Working Paper: CEPR ID: DP18683
Authors: Peter Egger; Corinne Jones
Abstract: The attention on multi-product firms to date has mainly focused on firm scope, rather than what firms produce. We study what firms produce by developing a heterogeneous firm model accounting for correlations in production efficiencies across products and correlations in consumers' tastes across products and destinations. Our estimation strategy centers on a novel extension of generalized score matching. We fit the model to Chinese exports data, focusing on firms exporting products with ISIC codes C19 (leather, leather products, and footwear) and C18 (wearing apparel, dressing and dyeing of fur). We document clustered co-production structures consistent with correlated efficiencies across products.
Keywords: multiproduct exporters; coexportation; correlated technology shocks
JEL Codes: F12; L25; L11; C34; C38; F14
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| joint production clusters among products with shared production technologies (L23) | coexportation frequency (F10) |
| correlated efficiencies across products (C10) | coexportation frequency (F10) |
| efficiency shocks of firms for products in different clusters (F12) | production efficiencies are not uniform across different product types (D24) |
| correlated technology shocks and appeal shocks (E32) | likelihood of joint production (D20) |