اولویت های فردی، سازمان و رقابت در مدل مفاد تشویقی R & D

Understanding the organization of R&D activities requires the simultaneous consideration of scientific workers’ talent and tastes, companies’ organizational choices, and the characteristics of the relevant industry. We develop a model of the provision of incentives to corporate scientists, in an environment where (1) scientists engage in multiple activities when performing research; (2) knowledge is not perfectly appropriable; (3) scientists are responsive to both monetary and non-monetary incentives; and (4) firms compete on the product market. We show that both knowledge spillovers and market competition affect the incentives given to scientists, and these effects interact. First, high knowledge spillovers lead firms to soften incentives when product market competition is high, and to strengthen incentives when competition is low. Second, the relationship between the intensity of competition and the power of incentives is U-shaped, with the exact shape depending on the degree of knowledge spillovers. We also show that the performance-contingent pay for both applied and basic research increases with the non-pecuniary benefits that scientists obtain from research, while the fixed component decreases. We relate our findings to the existing empirical evidence, and also discuss their implications for management and public policy.

The management of scientific workers and the design of effective incentives for them are considered key determinants of competitive success,1 but present numerous challenges for companies. A major organizational decision concerns whether to provide high-powered incentives based on the scientists’ performance, or to soften incentives instead and let the researchers’ quest for reputation drive their effort. Another difficulty is how to measure performance in the first place, as research is a complex activity with no necessarily immediate returns (Holmstrom, 1989). A further set of issues regards how the characteristics of the markets where a company operates, and in particular the level of competition and knowledge appropriability, affect the type and strength of incentives. Understanding how companies motivate scientific workers is of importance also for policy makers. Key industrial policy questions include how to design competition laws and intellectual property regimes that elicit incentives to innovate for firms, and therefore affect the types of incentives companies offer to their researchers, while not curbing the dissemination of knowledge. In fact, these issues point to broader challenges for both scholars and practitioners. All major organizational problems require the considerations of multiple levels of analysis: individual characteristics such as talent and tastes (Sauermann et al., 2010 and Stern, 2004); organizational capabilities and structure, including the incentive system (Henderson and Cockburn, 1994, Holmstrom, 1989 and Holmstrom and Milgrom, 1994); and the characteristics of the relevant industry, in particular the competitive pressure (Porter, 1980, Raith, 2003, Schmidt, 1997 and Turner et al., 2010). Although the importance of all of these dimensions is often recognized, research that tries to integrate them in one framework is limited. In this paper, we develop a model to show that not only all of these dimensions affect the determination of incentives to company scientists, but that these different factors interact in interesting ways. The model is developed in Section 2, and includes four key aspects. First, scientists engage in multiple, different activities (Cockburn et al., 1999). Second, the outcome of research activities, knowledge, is only imperfectly appropriable (Arrow, 1962 and Spence, 1984). Third, scientists are responsive to the provision of monetary incentives, but they also care about non-material outcomes, such as their reputation among peers (Dasgupta and David, 1994 and Merton, 1973). Fourth, the provision of incentives to scientists, and to all workers in general, is likely to depend on the conditions that a firm faces in the product market, such as the intensity of competition (Raith, 2003 and Schmidt, 1997). In the model, two firms compete in an industry by offering differentiated products, and design incentives for their scientists (simplified to be a single agent per firm) to invest in cost-reducing research. Scientists engage in two types of efforts. The first kind of effort – which we call applied (or proprietary) research – does not provide non-pecuniary benefits to the scientists and does not generate knowledge spillovers to the rival firm; the second kind of effort – we call it basic (or open) research – provides non-pecuniary benefits to scientists but spills over to the rival firm. The firm's owners offer a wage contract to the scientists contingent on observable outcomes. The outcomes can include, for example, patents and scientific articles. In Section 3 we discuss the results of the model, characterizing the optimal incentive contract for the scientists. The first set of results highlight how the provision of incentives for basic and applied research depends not only on the intensity of competition and the degree of knowledge spillovers, but also crucially on the interaction between these two environmental conditions. High knowledge spillovers do not necessarily reduce the incentives to perform research: if competition is low, then firms provide high-powered incentives for both basic and applied research, since their dominant position in the product market reduces the negative effects of spillovers while allowing firms to enjoy each other's produced knowledge. With high competition, not only do we derive that incentives for basic research effort decrease as spillovers become more pervasive; we also show that it is optimal to mute incentives for applied research effort, even if it does not generate spillovers. In turn, the impact of product market competition on the strength and direction of R&D incentives depends on the degree of knowledge spillovers. If knowledge spillovers are low, firms provide the strongest incentives for basic and applied research both when they face very little competition (since cost reduction through R&D has a bigger absolute impact on