- The ERS Agricultural Exchange Rate Data Set
- How the ERS Agricultural Exchange Rate Data Set is Different
- How the Exchange Rates are Derived
- Real Exchange Rate Definition
- Commodity Exchange Rate Index Defined
- Derivation of a Sample Real Trade-Weighted Exchange Rate Index
- Nominal Exchange Rates
- Consumer Price Index
- Trade Weights
- Why Use Real Exchange Rates and Exchange Rate Indices?
- How to Interpret Movements in Exchange Rates
- Why Use Commodity or Product-Specific Exchange Rates Rather than Broad Measures?
- Country and Currency Coverage
- Commodity Coverage
- Data Exceptions
The Economic Research Service’s (ERS) Agricultural Exchange Rate Data Set presents real and nominal exchange rates for the United States’ most important trading partners, and commodity-specific trade-weighted exchange rate indices for the agricultural commodities and products that represent the largest share of exports and imports. Once the United States went off of the gold standard in 1971 and the multilateral Bretton Woods system of fixed exchange rates ended, exchange rates became determined by the cumulative actions of other countries’ central banks and the behavior of foreign exchange markets. Without gold as an official reference, the dollar became the primary global reserve currency and the United States could no longer independently set its exchange rate.
A trade-weighted or “effective” exchange rate is a measure of the average exchange value of a currency against a basket of other currencies, where each currency’s influence reflects its importance in trade with that country. In the ERS Agricultural Exchange Rate Data Set, our trade-weighted or effective exchange rates are an average measure of the value of the U.S. dollar relative to the currencies of major U.S. trading partners. Since U.S. trade partners differ from commodity to commodity and product to product, and changes in exchange rates vary substantially across countries, the effective exchange rate that describes the relative value of the U.S. dollar also differs by commodity. Thus, instead of having a single exchange rate against gold (at say $35 an ounce), the U.S. has different exchange rates with each country and different effective exchange rates for each commodity/product (for the purpose of this documentation, “commodity” represents both commodities and products). For reference purposes, we also present a total agricultural trade-weighted exchange rate and a total merchandise trade-weighted exchange rate that are broad measures of the U.S. exchange rate.
In the following, we will present:
- How we derive our exchange rates;
- Which countries are included and our criteria for choosing them;
- Which commodities are covered and how we choose them;
- Why we present both real and nominal exchange rates;
- How to interpret movements in exchange rates; and
- When and why we suggest using commodity-specific exchange rates rather than broad measures.
All of the exchange rates in the ERS Agricultural Exchange Rate Data Set are presented in terms of local currency per dollar. This means that a rise in a country exchange rate or a commodity exchange rate index implies an appreciation of the dollar or a depreciation of the local currency(ies). Similarly, a decline implies a depreciation of the dollar or appreciation of the local currency(ies).
The ERS Agricultural Exchange Rate Data Set is the only publicly available data set that presents both nominal and real commodity-based exchange rates on a monthly and yearly basis over an extended historical period. Exchange rates extend from January 1970 to the most recent (available) month. Annual exchange rates and exchange rate indices are averages of the monthly exchange rates and indices.
There are many sources for exchange rate data. Banks and foreign exchange services typically present current exchange rates useful for converting currencies for travel. Other data sets offer a large number of country exchange rates against the dollar or other third-country currency (such as the euro, the pound, or the yen). Almost all of these sources provide only nominal exchange rates for a limited historical period and a limited set of countries. The International Monetary Fund (IMF) publishes data on official nominal exchange rates by country in their International Financial Statistics (IFS) database (IMF, 2014). We use these nominal rates in our data set and augment them with rates from the Pacific Exchange Rate Service (University of British Columbia, 2014) since current rates are not available in the IFS. To obtain real exchange rates, we also need a price index for each country. Fortunately, the IFS database also includes comprehensive data on country consumer price indices (CPIs).
To derive real exchange rates from nominal rates, both the local currencies and the dollar must be adjusted for price changes using a relevant price index. Technically, since the country exchange rates reflect a broad range of goods and services in a country, the gross domestic product (GDP) deflator, as the broadest measure of price changes in an economy, is an appropriate price index to use. Another appropriate price index is the Producer Price Index (PPI), since traded goods prices are not final goods prices. Unfortunately, neither GDP deflators nor PPIs are available on a monthly basis for the full set of countries included in the data set. Therefore, for practical reasons, we use the more widely available CPI indices to derive the real exchange rates.
