An important challenge in economics is to explain what causes firms to exit. The standard answer is low profitability. This answer leaves, however, open a number of questions as profitability partly depends on factors that are difficult to measure, for example, demand conditions, efficiency and market structure. In addition, due to market imperfections and regulations, other variables may also play a role in explaining exits.
In the empirical literature, several variables have been used in order to explain exits. These include plant size, see Mata, Portugal and Guimarães (1995); capital intensity, see Boeri and Bellmann (1995); financial leverage, see Dimitras, Zanakis and Zopounidis (1996); productivity, see Olley and Pakes (1996); capital vintage, see Salvanes and Tveterås (2004), and demand shocks, see Foster, Haltiwanger and Syverson (2005). The purpose of the present paper is to identify, through a structural microeconometric model with heterogeneous firms, imperfect competition and adjustment costs, some key determinants of exit behavior.
While traditional exit rules were myopic, such as Marshall (1966), modern theories assume that the exit decision is based on both present and expected profits. In its most refined version, the exit rule is derived from stochastic dynamic programming (SDP): The firm will stay operative as long as the expected present value of continuing production exceeds the value of closing down; see e.g. Hopenhayn (1992). Although SDP provides an attractive general framework for interpreting and analyzing intertemporal decision problems, huge computational requirements put sever limits on the number of state and decision variables that can be included in the analysis. Because of the obstacles involved in applying SDP, the literature offers alternative approaches. It is an empirical question, however, whether the SDP exit rule applied on a model of strictly simplifying assumptions provides a better explanation of observed exits than exit rules based on alternative approaches.
For optimal stopping problems, a frequently used approach is that of Stock and Wise (1990), who analyze the retirement decisions of older employees. In our setting, the Stock and Wise rule says that a firm finds (at t) the exit time that maximizes expected profits, given the current information set of the firm. If the “optimal” time to exit is now, the firm exits. Alternatively, if the “optimal” time to exit is not now, the firm continues production and in the next period (t + 1) reexamines whether it should exit immediately or continue production, and so on. In the present paper we use the Stock and Wise approach because it enables us to specify a much richer model of firm behavior than if we had relied on SDP. Furthermore, it incorporates the important forward-looking aspect of the exit decision problem.
In our model, each firm produces a version of a differentiated good. The demand function depends on the prices of the firm’s product and those of the competitors. Firms set prices simultaneously in a non-cooperative game (Bertrand competition). Production requires input of labor, materials and capital. The latter is a quasi-fixed factor. The production function incorporates both Hicks and non-neutral technological progress.
