Abstract: As an irreplaceable input factor and consumer product in the economy, the large oil price fluctuations may severely impact the economy and financial markets. The impact of oil prices on the economy should be reflected by the stock market. However, different types of oil price shocks affect the stock market in different directions, so the total stock price is always weakly correlated or uncorrelated with the oil price, which is known as the “oil price β puzzle”. In addition, different industries may react differently to oil price shocks. For example, a positive oil price shock can be positive for oil-producing firms in terms of increased production profits, but it may be negative for the consumer industry in terms of increased production costs. Therefore, the impact of heterogeneous oil price shocks on different industries cannot be ignored when studying the impact of oil price shocks on the stock market.
In order to investigate the dynamic impact of oil price shocks on the stock markets, there are four main steps in the study. First, with reference to a novel crude oil price decomposition method, a SVAR model is used to decompose oil price shocks into supply shocks, demand shocks, and risk shocks, based on the futures price of WTI, the VIX index, and the World Composite Oil and Gas Producers Stock Index. Second, based on the WIND industry index, the Generalized Forecast Error Variance Decomposition (GFEVD) method is used to derive a spillover index between heterogeneous oil price shocks and stock sector volatility, to test whether there is a risk transmission between them and identify the main factors of oil prices affecting the stock market by ranking the spillover effects. Third, using heterogeneous oil price shocks as the main systematic risk factor (pricing factor), a time-varying parameter multifactor asset pricing model is constructed and estimated using quasi-Bayesian local likelihood. The “oil price β puzzle” in the Chinese stock market is explained in terms of the heterogeneity of oil price shocks, the different direction and magnitude of the impact on the industry, and the time-varying coefficients. Fourth, the ARIMAX (p,d,q) model is constructed to further analyze the predictive effects of heterogeneous oil price shocks on stock market sectors.
The research results mainly include the following five aspects.(1)Oil price shocks affect the Chinese stock market mainly through risk shocks, followed by demand shocks and supply shocks. The effects of heterogeneous shocks on Chinese stock market sectors are significantly differentiated. The financial sector is most affected by supply shocks, the communication services sector by demand shocks, the energy sector by risk shocks, and the energy sector by total shocks. The impact of risk shocks on Chinese stock market sectors has a greater uncertainty. (2)The constant-parameter regressions show that the total oil price shock is insignificant for all stock sectors except the energy sector, suggesting that the “oil price β puzzle” does exist. The supply shocks significantly negatively affect the financial and real estate sectors, the demand shocks significantly positively affect the energy and financial sectors, while the oil price risk shocks affect all sectors significantly. (3)Time-arying parametric regressions show that extreme events are reflected in the dynamic coefficients, and most stock market sector responds significantly negatively to supply shocks and positively to demand shocks most of the time. (4)The main reason for the “oil price β puzzle” is that heterogeneous oil price shocks affect sectoral returns in different directions, the same oil price shocks have alternating positive and negative effects on sectoral returns, and total oil price shocks affect sectoral returns in different directions at the same time. (5)Heterogeneous shocks have different predictive effects on stock market sectors. In particular, aggregate price shocks can have a short-term positive predictive effect on some sectors and supply shocks have an insignificant predictive effect on the vast majority of sectors, while demand shocks positively affect sectoral returns in the longer term and risk shocks have a short-erm negative effect on sectoral returns.
According to the research conclusions of this paper, we can get some inspirations. Industries, financial regulators and stock investors should identify the types of shocks behind oil price shocks to make accurate decisions. For example, the industries can cope with the adverse effects of oil price supply shocks through commodity price adjustments and business strategy adjustments, and equity investors can take reasonable investment decisions based on the predictive effects of heterogeneous shocks on stock market returns, such as increasing holdings of equity assets in the case of positive oil price demand shocks while decreasing holdings in the case of larger risky shocks. Regulators should pay special attention to the adverse effects of risky shocks, especially on the financial sector, and take the necessary regulation to prevent the systemic financial risks.

Key words: heterogeneous oil price shocks, stock market, spillover index, multi-factor asset pricing model, time-varying parameter regression

摘要： 本文从行业视角出发,基于溢出指数模型和常量及时变参数多因子资产定价模型,全面分析了国际油价的供给冲击、需求冲击和风险冲击对中国股市的动态影响。研究表明,不同驱动因素所致的油价冲击(异质性冲击或结构性冲击)对股市各行业的风险溢出效应不同,风险冲击是石油价格影响行业股市的核心因素;常量参数多因子模型下仅风险冲击对所有行业影响显著,时变参数多因子模型下供给冲击对金融、材料、可选消费和公用事业等行业负向影响显著,需求冲击对所有行业正向影响显著,风险冲击对所有行业负向影响显著;“石油价格β之谜”源于异质性石油价格冲击对各行业的影响方向不同、同一冲击对同一行业正负影响的交替出现和同一时期总冲击对各行业的影响方向不同;需求冲击对各行业股市收益率具有长期正向预测功能,风险冲击则在短期内具有负向预测功能。

关键词: 异质性石油价格冲击, 股票市场, 溢出指数, 多因子资产定价模型, 时变参数回归