Journal of Systems & Management ›› 2022, Vol. 31 ›› Issue (3): 453-466.DOI: 10.3969/j.issn.1005-2542.2022.03.004

Previous Articles     Next Articles

Production Network and Industry Stock Returns:Spatial Factor Model Based on Time-Varying Network Dependent Parameters

SHI Zelong, FU Qiang, LI Shan   

  1. School of Economics and Business Administration,Chongqing University,Chongqing 400044,China
  • Online:2022-05-28 Published:2022-06-08

生产网络与行业收益率:基于时变网络依赖参数的空间因子模型

石泽龙,傅强,李山   

  1. 重庆大学经济与工商管理学院,重庆 400044
  • 作者简介:石泽龙(1985-),男,博士生。研究方向为系统性金融风险及资产定价。
  • 基金资助:
    国家自然科学基金资助项目(71373297)

Abstract: A complex and closely related production network has been formed in the industry through the input-output relationship, which has an important impact on the financial market. Meanwhile, the traditional factor pricing model ignores the influence of the production network, which can easily lead to model misspecification and confusion of systemic risk sources. Therefore, this paper constructs a spatial factor pricing model based on the time-varying network dependent parameters of the production network, by embedding into the traditional factor pricing model the time-varying production network constructed based on the input-output table released by the National Bureau of Statistics. By using this new pricing model, this paper studies the impact of the interaction between production networks and systemic risk factors on the rate of return of industries in China, by discomposing the total risk exposure of industry returns to factors into direct risk exposure and indirect risk exposure. The results show that compared with the traditional factor models, the spatial factor pricing model is more applicable to the rate of return and asset pricing of the industry in China. Compared with developed countries, China’s industrial structure has undergone significant changes in the past two decades, and the production network formed based on input-output relationship has also changed greatly. Therefore, the assumption of setting time-varying parameters as the spatial dependence parameters is more applicable to the economic reality in China, which will produce better empirical results. The decomposition of factor risk exposure also indicates that the spillover effect of the production network can amplify the risk exposure of industry returns to factors.

Key words: production network, time-varying spatial factor model, risk exposure decomposition, industry stock returns

摘要: 行业间通过投入产出关系形成了复杂且紧密关联的生产网络,对金融市场有着重要影响。同时传统因子模型忽视了网络结构所产生的影响>,易产生模型误设及混淆系统性风险来源的问题。因此,依据国家统计局发布的投入产出表构造时变的生产网络,将其嵌入到传统因子模型中,构建了基于生产网络的时变网络依赖参数的空间因子模型,进一步将行业收益率对因子的总风险暴露分解为直接风险暴露和间接风险暴露,研究了生产网络与系统性风险因子之间的相互作用对中国行业收益率的影响。研究表明:相比传统因子模型,采用空间因子模型更符合中国行业收益率与资产定价的实际;与发达国家相比,近20年来中国产业结构发生了重大变化,依据投入产出所形成的生产网络也发生了较大的变化,研究中假设空间依赖参数为时变参数更符合中国经济现实且实证业绩更佳;通过对因子风险暴露的分解,发现生产网络溢出效应能放大行业收益率对因子的风险暴露。

关键词: 网络, 时变空间因子模型, 风险暴露分解, 行业收益率

CLC Number: