This thesis comprises three essays. The first essay studies operational implications of yield uncertainty in mergers and acquisitions. During past few decades, M&A activities have been widely observed in the manufacturing industries, such as integrated circuit, bio-pharmaceuticals. As an important business strategy, the merger effects have been discussed from perspectives of market competition, cost reduction, price collusion, without concerning any production characteristics. To fill this research gap, the essay focuses on the background of yield uncertainty, and explores how the firms' merger behavior would affect their production decisions and equilibrium performances under a Cournot competition market. The analysis reveals that, the post-merger firm realizes a lower yield uncertainty than those non-participating firms after a horizontal merger. We term this phenomenon as yield uncertainty effect, and study its different interactions with the traditional merger effects, i.e., market competition effect and cost synergy effect. The results indicate that, yield uncertainty is an important determinant for merger behavior, and it exists robust threshold structure, such that any small merger in the Cournot competition market might be profitable even without cost synergy benefit. Besides, the yield uncertainty effect and cost synergy effect complement with each other, and both have opposite impacts compared to the market competition effect, on the equilibrium outcomes. This interplay reveals novel insights that distinct from the traditional merger theory. For example, when facing high yield uncertainty, the firms can still benefit from merger behavior even if the cost synergy effect is not significant, leading to a higher total output and a lower retail price in the post-merger market. Meanwhile, the consumer surplus and social welfare can also be improved.

The second essay considers the firms' estimation behavior under yield uncertainty, and studies its implications on their production decisions and market performances. Apparently, the risk of yield uncertainty is a severe challenge to the firms' development, due to which some firms estimate their future yield fluctuation when deploying production policy. During this process, their estimation behavior might suffer from overconfidence or underconfidence biases for limited knowledge, past experience or different personality. Therefore, the essay intends to study how these estimation biases would affect the firms' production decisions and equilibrium profits, as well as the interests of other relevant parties under both symmetric and asymmetric markets. The results indicate that, an overconfident (underconfident) firm would behave aggressively (conservatively), and increase (decrease) its target production quantity. In both cases, we derive a unique threshold of the estimation bias, below which the firms can achieve higher profit levels than the benchmark case. Specifically, in the symmetric case, there exists an overconfident bias interval, during which the firms' profits, consumer surplus and social welfare can be improved simultaneously. Furthermore, under the asymmetric case, a more biased firm might perform better than its less biased rival, even if the less biased firm is strategically unbiased.

The third essay studies inventory and service level management for competing firms, in the presence of strategic and heterogeneous customers. Due to limited supply, the risk of product unavailability has been a key factor in attracting market demand, such that customers would strategically pace their choices when gaming with multiple firms. Incurring high searching cost, these strategic customers are increasingly concerned about the inventory service level before making their purchase decisions. In this essay, service level refers to the probability that an individual customer would get the product if he/she pays a visit. Therefore, we use the Hotelling model to characterize the strategic customer behavior, based on which the firms' profit optimization problems are formulated under both centralized and decentralized markets. The results suggest that, the optimal solutions exist uniquely under different market structures. Compared to the traditional newsvendor model, the firms should provide a higher service level when facing strategic customers. Nevertheless, if a firm is irrationally biased to ignore product availability effect, it would provide a lower service level for a large product value, but a higher one for a small value than a rational firm that internalizes the effect. In this case, the firm's irrational behavior does not necessarily bring adverse effects. Under the decentralized market, the biased decision might alleviate intensified market competition, so that all the participants' profits can obtain Pareto improvements.