基于copula函数的花卉拍卖相关性研究

发布时间：2018-06-09 01:11

本文选题：Copula函数 + t-Copula-GARCH模型　；参考：《昆明理工大学》2015年硕士论文

【摘要】：我国花卉业的发展一直是以传统的花卉交易市场作为主流的,直到2002年11月12日,我国第一个花卉拍卖市场昆明国际花卉拍卖交易中心正式运营,预示了新的交易方式的产生。针对这种全新的交易方式,我们可以对成交量和成交价格等交易数据进行研究。花卉的成交量和成交价格具有着一定的关系,同一品种不同等级的花卉成交量之间以及同一等级不同品种的花卉成交量之间也具有着一定的关系。研究这些相关性,能够给买家、卖家、拍卖市场提供一定的指导意义,从而保证买家买到更满意的花卉,卖家获得更多的收益,拍卖市场更好的运营壮大。目前比较传统的相关性分析方法是Granger因果关系分析和线性相关系数的分析,但这两种分析方法只适用于有线性相关关系的变量之间,如果面对具有非线性相关关系的变量继续使用线性相关系数来度量两者之间相关性,则可能会出现误差甚至得出错误的结果。copula函数是一种将联合分布同它们各自的边缘分布连接在一起的函数,由它推导出来的相关性度量方法在线性和非线性单调增变换下都保持不变,具有更广的应用范围和实用性,在描述变量间的相关结构及相关性方面拥有更多的优势。因此,本文运用copula函数,选择花卉成交量和花卉成交价格这两个变量,对成交量和成交价格之间的相关性进行了研究,同时也选取了同一等级的卡罗拉和满天星的成交量进行研究,通过相关性的研究可以提醒花卉拍卖市场卖家在交易的时候关注某一变量出现上涨和下降的情况,给出了具体影响的概率值,为卖家进行交易分析提供了量化指标,希望能够为卖家进行交易分析时提供一定的理论依据,提高自己的收益。本文在介绍Copula函数理论和拍卖理论的基础上,通过建立copula模型来度量花卉成交量和成交价格之间的相关性。首先建立时间序列模型来分别确定花卉成交量和成交价格的边缘分布,参照模型估计的结果对样本数据进行适当的处理,然后选用GARCH模型来模拟花卉成交量和成交价格收益率的边缘分布,并评价了模型的拟合效果；文中选择用Frank Copula、二元正态Copula和二元t-Copula函数来表示花卉拍卖的成交量和成交价格之间的相关结构,接下来引入经验Copula函数,得出了二元t-Copula函数具有最好的拟合效果。为了尽可能准确地度量花卉的成交量和成交价格之间的相关结构及相关性,本文经过不断筛选最终确定使用t-Copula-GARCH模型,得出拍卖市场花卉成交量和成交价格之间存在负相关关系,最后以卡罗拉和满天星为例子,得出同一等级的卡罗拉和满天星的成交量具有密切的正相关关系。这些结论同时具有理论和现实意义,为花卉拍卖市场交易者进行花卉交易分析提供了一些有用的建议。
[Abstract]:The development of flower industry in our country has always been based on the traditional flower trade market. Until November 12, 2002, Kunming International Flower auction Center, the first flower auction market in our country, was officially put into operation, indicating the emergence of a new trade mode. In view of this new trading method, we can study the transaction data such as volume and transaction price. There is a certain relationship between the transaction volume of flowers and the transaction price, the volume of flowers of the same variety of different grades and the volume of different varieties of flowers of the same grade also has a certain relationship. The study of these correlations can give buyers, sellers, auction market to provide certain guidance significance, so as to ensure that buyers buy more satisfied flowers, sellers get more income, auction market better operation and growth. At present, the traditional correlation analysis methods are Granger causality analysis and linear correlation coefficient analysis. If a variable with a nonlinear correlation continues to use linear correlation coefficients to measure the correlation between the two, The copula function is a function that connects the joint distribution with their respective edge distributions. The method of correlation measurement derived from it is invariant in both linear and nonlinear monotone increasing transformations, and has a wider range of applications and practicability, and has more advantages in describing the correlation structure and correlation between variables. Therefore, using copula function, this paper selects the two variables of flower turnover and flower transaction price, and studies the correlation between the transaction volume and the transaction price. At the same time, we also selected the same level of Corolla and starry stars to study the transaction volume. Through the correlation study, we can remind the sellers of the flower auction market to pay attention to the rise and fall of a certain variable in the course of trading. The probability value of specific influence is given, and the quantitative index is provided for seller's transaction analysis. It is hoped that it can provide certain theoretical basis for seller's transaction analysis and improve his own income. Based on the introduction of Copula function theory and auction theory, this paper establishes a copula model to measure the correlation between the volume of flowers and the transaction price. Firstly, a time series model is established to determine the marginal distribution of the volume and price of flowers, and the sample data are processed appropriately according to the estimated results of the model. Then the GARCH model is used to simulate the marginal distribution of the volume of flowers and the rate of return of the transaction price, and the fitting effect of the model is evaluated. In this paper, Frank Copula, binary normal Copula and binary t-Copula function are chosen to express the correlation structure between the volume and the transaction price of flower auction. Then, the empirical Copula function is introduced, and the best fitting effect is obtained by introducing the bivariate t-Copula function. In order to measure the correlation structure and correlation between the volume and price of flowers as accurately as possible, this paper determines the use of t-Copula-GARCH model after continuous screening, and concludes that there is a negative correlation between the volume of flowers and the transaction price in the auction market. Finally, with the example of Carola and full Star, it is concluded that there is a close positive correlation between the turnover of Carola and the full Star of the same class. These conclusions are of both theoretical and practical significance, and provide some useful suggestions for the analysis of flower trade in flower auction market.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F326.13;F724.59

【参考文献】