The slow rate of growth in U. S. labor productivity in the last 25 years, in contrast to much faster growth in the previous half century, constitutes the much discussed `productivity slowdown.` Thus far the slowdown has eluded explanation by any theory based on a single cause, such as the behavior of energy prices, capital investment, infrastructure investment, government regulation, and educational test scores. More recently attention has focussed on the possibility that mis-measurement of price changes may partly or largely explain the slowdown. It has been conjectured that output growth is now more difficult to measure than previously, partly because computers are producing unmeasured intangible benefits, and as a result that inflation has been overstated and productivity growth understated more in recent decades than prior to the slowdown. The novelty in this project is to turn the presumed direction of bias upside down by exploring the possibility that inflation may have been understated (or overstated by substantially less) in earlier decades prior to the slowdown. This possibility, which has not heretofore been systematically investigated, implies that productivity growth may have been overstated in the golden era (spanning the period between 1915 and 1972) when measured productivity growth was rapid. If this hypothesis is supported by this research, it would explain the slowdown at least in part by showing that productivity growth in recent decades does not compare so badly with that in the golden era as has previously been thought. A set of research criteria developed for this project leads to the choice of shelter services (both rent and homeownership) as the central focus of the research, the aim of which is to develop new indexes of price and quality change for shelter over the full period from World War I to date. The project uncovers several sources of data suggesting that the official CPI for residential rent, which is the basis for the deflation of home ownership in the national income accounts, may substantially understate inflation in quality adjusted rents, particularly between 1930 and 1970. A useful byproduct of the research on the quality of residential shelter will be to provide new information to reassess the puzzle of negative productivity growth in the construction industry over the past three decades. The study of shelter prices will use several techniques to compile data on housing quality and prices. Several little-known studies covering the 1930 1970 period can be extended forward and backwards, using data on quality characteristics of rental apartments and houses from industry sources and consumer expenditure surveys for the earlier years, and existing government data from the Census of Housing and Current Housing Reports for the more recent period. Data will be analyzed using the analysis of variance or the hedonic regression approach, combining new estimates with coefficients from previous studies. The research also provides evidence that the CPI may understate the historic inflation rate for apparel, another important component of consumer expenditure. Comparisons of average prices paid with matched model indexes from the same data source suggest the possibility of a much more rapid rate of apparel inflation than in the matched model indexes or in the CPI. If this possibility is validated by a hedonic regression study of apparel prices over long historical intervals, it would augment the research on shelter prices in altering the standard historical data on U. S. output and productivity growth, potentially reducing the growth of productivity between World War I and 1970 and thus explaining part of the post 1970 productivity growth slowdown. The apparel research may also imply that previous matched model indexes for other products, as well as the many CPI components based on linking across model changes, may need to be reexamined.
