The iCPI team is jointly instructed by Professor Taoxiong Liu and Professor Ke Tang from the Institute of Economics, School of Social Sciences, Tsinghua University, and Professor Bin Xu, Department of Computer Science, Tsinghua University. The members have economics and computer backgrounds, including some engineers, doctoral and master students.
In the era of digital economy, this project aims to explore the design and application of real-time high-frequency price indicators with online big data, and to nowcast inflation and monitor the macroeconomic conditions in real time, thus providing a faster and more accurate reference for macroeconomic decision-making.
1.Goods Basket
(1) The goods basket of iCPI follows the latest CPI basket of the National Bureau of Statistics of the People’s Republic of China, which consists of the main index, 8 divisions, 27 groups and 262 classes.
(2) Specifically, the 8 divisions comprise: Food, Tobacco and Liquor; Clothing; Residence; Household Articles and Service; Transportation and Communication; Education, Culture and Recreation; Health Care; Other Articles and Services.
2.Data Collection and Cleaning
(1) Based on the research methodology of big data, we collect daily price data from China’s main E-commerce platforms.
(2) Computers automatically collect the required data at the specified time, website, and location. Prices are collected once a day and stored in a dedicated database after collection and cleaning.
3.Index Calculation
(1) The methods of determining iCPI weights are consistent with the official CPI, that is, the weights at all levels should be determined according to the consumption expenditure structure of households. We deduce the weights of different levels by synthesizing a variety of literature and international CPI compilation methods.
(2) The calculation methods of iCPI are also consistent with the official CPI. We calculate the daily, weekly, ten-day, and monthly price indices at all levels, including the main index, 8 divisions, 27 groups and 262 classes.
4.Online Publishing
(1) iCPI is generated in automatic computation procedures including data collection, data cleaning, and final processing to online publishing, which avoids human intervention and improves its validity.
(2) iCPI published through our website is updated every day including daily, weekly, ten-day and monthly indices.
(3) iCPI can also be downloaded in databases, including Bloomberg, Wind and CEIC.
(4) More details are shown in the paper: Liu T X, Tang K, Jiang T F, Zhang L, (2019). Design and Application of Novel CPI Based on Online Big Data [J]. The Journal of Quantitative & Technical Economics, 9:81-101.