| 裴丹,陈林,韩胜飞.网络型自然垄断产业纵向分离改革绩效评估[J].数量经济技术经济研究,2023,(1):192-212 |
| 网络型自然垄断产业纵向分离改革绩效评估 |
| An Impact of Structural Separation Reform on China's Mobile Telecommunications Industry: An Empirical Study Based on a Small Sample Difference in Differences Model with Randomization Inference |
| |
| DOI： |
| 中文关键词: 数字基础设施 纵向分离 自然垄断 随机推断 小样本双重差分模型 |
| 英文关键词: |
| 基金项目:本文获得国家社会科学基金重点项目（20AZD050）、广东省哲学社会科学规划重大基础理论研究课题（GD21ZDZYJ01）、广东省自然科学基金面上项目（2022A1515011107）、中央高校基本科研业务费项目（JDGTT202117）和国家人才项目的资助。 |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 中国铁塔股份有限公司的成立，是中国移动通信业的一次纵向分离式改革，也是国家对信息网络产业的“统一大市场”式改革。近期党的二十大报告提出加快建设数字中国、加快建设网络强国，而移动通信业正是“数字中国”和“网络强国”的基础设施领域之一,为考察其改革成效，并在“十四五”时期更好地指导数字信息基础设施建设，本文使用鲍莫尔（Baumol）的成本次可加方法实证测算铁塔设施的自然垄断属性，并剖析成本属性影响行业产出绩效的具体机理，然后在国内首次构建基于随机推断的小样本双重差分模型（Randomization Inference Differences-in-Differences，RI-DID）评估此次纵向分离改革的政策绩效。实证结果表明，信息基础网络上游环节的通讯铁塔设施有一定的非自然垄断属性，垄断式的经营因推高成本而导致下游运营商产出显著下降。从政策含义看，在网络型自然垄断产业或非自然垄断的网络环节，纵向分离改革需审慎推进。从方法论看，本文一是改良了成本次可加模型的数据选取方式，避免因不区分上下游业务环节及垄断利润过高造成自然垄断属性误判；二是引入随机推断法，在一定程度上控制了双重差分法的小样本学术难题，有助于双重差分模型在政策绩效评估中的进一步科学应用。 |
| 英文摘要: |
| China Tower Co., Ltd. was established in 2014 to separate China's mobile communications infrastructure sector from mobile operators. The government press release at the time stated that the major reasons for creating a mobile tower monopoly were to reduce duplicate investments, increase cost efficiency, and mandate mobile tower sharing among operators. This article analyzes the impact of that industrial structural reform on industry output, and theorizes on the associated causes. The study also provides a solution to the small sample problem, which scholars face when evaluating policies related to the telecommunications industry and other similar industries. We start by using the translog cost function and Seemingly Unrelated Regressions (SUR) to evaluate whether the upstream mobile tower facilities are cost sub-additive. We then apply a Randomization Inference-Differences in Differences (RI-DID) estimator to both company level and industry level data to test the impacts of the reform on output. Theoretical models are used to explain how the change in the upstream cost can affect industry output under an oligarchy market structure. We find the cost functions of mobile tower facilities are not sub-additive. This indicates that constructing and maintaining mobile tower facilities do not follow a natural monopoly cost structure. As such, monopolistic operations increase total costs. Our theoretical models indicate that if the upstream cost of any mobile operator increases, it decreases industry output. The results of our two RI-DID models indicate that the 2014 reform was associated with a statistically significant decrease in mobile telecommunications output. The theoretical and empirical analysis indicates that the decrease in output is due to the increased upstream cost. The main causes of duplicate construction include a lack of planning and regulation. Thus, we recommend that more competition be introduced to the upstream mobile tower construction and maintenance industry. In addition, telecommunications laws are needed to regulate China Tower and mobile operators. Furthermore, the government has indicated that a similar structural reform may be applied to other similar industries. As such, we recommend that structural reforms, such as vertical separation, should not be enforced across all network industries before performing cost sub-additivity tests. When the cost function of an industry is not sub-additive, other regulations (rather than structural separation) should be considered to promote market competition and increase efficiency. This paper makes three key improvements compared with previous studies. First, we separate upstream wholesale market data from downstream retail data, and only test if the upstream network is cost sub-additive. This approach is effective when studying industries with a fixed network. Second, compared with other Chinese studies that have applied a cost sub-additivity test, we use the total number of users, instead of revenue, as output indicators. This avoids the potential influence of monopoly profit on regression outcomes, and reflects the hypothesis that user data are a better indicator for output in the digital era. Third, we apply a RI-DID model to solve statistical problems associated with small sample regressions in oligarchy markets, where the total number of companies is usually small, making it difficult to evaluate the impacts of industrial policies. We combine the cost sub-additivity test and RI-DID models to evaluate the impacts of structural reforms and other industrial policies. This provides both theoretical and empirical evidence concerning the outcomes of the reform. Further, the methodologies used in this study can inform the evaluation of industrial policies for other similar industries in China, where data are limited due to a small number of market players. We also address the small sample problem in DID regression, which should help advance the use of RI-DID models in other empirical studies of oligopoly markets. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
