In behavioral economics, the endowment effect refers to the phenomenon that people usually demand higher prices to relinquish an object they own than they would be willing to pay to acquire the object they did not own (Kahneman, Knetsch, Thaler, 1990). For this phenomenon, dozens of studies have found the significant disparity between willingness to pay (WTA) and willingness to accept compensation (WTA), and this disparity is believed as an important reason to account for this effect. However, the WTA-WTP disparity cannot adequately interpret the endowment effect because sellers or buyers may be irrational traders. Therefore it is necessary to compare the endowment effect with the rational behavior. For this reason, in the present study we introduced the neutrals whose evaluation was regarded as reference price for reinterpreting endowment effect. In addition, one notable weakness of the previous researches in this domain is that the time period of the ownership is ignored. Although endowment effect can come from an instant valuation, the ownership in real life is dynamic and it may change with the time. Thereby, it is necessary to weigh the endowment effect on a dynamic time of change. Considering that, we inspected the changes of the WTA-WTP disparity in two continuous time periods including "past-present" and "present-future". In this study, we posited that valuation of sellers or buyers was decreased both in the "past-present" and "present-future" time periods, whereas WTA-WTP disparity was unchanged. When the valuation of neutrals as reference price (market price, MP) was introduced, WTA-WTP disparity would be divided into two parts: WTA-MP and MP-WTP. We inferred that MP would decline into a range between WTA and WTP, and the variation tendency of WTA-MP would be different from MP-WTP in a given time period. A total of 674 undergraduates and 35 computer sellers participated in three experiments. In Experiment 1, the participants were randomly assigned to each of 12 scenarios: 3 roles (seller, buyer, and neutral) × 4 time periods from past to present (3 months, 6 months, 12 months, and 24 months). Participants were asked to fill in expected price for a bike after reading a description including its cost price and usage time. In Experiment 2, participants were randomly assigned to each of 8 scenarios: 2 roles (seller and buyer) × 4 time periods from present to future (3 months, 6 months, 12 months, and 24 months). Besides that, 35 computer sellers were specifically assigned to all of 4 time scenarios as the neutrals. They were asked to fill in expected price for a computer after reading a description including its cost price, usage time and configuration. The results revealed that: (1) in all three experiments, selling price was always higher than the buying price, and this gap (WTA-WTP) was even unchanged both in the "past-present" and "present-future" time periods. (2) When valuation of neutrals was introduced as reference price, selling price was closed to reference price in three months, and their gap (WTA-MP) gradually increased following the time prolonging of the ownership. Whereas buying price was closed to reference price in a long time period of two years, and their gap (MP-WTP) gradually decreased following the time prolonging of the ownership. To sum up, the present study showed that endowment effect in trade stably existed even in a long time period. Whereas the introduction of the neutrals, which take the endowment effect into a dynamic time course, could provide a new perspective for the contestable previous viewpoints.