NBER Panel on Research on Price Index Measurement:Agendas for the Next Twenty Years

REMARKS BY ZVI GRILICHES
Harvard University and the National Bureau of Economic Research

NBER Panel on Research on Price Index Measurement: Agendas for the Next Twenty Years

July 21, 1999

Let me begin by stating that I believe it is time to forget about old battles, and instead ask, where do we go from here? The difficult tasks facing us we must do immediately, although the impossible ones may take some time.

I think it is important we congratulate the Bureau of Labor Statistics for undertaking, disseminating and publishing a great deal of very important price index research over the last twenty years. Both through the NBER's Program on Technological Progress and Productivity Measurement, and through the joint NBER - Conference on Research in Income and Wealth conferences and workshops, there have been extensive interactions among the government statistical community that creates and publishes price statistics, and users of these statistics in academia and in the public and private sectors. We all benefit from these interactions, and it is important they continue and grow.

In general, I agree with most of the remarks by Charles Schultze. In particular, I think it is important to take into account the fact that different price indexes can and will be used for different purposes. While construction and publication of different indexes may be feasible for the BLS today, publication of multiple indexes can create political difficulties. For example, if the BLS published a large number of price indexes (more than it does now), which, if any, would be suitable for cost-of-living adjustments for Social Security payments? In my Congressional testimony at the time of the Boskin Commission report, I spoke about problems with using the current CPI as a measure for indexing Social Security payments, and suggested that median nominal income per capita or per household may be a more appropriate index. This is an important issue for research, one that needs more attention in the near future.

There are two big remaining questions on which research should focus. The first is the difficult one, and concerns boundaries. What is the commodity and consumer space over which prices are measured? Among other issues, here I think it is important to recognize the link between BLS price measures and the GDP accounts at the BEA, and that there should be consistency between national accounts and price measures. Two examples come to mind. The first involves treatment of auto emission controls. If the costs of mandated auto emission controls are taken into account when the BLS creates price indexes for, say, automobiles, then the quantity benefits of these controls should show up somewhere in the national GDP accounts. The second example of consistency issues involves the medical area. Here the additional issue is that the BLS prices only the out of pocket payments by consumers in its medical CPI, plus making a general health insurance adjustment based on insurers' retained earnings. However, there is no price index that measures the "purchases" and purchase prices paid by third party payers, such as private insurance companies and governments. In the GDP accounts, there is an attempt to measure the real output of the medical sector, but the price indexes used to deflate expenditures are not right. There's a big need to focus on conceptual and practical consistency issues between BLS price measurement and the BEA's GDP measurement.

The impossible topic is quality change. Medical price measurement is again an example of the very difficult issues we face. Research needs to focus on the implications for price measurement of heterogeneity in access to medical care (now even available on the internet); how the absence of a free market affects the valuation, diffusion and use of medical technology; how to incorporate the value of life into price index calculations evaluating the benefits of new products and procedures that decrease mortality; and how the presence of various comorbidities affects both marginal benefit and marginal cost calculations. There is considerable room for reasonable discussion and debate concerning choice of boundaries in assessing quality change, and I think there is good reason to encourage construction and publication of various experimental price indexes, by BLS and others. As an example outside the medical care area, you might ask yourself if GDP should be unaffected whether Apollo landed and safely returned to earth, or not?

Over the longer term, I would like to see price index researchers begin to incorporate more of the research from the non-economics community. One example is the outcomes and cost-effectiveness research from medical and health policy researchers. Another involves environmental data. Data sets need to be expanded to encompass both economic and non-economic measurement.

I would also like to see the launching of a research institute devoted to researching price measurement topics, jointly funded by the BEA and the BLS, and with a strong institutional support not only from them, but also from the NSF.