Information, Belief, and Health Behavior: Evidence from China

Individuals with imperfect information may make suboptimal choices, but providing more information may not effectively improve decision making if the information is not turned into updated belief. We build a Bayesian updating model to illustrate this phenomenon and use a unique Chinese survey that provides data on information shock, belief updating, and corresponding behaviors to test it. We find that when individuals receive signals about their hypertension status, behavioral changes, such as quitting smoking and take medication, are more likely if the new information leads to updated belief. Furthermore, we find heterogeneous effects across subgroups of individuals: Males are more likely to quit smoking and taking medication after belief updating; rural people are more likely to quit smoking but less likely to take medication, possibly due to lack of affordability or accessibility to medical services. We find no significant impacts on drinking.