This dissertation comprises three chapters that empirically investigate various kinds of market inefficiencies arising from seller misbehaviors. The first two chapters focus on physicians' overtreatment in healthcare markets. A standard reputation system falls short of effectively curbing overtreatment due to the credence-good nature of such a market: patients cannot tell whether a high-cost treatment recommendation (versus a less costly and complex treatment) is necessary even after the service is completed. In the first chapter, I propose a novel solution to reinstate the function of reputation by combining a reputation mechanism with patient search for second opinions. I conduct a laboratory experiment to test this new mechanism. In the second chapter, motivated by newly-emerged public information, the aggregate count of each type of treatment provided by a health provider, I experimentally investigate whether such information could reduce overtreatment and replace patients’ reliance on costly searches in repeated interactions. In the third chapter, the focus shifts to market power in wholesale electricity markets. In this study, my co-author and I utilize nuclear refueling outages as a test for the exercise of market power by energy suppliers in the New England market during the years 2016-2018.