Totals 152,182 242,134 394,316 100.00
European Subindex United Kingdom 12,210 15,044 27,254 23.102 West Germany 9,084 17,810 26,894 22.797 France 6,037 8,516 14,553 12.336 Italy 4,375 8,504 12,879 10.917 Netherlands 7,554 4,329 11,883 10.073 Belgium 5,301 3,287 8.588 7.280 Switzerland 2,563 3,199 5,762 4.884 Spain 2,561 2,628 5,189 4.399 Sweden 1,542 3,427 4,969 4.212 Totals 51,227 66,744 117,971 100.00
Asian Subindex Japan 23,575 60,371 83,946 54.746 Taiwan 4,775 14,772 19,547 12.748 Korea 5,983 10,027 16,010 10.441 Hong Kong 3,062 8,899 11,961 7.800 Singapore 3,675 4,121 7,796 5.084 Australia 4,793 2,899 7,692 5.016 China 3,004 3,381 6,385 4.164 Totals 48,867 104,470 153,337 100.00
Asia-excluding-Japan Subindex Taiwan 4,775 14,772 19,547 28.169 Korea 5,983 10,027 16,010 23.072 Hong Kong 3,062 8,899 11,961 17.237 Singapore 3,675 4,121 7,796 11.235 Australia 4,793 2,899 7,692 11.085 China 3,004 3,381 6,385 9.201 Totals 25,292 44,099 69,391 100.00
* Trade weights reflect total trade in 1984. Each weight is defined as a country's total trade with the United States (exports plus imports) divided by total trade of the United States with the eighteen countries. Exports and imports are given in millions of U.S. dollars.
Source: Direction of Trade Statistics, International Monetary Fund, 1984; Federal Reserve Bank of Atlanta.
The level of the real Atlanta Fed dollar index indicates what the value of the nominal index would be without the presence of inflation. A positive difference between the nominal and the real dollar index indicates that cumulative average inflation in the eighteen countries has been above U.S. inflation and that the nominal index has been pushed upward by the presence of inflation (in the countries in the index) in excess of U.S. inflation. A negative nominal-real differential means that cumulative average inflation in the eighteen countries has been below U.S. inflation and that the nominal index has been pulled down by disparities in price pressures. Alternatively, if all the countries in the index had the same inflation rate, even if it were not zero, the nominal and real indexes would be identical and the correlation between the two measures would be 1.0.
Chart 1 shows that while the nominal and real indexes have moved together, the nominal index remained above the real index for much of the 1980s, when the weighted average of inflation for the eighteen currencies in the Atlanta Fed index exceeded that of the United States. Charts 2-5 show the experience for the subindexes, which is mixed. Since both the real and nominal dollar indexes are averages of currency movements, it is to be expected that the individual subindexes, both real and nominal, will experience more variability than the aggregated indexes (see Chart 6 and Chart 7).
The largest differentials between the real and nominal subindexes occur in the European subindex (see Chart 2). From 1982 to 1986 the greatest regional inflation outliers were in Europe. The chief culprits were Italy and France, where inflation averaged 9.4 percent and 5.2 percent, respectively, well above that of West Germany (Economist Intelligence Unit 1989, 191). Since 1986 closer alignment of European economic policies has been reflected in the uniformly lower inflation rates in these countries, a tendency reinforced by their continued participation in the Exchange Rate Mechanism of the European Monetary System (EMS). More recently, inflation in the United Kingdom has exceeded the EMS average by margins substantial enough to skew the European index. The average inflation rate for EMS countries is currently around 4.0 percent annually.5
Of the four subindexes, only the real Asian subindex lies above its nominal counterpart, reflecting the large weight of Japan (roughly 55 percent of the total; see Chart 3), which has had very low inflation relative to the United States over the period. The Japanese influence on the Asian subindex is clear when the Asia-excluding-Japan subindex is considered (see Chart 4). Unlike the Canadian and European averages in the past three years, the divergence of inflation differentials in the Asia-excluding-Japan subindex has widened, undoubtedly because of the increased inflationary pressures in Australia, Hong Kong, and China. According to the Asian and Asia-excluding-Japan subindexes, inflation rates in Japan have also been substantially lower than inflation rates in other Asian countries.
