BRIEFING ON THE BLS REACTION TO THE FINAL REPORT OF THE ADVISORY COMMISSION TO STUDY THE CONSUMER PRICE INDEX - December 19, 1996

The Commission recommends that the Bureau of Labor Statistics (BLS) adopt production of a cost-of-living index as its objective in measuring consumer prices (Recommendation i from the Executive Summary).

The BLS already operates within a cost-of-living framework in producing the Consumer Price Index (CPI) and will continue to do so. As is stated in the BLS Handbook of Methods, this framework guides operational decisions about the construction of the CPI. We would expect to adopt changes to the CPI that moved it closer to being a COLI, provided there were not compelling associated disadvantages, in terms of timeliness, understandability, reproducibility, objectivity or variance, for example.

It should be recognized that the CPI is limited in scope. In economic theory, there are many alternative cost of living indexes. Any COLI reflects the ability of consumers to substitute among goods and services as relative prices change. But a COLI can be defined either net or gross of income taxes, including or excluding changes in services provided by governments, or even including or excluding changes in the natural environment. The CPI excludes income taxes and the services paid for through taxes, and it excludes the effects of AIDS and other environmental factors mentioned in the Commission's Report. Saying that the BLS operates within a cost-of-living framework in producing the CPI does not carry any implication regarding the appropriate scope of the index.

Construction of the CPI Subindexes

The Commission recommends adoption of the geometric mean formula below the stratum level, arguing that it more accurately reflects consumer substitution (Recommendation v). The Commission's estimate, drawn from BLS research, is that this step would slow the rate of growth in the CPI by 0.25 percent per year.

BLS is about to begin producing a monthly experimental geometric mean index for users to see and evaluate.

In some strata, assuming a high degree of substitutability seems reasonable (e.g., bananas, or cakes, cupcakes and cookies).

There are other strata, however, in which the assumption about substitutability embedded in the geometric mean formula seems less obviously appropriate (e.g., prescription drugs, or physicians' services).

A complete list of the CPI item strata and a brief description of the geometric mean issue (with a list of examples of likely high-substitution and lower-substitution strata) have been included in your briefing package.

Following a period of evaluation, the BLS will make a decision regarding adoption of the geometric mean formula in producing the CPI. Any such adoption most likely would not apply to all strata.

Aggregation of the CPI Subindexes

Superlative measures use special formulas and current expenditure data. Under certain assumptions about, for example, market equilibrium, they accurately measure changes in the cost of living by taking account of how consumers are actually substituting in response to relative price change. Recent BLS research cited by the Commission indicates that superlative measures constructed to account for cross-strata substitution have risen by about 0.15 percent per year less rapidly than the CPI.

These superlative indexes cannot be produced in real time precisely because they require current expenditure data, which is impossible to collect and process on a monthly turnaround.

The Commission proposes that the BLS replace the formula currently used to aggregate the component subindexes to form the overall CPI with a geometric mean formula using annually-updated expenditure weights (Recommendation iii). Such a measure would not be a true superlative index. The Commission's apparent intent is to suggest a measure that could be produced each month on a timely basis and would more closely approximate, at least empirically, the movements of a true superlative.

The specific measure proposed almost certainly would understate the true change in the cost of living.

People often assume that substitution bias would be solved by updating the CPI expenditure weights more often. This is not the case, either theoretically or empirically.

More frequent weight updates might well be desirable in their own right, though annual weight updating could create its own problems (under certain circumstances, it could lead to a bias called chain drift).

Our initital reaction to the idea of adopting an ad hoc formula to correct for substitution bias in the CPI is negative. We would strongly prefer to devote our energies to producing a true superlative index, necessarily an after-the-fact measure but something we could comfortably defend as an accurate cost-of-living index.

The Commission also recommends production of an annual supplemental measure that would use a superlative formula (Recommendation iv). It would be published with a lag and subject to revision as new data become available.

Such a measure probably would be similar to several experimental superlative indexes that we have been publishing since 1993 (using Fisher and Tornquist formulas, chained and fixed-base weights).

We are receptive to the spirit of this recommendation.

Quality, New Goods and New Kinds of Outlets

Nearly two-thirds (0.7 percent per year) of the 1.1 percent per year overall bias in the CPI as estimated by the Commission comes from alleged deficiencies in the treatment of changes in the quality of goods and services and the emergence of new goods (0.6 percent) and the emergence of new kinds of outlets (0.1 percent).

Impact of Actual Adjustments Made for Quality Changes

Some seem to have the impression that the BLS makes no adjustments for quality changes in computing the CPI. This is not a correct impression.

The latest estimates of the impact of quality adjustments on the CPI cover the year 1995 and measure the effect of all such adjustments made in the routine compilation of the goods and services portion of the index (approximately 70 percent of the total index, with shelter the principal exclusion). During 1995, more than half of the price change reported for goods and services items was adjusted out: the price increase in this part of the index was 4.7 percent over the year before quality adjustment versus 2.2 percent over the year after the quality adjustments applied to produce the official index.

