Are Technology Improvements Contractionary?

This 2006 American Economic Review paper (circulated as NBER Working Paper 10592, June 2004) asks a question its own abstract answers in one word – “Yes” – are technology improvements contractionary? The authors construct a “purified” measure of aggregate technology change that corrects the standard Solow residual for varying utilization of capital and labor, non-constant returns to scale, imperfect competition, and aggregation effects, building on the production-function methodology of Basu and Fernald (1997) and Basu and Kimball (1997) – an identification strategy that relies on entirely different assumptions from, and is explicitly offered as a complement to, Galí’s (1999) long-run-restriction structural VAR approach to the same question. Using this measure on postwar U.S. data, the paper’s central finding is that when technology improves, input use (hours, utilization, and especially non-residential investment) falls sharply on impact while output barely moves, and only with a lag of one to several years do inputs, investment, and output recover and then rise strongly – a short-run contraction followed by a long-run expansion. The authors show this pattern is “clearly inconsistent with standard parameterizations of real-business-cycle models,” which predict inputs and output rise immediately after a technology improvement, but is qualitatively consistent with sticky-price models, in which a positive technology shock that is not immediately offset by expansionary monetary policy reduces firms’ demand for labor and capital in the short run even as it raises full-employment output. The paper argues this amounts to a “robust stylized fact” because four methodologically distinct approaches – their own production-function measure, Galí’s SVAR, estimated structural DSGE models, and Shea’s (1998) patent/R&D-based technology proxy – all independently reach the same short-run-contraction conclusion.

Returns to Scale in U.S. Production: Estimates and Implications

This 1997 Journal of Political Economy paper (circulated as Federal Reserve International Finance Discussion Paper 546, March 1995) asks how large deviations from constant returns to scale and perfect competition really are in the U.S. economy – a question central to models in which increasing returns and markups amplify business-cycle shocks or generate multiple equilibria. Using Dale Jorgenson’s 34-industry dataset of the U.S. private business economy for 1959-89, the authors estimate that “a typical (roughly) two-digit industry in the United States appears to have constant or slightly decreasing returns to scale,” which at face value implies small markups – but this finding comes bundled with three puzzles: returns-to-scale estimates rise sharply when industries are aggregated into broader sectors; several individual industries show statistically significant decreasing returns, which (absent large observed profits) implies the economically nonsensical conclusion that firms price below marginal cost; and estimates built from value-added data diverge substantially from estimates built from gross-output data. The paper’s central theoretical contribution is to show that all three puzzles can be artifacts of aggregating over heterogeneous, imperfectly-competitive producing units: because firms with different returns to scale have different market power and hence different marginal products for the same input, reallocating inputs toward high-return-to-scale, procyclically-growing firms during a boom raises measured aggregate productivity and apparent aggregate returns to scale even when every individual firm’s own technology is unchanged. After deriving and subtracting an estimated “reallocation” correction term from the aggregate data, the authors find the private economy as a whole is consistent with essentially constant returns to scale (around 1.01-1.03 across gross-output and value-added measures), with durable-goods manufacturing still showing statistically significant mild increasing returns and non-durables mild (statistically insignificant) decreasing returns. A simple stylized two-firm model in the paper’s final section shows this reallocation is not merely an econometric nuisance but can itself be a real macroeconomic mechanism – though whether it is safe to ignore heterogeneity and calibrate a representative-firm model with aggregate data depends on the economy’s non-production institutions (e.g., how labor and capital markets allocate factors across heterogeneous firms), so the paper explicitly declines to offer a general answer to that calibration question.

Susanto Basu | Macro Paper Warehouse

Are Technology Improvements Contractionary?

Returns to Scale in U.S. Production: Estimates and Implications