Real exchange rates are defined as:
where NXRi is the nominal dollar exchange rate in terms of country i’s local currency (LCi) and LCCPIi and USCPI are the CPI for country i and the United States, respectively. More specifically:
which can be written as:
The real exchange rate for country i can thus be seen as the ratio of the product of the local currency value and the U.S. CPI to the product of the value of 1 dollar and the local currency CPI. It is thus clear that relative rates of inflation are critical for determining real exchange rates: a real appreciation (depreciation) of a currency represents an increase (decrease) in its relative value against another currency such as the U.S. dollar. Thus, an appreciation implies that less of that currency is needed to purchase another currency—i.e., a declining (increasing) ratio of a local currency to the U.S. dollar is an appreciation (depreciation) of that currency. A higher rate of inflation in the local currency than the U.S. dollar results in a real appreciation of that currency relative to the dollar unless there is a nominal depreciation to offset the difference.
We publish three types of commodity exchange rate indices. Two are based on exports and one on imports. The market exchange rates describe the relative value of the U.S. dollar in key export markets for each of the commodities, products, and product aggregates that it covers. The market indices use a country’s share in U.S. exports as weights. These indices measure the impact that changes in exchange rates have on markets for our agricultural exports.
Changes in the competitive exchange rate indices reflect the trends in the competitiveness of U.S. agricultural exports in world markets. The competitive indices use each country’s share in global exports of the commodity, product, or aggregate to calculate weights. If the value of a competitive index declines, it suggests that the U.S. is becoming more competitive relative to other exporters of that commodity or product. The market and competitive indices are calculated for a comparable set of products. When U.S. markets appreciate (a depreciation in the US index), the U.S. industry becomes more competitive relative to the foreign domestic industry. When a U.S. competitor’s index appreciates, the U.S. industry becomes more competitive relative to competitor industries. A differential between the movements of the two indices points to different roots of U.S. competitiveness. Finally, the supplier exchange rate indices are weighted by U.S. imports. An appreciation in U.S. dollar terms (depreciation in local currency terms) in a supplier index suggests that import prices will come down, while a depreciation in dollar terms (appreciation in local currency terms) implies increasing import prices in dollars.
There are two ways to generate a trade-weighted index: arithmetic weighting and geometric weighting. An arithmetic index is derived by multiplying the trade weights by the respective currency, which is expressed as an index with a common base year. Commodity exchange rates in the ERS Agricultural Exchange Rate Data Set are geometric indices. An arithmetic index is simpler to calculate, but suffers from bias due to asymmetry in upward and downward movements—e.g., if an index calculated using an arithmetic average declines by 50 percent, it must increase by 100 percent to return to the initial level. A geometric index treats movements up and down in symmetric fashion and overcomes this bias. Figure 1 illustrates this bias, comparing arithmetic and geometric average-based indices for total merchandise trade.
Commodity exchange rate indices are calculated using a geometric average. First, we derive the log difference (LDit) between the country i real exchange rate at time t and its value in the base period (b) for all countries and periods.
The base period value is chosen as the monthly average for the base year. The base year in the current revision of the ERS Agricultural Exchange Rate Data Set is 2010. The data set has been rebased on a 5-year cycle starting with a 1995 base year, reflecting changes made in the IFS database and the World Development Indicators Database (World Bank, 2014).
Trade weights are country i’s share in the total exports or imports of all countries. We choose trade weights that lag behind the current period slightly to account for the fact that trade flow data are often significantly revised after they are initially published. In the current revision of the data set, we use the average of trade in 2014-16. The trade weight for country i in the product j commodity exchange rate index is:
where Ti(j) is the value of product j exports or imports between country i and the United States or the world during the base period, depending on the index. The denominator is the sum of trade flows for all countries in the data set. The real exchange rate indices (RXRIit(j)) are the antilog of the sum of the product of the weights and the log difference of the real exchange rates between time t and the base year, where the base is set equal to 100:
This equation defines a time series of indices for an individual commodity beginning in January 1970.
To demonstrate how the commodity trade-weighted exchange rate indices are derived, we use a simplified example of the U.S. Fresh and Processed Vegetable Market exchange rate index (we use only the top 11 countries in terms of export market share to derive the sample index). These 11 countries represent about 90 percent of all U.S. fresh and processed vegetables exports. We normalize the export shares of these countries so that they add up to one. Export shares are then used as weights in calculating the exchange rate index. The export shares used in the example, as well as those used to calculate the index in the ERS Agricultural Exchange Rate Data Set, are in table 1.
|Country||Normalized Export Shares*||Observed (Actual) Export Shares|
|Source: USDA, Economic Research Service.|
There are several things to note about the table. First, U.S. vegetable exports are highly concentrated: the top 11 markets account for approximately 90 percent of all U.S. exports and more than half of these exports go to Canada. A high degree of market concentration is not unique to fresh and processed vegetables; on average, approximately 75 percent of U.S. exports of agricultural commodities go to the top 10 markets.
