联系我们

论文查询结果	返回搜索

论文编号：	4515
作者编号：	1120080754
上传时间：	2012/6/13 13:16:47
中文题目：	预测市场的实验研究
英文题目：	Experimental Research on Prediction Market
指导老师：	李建标
中文关键字：	预测市场；个人所得税免征额；信息汇聚；赢者通吃合约；风险态度
英文关键字：	Prediction Market； Individual Income Tax Starting Point； Information Aggregation； Winner-Take-All Contract； Risk Attitude
中文摘要：	摘要行为经济学和实验经济学证据表明，人们在决策过程中往往偏爱于确定性事件或者确定概率的选择。奈特首次区分了风险和不确定性，认为风险是可度量的不确定性，不确定性是不可度量的风险。预测市场，作为一种新兴的预测工具，在不确定性和风险之间架起了一座桥梁。在信息多元化和行为异质性的不确定性世界中，对未来未知事件的估计和准确预测对于个人和组织至关重要。大量证据表明预测市场机制作为区别于传统预测工具的一种新型预测方法有着惊人的预测能力。预测市场机制的研究对推进经济理论发展和推广实践应用都有重要的意义。一方面预测市场验证了市场汇聚分散信息、传播信息和处理信息的能力；另一方面，预测市场为市场监管部门和市场运行主体提供了一种全新的信息收集方式。预测市场不仅能够对目标预测事件的期望值进行点估计，而且还能够对事件的概率分布做出预估。相对于其他预测工具，具有预测准确、反应即时、激励性强、洞见性、易扩展等优势，这在发达国家的政治选举、公司决策、公共事件的预测中已经得到证实。如果将预测市场实验移植到我国财税政策的预测中，市场能否真实地汇聚分散在大众中的信念？预测绩效是否比其他预测方法更加准确？哪些事件作为预测市场的预测目标更为恰当？市场中单个交易者预测准确性与交易者的哪些特征有关？在目前的交易方式中哪种收益支付方式能够有效地激励交易者在市场中真实地揭示自己的信念？基于此思路，本研究自2011年5月8日起实施了为期53天的个人所得税免征额预测市场实验，针对大众对个人所得税免征额的几个猜测值，在该市场中设计了5个赢者通吃型合约，这5个合约覆盖了个人所得税免征额的所有可能区间，5个合约在连续竞价双向拍卖市场中进行单独交易。2011年6月30日，全国人民代表大会常务委员会公布了个人所得税免征额为3500元，市场结束。市场对个人所得税免征额产生了较为准确的预测，合约B在市场结束前的几天中价格显著高于其他合约，市场交易得出的结果显著高于专家意见、名人微博等预测。为什么预测市场能够将分散在大众中间的信息以证券价格的形式汇聚？观察到预测市场在预测我国财政公共政策方面的预测能力后，本文将“前定”事件结果的预测市场移植到实验室中，将其他不相关的因素剥离掉，在第四章和第五章中设计了离散型事件结果的抽球预测市场实验和连续型事件结果的猜数预测市场实验，在抽球预测市场实验中考察不同信息抽取方式对预测市场价格收敛速度的影响，在猜数预测市场实验中考察信息强度和被试经验对预测市场信息汇聚速度的影响，在两个实验室预测市场中同时考察了交易者风险态度对个体预测准确性和个体交易量的影响，发现交易者风险态度对交易量有正向影响，与交易准确性并不相关。本文的主要创新有三点：第一，针对结果连续型事件，本文原创了私人信息强度不断增加的猜数预测市场实验，将预测市场实践中不相关的因素剥离掉，在市场中严格控制信息的强度和被试的经验，通过双向拍卖机制决定证券的价格，一部分交易者获得有关预测结果的私人信息，在实验过程中不断增加私人信息的数量，观察各只证券（彩票）价格对信息的反应程度和信息对价格的调整作用。第二，在对国外预测市场理论提炼的基础上，在实验室中抽象了结果离散型事件预测市场的预测过程，通过设计抽球实验，在实验中考察不同的信息抽取方式对市场收敛有效预测的影响，真实地再现信息不断揭示、汇聚的过程，通过双向拍卖交易机制来确定证券的价格，证券的最终赎回价格与装入箱子的真实组合有关，与事件相关的证券的价格会对私人信息与干扰信息的减少做出不断调整。第三，本文设计了中国具体经济情境的预测市场实验——个人所得税免征额预测市场实验，对比网站调查、专家微博预测等方法来检验预测市场对中国具体经济情境的预测能力，给出了市场对个人所得税免征额发生概率分布的预期，探讨了预测市场应用过程中的激励方式等问题。在微观层面，本文探讨了交易者风险态度、性别和工作经验对个体交易量和预测准确性的影响。 关键词：预测市场；个人所得税免征额；信息汇聚；赢者通吃合约；风险态度
英文摘要：	Abstract The evidence from behavioral economics and experimental economics show that people tend to prefer to choose the certainty and certain probability. Frank H. Knight distinguished the risk and uncertainty. And he asserted that the risk is measurable uncertainty, while uncertainty is the immeasurable risk. As a novel prediction tool, prediction market establishes a bridge between risks and uncertainty. In an uncertain world with diverse information and heterogeneous behaviors, accurate predictions are essential to individuals and organizations. Enormous evidence shows that the mechanism of prediction market as a new prediction means has surprising predictive ability compared with traditional prediction tools such as surveys. Research on prediction market has important implication for both development of economic theories and explicit application. On the one hand, prediction market tests the market capacity to aggregate diverse information in classic economic theories. On the other hand, prediction market provides a new method of information extraction for market regulation departments and market participants. Prediction market could not only make point estimation to the expectation value of aimed events, but also estimate the events’ probability distribution. Compared with other prediction tools, it is of much accuracy, immediacy incentive and high scalability. Furthermore, it has been proven to be available in the political election, business decision and public events of the developed countries. Applying the prediction market experiment to predicting the fiscal and taxation policy in our country, could the market reflect the real belief in the public? Is it more effective than other methods? Which events are more appropriate to be the target of prediciton? Which properties of the traders are associated with the accuracy of prediction about the very trader? Which payoff mode could encourage traders to reveal their real beliefs in the present trade mode? Based on this train of thought, we established the 