N ONE of the procedures used in the measurement of investment has generated more controversy than the deflation of construction expenditures, for the measured rate of growth of real investment in structures in the United States varies over a wide range depending on which price index is chosen. Deflators proposed for United States construction range from the official Department of Commerce Cost (CCC) used in the United States National Accounts, which implies that real investment in nonresidential structures grew by 228 per cent from 1919 to 1966, to the Bureau of Public Roads (BPR) Composite Highway Index suggested in [7], which implies a much higher 19191966 increase of 47& per cent. Far from achieving a reconciliation between these two alternatives, conflicting conclusions were reached in the most recent detailed studies of the subject. Dacy [2] [3] has proposed a method which yields a 1947-1963 price increase much closer to that of the BPR highway series than that of the CCC index, while Kendrick, R. A. Gordon, and other have cited evidence supporting the much faster long-run upward trend of the CCC.1 A resolution of this conflict would allow us to evaluate and improve United States capital stock data, which have recently been criticized for ignoring an alleged bias in the construction price deflators.2 Can the apparent decline in the United States capital-output ratio between the 1920's and 1950's, for instance, be partly explained by defects in the official construction price indexes? In addition, improved information on construction prices would yield new output and productivity indexes suitable for testing the commonplace proposition that construction is a primitive industry in which efficiency improvements are almost as rare as in barbershops and the United States Post Office.3 This paper examines primary source data on the subject and proposes a new construction price index for the period 1919-1966. The new index is based partly on actual buyers' prices paid for several structural components and partly on an improved version of Dacy's indirect method using aggregative data for the entire contract construction industry. To anticipate the conclusion of the paper, the new index exhibits a long-run trend which is, appropriately enough, halfway between those of the CCC and BPR indexes and suggests that the official data overstate investment in structures in the mid-1920's by more than 35 per cent. The paper also suggests that productivity advances in construction have been surprisingly rapid in the postwar years, in contrast to an apparent stagnation of efficiency in the prewar period.

This paper summarizes the book and assesses the reviews contained in the four contributed papers. Gregory Clark provides convincing arguments that extend the book's forecast that future technological change will be slower. Nicholas Crafts shows declining mortality and shorter work hours greatly increase welfare-augmented TFP growth during 1929-50. Benjamin Friedman points out that optimism about future technological change casts doubt on future employment growth, while technological pessimism implies employment optimism. Daron Acemoglu and co-authors emphasize the institutional environment that influences the timing and magnitude of innovation.

Secular stagnation on the supply side takes the form of a slow 1.6 percent annual growth rate of US potential real GDP, roughly half the 3.1 percent annual growth rate of actual real GDP realized from 1972 to 2004. This slowdown stems from a sharp decline in the growth rate of aggregate hours of work and of output per hour. This paper attributes the productivity growth decline to diminishing returns in the digital revolution that had its peak effect business hardware, software, and best practices in the late 1990s but has resulted in little change in those methods over the past decade.

A long tradition in macroeconomics dating back to Arthur Okun (1965) and Walter Oi (1962) regards cyclical productivity fluctuations as an artifact, a residual generated from the incomplete and lagged response of employment and labor hours to demand-driven fluctuations in real output. In Okun’s version a one percent decline in output relative to trend is divided up into a reduction of 1 ⁄3 point in productivity and 2 ⁄3 point in aggregate hours. The latter is further subdivided into a reduction of 1 ⁄3 point in the employment rate, with the remaining adjustment taking the form of lower hours per employee and in the labor force participation rate (hereafter LFPR). Yet this tradition of regarding cyclical productivity fluctuations as a byproduct of demand-driven output cycles has been almost forgotten over the past three decades as a result of widespread adoption of the real business cycle (RBC) model in which productivity shocks are treated as exogenous, as unexplained, as unrelated to aggregate demand, and as the sole driver of business cycles. Even in the more enlightened modern macro work on Dynamic Stochastic General Equilibrium Models, aggregate demand and sticky prices have reappeared, but most recent papers still include an autonomous “technology shock” as one of several causes of short-term business cycle fluctuations. Revisiting and Rethinking the Business CyCle †

It is now 25 years since the growth rate of labor productivity and of multi-factor productivity (MFP) decelerated sharply both in the United States and in most other industrialized nations. This ‘‘productivity slowdown’’ has eluded many attempts to provide single-cause explanations. Slow productivity growth in the past 25 years echoes slow productivity growth in the late 19th century. Perhaps both were normal, and what needs to be explained is not the post-1972 slowdown, but rather the post1913 ‘‘speedup’’ that ushered in the glorious 60 years between World War I and the early 1970’s in which U.S. productivity growth was much faster than before or after. This paper makes a sharp distinction between MFP growth calculated from inputs that combine simple measures of labor hours and the capital stock and growth based on measures that adjust for the changing composition of labor and capital. The first step toward an understanding of long-term trends is to compare like with like, splicing MFP data based on unadjusted inputs prior to 1950 with post1950 data based also on unadjusted inputs, as contrasted to the composition-adjusted inputs that are now desirably incorporated into our official MFP measures. The MFP record prior to 1929 still rests largely on the monumental work of John Kendrick (1961) which, however, is based almost entirely on input quantities that lack any adjustment for changes in composition. Edward Denison ( 1962, 1985 ) and Zvi Griliches (1960) pioneered the development of composition adjustments for labor input. Dale Jorgenson and Zvi Griliches (1967) introduced a framework that treats the problem of composition adjustment in both labor and

Comparative economics is not the same as international economics. In fact, the word evokes the phrase systems, until recently a dull chapter dutifully included at the end of almost every Principles book. The collapse of communism has left a single system, the mixed economy of capitalism-cum-government in all of the infinite variety of possible combinations. Since the world has converged on this single system, one can snatch the adjective from its old mate systems and attach it to a whole new concept of how economics can and should be taught, that is, comparative economics.