| 吕越,谷玮,尉亚宁,包群.人工智能与全球价值链网络深化[J].数量经济技术经济研究,2023,(1):128-151 | | 人工智能与全球价值链网络深化 | | Artificial Intelligence and Deepening of the Global Value Chain Network | | | | DOI： | | 中文关键词: 全球价值链 网络人工智能 增加值贸易 社会网络分析 | | 英文关键词: Global Value Chain Artificial Intelligence Value-added Trade Social Network Analysis | | 基金项目:本文获得国家社会科学基金重大项目“新发展格局下中国产业链供应链安全稳定战略研究”（21&ZD098）、国家自然科学基金面上课题“全球价值链、创新驱动与制造业‘低端锁定’破局：成因、机制及应对策略”(71873031)、国家自然科学基金面上课题“全球疫情大流行下国际国内价值链重构对中国的影响和应对策略”（72073025）、对外经济贸易大学杰出青年学者资助项目（20JQ02）、国家社会科学基金重点项目“依托国内市场优势、强大国内生产网络与贸易强国建设”（21AZD024）的资助。 | | | 中文摘要: | | 发展以人工智能为代表的高新技术是中国推动高质量对外开放、实现更高水平融入全球价值链分工网络的重要依托。本文基于Melitz（2003）和Bai等（2019），在异质性企业出口决策模型中引入人工智能，将企业出口模型拓展至企业增加值出口模型，并实证检验了人工智能发展对GVC网络深化的影响和内在机制。本文研究结果显示，各国人工智能产业的进步能显著促进GVC网络的深化。人工智能对GVC网络的积极影响主要是通过劳动力替代和缓解资源错配实现。相比于发达国家，人工智能对深化发展中国家GVC网络的促进效应更强；相比于高出口依赖型国家，对低出口依赖型国家的积极效应更加突出。人工智能除了影响各国GVC网络的深化外，还能延长GVC长度，增强GVC竞争力，以及推动各国向GVC上游攀升。这一发现对当前“双循环”新发展格局构建——“内循环为主、外循环赋能”具有重要的战略指引。 | | 英文摘要: | | China has recently achieved rapid developments in its manufacturing industry, by maximizing the advantages of its labor force to introduce many processing and assembly jobs. It has also deeply integrated itself into the global value chain (GVC) network. However, this growth has come with increased labor costs. As such, many countries are seeking new technological upgrades and breakthroughs as a primary strategy for developing individual GVC networks. In this context, many countries have successively formulated transformation and upgrade plans for manufacturing industries, especially industrial robots using artificial intelligence (AI), considered critical for future manufacturing development. This highlights the fact that promoting high-quality development and increasing integration into the GVC network requires the high-tech support provided by AI.This research empirically studies the impact of AI on the GVC network and analyzes the relevant mechanism associated with the impact. Based on Melitz (2003) and Bai et al. (2019), this paper introduces AI and misallocation of resources into the export decision-making model of heterogeneous enterprises, which describes the conditions of firms' export decision first introduced by Melitz (2003). The study also analyzes how AI influences enterprise value-added exports. On this basis, we use industrial robot data from various countries, released by the International Robot Federation from 2000 to 2014, and measure the GVC network index from the perspective of value-added trade. We also empirically test the development of AI in different countries with respect to the deepening of GVC networks.The results show the following. (1) The progress of the AI industry in different countries can significantly deepen GVC networks. (2) The positive impact of AI on the GVC network is mainly achieved through labor substitution and by mitigating the misallocation of resources. (3) AI has a stronger impact in deepening GVC networks in developing countries compared with developed countries. There is also a more prominent positive effect for low export-dependent countries compared to high export-dependent countries. In addition, the financial crisis may somewhat weaken the positive role of AI in deepening the GVC network. The effect of AI on GVC network centrality is more substantial in labor-intensive and technology-intensive industries, and is not significant in capital-intensive industries. (4) AI can also extend the length of GVCs, and enhance GVC competitiveness and the degree of upstream reach. Based on previous studies, this paper makes the following key contributions. (1) In terms of research themes, this paper explores the impact of AI on GVC networks. Previous studies have focused on analyzing specific facts about GVC networks. In contrast, this paper concentrates on the internal mechanism behind the deepening of GVC networks. The study also significantly enriches the examination of AI impacts. (2) In terms of theory, this paper extends the heterogeneous firm exports model of Melitz (2003) and analyzes the mechanism by which AI influences GVC networks, by introducing AI and misallocation into firms' export decisions. This enriches existing theory. (3) In terms of data, this paper combines industrial robot data released by the International Robot Federation (IRF) at the industry level in each country from 2000 to 2014 with the University of International Business and Economics (UIBE) GVC Indicators database, the World Development Indicators (WDI) database, and the Worldwide Governance Indicators (WGI) database. Based on the UIBE GVC Indicators database, this paper constructs and measures country-industry-level GVC network indicators, based on value-added trade. It also measures different dimensions of GVC network indicators, such as the average degree, global efficiency, centralization of eigenvector centrality, reciprocity, assortative characteristics, and the global clustering coefficient. | | 查看全文 相关附件： 下载数据代码附录 |
| | |
