Is Capital Reallocation Really Procyclical?

Abstract

Aggregate reallocation is procyclical. This empirical observation is puzzling given the documented fact that the benefits to reallocation are countercyclical. I show that this procyclicality is entirely driven by reallocation of bundled capital, which is highly correlated with market valuation and bears no consistent relation to measures of productivity dispersion. Reallocation of unbundled capital, on the contrary, is countercyclical and highly correlated with dispersion in productivity growth, both within industry and across industries. To rationalize these facts, I propose a heterogeneous agent model of investment featuring two distinct used-capital markets and a sentiment component. In equilibrium, unbundled capital is reallocated for productivity gains only, whereas bundled capital is also reallocated for real, or perceived synergies in the equity market. While equity overvaluation negatively affect total factor productivity (TFP) by encouraging excessive trading of capital, such an adverse impact is largely offset by eased frictions to reallocation in the unbundled capital market.