Computing Price-Cost Margins in a Durable Goods Environment

This paper develops and executes a new method for calculating price-cost margins in a durable goods environment. We study digital camcorders, a new good during the time period of the data set. We analyze how margins differ across products, firms and time. We are particularly interested in the extent to which falling marginal costs explain falling prices, as opposed to alternative explanations such as intertemporal price discrimination or increased competition. Although price-cost margins are an important element of many studies in industrial organization, margins are challenging to compute because it is unusual to observe an accurate measure of marginal cost. Rather, researchers typically estimate marginal cost in the context of a structural model of competition. A popular empirical method associated with Bresnahan (1981) and Berry, Levinsohn & Pakes (1995) (henceforth, BLP) proceeds by estimating demand, assuming an equilibrium concept to describe interactions between firms, and then calculating marginal revenue at each product. We can interpret the result as marginal cost, as optimizing behavior implies that marginal revenue equals marginal cost. However, taking this approach in a dynamic environment brings up substantial new challenges. In particular, the relevant marginal revenue to consider is a dynamic one, the change in the present discounted value of the infinite stream of future revenue. For a durable good such as digital camcorders, a price change today affects future revenue in two ways. First, the price change affects consumer purchases today and thus affects consumer holdings