基于聚类动态面板引力模型的国际贸易网络研究

本文提出一种针对网络型数据的聚类动态面板引力模型,用于国际贸易流量网络的研究.该模型假设各贸易国分属于不同的潜在类别,各国间贸易流量对应的模型系数由出口国和进口国所属的类别决定.提出使用马尔可夫链蒙特卡罗方法对模型参数以及各贸易国所属的潜在类别进行贝叶斯估计.对2001-2015年60个国家间的贸易流量数据进行了实证分析.结果表明,所提出的模型能够对贸易国进行聚类,有效地提高贸易流量预测的精度.所提出的聚类动态面板引力模型可以被广泛的应用于其他动态网络型数据的研究.

Dynamic panel gravity model with latent clusters for the international trade network

This paper proposes a dynamic panel gravity model with latent clusters (DPG-LC) to study the international trade network. In the model, the countries are divided into several latent groups. The model coefficients depend on the group labels of the exporting country and the importing country. The Markov chain Monte Carlo method (MCMC) is used to estimate the model under the Bayesian framework. The model is applied to the international trade data of 60 countries from 2001 to 2015. The results show that the DPG-LC model can effectively improve accuracy of predicting trading volumes. The proposed model can be widely applicable to study other dynamic network data.