To calculate the U.S. export exchange rate index for fresh and processed vegetables, we first create a table of real exchange rates for each country over time (table 2). For this example, we use annual exchange rates from 1970-2014 at 5-year intervals except for 2014, the last year of data in the data set. However, the process is the same whether the data set has 2 or 220 countries, and the time interval is a year, a quarter, a month, or a day.
Next, we derive a table of log differences between the real exchange rate at time t and the base period as described above. We then sum the product of the log differences and weights (this is equivalent to multiplying a vector and a matrix). We then convert these values to index levels using the exponential function. The Excel formula used to derive the commodity indices is:
for countries i = 1..I, for commodity j, at time t.
The primary source of nominal exchange rate and CPI data is the IFS database. Alternative data sources are used where they are considered to be more accurate or up to date. Where data are unavailable, interpolation is used to ensure the data set is complete and current. Where available data are considered inaccurate or misleading, various methods are used to obtain a data set that is representative of global market conditions. These cases are described in the Data Exceptions section. Trade weights are derived using 2014-16 data from the Global Agricultural Trade System (GATS) database managed by the USDA’s Foreign Agricultural Service (FAS) and from FAOSTAT, which is managed by the Food and Agriculture Organization (FAO) of the United Nations.
Nominal exchange rates going back to 1970 are mostly obtained from the IFS database. For most countries, we use the monthly average market rate in units of each country’s national currency per U.S. dollar. Where the current monthly average exchange rate is not reported in the IFS, the most recent averaged daily rate reported by the University of British Columbia, Pacific Exchange Rate Service is used.
Nominal exchange rates are converted to real rates using monthly CPIs. The CPI base year is 2010 for all countries. The CPI is obtained from the IFS database for all countries except Taiwan. When CPI data is not available up to the current month, values are projected forward using a 6-month moving average in order to have a complete data set for every country and commodity. CPI values for China and Russia are only available in annual percent-change terms. We convert the annual percent change into monthly percent changes and create an index with a 2010 base year.
U.S. export and import weights for the market and supplier indices, respectively, are calculated based on bilateral trade flow data from the GATS database. The GATS database collects and categorizes U.S. Census data on trade in agricultural and agriculture-related products. World export weights for the competitive index are calculated from each country’s commodity-level world trade flow data, which are obtained from the FAOSTAT database.
The data described above allow us to report a complete monthly series of nominal, real, and trade-weighted exchange rates for 79 countries from January 1970 to the present month. Table 3 provides a complete list of the countries and currencies covered by the exchange rate database, listed by region.
Several criteria are used in selecting countries included in the data set. First and foremost, the countries included must reflect the bulk of U.S. agricultural exports for each covered commodity. In all cases, more than 90 percent of U.S. exports are represented by the countries chosen. A second objective is to have representative coverage of regions around the world. For example, African countries were selected based on their overall economic importance. Lastly, data availability is a constraint on country selection. Only countries with available data on nominal exchange rates and CPIs sufficient to derive a relatively reliable and complete data series were considered.
Real, trade-weighted exchange rate indices are calculated for major U.S. agricultural exports and imports and for product aggregates. The commodities and product aggregates covered are listed in table 4, and aggregates are in bold print. Most product aggregates for market and supplier exchange rate indices are defined as in the BICO HS-10 level used in the GATS database. The exceptions are the supplier indices for consumer-oriented and agriculture-related products, where we use the BICO HS-6 definition. Product aggregates for competitive exchange rate indices are calculated from the total global exports of the individual commodities listed within each category. For example, the weights for global bulk exports are calculated from the sum of each country’s exports of corn, cotton, rice, soybeans, tobacco, and wheat only.
Commodities were chosen to represent the most important U.S. agricultural exports and imports. Commodity coverage for competitor exchange rate indices is intended to duplicate the set of market exchange rate indices as nearly possible. This facilitates comparison of exchange rate changes in U.S. export markets with competitor exchange rates in order to evaluate changes in the U.S.’s competitive position in export markets. No comparable commodity to distiller’s grains exists in the FAO data set, so it is omitted from the set of competitor exchange rate indices.
In order to maximize its value for understanding patterns in global currency markets and their effect on the U.S. agricultural sector, the ERS Agricultural Exchange Rate Data set is published without missing values. To assemble such a data set occasionally requires us to modify underlying data or fill in missing values. In some cases, there are simply gaps within a series. In these cases, we fill in missing values using arithmetic interpolation. We also modify our approach to calculating commodity exchange rate indices to address instances of hyperinflation. In these cases, observed exchange rates provide a misleading picture of global markets. We discuss our approaches to addressing these issues in more detail below.