53-days prediction market experiment about the individual income tax starting point on May 8th in 2011. Besides, we designed 5 winner-take-all contracts in this market, and the 5 contracts covered all possible ranges of the individual income tax starting point. And all the contracts were traded respectively in the double continuous auction market. It was announced the individual income tax starting point was 3,500 RMB by the standing Committee of the National People's Congress, and at the same time the market was over. It turns out that the market is more accurate, and the price in Contract B is evidently higher than other contracts. In addition, the conclusion from the market trades is also more accurate than the blogs of the famous and the votes in web. Moreover, the incentive mechanism according to the rank of the traders could reveal the real beliefs of the traders. The risk attitude affects the trade quantity notably positively, while this has no evident effect on the accuracy of the individual prediction. Why the prediction market can aggregate the information dispersed in the public in the form of the stock price? After realizing its ability to predict the fiscal and public policies in our country, we remove the prediction market to the lab, and get rid of the irrelevant factors. In chapter 4th, we design the ball-drawing prediction market experiments which are discrete events. In these experiments, we mainly study if different information extraction methods would affect the speed of information aggregation. Further, in chapter 5th, we apply numbers-guessing experiments which are continuous ones to observe what effect information intensity and subject experience make on the speed of information aggregation. In both markets, we study the impacts that traders’ risk attitudes have on the individual prediction accuracy and trading volume. And then we find that risk attitudes influence trading volume positively, but are irrelevant to the accuracy. Several potential innovations in this dissertation are as following: First of all, we originate the gradual increase of the intensity of private information in the numbers-guessing experiments, and get rid of the irrelevant factors in the prediction practice. We strictly control the intensity of information and the subjects’ experience. The stock price is still decided by the double auction mechanism. Some traders have the private information about the outcome of prediction, and then we increase the quantities of the private information gradually to observe the response of every share price and the adaption information makes to the price. Secondly, on the basis of refining the foreign prediction markets theories, we abstract the procedure of discrete event outcome prediction market. Through designing the ball-drawing experiment, we study the impact that the different information extraction methods have on the effective market aggregation prediction. In addition, it is a true representation about the revelation and aggregation procedure of information. The stock price is decided by the double auction mechanism, and the final liquidation price is associated with the combinations of balls put into the box. And the relevant stock price could adapt to the individual information and the reduction of disturbance one. Thirdly, we design the prediction market experiments in the frame of China economic environment——individual income tax starting point experiments. We compare it with the investigation in web and experts’ speculation in their blogs, and give the expectation of the probability distribution about individual income tax starting point. Besides we probe into the incentive methods in the market and etc. In the micro-level, we investigate how the traders’ risk attitudes, gender and working experience influence individual trading volume and prediction accuracy. Key Words: Prediction Market； Individual Income Tax Starting Point； Information Aggregation； Winner-Take-All Contract； Risk Attitude
查看全文：	预览  下载(下载需要进行登录)