Choosing the wrong theoretical model to describe an industry’s behavior may lead to biased estimates of the degree of market power. This paper presents a two-step, data-driven methodology to reduce the risk of mis-specification bias. In the first step, a sliced inverse regression identifies the significant factors that affect the industry’s pricing behavior. In the second step, a non-parametric regression of price on the SIR factors estimates the link functions. The output of the algorithm offers useful information to identify the model that best approximates the industry’s pricing behavior. In addition, the output of the algorithm is used to develop a post-estimation test for model specification.
Estimating Market Power with Weak A Priori Information: An Exploratory Approach to the Model-Specification Problem
RUSSO, Carlo
2012-01-01
Abstract
Choosing the wrong theoretical model to describe an industry’s behavior may lead to biased estimates of the degree of market power. This paper presents a two-step, data-driven methodology to reduce the risk of mis-specification bias. In the first step, a sliced inverse regression identifies the significant factors that affect the industry’s pricing behavior. In the second step, a non-parametric regression of price on the SIR factors estimates the link functions. The output of the algorithm offers useful information to identify the model that best approximates the industry’s pricing behavior. In addition, the output of the algorithm is used to develop a post-estimation test for model specification.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