Russian nominal exchange rate and CPI—There are no available series for the Russian nominal exchange rate before 1992 since the country was then a part of the USSR. In the ERS Agricultural Exchange Rate Data Set, we developed a smoothed series going back to historical exchange rates and inflation rates in the Soviet Union. This is done to prevent the ruble’s instability during the country’s transition from distorting the trade-weighted exchange rate data set in the averaging process. Between 1988 and 1992, the ruble underwent a major real appreciation. Because the inflation rate reached into the thousands of percent, the real ruble most likely appreciated several thousand times, making it unworkable as a currency for exchange in trade. By the end of the transition period, it appears that those in charge of macroeconomic policy recognized the problem and began to move toward a workable real exchange rate.
Hyperinflation—Peru, Nicaragua, Ghana, and Mozambique all had periods where inflation exceeded 1,000 percent in the early years of their series. Instances of hyperinflation present major challenges for the development of real commodity trade-weighted exchange rate indices. Since we use fixed trade weights, including a country with hyperinflation can create a major distortion of the overall indices. Even if these countries account for a very small share of trade, an appreciation exceeding one thousand percent will dominate the change in the indices. During the periods of hyperinflation in such cases, the currency is unlikely to be used for exchange in international trade. Therefore, to preserve the relevance of the exchange rate index, we modify the existing data of these countries to reduce the temporary swings in real exchange rates from dominating the aggregated series.
The Euro Zone—Prior to 1998, the countries of the Euro Zone issued their own national currencies (French francs, German deutschmarks, etc…). To develop a continuous series of exchange rates for the Euro Zone countries, we converted these national currencies to euros based on their exchange value on the conversion date, then used the percent changes in national currencies to present the entire time series in euro terms. To create a time series of nominal exchange rates for the euro going back to 1970, we use a GDP-weighted average of changes in the Euro Zone national currencies. The euro CPI series was similarly derived as the GDP-weighted average of the Euro Zone countries.
CPI series for Australia, New Zealand, and Bangladesh—These countries only publish a CPI on a quarterly basis. We treat the quarterly values as the 2nd month of the quarter and arithmetically interpolate the missing values.
Taiwan—Taiwan is a major U.S. trading partner, but data for Taiwan is not available in the IFS database. However, the Taiwan Government’s Directorate of Budget, Accounting and Statistics makes the CPI and nominal exchange rates available online (Republic of China, Taiwan, 2014). We use the Taiwan exchange rate reported by the Board of Governors of the Federal Reserve System (2014) as a source for nominal exchange rates and the Taiwan government as a source for the CPI.
The nominal or current exchange value of foreign currencies in U.S. dollars is the appropriate exchange rate for the purposes of travel or making current purchases of goods and services in foreign currencies. However, the ERS Agricultural Exchange Rate Data Set is primarily intended to expose the long-term pattern of country and commodity exchange rate movements over a historical period. For countries with similar inflation rates, such as the United States and European countries, the nominal exchange rate will provide an approximate indication of the long-term pattern. However, nominal exchange rates provide a misleading indication of exchange rate movements over time when inflation rates differ. In such cases, the real exchange rate, which corrects for price movements, is more appropriate.
To see how using the nominal exchange rate to understand movements in currencies may lead to inaccurate economic analysis, consider an example. Suppose country has an inflation rate averaging 25 percent per year. The average annual U.S. inflation rate over the past 20 years has been approximately 2.5 percent, which implies a difference of 22.5 percent per year. If the nominal exchange rate between country ’s currency and the U.S. dollar remained constant over a 20-year period, the real exchange rate would have appreciated by 22.5 percent per year on average. In this case, the real exchange rate would reveal that the price-adjusted purchasing power of the foreign currency appreciated more than 25 times relative to the dollar over that period, which would mean that country has a highly over-valued exchange rate. In turn, country would have experienced a drastic decline in their ability to export and earn foreign exchange while domestic goods would have become expensive relative to foreign goods, creating an enormous incentive to import from overseas. In such a circumstance, the likelihood of a foreign exchange crisis in country would be very significant. In this case, the constant nominal exchange rate would mask country’s underlying economic position.
Changes in exchange rates over long periods of time represent broad movements in prices for traded commodities and products. Exchange rate intervention is the most powerful policy instrument countries have to affect relative prices and the competitive position of exporters and importers in their economy. By executing a currency devaluation or revaluation, a country’s central bank simultaneously changes all measured border prices by a comparable percentage. The final change in traded prices will differ depending on initial demand and supply responses, border protection measures, and the dynamic process that follows in both exporting industries and import-competing industries, which will compete for resources under changing relative prices.
