Price and Quantity Measurement Research Priorities
Comments by Katharine G. Abraham, Commissioner, U.S. Bureau of Labor Statistics, on
Price and Quantity Measurement Research Priorities
National Bureau of Economic Research Panel Discussion
July 21, 1999
Were experts in the field to be polled, I believe there would be substantial agreement among them regarding the most fundamental problems of price and quantity measurement. One major problem area is service sector measurement. Measurement can't begin without an output definition, and, for many services, there is no consensus regarding the appropriate definition. Customization of output is an important complication. If no two customers purchase exactly the same product, what rules can be used for defining the output that has been produced? This isn't a problem that is restricted to the service sector - it long has been important in construction and may be increasingly important in goods production more generally - but it certainly is a very serious problem for service sector measurement. Additional conceptual problems arise because, as has been noted by Zvi Griliches (Griliches, 1992), the output of a service activity like medical care or teaching depends critically on consumers' characteristics and the nature of their participation in the service delivery process. More globally, price and quantity measurement are complicated by technical change, which may lead both to changes in the characteristics of goods and services and to the development of wholly new goods and services.
The nature of these problems is such that no general solution to them is possible. Rather, much of the work to improve our price and output measures must proceed on an industry-by-industry, product-by-product basis. That's not to say there won't be spillovers from studies in one industry or product that inform thinking about another, but rather that each ultimately will require individual analysis.
Perhaps the best way for me to contribute to this discussion of priorities for research on price and output measurement is to talk about the research currently being done at, or supported by, the Bureau of Labor Statistics (BLS), particularly research that relates to the fundamental problems I alluded to a moment ago. I should acknowledge that the agenda implicit in this research probably is not a 20-year agenda. By and large, the projects in which we're investing are projects that we would hope might pay off in the form of improved measurement methods that could be implemented within the next five to ten years.
One important thing we're involved in that might not generally be considered research is a major interagency - indeed, international - effort to develop a comprehensive product classification structure (Office of Management and Budget, 1999). In addition to staff from the BLS, the Bureau of Economic Analysis (BEA) and the Bureau of the Census, Canada and Mexico also are engaged in this effort, which follows work done to develop the new North American Industry Classification System. I mention product classification in the present context for the reason that, if done well, a service sector product classification structure will go a long way towards defining the output of the service sector in an operationally useful fashion.
Secondly, as part of the Consumer Price Index (CPI) Improvement Initiative for which the BLS first received funding in 1998, we are working to expand the use of hedonic quality adjustment methods in the construction of the CPI. Hedonic methods already are used in producing the CPI housing, apparel, television, and computer indexes. We are working to develop hedonic models for a range of additional products. Those selected for model development work in 1999 include telephones, VCRs, DVD players, camcorders, refrigerators, microwave ovens, washers, dryers and audio products (Fixler, Fortuna, Greenlees, and Lane, 1999). In each case, it remains to be seen whether an acceptable hedonic model can be developed, but that is our goal.
Thirdly, in the medical care arena, working collaboratively with National Bureau of Economic Research (NBER) researchers, we are exploring the use of third-party databases to identify shifts in hospital treatment patterns that may affect what we should be pricing (Bureau of Labor Statistics, 1999). Under standard Producer Price Index (PPI) procedures, a hospital procedure selected for pricing would continue to be priced until the next time the survey sample was redrawn. In this case, however, we are seeking a way to identify new treatments that are now competing with older treatments and, where such new treatments can be identified, to begin substituting toward them in our pricing sample to reflect their current period usage. An important part of this work will be deciding whether and how to compare the prices of the old and the new treatments in constructing our measure of hospital services price change. I might add that, more generally, the BLS has been a significant financial supporter of NBER research on measurement of medical care prices.
We also are seeking to support research on price measurement in other challenging sectors. For example, we plan to commission outside research on telecommunications services that will address such questions as the appropriate output concept for this industry, the best treatment of frequently changing service plans, and indicators that might be useful for tracking changes in the quality of the service offered. To take another example, we hope to commission outside research on college education prices. That research would explore strategies for capturing transaction prices (tuition less the value of the financial aid package) rather than list prices (tuition), as well as possible approaches to adjusting for changes in the quality of the educational product.
