Working Paper: CEPR ID: DP16022
Authors: Joseba Martinez
Abstract: This paper develops a model of automation as an embodied technology. The gradual discovery and obsolescence of technologies gives rise to a distribution of capital with varying degrees of automation. I derive conditions under which, aggregating over heterogeneous production units, output can be represented as a CES production function, the parameters of which are determined endogenously by the distribution of technology. Through the lens of the canonical model, I show how the distribution of automation technology determines its aggregate effects; in the long run, only the distribution of technology matters. The transition dynamics of the economy in response to an increase in frontier automation technology are consistent with notable micro and macro US stylized facts of recent decades: at the firm level, a fall in the labor share driven by reallocation towards low labor share establishments; at the macro level, slowing total factor productivity growth and a fall in the real interest rate.
Keywords: automation; factor shares; vintage capital; aggregation
JEL Codes: O4; E1; E22; E25
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| increase in the most automated capital (E22) | decrease in labor share (E25) |
| scrapping the least automated capital (P19) | increase in labor share (J39) |
| distribution of automation technology (L63) | aggregate effects on labor share (E25) |
| distribution of automation technology (L63) | aggregate effects on productivity (O49) |
| increase in frontier automation technology (O31) | transition dynamics in the economy (P23) |
| reallocation of economic activity towards low labor share firms (E25) | changes in labor share (E25) |
| automation (L23) | observed fall in the aggregate labor share (E25) |