Path-dependent Preferences and Polarized Public Response to Pandemic

Description

In this research project, we ask a fundamental question how individuals or households (henceforth, agents), given their varying backgrounds such as age, culture, education, ethicality, attitudes toward fatality risk and so on would optimally respond to a pandemic---in a liberal society where everyone is rational and well informed. We aim to develop and analyze a comprehensive, dynamic model of decision making under pandemic uncertainty that is sufficiently general while beingtractable, and to investigate the model predictions both theoretically and empirically.Compared to the existing literature, our model will exhibit the following novel features.• Asymptomatic as well as symptomatic transmissions of the disease• Path-dependent infection probability, affected by one’s past and current actions• Heterogeneous agents with observable and unobservable traits (or private information)• Stochastic transmission rate of the disease• Decision problems as a dynamic game with incomplete informationTo our knowledge, the above elements of consideration have not been incorporated in the analysis of pandemic-related literature to date. The present project will contribute a first step along these dimensions of reality.We will derive a sequential public-response equilibrium (akin to perfect Bayesian equilibrium in this context) and characterize agents’ optimal actions in each period during the pandemic. We will show that the common assumption of a concave matching function for modelling disease transmission implies polarized public responses, i.e., part of the agents will find it optimal to choose “safety-first” actions in each period to minimize their infection probability whereas the rest of the agents choose “life-as-usual” actions with zero protection. A dynamic comparative statics analysis will show agents who are more likely to get sick or who expect to suffer more once infected tend to choose safety-first actions, while the reverse holds true for agents who are less prone to sickness and sufferance. A free-riding problem, however, will prevent all agents from choosing safety-first actions even when it is socially optimal to do so.To show the validity of our model, we will investigate people’s responses to the pandemic in terms of various preventative measures in an empirical analysis. We plan to utilize two data sources: (a) a survey dataset collected from 29 countries and regions (b) an individual mobility dataset obtained using location records of mobile devices. The first dataset allow us to analyze what individual characteristics determine people's responses to the Covid-19 pandemic, while the second dataset is useful to obtain the distribution of heterogeneous choices of preventative behaviors and its impacts on the Covid-19 pandemic.The long-term impact of this project will be twofold regarding academic influence and policy implications. Conceivably, the general model will encourage many follow-up academic papers either sharpening or further extending the scope of the main results generated from this project. From the policy point of view, a social policy that is not well aligned with people’s preferences will cause frictions in the society and is not likely to be effective. Understanding people’s behaviors when they are free to act will provide valuable insights in the search for socially optimal objectives and the best ways to implement optimal policies.