A final important area of BLS research activity relates to the use of scanner data (Bradley, Cook, Leaver and Moulton, 1997). We currently are engaged in a test effort to produce real-time CPIs for certain products in certain geographic areas (specifically, in the first instance, breakfast cereals in the New York metropolitan area) using scanner data as an alternative to data collected by our field economists. So far, that test seems to be going well. Ultimately, of course, scanner data should be helpful for dealing with shifts in product mix and the emergence of new products.
The above list outlines work that is relevant to addressing at least a piece of the challenge I outlined at the beginning of my comments. I've omitted a substantial amount of work we're doing or supporting on less glamorous things where, it turns out, there also is plenty to be done. That includes work on things such as small sample bias in index number construction and seasonal adjustment in cases where quantities sold are highly seasonal even though prices may not be. (See, for example, McClelland and Reinsdorf, 1997, and Diewert and Feenstra, 1999.) We're also engaged in research in a number of related areas, including the development of inter-area cost of living indexes, democratic (person-weighted) as compared to plutocratic (dollar-weighted) indexes, and the construction of separate indexes for population subgroups. (See, for example, Kokoski, Moulton and Zieschang, 1996; Kokoski, 1997; Amble and Stewart, 1994; and Garner, Johnson and Kokoski, 1996.)
Although there is a substantial amount of research work underway at the BLS on a fairly wide range of issues, our in-house resources are limited, as are the resources of our colleagues at BEA and the Census Bureau. With so much to be done, there is certainly ample scope for useful academic research on fundamental price and quantity measurement problems. Indeed, we at the statistical agencies are counting on your help. From my perspective, there are two broad areas in which academic research can be particularly helpful. First, given the importance of case-by-case analysis, there is great need for careful studies of individual industries and products. Second, because the programmatic uses of BLS data push us to focus on improvements to our current data collection methods, we devote relatively little effort to historical research. Our energy goes principally into the development and implementation of new, and hopefully improved, methods, rather than towards asking how our view of the past might have been affected had those new methods been adopted earlier. This obviously is something the BEA must deal with, but their research resources are even more limited than those of the BLS. Academic research oriented towards producing better historical price and output series could fill an important void.
Amble, N. and K. Stewart. 1994. Experimental Price Index for Elderly Consumers. Monthly Labor Review 117 (May): 11-16.
Bradley, R., B. Cook, S.G. Leaver, and B.R. Moulton. 1997. An Overview of Research on Potential Uses of Scanner Data in the U.S. CPI. Paper presented at the Third Meeting of the International Working Group on Price Indexes, Washington, D.C., April 16-18.
Bureau of Labor Statistics. 1999. Medical Care Quality Adjustment Initiative in the Producer Price Index. Unpublished document. Washington, D.C.
Diewert, W. E., and R.C. Feenstra. 1999. International Trade Price Indexes and Seasonal Commodities. Report to the Bureau of Labor Statistics. Washington, D.C.
Fixler, D., C. Fortuna, J. Greenlees and W. Lane. 1999. The Use of Hedonic Regressions to Handle Quality Change: The Experience in the U.S. CPI. Paper presented at the Fifth Meeting of the International Working Group on Price Indexes, Reykjavik, Iceland, August 25-27.
Garner, T.I., D.S. Johnson, and M.F. Kokoski. 1996. An Experimental Consumer Price Index for the Poor. Monthly Labor Review 119 (September): 32-42.
Griliches, Z. 1992. Introduction. In Output Measurement in the Service Sector, ed. Z. Griliches, 1-22. Chicago: University of Chicago Press.
Kokoski, M.F. 1997. An Alternative Aggregation for Price Indices: Democratic versus Plutocratic Indices. Working paper. Bureau of Labor Statistics: Washington, D.C.
Kokoski, M.F., B.R. Moulton, and K. Zieschang. 1996. Interarea Price Comparisons for Heterogeneous Goods and Several Levels of Commodity Aggregation. Bureau of Labor Statistics Working Paper No. 291. Washington, D.C.
McClelland, R. and M. Reinsdorf. 1997. A Note on Estimating Small Sample Bias in Two Price Index Formulas. Working paper. Bureau of Labor Statistics: Washington, D.C.
Office of Management and Budget. 1999. Economic Classification Policy Committee; Initiative to Create a Product Classification System, Phase 1: Exploratory Effort to Classify Service Products. Federal Register, April 16, 1999, 18984-89.