全国大市场中贸易流量的省际行政边界效应——来自地级市增值税发票的证据

The Provincial Administrative Border Effect of Trade Flows in the National Unified Market:Evidence from Chinese Prefecture-level VAT Invoice Data

建设全国统一大市场是构建现代经济体系、推动高质量发展的必然要求,而行政边界效应则是造成市场分割的重要原因。与现有文献不同,本文基于2017年中国地级市间增值税发票计算的城市间贸易流量数据,在因果推断框架下对我国省际行政边界效应进行了估计和分解。结果显示,我国省际行政边界效应平均约为4.45倍,即省内相对贸易份额约为省际相对贸易份额的4.45倍。这个结果明显小于欧盟内部的国家间的行政边界效应。中我国东中西部的省际行政边界效应逐次递增,并且相比于商品市场,劳务、服务等要素市场的边界效应更大。本文也用现有文献的主流方法进行了估计,并与文献中已有的估计结果对比。本文采用的数据和估计方法在一定程度上缓解了使用省际贸易流量数据和引力模型估计可能存在的偏误,有助于科学评价我国统一大市场建设进展和经济体制改革成效。

The Border Effect is one of the most interesting topics in the literature due to its great practical significance to the construction of market integration.The estimation of border effect is mainly based on the gravity model,the law of one price,or the degree of specialization,and the data used in the corresponding study are usually the trade flow data,price data,input-output tables or micro enterprise data.However,several biases may underline the estimation of provincial border effect by the provincial trade flow and classical gravity model.First,the gravity model neglecting the geographical factors other than distance may result in serious overestimation of border effect,especially in the regions with large geographical differences.Second,the inclusion of intra-regional trade in the model leads to overestimation of border effect theoretically.Third,the underestimation may appear in the gravity model because of the problem of bad control.Finally,it may also underestimate the border effect using the provincial trade flows compared with the prefecture-level trade flows.In this paper,we resort to the causal inference framework to estimate and decompose the provincial border effect in China by a novel dataset constructed by the value-added tax invoices,which contains all the trade flows between prefectural cities in 2017.We combine the propensity score matching method and a series of geographical variables,including the distance,location remoteness,altitude,river basin,and ruggedness,to obtain highly comparable intra-provincial and inter-provincial prefecture-level trade pairs.We find that the average provincial border effect is about 4.45 in China,that is,the trade between the intra-provincial cities is 4.45 times as much as that between the inter-provincial cities.Obviously,it is a big obstacle for cross-border trade in China,but much smaller than the national border effect in the European Union.Furtherly,the provincial border effect in the eastern region is significantly lower than that in the central and western regions.Concretely,it is 2.77 in East China,3.97 in Central China,and 7.76 in West China.We also find that the border effect in the labor and service market is much larger than that in the commodity market.The border effect in service and labor is 12.26 and 5.47 respectively.While it is ranging from 3.18 to 4.27 in the commodity market such as industrial goods.The intangible products with lower transport costs,including intangible assets and real estate,have the lowest border effect of 1.81 and 2.13.Therefore,the degree of integration of the commodity market is higher than that of the factor market in China,which is consistent with previous literature.We also try to estimate the provincial border effect by the currently popular methods in the literature and compare the results with ours.To some extent,our data and method can rectify the potential biases caused by the gravity model setting and provincial trade data as mentioned above.Our work will contribute to evaluate the process of market integration and the performance of economic reform in China.The accurate estimation of the provincial border effect is not only out of the academic interest,but also due to the policy-making demand.For example,if the geographical obstacles effect is wrongly mixed to the border effect,the government may unilaterally overemphasize the provincial coordination policies,but neglect the construction of transportation infrastructure.Besides,the decomposition of the border effect in different regions and industries can provide an important base for exploring the causes of the border effect and building the effective channels to improve the national unified market.We also provide some suggestions to the policy makers.It is necessary to establish a coordinated mechanism across regions,gradually eliminate administrative barriers to public services,and further exploit the advantages of richer nature resources,lower factor costs,and larger market potential in the central and western regions,strengthen their ability to undertake industrial transfer,to accelerate their industrialization and urbanization processes.