基于在线大数据的中国商品价格粘性研究

The Stickiness of Online Prices in China

近年来,数字经济迅速发展,同时宏观经济的短期波动变得愈加频繁,理解宏观经济短期分析的基础--价格粘性、减少宏观政策时滞变得愈加重要。本文利用来自100多个网站的高频价格大数据(囊括CPI篮子的8大类、46中类、262子类的1970多万条日度商品价格),测度中国商品价格粘性程度、识别价格调整模式,并将微观测度结果用于测算货币政策有效性。研究发现:中国总体上调价较为频繁(调价周期小于2个月)、调价幅度较大(14%-20%)、部门异质性明显、不对称性显著(上调频率及上调幅度更大);调价模式是异质性的时间相依(TDP)和状态相依(SDP)相结合;调价频率(负向)、调价大小的峰度(正向)、异质性部门数量(正向)等均会影响货币非中性程度,相同的货币冲击对各行业的影响有显著的异质性。本文为洞察数字经济对商家调价行为和宏观经济动态的影响抛砖引玉,对大数据时代的货币政策制定和通货膨胀管理具备一定的参考意义。

Summary:The rapid development of the digital economy and big data technology has profound impacts on the economy and society.Along with this rapid development,short-term fluctuations in the macroeconomy and financial system have become more frequent.To reduce macroeconomic policy lags,it is important to understand the micro-foundations of the macroeconomic short-term fluctuations,namely price stickiness.However,due to the lack of price data,empirical evidence of price stickiness in China is relatively scarce.Online big data provide us with a new opportunity to study this important issue.In this paper,we measure static price stickiness indicators and identify the dynamic price adjustment patterns in China using a unique daily online price dataset.The data come from the iCPI project(www.bdecon.com)of Tsinghua University and contain online prices from more than 100 websites covering the whole basket of goods used in the Chinese Consumer Price Index(CPI),with over 19 million price records broken down into 8 divisions,46 groups and 262 classes.We combine the empirical evidence with a sufficient statistic for the real effects of monetary shocks(Alvarez et al.[KG-3],2016)to measure the monetary non-neutrality of different heterogeneous sectors.We also explore the application of weak online price stickiness to construct a high-frequency online consumer price index and nowcast inflation.We find that online prices in China are not very sticky:the average price change duration is less than two months(about 45 days),lower than those found by most other studies.The weighted average absolute price change is about 14%,higher than what is found in the literature.The overall price-adjusting frequency is about 17%higher during price surges relative to price drops.The magnitude of the price increases is about 38%higher on average than that of the price decreases but with heterogeneity by class,consistent with the findings of Nakamura&Steinsson(2008).Additionally,the non-parametric survival analysis reveals that the hazard rate of price changes inclines slightly upward with some peaks.A semi-parametric Cox proportional model shows that both time-dependent and state-dependent variables affect the price adjustment probability.These results indicate that the price adjustment pattern in China follows a typical combination of the time-dependent pricing model(TDP)and the state-dependent pricing model(SDP).The price change frequency,kurtosis of the price change size,price adjustment mode and heterogeneity can all affect monetary non-neutrality.For example,monetary non-neutrality is positively and negatively linked to the kurtosis of the price change size and the price change frequency,respectively,and the same monetary shock can have heterogeneous impacts on different sectors.Considering more heterogeneous sectors improves the explanation for economic fluctuations.We make three contributions.First,we use a unique high-frequency micro price dataset from China that has full coverage of the CPI basket and comes from multiple platforms,implying that our data are representative of the Chinese economy.Second,we connect the microeconomic evidence and macroeconomic applications of price stickiness in China and explore the influence of the rapid development of the digital economy on price adjustments and monetary non-neutrality.Third,we provide new empirical evidence and a micro-foundation for a better general equilibrium price adjustment model that fits Chinese facts.This paper provides insights into the impacts of the digital economy on price adjustments and macroeconomic dynamics.These findings have important implications for monetary policy formulation and its transmission mechanisms,the development of online real-time high-frequency price indicators and the improvement of inflation prediction timeliness in the digital economy and big data era.