The nominal-real gap on the Canadian subindex is considerable since Canada has a large cumulative inflation differential against the United States (see Chart 5). This gap appears in part because Canada is the sole country within its subindex so that, unlike the other subindexes, no other countries balance out Canada's higher inflation.
Of the four subindexes, three show large positive divergences from the U.S. retail price history. Over the period as a whole, the European, Canadian, and the Asia-excluding-Japan prices did not always move proportionately with U.S. price developments. The cumulative divergence above the total U.S. consumer price index from 1981 to 1989 is roughly 6.5 percent for the European, 12.9 percent for the Canadian, and 7.0 percent for the Asia-excluding-Japan countries. The cumulative inflation differential for the countries in the Asian subindex was actually 9.3 percent below total U.S. inflation for the period.
The difference between the levels of the overall real and the nominal dollar index is the amount of aggregate bias introduced by using the nominal dollar index to approximate a real dollar index. According to the real dollar index, the dollar's appreciation from January 1981 to the March 1985 peak and its subsequent descent were not as large as portrayed by the nominal index. Since mid-1986 the difference between these two indexes has decreased markedly, indicating that inflation rates in the eighteen countries covered by the index have tended to converge toward U.S. rates and have eliminated much of the cumulative inflation discrepancy reflected in the difference between the indexes.
Comovement of Indexes
For the purposes of this study the comovement of the real and nominal indexes and subindexes is of greater interest than the differences between levels of the nominal and real indexes (and subindexes) or the changes in the aggregate real index. The comovement, or correlation, of the indexes indicates the degree to which changes in the nominal index (subindex) reflect changes in real values. Since it is changes in the exchange value of the dollar that are thought to be important in explaining movements in international trade, the correlation of the changes in the indexes are of particular interest, and the strength of that correlation is a measure of the nominal index's resilience to inflation differentials. The higher the correlation, the better is the nominal index at capturing real movements and the smaller is the inflation distortion of the nominal index.
Using correlations of first differences (changes) of levels shows that the nominal dollar index closely shadows the real or price-deflated dollar index. (See Table 2 for correlations.) This result could be inferred from reference to Chart 1, which shows close comovement during 1981 and from mid-1987 to December 1989. The close correlation of the first differences over the entire sample period (1981-89), 0.987, indicates that the nominal index remains a reasonable proxy for the real dollar index.6
Correlation Coefficients between Changes in Nominal and Real
(monthly data, January 1981 to
Index or Subindex Correlation
Nominal Atlanta Index,
Real Atlanta Index
0.987 Nominal European,
0.996 Nominal Asian,
0.977 Nominal Asia-excluding-Japan,
0.923 Nominal Canada,
Two inflation outliers serve as case studies for examining the sensitivity of the nominal Atlanta Fed dollar index to exceptions to the ideal criterion of uniform inflation among countries in the index. Canada, which has the single largest weight of any country in the index, experienced inflation that was moderately but persistently above the United States' for most of the 1980s. China, which has a very small weight, showed a much larger inflation differential than the United States over the same period.
The Canadian inflation rate peaked in 1982 and since then has been relatively stable, although consistently above U.S. rates. The level of the nominal Canadian subindex diverged markedly from the level of the real Canadian subindex starting in late 1983, and the divergence persisted and grew over the next three years, peaking in late 1986 and shrinking slightly thereafter. The nominal-real subindex differential has been relatively steady since mid-1988.
