Experimental Design
In this study, we analyze individuals’ economic expectations during the COVID-19 epidemic in China using an online longitudinal survey of approximately 1,900 individuals from seven provinces in China conducted from late February to mid-March of 2020. In each wave, we ask incentivized questions regarding respondents’ expectations on GDP growth rate in the first quarter of 2020, the period that Chinese economy was heavily affected by the COVID-19 epidemic, as well as their expectations on a few other economic indicators.
We carry out the research in two stages using both quasi-natural experiment and randomized controlled trials (RCTs). In the first stage of the analysis, we study the impact of the severity of COVID-19 epidemic on individual’s economic expectations. Specifically, we match individual’s revision of economic expectation between two waves with the number of COVID-19 new cases in the city that he/she lives in. Given that we conduct three waves of survey on the same respondents, we can measure each respondent’s revision on economic expectation twice. Using an individual fixed effect model, we can then study the causal impact of the number of COVID-19 new cases at the city level on individual’s revision of economic expectation, assuming that the number of COVID-19 new cases between the two waves (approximately 6-7 days) in a city is an exogenous information shock to individuals.
In the second stage of the analysis, we study the impact of economic recovery during the COVID-19 epidemic on individuals’ economic expectations from two perspectives. First, we elicit respondents’ belief on the current level of economic recovery in China, measured as a percentage compared to the normal level of economic activities (i.e., the so-called work resumption rate in China), and examine whether respondents’ perceived work resumption rate would affect their economic expectations based on an individual fixed effect model. Second, we implement a randomized controlled trial with information treatment in the third wave of the survey. More specifically, after eliciting respondents’ pre-treatment economic expectations and perceived work resumption rate, we randomly divide the respondents into five groups. In the baseline group (T0), we provide the current level of work resumption rate in China measured by the AI experts at Tsinghua University. In the other four treatment groups, we provide additional information on the work resumption rate in some selected cities, including the major city with the highest (T1) and lowest (T2) levels of work resumption rate, Beijing (T3), and the provincial capital of the individual’s residing province (T4). After the information treatment, we ask the respondents to revise their economic expectations. We aim to understand whether individuals revise their economic expectations according to the perception gap of the economic recovery level, and whether the way that the information is presented would affect their revision of economic expectation.