This paper provides new price indexes for medical care spending based on cost per case for individual diseases using data from nationally-representative surveys of medical expenditures in the civilian non-institutional population in 1980, 1987, 1997 and 2006. Several studies have provided indexes like these (called Medical Care Expenditure indexes, or MCEs) for different patient populations in recent years. Our contribution is to provide these price measures for a more comprehensive population of patients during the earlier periods (1980-87 and 1987-97).
Our indexes are similar in coverage to the official statistics in that they are constructed for a comprehensive list of conditions and patients. Comparing our MCEs to the official PPIs, we find that our indexes sometimes show slower price growth and sometimes faster price growth than the official statistics. For the earliest time periods, our disease-based price indexes grow 8.9% from 1980-87, very close to the 9.0% price increases currently in the national accounts; over the period 1987-97, our indexes grow 3.8%, substantially slower than the 6.0% growth rate in the national accounts. For the most-recent time period, we confirm existing findings that disease-based price indexes show faster growth than the official statistics in that period (4.9% vs. 2.7%). The fact that disease-based price indexes can show slower or faster growth than the official statistics is not well appreciated. For all price indexes, the prior is that the official statistics have an upward bias, in part because of the well-known substitution bias problem with the Laspeyres-type formula used in official statistics and also because of problems with adequately accounting for quality change. For medical care indexes specifically, early case studies empirically quantified an upward bias for several important conditions, owing to both lack of quality adjustment in the official statistics (Cutler heart attacks) and shifts in the utilization of treatments across industries (Berndt’s depression study; Cutler; cataract). However, recent work for medical care price indexes has demonstrated that there are other sources of bias that could work in the other direction and potentially offset these upward biases. In this paper, we (i) review the potential sources of bias in the official statistics and their relevance for medical care price indexes, (ii) summarize DLS’s argument, (iii) talk about insurance as a potential wedge in the two price measures, and (iv) discuss historical trends in the medical care sector over the 1980-2007 period to assess whether our findings square with developments in this sector over that time period.
October 04, 2014