Movements in exchange rates are a leading indicator of movements of U.S. exports. In examining long-term real exchange rate movements, it is clear that there have been substantial swings and relatively long periods of both appreciation and depreciation. Between 1980 and 1985, the real dollar appreciated 57 percent, undermining the competitiveness of U.S. agricultural exports and contributing to low commodity prices. Conversely, between 2002 and 2011, the dollar depreciated 30 percent, resulting in record agricultural exports between 2010 and 2013.
Many researchers use broad measures of exchange rates such as the Federal Reserve Board Broad Exchange Rate Index, which is a real trade-weighted index of total merchandise trade. Such a measure is useful as an overall indicator of exchange rate change. A country’s weight in the index is an average of the share it represents in U.S. exports, its share of global exports, and its share in U.S. imports. Such an index, however, is a poor measure of what is happening to the competitive position of the United States relative to its competitors in merchandise trade. It is an even poorer measure of what is happening to exchange rates relevant to trade in a specific commodity, such as U.S. soybean or wheat exports. As an example, the U.S. wheat and soybean export-weighted real exchange rate indices are presented in figure 2. The underlying patterns are substantially different—there are periods of time when the real exchange rate index is significantly higher for wheat than soybeans, and vice versa.
The ERS Agricultural Exchange Rate Data Set provides 79 monthly and annual real country exchange rates and 49 real commodity trade-weighted exchange rate indices relevant to U.S. agriculture. Because the commodity indices are derived by weighting exchange rates of relevant countries for each commodity, they comprise a much better measure of how currency movements will affect U.S. agricultural trade than a broad measure of the effective exchange rate.
Exchange rates are, arguably, the single most important price for any economy with significant trade. Knowing exchange rate movements is critical for predicting movements in both U.S. exports and imports. ERS research has demonstrated that, while long-term growth in U.S. agricultural exports is mainly driven by growth in foreign income, year-to-year variation in U.S. agricultural exports is largely driven by changes in exchange rates (Shane et al., 2008). When the value of the U.S. dollar has been high relative to our trading partners’ currencies (periods of appreciation), U.S. agricultural exports have been restrained. In contrast, when the value of the dollar has been relatively low (periods of depreciation), exports have hit highs.
The U.S. agricultural economy is heavily dependent on exports for growth. In commodities such as wheat, approximately half of all production is exported. In total, approximately 20 percent of all production is exported. Periods where the dollar’s value has been relatively low have been good times for U.S. agriculture. A depreciated dollar results not only in strong exports, but in relatively high prices for agricultural commodities and products. Thus, strong exports of U.S. agriculture have also resulted in high agricultural income. In a world where communication is almost instantaneous (even to the most remote areas), and where ever-improving transportation services have brought down the share of costs associated with transportation, exchange rates and trade will continue to be very important measures of the success of U.S. agriculture.
Federal Reserve System, Board of Governors. 2005. Indices of the Foreign Exchange Value of the Dollar, Federal Reserve Bulletin, Winter 2005. (accessed August 2014).
Federal Reserve System, Board of Governors. 2014. Foreign Exchange Rates-H.10. (accessed August 2014).
International Monetary Fund. 2014. International Financial Statistics, Washington, DC. (accessed August 2014).
Orden, D., and P. Fackler. 1998. “Identifying Monetary Impacts on Agricultural Prices in VAR Models,” American Journal of Agricultural Economics 71(2):495-502.
Republic of China, Taiwan, National Statistical Bureau. 2014. Consumer Price Indices, Directorate of Budget, Accounting and Statistics, Taipei City. (accessed August 2014).
Schuh, G.E. 1974. “The Exchange Rate and U.S. Agriculture,” American Journal of Agricultural Economics 56(1):1-13.
Shane, M., T. Roe, and A. Somwaru. 2008. “Exchange Rates, Foreign Income and U.S. Agricultural Exports,” Agricultural and Resource Economics Review 27(2):160-75.
United Nations, Food and Agriculture Organization. 2014. FAOSTAT, Rome. (accessed August 2014).
University of British Columbia. 2014. Pacific Exchange Rate Service, Sauder School of Business. (accessed August 2014).
U.S. Department of Agriculture, Foreign Agricultural Service. 2014. Global Agricultural Trade System (GATS). (accessed August 2014).
World Bank. 2014. World Development Indicators Database, Washington, DC. (accessed August 2014).