行政审批改革对制造业出口技术复杂性的影响分析

The data on administrative approval reform are sourced from the county-level administrative approval center database developed by Professor Xu Xianxiang’s research team at Lingnan College, Sun Yat-sen University. This dataset covers the period from 1995 to 2015. Administrative approval centers were established nearly every year after 2001 (Fig.1). Due to limitations in the data from the "China Industrial Enterprise Database," this study selects the sample period from 2001 to 2013 to examine the impact of administrative approval system reform on the technological complexity of manufacturing exports.The data processing and matching process for the technological complexity of manufacturing exports involve three main steps:Step 1: Calculate the Dependent Variable, Export Technological Complexity. To calculate the technological complexity of products, as specified in formula (14), global per capita GDP data and bilateral trade data from various countries are required. The BACI Bilateral Trade Statistics Database compiles bilateral trade data and values for HS6-coded products from 246 countries and regions worldwide. This dataset is then matched with the per capita GDP data of these countries and regions, sourced from the World Bank database, to compute the technological complexity of HS6-coded products. Subsequently, the technological complexity of HS6-coded products is matched with data from the China Customs database, and the manufacturing export technological complexity at the enterprise level is calculated according to formula (15).Step 2: Match Customs Data with Industrial Enterprise Data. Before matching the customs data with the industrial enterprise database, we exclude samples that lack enterprise names, destination country names, or product names. Additionally, we remove samples with a single transaction size below $50, or a quantity less than 1, as well as those involving agricultural or resource products and trade intermediaries. Following the methodology of Su et al. (2018) [29], we first match the data based on enterprise names and years. For the remaining samples, we proceed with matching using the last seven digits of the phone number and the postal code.Step 3: Processing the Industrial Enterprise Database. Following the methodologies outlined by Nie et al. (2012) and Brandt et al. (2012) [30][31], we processed and filtered the samples from the industrial enterprise database. Specifically, we undertook the following steps:(1) Outlier Removal: We removed outliers and missing values related to industrial value-added, sales revenue, and the number of employees. This included eliminating negative values, instances where fixed capital exceeded current capital, and enterprises that did not meet the criteria for large-scale enterprises.(2) Exclusion of Specific Industries: We excluded enterprises classified under the "mining" and "electricity, gas, and water production and supply" industries, as identified by their 2-digit industry codes.(3) Geographical Exclusion: We also excluded industrial enterprises located in the four municipalities of Beijing, Shanghai, Chongqing, and Tianjin.