Despite the large and persistent differential in the subindex levels, changes in the real Canadian dollar subindex and the nominal Canadian subindex tended to move together. The changes in these indexes are highly correlated from 1981 through 1989, at 0.941. This figure suggests that the change in the nominal index is a reasonably good measure of the change in the real index, notwithstanding the difference in their levels. In the case of Canada, a moderate but long-standing inflation differential in an important trading partner causes a widening gap between the nominal and real subindex. The nominal and real subindexes continue to move together, however, since the inflation gap does not worsen. In this way, the usefulness of the nominal index is not badly compromised by the different inflation experiences of the United States and Canada.
China is a contrasting case. Because its trade weight in the overall Atlanta Fed index in the Asian subindexes is quite small (0.0162), a very large inflation differential would be required between China and the United States to drive the nominal and real indexes and subindexes apart. The China-U.S. inflation differential has been substantial over the last several years, although high inflation is relatively new to China. During the thirty years prior to 1980, the Chinese government was able to hold inflation to about 3.0 percent per year, excluding a brief period in the 1960s.7 Stable inflation was achieved by regulating supply and demand through quotas, rationing, and price controls.
In the late 1970s, the Chinese government decided that a slow move toward a more market-oriented economic system would help promote growth. In December 1978 China began a program of economic liberalizations that raised real national income by an annual average of 9.9 percent in the 1980s (see Cheng 1988a, b and Bank of Japan 1989). The economic reform occurred in two segments: the rural liberalizations begun in 1978 and the urban industrial liberalizations implemented in 1984. Under these reforms, farms and businesses were allowed to sell anything produced over-and-above official production quotas at "market" prices, which were generally higher than state-set prices. Production increased dramatically, but at the same time the practice of allowing firms to determine their own financing needs kindled inflation. The result was a wave of domestic borrowing that greatly increased the money supply. A by-product of reform has been increased inflationary pressures. Retail prices soared, increasing 8.8 percent in 1985, 6.0 percent in 1986, 7.3 percent in 1987, 18.5 percent in 1988, and around 16.0 percent in 1989, with even higher rates in urban centers.
The rising inflationary pressures in China were of particular concern since the exchange rate was fixed for a long period after mid-1986.8 Unlike free-floating exchange rates, fixed exchange rates often do not reflect domestic economic and political policies. Countries fix exchange rates for a variety of reasons such as countering the effects of high domestic inflation, protecting nascent manufacturing sectors, or discouraging imports. In a country with fixed rates, devaluation is often regarded as an indication of the government's inability to manage the economy effectively. As a result, governments frequently wait until exchange rates are grossly out of line before devaluing or revaluing, and they do so in large strokes. Thus China developed the characteristics—high inflation and use of unofficial exchange rates—that earlier had led to the exclusion of other important trading partners (Mexico, for example) from the Atlanta Fed's dollar index. The exchange rate remained unchanged at 3.341 yuan per dollar for roughly four years. This exchange rate was maintained from mid-1986 through 1989 despite accelerating price pressures, yielding an effective or "real" appreciation of the yuan. Much later, in December 1989, a 21 percent devaluation was announced.
China is an example of a clear inflation outlier. However, its very slight weight in the weighted overall dollar index minimized the effect of its large devaluation (after several years of high inflation) from having a large impact on the overall Atlanta Fed index or on the Asia and Asia-excluding-Japan subindexes. The cumulative divergence in the real and nominal indexes is not large, and the correlation between changes in the nominal and real indexes is high—0.977 for the Asian subindex and 0.923 for the Asia-excluding-Japan subindex.
Given that the average inflation rates for the eighteen countries included in the Atlanta Fed dollar index are still roughly in line with U.S. inflation rates and that the difference between the nominal and real dollar index is relatively modest, the nominal Atlanta Fed trade-weighted dollar index remains a reasonable proxy for a real or price-deflated dollar index. The incidence of substantially higher inflation in China and of a sustained moderate inflation differential in Canada, for example, has been balanced by the inclusion of low-inflation countries within the Atlanta Fed dollar index and by China's relatively small weight in the index. It should be noted, however, that there is no feature in the composition of the index which ensures that it will remain free of distortions arising from inflation disparities in these or other countries. The presence of inflation outliers naturally has a larger impact on the subindexes because of the larger weight of individual countries in the subindexes. Within the Asia-excluding-Japan subindex China has a weight of 10 percent. If a country's weight in both the total and the subindex were relatively large and it continued to inflate excessively, it would be reflected in a divergence of the real and nominal indexes and in a decline in the measures of comovement.
