A serious policy debate is currently raging in the U.S. on whether we are mismeasuring the level of GDP, the rate of productivity growth and the inflation rate.

In December 1995, the U.S. moved to a new system of measuring GDP and the GDP deflator: from a fixed-weight method to a chain-weight method.

The problem with the traditional (fixed-weight) method of measuring output was that too much weight was given to production of the goods (such as computer and microchips) whose price has fallen over time. A Short Technical Note explaining the difference between fixed and chain weights is presented in the NY Fed Web site.

However, the move to the new method led to some surprising implications that have become a contentious source of debate among economists and policy makers. Using the new chain-weight method to measure output implies that:

1. GDP growth after 1987 is lower than previously estimated (2.2% per year in the 1990s against the 3.0% obtained using the fixed weight method).

2. Productivity growth is lower than previously estimated (1.4% per year in the 1990s against the 2.2% obtained using the fixed-weight method).

These numbers look dismal because many economists believed that the process of corporate restructuring, reengineering, down-sizing of the last decade had led to a major resurgence of productivity. The new chain-wighted numbers seem to deny that such productivity resurgence ever occurred.

Several economists had argued that, even with the new chain-weight method, there serious problems of mesurement that tend to underestimate output and overestimate inflation:

1. Quality changes are not correctly measured (examples: computers, light) leading to under-estimate of the product of industries where such quality changes occur.

2. Major productivity growth in the service industries (ATM’s, telecommunications, quality of health care) not measured by standard GDP measures.

An interesting New York Times article discussing the problems in measuring GDP is "Counting the Wealth of Nations: G.D.P.'s Accuracy Is Under Attack From All Sides.
 

On a separate but related front, many economists had argued for a long time that the (CPI) inflation rate (that is calculated by Bureau of Labor Statistics at the Department of Labor) is probably overestimated because of a number of biases in its measurement. A formal Commission headed by Stanford Economist Michael Boskin studied these issues and presented the results of its report on December 5, 1996. The Commision argued that that inflation is likely to be overestimated by about 1.0% to 2% per year. The sources of this bias in CPI inflation are:

1. Substitution bias (0.2 - 0.4 points per year)

2. Oulet bias (0.1 - 0.3 points)

3. Quality changes (0.2 - 0.6 points)

4. New products (0.2 - 0.7 points)

5. Formula bias (0.3 - 0.4 points)

Total bias is: 1.0 to 2.0 % per year.

The conclusions of the Boskin report have led to a wide debate on whether we are mismeasuring inflation. The Washington Post home page has a Page on The Cost-of-Living Debate with many useful links to articles, documents and alternative views of the Boskin Commission results. See also the official Bureau of Labor Statistics (BLS) home page on the debate for the BLS reply to the Boskin report criticism of its measurement of inflation. For the views of Krugman on the CPI controversy see his article on Slate. See also New York Times articles on the Boskin Commission, the political difficulties in implementing the Boskin Commission recommendations and a critique of the Boskin report from a liberal think-tank. For a Wall Street view of the report, see Stephen Roach's (chief economist at Morgan Stanley) comments on November 27 and December 4, 1996.

The Fed's view of the Boskin report has been presented by Chairman Alan Greenspan in his testimony before the Senate Finance Committee on January 30. He recommended a two track approach to solving the CPI problem. First, he suggested that the BLS address the measurement issues addressed in the Boskin report. He wants Congres to give the BL sufficient resources to update the CPI market basket annually and to improve its efforts at measuring quality changes. Second, he recommends the establishment of an independent national commission that would annually determine the cost of living adjustments to be applied in all government programs (including social security). His views were recently elaborated in a speech in November 1997.

The results of the Boskin study (and those of Nakamura, a Philadelphia Fed cconomist) have important consequences for the estimates of real wage and income growth in the last two decades. They imply that previous studies suggesting a secular fall in median real wages might be incorrect. For a summary of Nakamura's results see a NYT article and a Washington Post article . They have also implications for the controversy on the Increasing Income Inequality in the US.

The debate on whether the output and productivity of the service sector is correctly measured has also reached the Federal Reserve Board. In the Minutes of the September 24, 1996 meeting of the FOMC, several FOMC members expressed such an opinion that has been endorsed in public by the Fed Chairman Greenspan. Also, the fact that low unemployment and full capacity has not been associated with with higher inflation has been explained in part by the fact that productivity growth is likely to be underestimated. In words of the minutes:

"Widespread reports suggested major gains in productivity in numerous industries, induced in recent years by business restructuring and related activities and by large capital investments that had introduced increasingly productive equipment. Although currently available measures of productivity for the economy as a whole showed only weak gains, sectoral disaggregation of the data gave reasons to question the productivity measurements. Productivity had increased fairly sharply in manufacturing, and the slowdown in overall productivity since 1973 had been concentrated in the service areas of the economy. Indeed, measured productivity in noncorporate businesses--largely services--had displayed a negative trend for many years. This result was implausible and suggested considerable error in estimating output and prices for many services. Consequently, it was likely that actual productivity growth was higher than the current measures indicated. By the same token, the rate of price inflation was lower than had been reported, consistent with the findings of a number of studies of distortions in published price data."

The current debate on the chain-weight system of measuring GDP and the biases in measuring the inflation rate is also tightly related to the question of whether we are correctly measuring productivity growth and whether there has been a resurgence of productivity growth in the 1990s after the dismal productivity experience in the 1973-1990 period. On this debate and the related issue of the productivity slowdown see the home page on the controversy Productivity Growth, Its Slowdown in the 1973-90 period and its resurgence in the 1990s: Truth or a Statistical Fluke?