Learning from Failure Across Products

Working Paper: CEPR ID: DP13140

Authors: Johanna Glauber; Tobias Kretschmer

Abstract: Learning rates for the same or similar products differ significantly across firms. One reason for this heterogeneity may be that most firms are multiproduct firms and that they learn both within and across the products they produce and sell. Moreover, the organization of production and the technological design of (families of) products may affect the extent of learning across products. We study learning from failures within and across products in the US automotive industry, using safety recalls as a particularly costly form of product failure. We find that firms indeed learn from failure across products, but learning is faster if products use a common technological platform or are produced in a common plant. Severe product failures involving a supplier also lead to increased across-product learning. Our results shed light on the characteristics of firms’ learning from failure across products and extend the existing literature, which typically studies learning at the firm or the product level and thus misses out on an important channel for learning in manufacturing contexts.

Keywords: organizational learning; learning from failure; product recalls; automotive industry; multiproduct firms

JEL Codes: L23; L62; M11; L11


Causal Claims Network Graph

Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.


Causal Claims

CauseEffect
Cumulative number of past recalls for a specific car model (C29)Number of recalls for that same model in subsequent years (C20)
Cumulative recalls of other models (C39)Recalls of the focal model (E17)
Cumulative number of past recalls for a specific car model (C29)Learning from product recalls within products (D18)
Cumulative recalls of other models (C39)Learning from product recalls across products (D18)

Back to index