The Change Points Diagnose Of Oil Price And Stock Market Relationship Based On Bayesian PHMM Model

Change is the eternal theme of financial markets, policy change, the financial crisis, institutional change event shocks often have huge impact on the financial markets, it may cause structural changes in the relationship between financial variables. In the financial economics studies, if you do not consider the impact of these events change point on the financial relationship between the variables, it is easy to cause decision-making bias, and lead errors to the financial variables relationship assessments. Under the background of large T panel data, the smooth data process could not guarantee variable relationship didn’t exist structural changes, and the unobserved heterogeneity between the cross-section of non-time-varying hypothesis is difficult to establish, using traditional panel data model to the relationship between financial variables change point’s study is difficult to solve the problems above, for the unobserved heterogeneity between the panel data cross-section of non-time-varying hypothesis caused censored variable bias and the inference invalid problems. By introducing the Hidden Markov method to construct the time-varying heterogeneity and the structure of qualitative variability relationship changes situation, using Bayesian Theory to statistic the model parameters, design MCMC sampling algorithm and define Bayesian factor to solve the model selection problem (i.e. variable dot diagnosis’problem), this paper constructed Bayesian fixed effects and random effects panel data Hidden Markov model to describing the qualitative variation, capture change points which may exist in the financial system environment and investigate their financial variables non-systemic and systemic effects. Finally, take oil price and stock market relationships transition points as research objects, select international crude oil price index,13major industries stock market index and China’s interest rate index data, investigate the financial system of environmental changes systemic and non-systemic effects on the oil prices and the stock market relationships.The results indicated that the presence of long-term stable relationship between international crude oil price and stock market, but the long-term stable relationship environmental will be affected by the financial system shocks, and13industries affected by the system environment can be divided into non-systemic and systemic effects, each individual industry has three to five change points, but the common change points are only three; and with large T panel data context, if considering the inter-industry time-vary ing unobserved heterogeneity, we can found that different system environment events will make different correlations in different time periods in oil price and stock market relationships, it verified the effectiveness of the Bayesian fixed effects and random effects panel data Hidden Markov model.