| 摘要: | 本論文乃是利用賽局理論所發展出來的均衡概念，處理在訊息不完全之下「信用評價制度與網路交易」的議題。我們分成三個主題討論，得到結果如下：第一，有關「網路交易信用評價機制的探討」。我們讓買賣雙方都有上網交易的選擇，且賣方的類型為其私有資訊，我們研究結果顯示，即使沒有「信用評價機制」，仍會有交易，只不過存有詐欺誘因的賣方，每期都會詐騙；而在「信用評價機制」下，存有詐欺誘因的賣方，第一期會以正常交易獲得正評價，取得信任後，再於第二期詐騙的信用作弊現象。因此，此機制要有作用，不在於評價本身，而在於網路上願意誠實交易的賣方多寡。第二，有關「資訊不對稱下均衡概念之關聯─以網路交易分析為例」。此內容建構在一個賣方與兩個買方的兩期網路交易模型上，我們讓賣方型態與交易期數為賣方的私有資訊，其中賣方類型假設有誠實型與一般型，誠實型賣方偏好交貨，一般型則偏好不交貨，因此，我們利用Kreps and Wilson（1982）發展的「信譽效果」概念，以對偶方式呈現我們的兩期模型。然而，我們的模型有別於連鎖店賽局只用直觀準則，就可完全刪除買方用不合理猜測所支撐出的均衡。我們證明此模型，必須進一步採用D1準則，甚至到策略穩定性要求，才有唯一的均衡路徑。這一部份的分析，是以往相關文獻上不需討論的，其間可清楚看出，在資訊不對稱下各均衡概念的差異。我們並應用Gambit軟體以數值模擬方式找出Nash均衡，具體呈現各個均衡概念的關聯。在得到唯一滿足策略穩定性要求的序列均衡路徑下，我們的研究結果顯示，「信用評價制度」的引進，的確在鼓勵賣方採取誠實行為上有其貢獻。第三，有關「交易期數資訊的揭露程度對網路交易的影響」。在前述文章的基礎下，我們讓交易期數依資訊揭露時點的不同，分為三種設定，即完全揭露、部份揭露與完全不揭露，以探討賣方交易期數這個資訊在不同揭露程度下之網路交易行為。我們研究的結果顯示，當賣兩期的可能性較高時，網路平台引進「信用評價制度」，不僅增進兩期市場的總購買量，而且提高兩期市場的總交貨量；不過，當賣方賣一期的可能性較高時，即使誠實型賣方出現的機率非常高，若在完全不揭露的期數資訊下，因為認為沒有第二期市場，故其對兩期市場總購買量的貢獻，反而低於沒有實施「信用評價制度」的結果。因此，引進「信用評價制度」，也無法保證一般型賣方會有動機採取交貨行為。This paper is theoretical analysis of game theory equilibrium concepts to discuss with incomplete information in reputation mechanism and online trading. Therefore, we develop models that aim to capture the reputation mechanism and online trading. We divided into three topics, the results are as follows:About the first topic, we discuss ‘the Exploration of Reputation Mechanism in Online Trading’. We assume seller and buyer a selection of online transactions, and seller’s type is his private information. We find that even if there is no "reputaion mechanism", there will continue to be traded and an opportunistic seller will cheat. After considering the class of binary reputation mechanism where buyers can only report the outcome of a transaction as ‘positive’or‘negative’, the reputation induce current cooperation using the threat of future punishment following negative reports, about the seller's current behavior. One important result of the topic, the mechanism is to be useful, not the mechanism itself, but sellers willing to honest transactions on the online trading. About the second topic, we discuss ‘Connecting Equilibrium Concepts in Asymmetric Information−the Market for Online Transactions’. The topic builds a two-period signaling game to study whether or not the ‘Ordinary’ long run seller has sufficient incentive to gain credibility for his second sale by delivering the good in the first period. We assume that short-run (one-shot) buyer is uncertain about the seller's type, and whether or not the seller will be in the market in the second period is also seller’s private information. This model can be viewed as a dual model of Kreps and Wilson (1982). Contrary to the uniqueness results in Kreps and Wilson, we have multiple equilibrium paths passing Intuitive Criterion. Hence, in order to obtain a unique equilibrium path, we must apply D1 Criterion and require strategically stable. Therefore, we can tell the discrepancy under each type of equilibrium concepts under asymmetric information. We also apply Gambit software to get numerical simulation of all pure strategy Nash equilibria of this model. In the unique strategically stable equilibrium, when the probability of an ‘Ordinary’ short run seller is sufficiently low, the outcome is that the ‘Ordinary’ long run seller has sufficient incentive to gain credibility for his second sale by delivering the good in the first period. Hence, concern for reputation can encourage sellers to adopt honest delivery behavior, thereby benefiting the market for online transactions.About the final topic, we discuss ‘Online Trading with Different Degrees of Trade Frequency Revelation’. The topic applies the model of the previous research to study the impact of different degrees of trade frequency revelation. Whether or not the seller will be in the market in the second period is either public information or seller’s private information. In the unique sequential equilibrium, when the probability of the second transaction is high, introducing a feedback system increases the total trade volume and shipping. However, when the probability of the second transaction is low, with high probability of ‘Honest’ seller, introducing a feedback system decreases the total trade volume if neither buyer nor seller knows the second transaction for sure. Hence, concern for reputation cannot encourage ‘Ordinary’ seller to adopt honest delivery behavior. |
