Dynamic Models of Innovation and Learning

This dissertation analyzes three dynamic models of innovation. The primary focus of the analysis of each model is to highlight the role of endogenous mechanisms---such as observational learning and product market competition---in determining firms' R&D incentives, the speed of innovation, and the efficiency of equilibrium investment.

In Chapter 2 of this dissertation, entitled "Learning and the Timing of New Technology Adoption," I develop a continuous-time model in which two asymmetric firms (incumbent and challenger) decide whether to adopt an innovation of uncertain quality; and if so, when. Early adoption is valuable, as it comes with the possibility of generating a first-mover advantage. However, early adoption comes at the price of greater uncertainty because the innovation’s true quality can be learned only by observing market performance. In particular, firms learn through a process of market experimentation by which the innovation’s true quality is revealed at a rate proportional to first-mover sales. Through this endogenous learning channel, the flow of information becomes inextricably linked to market structure, which I show has significant implications for equilibrium pricing incentives and the dynamics of innovation diffusion. In particular, my results provides an information-theoretic explanation for why established incumbents, despite possessing substantial innovative ability, are disproportionately slow to respond to disruptive innovation—commonly referred to as the Innovator’s Dilemma.

In Chapter 3 of this dissertation, entitled "The Role of Market Structure in Competitive Experimentation," analyzes the strategic incentives of firms to trade-off investment into new versus existing technology. Specifically, I develop a continuous-time model in which two firms choose how to allocate a stock of resources toward either R&D (to develop an possible innovation) or current production (to maximize short-term competitiveness). From a technical standpoint, this is the first model of strategic experimentation in which direct payoff externalities arise through endogenous product market interaction. In terms of empirical contribution, my model provides an explanation for (1) the sensitivity of results in empirical work on R&D competition to industry specification and (2) the seemingly paradoxical decisions of historically innovative companies (e.g. Polaroid, RCA, Xerox) to abandon R&D highly-promising new technologies. Finally, the model be used for policy analysis—namely, to evaluate the effect of horizontal mergers on innovation and welfare. Specifically, my results provide a micro-foundation for conceptual arguments made in recent high-profile merger cases (e.g. Dow/DuPont) based on an “innovation theory of harm," by which a statically pro-competitive merger may be viewed as dynamically anti-competitive due to its negative effects on dynamic innovation incentives.

In Chapter 4 of this dissertation, entitled ``Leadership and the Value of Persistence,'' which is co-authored with James Anton and Dennis Yao, Co-authored work with James Anton (Duke Fuqua) and Dennis Yao (Harvard Business School) expands the scope of my existing research to include organizational incentives for innovation and the economics of leadership. In a project called ``Leadership and the Value of Persistence,'' we seek to answer the following question: "What is the value of having a persistent leader in an organization?" To make this question precise, we develop and analyze a dynamic model of organizational innovation. In the model, a leader faces a sequence of projects and wants to motivate self-interested managers to work on them so that they can succeed and generate value for the organization. The key strategic tension is that individual managers only internalize the portion of project returns they receive, but the leader internalizes the project’s value to the entire organization. In equilibrium, leaders use persistence (i.e. the refusal to give up on failed projects) for several reasons. First, persistence helps overcome the problem of moral hazard in teams by creating dynamic incentives for self-interested managers to work on projects of low individual value but high organizational value. Second, we show that persistence can be used to credibly communicate private information about high value projects. Intuitively, a leader’s decision to persist after failure provides good news to managers about the project’s value which, in turn, motivates effort. However, persistence can also exacerbate the problem of moral hazard by creating strategic incentives for managers to "experiment with failure" by shirking on current projects in order to learn about the project’s value through the leader’s persistence decision.