This does NOT mean that the methods used in constructing the CPI capture quality change perfectly, only that the quality adjustment procedures currently in place have a profound impact on reported price change.

The Advisory Commission's Quality/New Goods Bias Estimates

Hard empirical evidence regarding the magnitude of biases from these sources is extremely limited. The Commission's estimates of these biases rely heavily on the members' best judgment as to the value to consumers of various marketplace developments, as opposed to a comparison of the CPI against any alternative measure that the Commission suggests the BLS might implement. (As one of the Commission members has commented, the numbers are "squishy")

Some examples of conjectured impact of quality change or new goods on the accuracy of the CPI follow (all estimated biases are upward; the Commission looks at 27 categories of the CPI and finds no examples of current downward bias):

Food. "How much would a consumer pay to have the privilege of choosing from the variety of items available (today) ... instead of ... the much more limited variety available 30 years ago? A conservative estimate of the value of extra variety and convenience might be 10 percent for food consumed at home other than produce, 20 percent for produce where the increased variety in winter (as well as summer farmers' markets) has been so notable, and 5 percent for alcoholic beverages where imported beer, microbreweries, and a greatly improved distribution of imported wines from all over the world have improved the standard of living. Increased variety and convenience in food away from home ... can also be credited with a 10 percent premium." (pp. 41-42 of the Report) The conjectured bias in this area arises not because of any alleged failure to measure prices accurately or to measure quality change successfully, but rather because the CPI does not account for an increase in the variety of products available.

House furnishings other than appliances: "Regarding housefurnishings other than appliances and video-audio products, there is no available research to provide guidance... There have been many new products in this area, including furniture and fabrics that are much less susceptible to damage by stains and childrens' accidents than was previously possible. This cateogry also includes soap and cleaning products, where substantial progress has been made. We view a bias rate of ... 10 percent over the past 30 years as conservative." (p. 50)

New and used cars: The Commission begins by noting that there has been a significant increase in the average age of cars on the road. All of this increase is assumed to be due to increased durability of cars. In addition, it is assumed that the CPI incorporates no adjustments for changes in durability. Given these assumptions, the Commission estimates that increased durability has imparted a 0.59 percent annual upward bias in the CPI over the recent past. No adjustment is made for the decreased incidence of defects as measured by the J.D. Power survey; the Commission thus believes its estimate to be conservative. (pp. 52-56) As noted by the Commission itself, some might question whether all of the increase in the average age of cars on the road reflects an increase in durability. In addition, it is clearly wrong to assume that the CPI does nothing to take increased durability into account. Your briefing materials contain a list of some durability-related quality adjustments to auto prices made in the CPI since 1992.

Apparel: The Commission compares the CPI to an index constructed using data on items from Sears catalogs that remain unchanged from one year to the next. The CPI went up 1.92 percent per year faster than the Sears index over the recent past. This figure is divided roughly in half to arrive at the Commission's estimate of a 1.0 percent per year bias since 1985 in the apparel component of the CPI. (pp. 50-51) No one would suggest that the CPI be constructed using data from a single retail catalog. In addition, given the importance of fashion in the apparel market, prices for items that remain unchanged from one year to the next may well show less increase than prices for other apparel items.

High-tech consumer goods (pp. 48-49) and medical care (pp. 57-59): These are areas where we agree there are important limitations to our quality-adjustment procedures. Improvements in those procedures are planned. Even after these previously-announced changes have been implemented, however, issues will remain that it may never be possible to resolve fully.

In general, the Commission's discussion of quality/new goods biases does not include explicit recommendations regarding the adoption of procedures to correct the problems it believes exist. Progress on the Quality Adjustment Front

As many have said before, dealing with the quality adjustment issue is the "house-to-house combat" of constructing price change measures.

Although it is not clear that we will ever find satisfactory means of dealing completely with quality change, we do believe further progress can be made and will continue to work toward this end.

Activities that could be undertaken in support of this effort in the reasonably near term, resources permitting, include:

Use of hedonics for some additional components of the index.

Adoption of more aggressive procedures for identifying new goods and incorporating them more promptly into the samples of items priced.

Expansion of resampling efforts might allow the BLS to capture the emergence of new outlets more quickly. Comparison of the prices charged by different retail outlets is complicated by the fact that different outlets offer different shopping environments. Research on the factors affecting consumers' choices about where to shop ultimately may be helpful in devising appropriate procedures for making such comparisons.

Conclusion

• The issues identified in the Commission's Report are not new problems. BLS has been aware of the measurement issues concerning the CPI and, over the years, has made efforts to study and advise users of the data about them.

• At the same time, we welcome new information and are confident that public debate will contribute to the development of new and improved methods for constructing the CPI.

• There are a variety of longer-run research activities that the Commission recommends. These include recommendations that we investigate pricing of vehicle services using leasing data; direct pricing of health insurance policies; increased price collection on weekends; and a set of sample design issues. We will pursue all of these issues as aggressively as time and resources permit.

---