加载中...
City Size, Internal Migration and Non-tradable Goods Variety: Evidence from Meituan-Dianping
作为中国文化最显著的特征之一,中华美食享誉世界,是软实力的体现。然而,关于中国餐饮业的研究仍显不足,鲜有文献从全国尺度考察多样性福利。本文试图研究中国不可贸易品的分布规律。现代经济学尤其是新贸易理论高度重视分布规律,因为多样性偏好是现代经济理论的基本假设。从克鲁格曼以来的研究逐渐确定,源自供给集聚效应的规模经济能够增加生产中的产品种类并提升多样性福利;相较之下,消费和不可贸易品的规律仍被误解。按照城市经济学,所有基于特定区位、具有差异化并由具有特定偏好的消费者光顾的服务或商品均可视为不可贸易品。由于不可贸易品具有运输成本异质性和低替代性两大特征,越来越多的学者认识到,不可贸易品多样性比其它指标更能代理城市不可贸易品福利。此外,学者们已确定不可贸易品是城市舒适性的主要来源之一,也是吸引人们居住在城市的最重要因素之一。因此,研究不可贸易品的分布规律对城市发展具有重要意义。中国庞大的人口规模和独特的人口流动模式也推动了本文研究。"十二五"和"十三五"规划强调大城市应限制人口流入,使中国人口具有独特的流动性。近期实证研究表明,产业结构随人口规模系统性变化。然而,经济与人口之间存在极端的空间错配,削弱了城市人口集聚效应和规模经济效应。因此,本文在限制人口流动政策背景下研究不可贸易品多样性福利的潜在损失。本文从美团-大众点评获取餐饮数据,其分类标准包括"品类"和"菜品",可代理城市多样性。与已有研究相比,本文有三点贡献:首次使用大数据代理中国城市不可贸易品多样性;讨论中国人口规模、人口结构与不可贸易品多样性之间的关系;估计人口流动限制政策下中国城市不可贸易品多样性福利损失。本文将2015年大众点评餐饮数据与第六次人口普查及城市层面土地数据结合,实证检验人口规模与结构对不可贸易品多样性福利的因果影响。研究发现,多样性对人口规模的弹性介于0.528至0.696之间,对流动人口的弹性介于2.19至3.56之间。即"流动人口"既是城市的一类特殊品类,也促进城市创造新品类。本文支持人口规模和结构多样性对不可贸易品多样性福利的正向促进效应。基于工具变量估计,本文通过数值模拟以不可贸易品多样性作为福利指标,估计不同对数正态分布参数下中国城市的潜在损失。结果表明,当前人口流动限制导致巨大的多样性福利损失,尤其对大城市损失显著,但对中小城市具有保护作用。
Abstract
As one of the most representative traits of Chinese culture, Chinese cuisines are world famous and a reflection of soft power. However, the literature on the Chinese cuisine industry is significantly insufficient and few papers have considered variety welfare across the whole country. We attempt to determine the distribution rule of non-tradable goods in China. Modern economics, especially new trade theory, pays more attention to the distribution rule as preference for variety is a basic assumption of modern economic theory (Armington, 1969). Studies from Krugman until now have gradually determined that the scale economy, which comes from the agglomeration effect of supply, can increase the number of varieties in production and promote variety welfare (Krugman, 1979, 1980, 1991a; Broda & Weinstein, 2006). In contrast, the rule of consumption and non-tradable goods remain misunderstood. According to the urban economy, all location-based services or goods that are differentiated and patronized by consumers with a specific set of preferences can be regarded as non-tradable goods. Due to the two main characteristics of non-tradable good, including transport cost heterogeneity and low substitution, more scholars have recognized that varieties of non-tradable goods can better proxy for a city's non-tradable goods welfare than other indexes. In addition, scholars have determined that non-tradable goods are among the major sources of a city's amenities (Glaeser et al., 2001) and one of the most important factors attracting people to live in a city (Chen & Rosenthal, 2008; Lee, 2010). Therefore, it is important to study the distribution rule of non-tradable goods for urban development. China's huge population and unique population mobility pattern also motivate our research. The 12th and 13th Five-Year Plans insist that big cities should limit population inflow, which has made China's population uniquely fluid. Recent empirical work has shown that industrial composition varies systematically with population size (Mori et