|1||For example, the Morgan index, the Federal Reserve Board of Governors index, and the Dallas Fed index each offer a different representation of the dollar's level and variability.|
|2||The most important criterion is that the average inflation rates of the countries included approximate that of the United States, so that the index will cover the largest proportion of U.S. trade possible without skewing the index by including large inflation outliers or countries with multiple exchange rates. See Rosensweig (1987).|
|3||An exchange rate is the price of one currency in terms of another currency. This relative price is determined not only by economic fundamentals—inflation, growth, and monetary policies—but also by political events and market expectations. In the case of multiple exchange rates it is difficult to choose the one most representative of a free market rate. See Rosensweig (1987) and Hervey and Strauss (1987).|
|4||It has been suggested that a measure of wholesale prices rather than consumer or retail prices be used to deflate exchange rates because some nontraded goods that are included in consumer or retail price indexes may skew a real or deflated dollar index (Harberger 1986). However, consumer price indexes were used as the inflation measure because they are generally available and more closely comparable among countries. Wholesale price measures were unavailable for two of the countries within the index—Saudi Arabia and Hong Kong.|
|5||Countries in the EMS but not in the dollar index are Denmark and Ireland. European countries in the dollar index but not in the EMS are the United Kingdom, Sweden, and Switzerland. Spain, which is also in the index, has only recently become an EMS member.|
|6||Objective criteria for goodness of fit in this correlation have not been established. However, criteria for other economic times series suggest that correlations above 0.90 are considered strong.|
|7||Low inflation is seen as one of the accomplishments of the Communist government since the hyperinflation of the late 1940s contributed to the downfall of the Nationalist government (Harding 1987, 279).|
|8||After being included in the Atlanta Fed dollar index, China soon began to fix the exchange rate of the yuan, setting it at a given number to the dollar.|
|Bank of Japan. "Recent Developments in the Chinese Economy: Economic Reform, Its Success and Problems." Special Paper no. 176, April 1989.
Cheng, Hang-Sheng. "Inflation in China." Federal Reserve Bank of San Francisco Weekly Letter, November 4, 1988.
______. "Monetary Policy and Inflation in China." In Monetary Policies in Pacific Basin Countries, 401-27. Boston: Kulwer Academic Publishers, 1988.
Economist Intelligence Unit. European Community: Economic Structure and Analysis. London: The Economist Intelligence Unit, 1989.
Harberger, Arnold. "Economic Adjustment and the Real Exchange Rate." In Economic Adjustment and Exchange Rates in Developing Countries, edited by Sabastian Edwards and Liaquat Ahamed, 371-423. Chicago: The University of Chicago Press, 1986.
Harding, Harry. China's Second Revolution: Reform after Mao. Washington, D.C.: The Brookings Institution, 1987.
Hervey, Jack L., and Strauss, William. "The New Dollar Indexes Are No Different from the Old Ones." Federal Reserve Bank of Chicago Economic Perspectives 11 (July/August 1987): 3-22.
International Monetary Fund. World Economic Outlook. Washington, D.C.: International Monetary Fund, 1990.
Rosensweig, Jeffrey A. "A New Dollar Index: Capturing a More Global Perspective." Federal Reserve Bank of Atlanta Economic Review 71 (June/July 1986): 12-22.
______. "Constructing and Using Exchange Rate Indexes." Federal Reserve Bank of Atlanta Economic Review 72 (Summer 1987): 4-16.
Weil, Gordon. "Exchange Rate Regime Selection in Theory and Practice." Monograph Series in Finance and Economics 1983-2. Solomon Brothers Center for the Study of Financial Institutions, Graduate School of Business Administration, New York University, 1983.