| New theoretical developments and recent experimental studies involving the sealed-bid k-double auction mechanism for bilateral bargaining under incomplete information have raised new questions about procedures that induce efficient bargaining behavior and about the applicability of extant adaptive learning models. It is now generally accepted that a theory of bargaining behavior for individuals who typically do not meet the stringent assumptions about common knowledge of rationality cannot be complete without systematic empirical investigations of the properties of the various mechanisms that structure bargaining. The aim of this dissertation is to critically explore the extent to which efficient bargaining outcomes can be achieved while dynamically accounting for individual behavior across repeated play of the game. In the first study, an endogenous bonus is introduced into the baseline single-stage game. Although theoretically doing so induces truth-telling behavior for both players, the experimental data provide very limited support. In the second study, the baseline game is extended by incorporating an additional, costless period of bargaining, thereby giving players an increased opportunity to reveal information about their respective reservation values. The data show that subjects quickly learn not to reveal information about their private valuation despite the increased opportunity to bilaterally improve efficiency. Finally, the third study investigates behavior sensitivity to variation in the trading parameter, k. Instead of following the historical precedent of setting k = ½, extreme values of k are invoked in an asymmetric information environment endowing a player with exclusive price-setting power. Although theoretical analysis suggests that expected profits for a seller (buyer) decreases (increases) in k, experimental results show that under conditions of dramatic information asymmetry, the observed share of the surplus is much smaller for the player with price setting power if countered with an information disadvantage resulting in poor support of the LES. Furthermore, the price setting power effectively counters the information disparity advantage demonstrated in previous studies. Results from a previously proposed reinforcement-based adaptive learning model not only demonstrate robust applicability across studies but also the model's ability to account remarkably well for the dynamics of play across iterations of the stage game. |
