Seminars

Abstract: This paper studies how experimental and survey measures of trust change over time for individuals and how these changes are correlated with changes in behavioral measures of trustworthiness, altruism, reciprocity, cooperation, and risk tolerance. In the Trustlab Japan project, panel data sets for different target populations that follow the same individuals with online experiments and surveys were collected by the Trustlab Platform in Japanese. This platform is internationally comparable and includes four incentivized tasks for the trust game, the public goods game, the dictator game, and a lottery choice. The data sets in the project are the only data which have both experimental and survey measures of trust for the same individuals for two or more points in time. This paper analyzes three panel data sets with at least 250 samples from this project. The survey measures of trust are less volatile over time than the experimental measures of trust and other behavioral measures. The change in the experimental measure of trust is positively correlated with each of the changes in social behavioral measures of trustworthiness, altruism, reciprocity, and cooperation with only a few exceptions of statistically insignificant correlations. The change in each of our two survey measures of trust is less positively correlated (both in terms of the smaller magnitude of correlation and of less frequent significance) with the changes in these social behavioral measures. The change in the experimental and survey measure of trust is sometimes positively correlated with the change in risk tolerance. There is evidence that part of these changes was caused by COVID-19 for people living in larger sized metropolitan areas. These results are consistent with a class of models in which social capital affects preferences and preference shocks also affect social capital.