profits), and when competition is very high (for competitive pressure makes any small cost reduction a proportionally large one, because profits are lower). Thus, the relationship between the intensity of competition and the power of incentives to scientists is U-shaped. In contrast, when there are high levels of spillovers, the strength of incentives is decreasing in the intensity of competition. A further implication of these findings is that incentives for basic and applied research are complementary only if either the level of product market competition or the degree of spillovers is low. The second set of results concern the impact of a scientist's non-monetary motivation to perform basic research, or taste for science, on her pay scheme. The response of scientists to steeper incentives is stronger when they also have high intrinsic motives to perform basic research. As a consequence, companies optimally provide stronger incentives to intrinsically motivated scientists, both for basic research and applied research, even if the latter does not generate non-monetary benefits to the scientists. We show, in contrast, that a trade-off can occur between the fixed component of pay and non-monetary rewards. An implication for empirical research is that studies of the determinants of incentives to scientists need to account for such environmental conditions as the degree of product market competition and of appropriability of knowledge, and need to analyze separately different components of wages, e.g. fixed and contingent pays, as they might respond differently to certain individual or environmental changes. We also interpret a number of existing empirical studies in light of our findings. The model in this paper is, to our knowledge, the first one to analyze the effects of product market characteristics on incentives for effort (in research activities) where effort is multi-dimensional and the agents have preferences or tastes for certain activities. The building blocks of the model here have been established by an extensive literature. There is, in particular, a vast literature investigating the relationship between competition and managerial efficiency (e.g., Raith, 2003 and Schmidt, 1997) and between competition and innovation (among the most recent contributions, see Sacco and Schmutzler, 2011, Schmutzler, 2010 and Vives, 2008). Baggs and de Bettignies (2007), moreover, link these two streams of literature by developing a model where they isolate the agency effect of competition from the direct pressure effect, which is present independent of agency costs. Some papers consider also the presence of knowledge spillovers in R&D investments (Qiu, 1997 and Spence, 1984).2 We contribute to these studies by showing that the impact of each of these two factors crucially depends on the other. Furthermore, a few papers have developed principal-agent models where agents are intrinsically motivated. Murdock (2002) considers a model where agents also have intrinsic motivations for the completion of projects, but the principal may prefer not to implement some of these projects if they have negative expected financial returns. Implicit contracts where the principal commits to implement the projects preferred by the agent may be socially superior and are more likely to be chosen by a principal when the agent's intrinsic motivation is higher (see also Manso, 2011).3 Murdock's model, therefore, studies the relationship between decision right allocation and the intrinsic motivation of agents, while our model analyzes the shape of the optimal incentive contracts as it responds to non-monetary motives of agents. Casadesus-Masanell (2004) and Ramalingam and Rauh (2010) also introduce intrinsic motivations in a principal-agent framework, in the form of adherence to social norms or altruism. These motivations are endogenously determined or institution-specific. In our setting, we consider scientists as responding to additional incentives or motivations as possibly defined by an “external” entity or organization, as represented by the scientific community. Besley and Ghatak (2005) develop a model of matching with intrinsically motivated agents and show that monetary and non-monetary incentives are substitutes. In their model, the reward of the principal is unaffected by the agent's intrinsic motivation, thus the principal exploits intrinsic motivation to save on the cost of high-powered incentives. In our model, the taste for science, through its impact on the desire to perform basic research, directly impacts the principal's payoff, thus leading to complementarity. Finally, Banal-Estañol and Macho-Stadler (2010) study the time allocation problem among different activities of a researcher who responds to both financial and scientific incentives, and show that higher financial incentives can lead a scientist to opt for riskier projects. Their study, however, abstracts from the analysis of the impact of knowledge appropriability and product market competition. Section 4 explores the managerial and policy insights from our results. In highlighting the interaction between the conditions in the product market and the ease of transmission of knowledge, our results inform R&D managers on the importance to look at their company's position in the product market and at the knowledge appropriability conditions for different types of activities when designing their internal incentive schemes. R&D managers, moreover, need to consider the different degrees of interest for monetary pay and for their reputations of their scientists. Scientists who are more eager to maintain their links to the scientific community even when employed by a firm, and are allowed to do so, are not necessarily “cheap.” Instead, these are the scientists who are given more powerful incentives for the performance of both basic and applied research. As for public policy, our analysis implies that IP protection rules should be determined in relation to the level of competition of each industry and that, in particular, antitrust legislation and IP protection are complementary instruments. For example, in industries where IP protection is very strong, competition on the product market should be particularly favored. Section 5, finally, offers concluding remarks.

