S-Space College of Engineering/Engineering Practice School (공과대학/대학원) Program in Technology, Management, Economics and Policy (협동과정-기술·경영·경제·정책전공) Theses (Ph.D. / Sc.D._협동과정-기술·경영·경제·정책전공)
Essays on the value of learning in strategic investment decision
전략적 투자 의사결정에 있어 학습의 가치
- 공과대학 협동과정 기술경영·경제·정책전공
- Issue Date
- 서울대학교 대학원
- Real options; Bayesian learning; Endogenous uncertainty; New market entry; Optimal learning; Trade-off between commitment and flexilbity
- 학위논문 (박사)-- 서울대학교 대학원 : 기술경영·경제·정책전공, 2017. 2. 이정동.
- Through three essays, this dissertation examines the implications of learning in strategic investment decision making under uncertainty by introducing the Bayesian real option model, which extends the existing real option theory. The existing real option theory mainly assumes exogenous uncertainty that is resolved through waiting and passive learning. However, in the real business environment, information about given investment opportunities is difficult to acquire through waiting. Therefore, this dissertation emphasizes active learning by assuming endogenous uncertainty that is resolved only through active information acquisition by the corporation as a major uncertainty source. This approach calls for a shift from the wait and see principle to the learning and see doctrine to explain the value of decision-making flexibility in the existing real option theory.
The first essay compares two approaches for new market entry under uncertainty: experience-driven planning, which blindly assumes existing knowledge and experience of a firm, and discovery-driven planning, which continuously updates the initial assumptions through information obtained by early entry into the market, without accepting existing knowledge and experience. The former is associated with the wait and see approach of the existing real option theory, and the latter is linked to the learning and see principle. Experience-driven planning is modeled as determining the optimal time of entry based on prior knowledge and experience. The discovery-driven planning invests a portion of the total opportunity at the present time, then modifies the initial assumption based on the information gained in the market, and finds the optimal time to acquire the remaining portion of the opportunity based on the modified assumption. However, in order to carry out discovery-driven planning, additional learning costs arise because the market is not yet sufficiently developed but the information is acquired from the market through early entry. The effects of maturity and relevance, which are the characteristics of the market in which the firm is seeking to enter, and the level of the core competence, which is a characteristic of the firm, on the relative value of these two defined approaches are analyzed. The value of discovery-driven planning is maximized when the market a firm aims to enter is highly irrelevant, but mature, and the core competency of the firm is low. On the contrary, experience-driven planning is advantageous when the firm targets an emerging, highly relevant market and the firm has a high level of core competence.
The second essay expands the discussion of the first essay and developed a model with sequential learning and flexibility in decision making about the innovation opportunities that the company has. As companies pay for learning and acquire information about a given opportunity, they can update the expectations of the opportunity from the learning outcome. Thus, the company has three options at each point in time: to stop learning activities and acquire opportunities, to continue learning activities, and to stop meaningless learning activities and give up opportunities. Under these assumptions, I assume that the value of a given opportunity is the Bellman equation with the expected value of the opportunity, which is continuously updated by learning activities and time. Then, using dynamic programming, I analyzed how the firms optimal behavior and the value of the opportunity change accordingly. From the analysis, higher prior uncertainty about the firms opportunity increases the value of the opportunity, while the uncertainty of the opportunity itself reduces the value of the opportunity. This result occurs because the coefficient that determines the variation of the posterior expectation over time is not an increasing function of the uncertainty about the opportunity, but an increasing function of prior uncertainty about the value of the opportunity. Moreover, it has been confirmed that even when the present value of a given opportunity is considerably negative, there is room to increase the value of an opportunity through an optimal learning strategy. In addition, a decrease in the value of opportunities relative to the increase in unit learning costs was relatively small. Further, higher prior uncertainty about the value of an opportunity has been shown to further increase the downward (posterior expectation) areas where continuous learning is the optimal behavior.
The third essay attempts to analyze the implications of learning effects on strategic investment decisions under uncertainty of the firm. Under uncertainty, companies fall into a dilemma of a trade-off between commitment and flexibility in strategic investment decisions. The reason is that the increase in uncertainty positively affects both commitment and flexibility. However, as can be seen in the first and second essays, the learning effect is created only through commitment, which represents the immediate action of a company. Therefore, increase in learning has a positive effect on commitment and a negative effect on flexibility. Thus, the effect of uncertainty on commitment and flexibility will vary with the magnitude of the learning effect. Through the theoretical model, I found the following results. If there is a learning effect over a certain scale, the increase in uncertainty is favorable for commitment. However, if the learning effect drops below a certain level, the increase in uncertainty favors flexibility. In addition, it was noted that the magnitude of the learning effect varies according to the type of investment and the environment in which the investment takes place, which empirically confirms the argument of the theoretical model. The type of investment is considered as R&D versus capital investment, while R&D investment environment signifies a comparison between high- and low-tech industries. In each case, the analysis of existing literature shows that the former investment is associated with more learning effects. From the analysis, it was found that uncertainty about R&D investment had the same threshold effect (negative impact below threshold and positive impact above threshold) as R&D investment only in high-tech industry.
Overall, the implication of this dissertation is that, under extreme and complex uncertainty, the company should not adopt a wait and see attitude but follow an active learning approach. To this end, companies can consider adopting active post-audit systems throughout the investment decision-making process. The main contribution of this dissertation is that it reflects on endogenous uncertainty, which receives little attention in the existing real option theory, by integrating Bayesian learning into real options theory. Future research should reflect on more advanced Bayesian learning techniques and develop a model to help companies make more informed decisions.