Journal Issue
Share
Article

Getting a truer measure of producer price inflation

Author(s):
International Monetary Fund. External Relations Dept.
Published Date:
November 2004
Share
  • ShareShare
Show Summary Details

Ever since the mid-1990s, when Alan Greenspan, Chair of the U.S. Federal Reserve Board, labeled the U.S. consumer price index biased, the search has been on for more accurate price indices. That effort is now bearing fruit with the release, in July, of the International Labor Office’s revised Consumer Price Index Manual and with the publication, in September, of the IMF’s new Producer Price Index Manual. Paul Armknecht, a Deputy Division Chief in the IMF’s Statistics Department, explains why statistical improvements in price indices are needed and what users can expect from the new producer price index (PPI) manual.

In a sense, Greenspan’s concern—that the index used to measure consumer prices was overstating inflation—was a fairly technical matter. But it was a technical issue with huge implications for the federal budget. Since the U.S. government uses the consumer price index (CPI) to index payments for such large-scale programs as social security, welfare, and federal pension payments, even a small increase in the CPI translates into significant sums for the U.S. budget.

Spurred by the Greenspan remark, the U.S. Senate Finance Committee convened an advisory commission, chaired by Michael Boskin, to examine the accuracy of the CPI. Its report, released in December 1996, suggested that the U.S. CPI overstated inflation by between 0.8 and 1.6 percent annually. Why this upward bias? Principally, the report said, because the market basket of products and shops had been kept fixed for long periods and because improvements in the quality of products were not being adequately captured.

Why the IMF took the lead in PPI

In the early 1990s, when the Soviet Union dissolved and a number of independent countries were being formed, the new national statistical agencies found themselves with no ready provider of technical assistance in price statistics. The International Labor Office, which has technical oversight for methods and procedures in the consumer price index, lacked the budgetary resources to provide assistance on producer prices. The UN Statistics Division, which produced the previous manual on producer price indices in industrial activities, had discontinued its work in this area for similar reasons.

That left the IMF, which needed measures of inflation to gauge economic performance, to provide technical assistance through its Statistics Department. Over the years, the IMF has continued to help countries worldwide develop the means to measure consumer and producer price inflation.

This potential bias in the CPI raised questions, too, about the accuracy of the PPI. CPI measures changes in the prices of goods (such as milk and appliances) and services (such as rent and auto repair) bought by households, while the PPI measures inflation for goods (for example, steel girders and cases of tomato paste) and services (for example, accounting and legal services) bought and sold by businesses. A true picture of the inflationary pressures facing an economy requires an accurate measurement of both the CPI and the PPI.

Fixing what’s broken

The issues first raised in the United States had ramifications for statistical offices around the world. In response, the international organizations concerned with inflation measures formed the Intersecretariat Working Group on Price Statistics. Atop that group’s agenda was the development of new manuals on best practices for price indices. The IMF’s Statistics Department, which is the primary provider of technical assistance on inflation-related statistical topics and is heading the effort to develop international standards for the PPI (see box), was asked to take the lead in improving producer price statistics.

Under its guidance, preparation of the new manual focused on recent developments in relevant statistical theory and practice as well as key changes in the economy and weaknesses that had been identified in current practices. In particular, the IMF Statistics Department identified six weaknesses:

Problems in standard fixed baskets. Advances in index number theory have pointed to weaknesses in the standard fixed basket (Laspeyres-type) indices commonly used to develop a PPI. The manual suggests that a better market basket is one that includes both current and comparison periods (symmetric index) in the calculation. Three such indices—the Fisher Ideal index, the Walsh index, and the Törnqvist index—offer more sophisticated approaches.

Low-level indices. Recent evidence suggests that there is a significant potential for upward bias in lowlevel indices where prices for detailed products, such as sheets of rolled aluminum, are first combined to produce an index. Currently, low-level indices are compiled as an unweighted, simple average of price observations. This generally gives more importance to products with larger price changes than to those with small or declining price changes. The manual recommends that geometric averages of price changes be used to help resolve this problem and that detailed product weights information be obtained to compute weighted averages.

Quality differences. At present, no, or inadequate, adjustments are made for quality differences among existing and new products. If no adjustments are made for quality differences, price differences due to improved quality are counted as price changes. Newer statistical models, such as “hedonic” regression analysis, permit compilers to measure the quality component and adjust for the quality contribution using standard index number methods.

Seasonal products. At present, seasonal price movements distort monthly price movements and make it more difficult to measure underlying trends. The manual recommends applying statistical methods to reduce the effects of seasonal changes and smooth the time series.

PPIs typically exclude services. While services are the fastest growing sector in most economies, it is often difficult to measure price changes in this sector, and most countries are still reluctant to include services in the PPI. To encourage the inclusion, the manual highlights the experience of a number of countries that have been successful in including more services.

Family of indices needed. Because inflation does not affect all segments of the economy equally or during the same time period, a single index number does not satisfy the needs of most users. The manual recommends using a family of indices to measure the PPI by industry, product, stage of processing, and the like. This allows the PPI to track inflationary movements throughout the chain of distribution in the economy (from primary products to intermediate goods and services to final demand).

Photo credits: Henrik Gschwindt De Gyor, Eugene Salazar, and Michael Spilotro for the IMF, pages 329–31, 336–38, 342, and 344; Tatyana Makeyeva for AFP/Getty Images, pages 332-33; Yuri Kochetkov for AFP/Getty Images, page 333; H. Salgado for World Council of Churches, pages 334–35; Jung Yeon-Je for AFP, page 339; and Grigory Dukor for Reuters, pages 340-41.

Reading the manual

The IMF’s Producer Price Index Manual pulls together both analysis and best practices in compiling more accurate data and is intended for both expert and general audiences. Chapters 1 and 2 are intended for general readers, with an overview and summary in the first chapter that can serve as a primer on index number theory and practice. The second chapter offers background on index number history and the uses of the PPI.

Chapters 3-13 are meant to provide guidance for compilers on coverage, construction, processing, and publishing the producer price index. Chapters 14-22 present the concepts and theory of the producer price index for professional economists, students of economics, and more advanced compilers.

Copies of both the Producer Price Index Manual: Theory and Practice (English) (French, Spanish, and Russian forthcoming) and the Consumer Price Index Manual: Theory and Practice (English) (French, Spanish, and Russian forthcoming) can be ordered from IMF Publication Services for $125.00 each. See page 335 for ordering details. The IMF’s Statistics Department is currently also developing the Export/Import Price Index Manual, which is expected to be available late next year.

Other Resources Citing This Publication