The main insight from this first set of results is not only that such characteristics as product market competition and knowledge spillovers matter in the determination of incentives to scientists, but also that they interact. In addition to this theoretical insight, a major implication for empirical analysis is that the structure of the industry and the IP regimes need to be controlled for when assessing the determinants of scientists’ pays and incentive structures. Cockburn et al. (1999), for example, find that incentives for basic and applied research are complementary in the pharmaceutical industry: when firms commit to high-powered incentives to obtain recognition in the scientific community, they also offer higher-powered rewards for applied activities. The authors use data at the level of research programs; arguably, different research programs refer to different final product markets, thus our model suggests an extension of the work of Cockburn et al., consisting of the estimate of the relation between basic and applied research incentives separately for each submarket, in order to account for potentially different competitive and knowledge-appropriability conditions. In a study of the wage determination of software developers, Andersson et al. (2009) find that wages are more responsive to performance in more “risky” industry segments, where riskiness is measured in terms of the 90/50 ratio of product line sales per worker. The authors offer a sorting explanation for their results. Firms in highly risky environments benefit more from having star workers. In order to attract them, firms offer a better pay, both in terms of fixed and performance related-wage. Our results point to additional (though not necessarily mutually exclusive) explanations. Consider, for example, the video game software developing/publishing segment, indicated by Andersson et al. as the riskiest in their sample. Williams (2002), moreover, reports that this segment has experienced increased concentration over the 1990s, up to a four-firm concentration ratios greater than 50% in the early 2000s. The IT-software online journal SoftwareMag.com publishes a list of the biggest software companies. Among the 100 biggest companies of this survey, only two declare “Database” as their primary product line; eight indicate software for financial applications, and nine indicate infrastructure/networking software. Among these three segments, Andersson et al. indicate “networking” as the riskiest, and “database” as the least risky. Therefore, some of the riskiest segments of the industries are also those where firms are of larger size, which in our model corresponds to lower competitive pressure. In addition, intellectual property protection in software is relatively weak, and knowledge spillovers are pervasive.15 If higher riskiness happens to be associated with the presence of larger firms, then our model predicts that companies offer higher powered incentives in less competitive product lines (where firms are larger). The above results provide also insights to the “endogenous spillovers” literature that studies the choice of how “close” or “open” firm decide to make their research (Katsoulacos and Ulph, 1998, Lambertini et al., 2004, Milliou, 2009 and Piga and Poyago-Theotoky, 2005). Within this literature, the paper that is closest to ours is Milliou (2009). In an environment with one-dimensional research effort, this paper shows that firms can decide to make their research publicly available, especially when competition is not too intense. We obtain a similar result; however, by modeling two-dimensional effort, with different degrees of openness or spillovers, we specify that, even if the absolute amount of open research may increase, the strength of incentives for open research relative to appropriable research always decreases when spillovers are higher. Distinguishing between absolute and relative investments in types of research can therefore have important implications.