The goal of this research project is to estimate the long-run expected returns to R&D and human capital investment by German firms using data from the Mannheim Innovation panel (MIP). The research will build upon recent papers by Aw, Roberts, and Xu (American Economic Review, 2011) and Peters, Roberts, Vuong and Fryges (working paper, 2013) that estimate dynamic discrete choice models of R&D investment. The latter paper develops a dynamic structural model in which a firm’s decision to undertake R&D impacts the probability of both product and process innovations and the realization of an innovation affecting future productivity and profits. Both papers estimate the dynamic decision rule for the firm’s choice of R&D which allows them to quantify the impact of R&D choice on the value of the firm. The model allows for persistence in firm-level productivity and recognizes that the returns to R&D will vary across firms with differences in their productivity, capital stock and R&D history. The model provides a structural counterpart to the Crepon, Duguet, and Mairesse (1998) framework that has been widely used to analyse data on R&D, innovation, and productivity derived from the Community Innovation Surveys. The advantage of the structural framework is that it provides a representation of the firm’s investment decisions that is consistent with the maximization of the long-run expected net payoff from the investment. Furthermore, the empirical model provides estimates of this long-run payoff and thus can be used to conduct counterfactual policy analysis.This research project will build on the papers mentioned above and make novel contributions in several dimensions:

1. R&D choice will be modelled as a continuous, rather than discrete, choice variable, which will allow us to measure the long-run payoff to variations in the firm’s R&D expenditure.
2. Firm’s investment in worker training will be incorporated as it represents another long-term investment by the firm that can impact its future productivity and profits, and we will estimate the long-run expected benefit to the firm of investing in worker training.
3. The model will be applied to the service industries. The service sectors are increasingly important in the German knowledge-based economy but their productivity growth and the role of R&D and human capital investments are less well studied in these industries. In the service industries, expenditure on worker training is generally larger than the firm’s expenditure on R&D suggesting that efforts to invest in human capital are an important source of firm productivity growth and performance.
4. Using the estimates from the structural dynamic model, we can conduct counterfactual policy analysis. For example, we can simulate how changes in R&D or worker training costs, such as would result from a subsidy, will affect the firms’ R&D and training decisions as well as their impact on productivity and firm value.