Working Paper: NBER ID: w5656
Authors: Timothy Dunne; John Haltiwanger; Kenneth R. Troske
Abstract: In this paper, we exploit plant-level data for U.S. manufacturing for the 1970s and 1980s to explore the connections between changes in technology and the structure of employment and wages. We focus on the nonproduction labor share (measured alternatively by employment and wages) as the variable of interest. Our main findings are summarized as follows: (i) aggregate changes in the nonproduction of labor share at annual and longer frequencies are dominated by within plant changes; (ii) the distribution of annual within plant changes exhibits a spike at zero, tremendous heterogeneity and fat left and right tails; (iii) within plant secular changes are concentrated in recessions; and (iv) while observable indicators of changes in technology account for a significant fraction of the secular increase in the average nonproduction labor share, unobservable factors account for most of the secular increase, most of the cyclical variation and most of the cross sectional heterogeneity.
Keywords: technology; employment; wages; nonproduction labor share
JEL Codes: J31; J24
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Technology adoption (O33) | Nonproduction labor share (J39) |
| Within-plant changes (Q16) | Nonproduction labor share (J39) |
| Economic downturns (E32) | Technology adoption (O33) |
| Observable indicators of technology (O30) | Nonproduction labor share (J39) |
| Unobservable factors (D89) | Variation in nonproduction labor share (J49) |