Working Paper: CEPR ID: DP5126
Authors: Andrew B. Bernard; Stephen Redding; Peter K. Schott
Abstract: Firms? decisions about which goods to produce are often made at a more disaggregate level than the data observed by empirical researchers. When products differ according to production technique or the way in which they enter demand, this data aggregation problem introduces a bias into standard measures of firm productivity. We develop a theoretical model of heterogeneous firms endogenously self-selecting into heterogeneous products. We characterize the bias introduced by unobserved variation in product mix across firms, and the implications of this bias for identifying firm and industry responses to exogenous policy shocks such as deregulation. More generally, we demonstrate that product switching gives rise to a richer set of industry-level dynamics than models where firm product mix remains fixed.
Keywords: deregulation; industry evolution; product choice; productivity
JEL Codes: D21; L11; L60
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Aggregation of product data (D20) | Biases in standard measures of firm productivity (D20) |
| Firms switching products (L19) | Distortion in measured productivity (D24) |
| Switching products (L15) | Capturing true productivity differences and variations in production technology (O49) |
| Observable firm-specific variations controlled for (L20) | Necessity of tracking product choices over time (D12) |
| Changes in weights of products in consumer utility (D11) | Fluctuations in measured industry productivity (D20) |