al., 2008; Mori & Smith, 2011; Hsu, 2012; Schiff, 2015). However, there is an extreme spatial mismatch between the economy and population that weakens the urban population agglomeration effect and scale economy effect (Lu, 2013). Thus, we research the potential loss of non-tradable goods variety welfare against the background of limiting population mobility policy. We acquire the cuisines data from Meituan-Dianping (dianping.com). There are two classification standards for cuisines, including categories and dishes, which can proxy for the varieties of a local city. Compared with other studies, ours makes three main contributions. First, we first use big data to proxy for the varieties of non-tradable goods in a Chinese city; second, we discuss the relationships between China's population size, population structure and varieties of non-tradable goods; third, we estimate the loss of non-tradable goods variety welfare in a Chinese city under the population mobility restriction policy. We combine the cuisines data from dianping.com in 2015 and consider the sixth census and land data at the city level to empirically test the causal relationship between population size and structure and the variety welfare of non-tradable goods. We find that the elasticity of variety in terms of population is between 0.528 and 0.696, while that in terms of fluid population is between 2.19 and 3.56. That is, the "fluid population" not only serves as a special category for the city but also encourages the city to create new categories. This paper supports the positive promoting effect of population scale and structure diversity on the variety welfare of non-tradable goods. Based on our estimation of the instrumental variables, we use varieties of non-tradable goods as welfare indicators through numerical simulations to estimate the potential losses of Chinese cities under different parameters of logarithmic normal distribution. The results show that the current limits on population mobility result in a huge variety welfare loss, especially for big cities, but have a protective effect on small and medium-sized cities.
城市人口规模显著提升不可贸易品多样性
以地级市截面数据估计,品类数量对人口规模的弹性为0.554(OLS),工具变量估计为0.528,即在1%显著性水平上,城市人口每增加1%,餐饮品类数量约增加0.53%,说明人口集聚带来的规模经济显著扩大了本地不可贸易品的多样性供给。
流动人口对多样性的促进效应更强
以流动人口比例作为人口结构代理变量,多样性对流动人口规模的弹性介于2.19至3.56之间,显著高于总人口的弹性,表明流动人口不仅自身构成特殊消费品类需求,还通过偏好多样性推动城市创造新品类,人口结构多样性对多样性福利的作用超过单纯规模效应。
人口流动限制造成大城市多样性福利损失
基于工具变量估计并采用对数正态分布数值模拟,在σ=1.75参数下,若人口流动受限,大城市多样性福利损失可达约499.92%,中等城市约166.75%,小城市约34.12%,表明限制人口流入的政策对大城市不可贸易品多样性福利的负面影响尤为巨大。
中小城市在人口流动限制下获得保护效应
数值模拟显示,人口流动限制对中小城市多样性福利的损失相对较小(约34.12%),甚至部分中小城市因人口回流而受益,说明统一的流动限制政策对城市体系产生非对称影响,大城市承担主要福利损失,中小城市相对受到保护。
核心解释变量
城市人口规模(pop)与人口结构(流动人口比例proportion),核心解释变量为地级市常住人口规模及流动人口占比
被解释变量
不可贸易品多样性,分别以餐饮"品类"(categories)数量和"菜品"(dishes)数量两类指标衡量
样本与数据
基于2015年美团-大众点评(dianping.com)餐饮商户数据,结合第六次全国人口普查城市层面人口数据及土地数据,样本覆盖全国1735个地级市观测,工具变量回归样本为1565个
识别方法 / 模型设定
采用工具变量法(2SLS)处理人口规模的内生性,工具变量基于1820年历史人口与地理适宜性指标构建,Kleibergen-Paap F统计量为64.3、54.03,超过Stock-Yogo 10%水平临界值7.03
内生性及稳健性检验
通过OLS与工具变量估计对比、替换被解释变量(品类与菜品)、加入GDP等控制变量、采用reduced-form回归以及不同对数正态分布参数(σ=1.75与σ=0.8449569)的数值模拟进行稳健性检验,结果保持一致
更多相关数据正在补充