Is the CPI or PCE the "Correct" Price Target for Monetary Policy? Both Are Flawed Price Measures

The economic profession and the Bureau of Labor Statistics (BLS) statisticians have been debating the proper construction of price indices for decades. It is probably a good time to renew the debate since published price indexes have a new use: the official price target of the Federal Reserve.

The debate has almost always centered on whether published price indexes such as the consumer price index (CPI) should be a price index or a cost-of-living index. A detailed discussion of the issues surrounding the difference between a price index and a cost of living index can be found in two government reports; "At What Price? Conceptualizing and Measuring Cost of Living And Price Indexes" by the National Academy of Sciences, and "Measurement of Homeowner Costs in the Consumer Price Index Should Be Changed" by the General Accounting Office.

The CPI was initially designed to be a "buyers" price index as it was meant to capture the price change of currently purchased goods and services. Yet, many found fault with the construction of the CPI arguing that it should be designed more along the lines of a cost of living index since it was used to adjust benefits in government social programs, as an escalator for wage contracts and to evaluate overall economic policy.

The measurement of owners housing costs has always been at the center of the price and cost-of-living debate. Critics argued that the initial design of measuring the change in house prices and interest rates did not reflect the average costs incurred by the consumer from month to month. Yet, when a group of academics and statistical experts first proposed a formal change to a rental base measure the BLS Commissioner disagreed arguing that type of measure "could not be justified in a price index but only in a measure of the cost of living." And, he further argued that if adopted it would make the CPI a hybrid price series, which could lead to ambiguity and not be suitable for all of the uses of the CPI.

Nevertheless, in 1981 after further review and public debate and a new Commissioner BLS formally shifted to the rental base measure for owners housing, with a start date of January 1983. The tipping point for the change appears to have stemmed from an important new use of the CPI. The Economic Recovery Tax Act of 1981 required that the tax brackets be adjusted every two years for cumulative change in the CPI. At that time many inside and outside of government felt that the "old" construction of the CPI overstated the rate of inflation and in order to prevent a substantial revenue loss to the Federal Government a new CPI design was needed.

Fast-forward to 2012, another important new use for price indexes occurred when the Federal Reserve formally announced it price-targeting scheme. Yet, instead of choosing the widely used CPI policymakers picked the CPI sister measure, the personal consumption expenditure (PCE) deflator.

What is odd about this decision is that there was no formal review or public debate. One would think a decision of this magnitude would have at least been analyzed and reviewed by a panel of experts inside and outside of the Federal Reserve.

Before BLS announced the change in the CPI in 1981 they began publishing, starting in January 1980, 5 alternative experimental measures of the monthly rate of change for different measures of homeowner costs and what the monthly change would have been the prior decade if each experimental measures had been in place. Thus, when the actual change in the CPI occurred in January 1983 users were not surprised.

Also, the PCE, like the CPI, is a hybrid measure as it includes market prices and non-market prices. Within both series, the non-market rent component creates two problems: first, it’s the largest single component and therefore can create the illusion of price stability, and second, the rent estimate misses important price signals from the all-important housing market.

Price statistics, as is also true with all other economic statistics, must always be as accurate as possible and relevant to its users. There is no absolute perfection in price statistical methodology as one price index that is good for one purpose might not be an accurate gauge for another.

The debate over the proper construction of price indexes is no longer academic as monetary policy decisions are now being based on the published price series. In my view, published price series are not suitable for the Fed’s price targeting scheme as they miss important (actual) prices from the real estate market. A new broad transaction based price index should be created to better inform policymakers of actual economy wide inflation. The statistical agencies should welcome the challenge as all statistics and statistical methodology must constantly be reviewed and updated so to be relevant to the public and policymakers. And policymakers should support the research project as they may find that when actual prices are employed economy-wide inflation is at or above their preferred target.

Viewpoint commentaries are the opinions of the author and do not reflect the views of Haver Analytics.