Urban population distribution in China: Evidence from internet population

Based on mobile internet user data, we construct an “Internet population” measure and reexamine spatial population distribution in China. The location based service (LBS) data of mobile internet uses is able to capture the accurate location of users' residence and solve the underestimation problem of missing migrants. We have three main findings. First, contrary to previous studies based on traditional population statistics, city size distribution of Internet population fits well into Zipf's law with a R2 of 90.7%. Second, the Internet population indicator is superior to traditional population statistics in explaining inelastic household consumption such as water consumption, electricity consumption, and garbage disposal. It suggests that the “Internet population” is a better proxy of actual city population. Third, the traditional population statistics systematically overestimate population in small cities and underestimate population in large cities. It indicates that the public resource distortions will continue to exist or even worsen off in China if the allocation process relies greatly on traditional population statistics. Although no measures are perfect, our new population measure provides important incremental information for future discussion.