Security Rating Analysis Based On Semiparametric Latent Variable Transformation Models

Up to now, China's a-share listed companies reached more than 2,700 enterprises. In the process of globalization, development of the financial sector are no longer confined to any one country. In order to solve the balance problems in the financial markets, and to help investors get a better deal with market risks, credit rating agencies have emerged and gradually grow. Rating agencies'role should be to protect the interests of investors and improve market efficiency, but it is regrettable that China Securities is characteristic of leading rating agency did not play its due role like a mature market.2015 is a fluctuation of Chinese stock market, a great year. In a very short period of time, the stock market rises and falls, a disorder in the market. Most of the investors have no clear market, blind investment caused a lot of damage to property, only on June 12 of last year's stock market value evaporate on trillion yuan. In this case, investor rating agencies are increasingly high requirements.With the development of economy, financial markets and financial products of the new development and more and more complex, the rating agencies must change and development. Variety of factors affecting the securities rating, different securities and rating standards for securities analysts at rating agencies also vary, how to eliminate the subjective factors of the different security agencies is an important issue. This article in hopes that latent variable model is proposed in the context of existing securities rating system to identify a potential security risks under index, measure of its stock rating, and find out the latent variable at the same time, an analysis of influence factors of securities ratings and forecasts.Innovation of the article is more than just simple ratings to certain securities institutions to study, but different rating agencies for the same stock rating combines the results of the study, which can influence the stock rating of the potential impact of unobserved factors, you can also get a new indicator is no longer dependent on the securities rating agency ratings. Semiparametric latent variable model was also used to analyze the different ratings, with rating results in discrete variable predetermined transformation model in discrete output variables of the study has some limitations, and deviation of model scenarios using Semiparametric latent variable model can very well avoid these problems.In the first part of this paper presents the research questions, and the full text of the research background, research significance and the results described in this paper and instructions. Second part sums up the securities rating history, domestic and foreign securities institutions for securities rated evaluation criteria, at home and abroad for securities rating and the rating value of related research brief. The third section is a relevant model and model estimates, about this part is divided into two chapters. The last part to summarize the entire article.