RESEARCH IN PROGRESS

“Competition, Product and Process Innovation: an empirical analysis”, December 2010

Understanding how changes in competition affect innovation and productivity is an old and important economic question. The main contribution of this paper is to separate the effects of competition on the two types of innovation: product and process. To achieve this I write down a model of firm behaviour and introduce explicit assumptions about unobserved heterogeneity that allow me to identify the causal effect. I use a sample of Spanish firms to estimate the model. The sample has particularly useful features to answer the question at hand - detailed information about the types of innovation and self reported measures of market structure that have the advantage of being close to the actual competition firms respond.
Overall, I find that competition, measured by the number of competitors or market shares, has negative effects on product innovation and no effects on process innovation. The results are in line with the predictions. By shifting demand, competition directly changes the optimality condition for product but not for process innovation. Thus, competition has no direct effects on process innovations or, as a consequence, productivity. Changes in competition operate indirectly on process innovation, namely through firm size.

“Sunk Costs of R&D, Trade and Productivity: The moulds industry case”, September 2010

Evidence suggests that trade improves industry productivity via (i) selection of the best firms and (ii) within firm productivity growth. While the selection effect has been well explained by the literature, the productivity growth is not accounted for in standard models. Trade liberalization creates economies of scale in the R&D process that can explain the observed growth. I estimate a model of industry dynamics with endogenous productivity, capital accumulation and aggregate uncertainty. I use data from the Portuguese moulds industry which experienced an exogenous trade shock in 1993 (establishment of the Common Market) and a significant increase in foreign demand afterwards. Firms exploited the increase in demand from other European countries to increase R&D spending. The final results confirm the trade-induced innovation effect.

“Solving Dynamic Games by Discretizing the State Distribution”, April 2009

After more than a decade of advances in theoretical models for industry dynamics, there has been a growth in applications using recently developed estimation methods for dynamic games. However, estimation methods often require equilibrium calculation either in the estimation routine and/or for the construction of counterfactuals, and it is well known that, due to the 'curse of dimensionality ', equilibrium calculations can be computationally demanding.

In this paper a method to circumvent the 'curse of dimensionality' is proposed. Using the result that under symmetry and anonymity the industry distribution fully characterizes the industry state, we can use quantiles to discretize and approximate this distribution. As the number of quantiles increases, the approximation becomes exact.

This resembles the widely used discretization method in solving dynamic programming problems with continuous state variables. In the dynamic game case, the continuous state variable is a distribution, defined over the unit interval. The main advantage of this approach is that the dimension of the state space becomes unrelated with the number of players. Using simulations it is illustrated that the methodology sucessfully reduces the cardinality of the problem while providing a good approximation.

“Identifying financial constraints in a dynamic structural model of R&D and investment: the US Iron and Steel industry” (joint with John Van Reenen, in progress)

Identifying financial constraints is difficult in the presence of unobserved future expectations of demand. In this paper we develop a model that allows for financial constraints in a framework where firms make capital investment and R&D decisions. By specifying a dynamic structure and solving through numerical simulation we model adjustment costs, R&D decisions and financial constraints simultaneously. Applying this to 35 years of firm-level panel data for the US Iron and Steel industry we provide evidence that costs of external finance are substantial, consistent with asymmetric information, even in a developed financial market. The average sunk cost of R&D is on the order of $194m - consistent with industry estimates of the typical costs of building an R&D lab.

“Production Functions with differentiated products and competition”

In this paper I address two common problems in the production function estimation literature. The first problem is input endogeneity and the second is the use of deflated sales as a proxy for output when there is imperfect competition. Using a demand system and allowing input demand to depend on the individual state variables as well as on the industry equilibrium I explain how to jointly recover the production function parameters and demand elasticity.