Essays on research and development, innovation and productivity
Open Access
- Author:
- VUONG, VAN ANH
- Graduate Program:
- Economics
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- September 14, 2011
- Committee Members:
- Mark John Roberts, Dissertation Advisor/Co-Advisor
Mark John Roberts, Committee Chair/Co-Chair
James R. Tybout, Committee Member
Coenraad Arnout P Pinkse, Committee Member
Spiro E Stefanou, Committee Member - Keywords:
- innovation
productivity
r&d - Abstract:
- This dissertation investigates firms' incentives to engage in research and development (R&D). In particular, I investigate empirically the link between R&D, innovation and productivity. This allows me to evaluate costs and benefits of R&D from a firm's perspective. Firms invest in R&D in order to raise the probability of developing new innovations. These innovations can be new processes that improve product quality or reduce the production costs for the firm. Alternatively, innovations can encompass the introduction of new products or an extension of the range of final goods that the firm supplies. Firm innovations generate improvements in a firm's performance, specifically they increases firm's productivity and profits. There is significant heterogeneity in R&D spending across firms and industries. Positive R&D expenditures are not necessarily observed for every firm at any point in time. So, the question is: Why do some firms invest in R&D and others do not? Also, what are the factors that affect firms' R&D investment decision? As with any other kind of investment the rate of return of R&D investment and the cost of R&D investment determine the firms' incentives to engage in R&D. In this thesis I develop an empirical model to estimate these returns and costs in order to understand the incentives firms face when they make their R&D choices. This in turn helps us understand the heterogeneity in firms' R&D efforts and productivity performances. Most studies on the effects of R&D assume a direct link between R&D and productivity, for example Lichtenberg and Siegel (1991), Hall and Mairesse (1995), and Doraszelski and Jaumandreu (2011). These studies treat the process between R&D and productivity as a black box, since the outcome of the R&D process, the innovation it produces, is not directly observable. This thesis extends these studies by modeling the link between R&D and productivity explicitly. In particular, I let R&D affect the realization of product and process innovation. Once realized, these two types of innovation are then treated as determinants of productivity. By breaking the direct link between R&D and productivity into two links, i.e. the link between R&D and innovation and the link between innovation and productivity, I am able to account for the different types of uncertainties associated with each link separately. The first link from R&D to innovation largely captures the uncertainty regarding whether R&D investment actually leads to an innovation. This uncertainty is sometimes referred to as R&D risk. The second link from innovation to productivity captures a very different type of uncertainty. Product innovations are typically associated with the risk that the market might reject these products. Process innovations on the other hand, are typically associated with the risk that the higher efficiency might not lead to lower costs or difficulties in their implementation. In contrast to previous models which assume a one-step direct link between R&D and productivity, my model distinguishes both links. Knowing more about the uncertainty inherent in each of the linkages will allow for a better understanding of the determinants of firms' R&D decisions. This is important when evaluating public policies, such as subsidies to R&D which can be undertaken to promote productivity growth. My model is based on the structure of Aw, Roberts, and Xu (2011) (ARX). They endogenize firm productivity by allowing the firm's investments in R&D to shift the future path of productivity. In contrast to ARX, I model the link between R&D and productivity in more detail by assuming that R&D can lead to process and product innovation which in turn can lead to productivity gains. My model features a single agent which makes a dynamic choice of whether or not to invest in R&D: The current R&D choice leads to a change in future productivity which in turn affects the future R&D choice. In that sense productivity is endogenous and the R&D choice is dynamic in my model. The second distinctive feature of this thesis is that I exploit a unique new data set,the Mannheim Innovation Panel (MIP), a rich data set that provides information on the innovation success of German firms. The MIP contributes to the Community Innovation Survey (CIS) which is available for many countries. Therefore, empirical studies using the CIS data for various European countries can be compared to studies using the German MIP data. The uniqueness of the data set lies, among other things, in the presence of variables identifying whether or not the firm introduces an innovation during a certain time period. Furthermore, it distinguishes between product and process innovation which makes it possible to separate the effects of different kinds of innovation on firms' productivity. The MIP has the additional advantage that it contains information on innovation expenditures. This means that in contrast to previous studies, I do not only account for R&D expenditures but also for costs pertaining to the link between innovation and productivity such as the marketing costs of new products. To my knowledge, this thesis presents the first model structurally formulating the links from R&D to innovation and from innovation to productivity that is applied to the CIS data. The key structural parameters estimated are those that describe the process of endogenous productivity evolution, including the effect of an innovation of the firms future productivity, and the costs of conducting R&D for both experienced firms and firms beginning their R&D investments. The empirical model includes an equation describing the firms dynamic demand for R&D investment and it allows me to measure both the benefits and costs of R&D. My empirical model produces the following results. First, product innovation as well as process innovation increase productivity. This result corresponds to economic intuition. Second, participation in R&D leads to a higher probability of a product or process innovation, implying that engaging in R&D leads to higher productivity. This result is in line with the empirical findings of the literature. Third, the firm's current R&D decision depends on productivity and on past R&D decisions. The idea here is that R&D investments and the productivity process are mutually dependent over time. Fourth, fixed costs of R&D are significantly smaller than sunk costs of R&D. This means that the firm R&D history is an important determinant of current R&D behavior: a firm that has chosen to invest in R&D previously has a lower cost of continuing than a firm that did not chose to invest in R&D previously. The latter firm has higher costs of entering the R&D process. The fifth finding concerns firm size: larger firms have higher R&D costs than smaller firms. This corresponds to the empirical pattern in my data that larger firms have a higher probability of investing in R&D than smaller firms. I use the empirical estimates to investigate two policy applications. The first policy simulation investigates the questions whether an R&D subsidy leads to an increase in productivity. This question is at the heart of many discussions regarding the costs and benefits of public subsidies. I find that a subsidy which leads directly to lower R&D costs leads to more innovation and higher productivity. The second policy experiment concerns the effect of competition on R&D investment. There are two schools of thought: The first states that only monopolies have an incentive to innovate in order to deter potential entrants whereas the second school of thought states that competition fosters R&D because market participants want "to escape competition." My results shows that when firms face less elastic demand for their output, as would occur when there are fewer substitute products available, the gain in average productivity and the demand for R&D both decline. This thesis contains the following chapters. Chapter 1 presents an overview of the literature on R&D, innovation and productivity. This chapter contains also a discussion of the results in the empirical literature that uses the CIS data. Chapter 2 contains the model while chapter 3 provides a detailed description of the data. Chapter 4 presents the estimation strategy. Chapter 5 discusses the empirical results and chapter 6 contains the results of the policy experiments. Chapter 7 concludes the thesis with a summary.