For the first essay, we generalize the classic Williams [1998 RFS] brokerage model by introducing the diffused effort. That is, the agent can cross-utilize effort spending on one listing to another one. Besides, the agent can manage heterogeneous housing assets. One counterintuitive finding in Williams' paper is the absence of the agency problem. As a special case in our model, we recover the agency problem. We examine the positive externality due to the diffused effort and show that it depends on the agent's inventory size. Hence there exists a trade-off between agents' effort spending on existing listings and on finding a new listing. For the second essay, we model a home seller's pricing decision under a generally defined prospect value function. We show a simple disposition effect is caused by reference dependence, but it only exists when the agent is risk neutral. Diminishing sensitivity will lead to a two-way disposition effect by generating a local reverse disposition effect, a range in which the seller's asking price decreases with increasing potential loss. Loss aversion tends to magnify the disposition effect and hence mitigates the reverse disposition effect. One direct implication is that acclaimed tests on loss aversion such as Genesove and Mayer [2001] and Pope and Schweitzer [2011] are likely invalid. We present evidence consistent with the model by using multiple listing service data from Virginia. Our findings suggest that studies which predominantly focus on a one-way disposition effect can be overly simplistic and misleading as it depends on the strong assumption of risk neutrality.