|Researchers:||Andrej Gill, Florian Hett|
|Category:||Household Finance, Experiment Center|
This project represents a follow-up extension of an existing SAFE research project which is currently being finalized. Within this existing project, we study the prevalence of time-inconsistency in households’ financial decision making. By collaborating with a fin-tech start-up, we are able to access detailed financial account data from approximately 50,000 clients. Using this transaction-level bank account data, we elicit whether people show signs of present bias. The crucial characteristic of this data set is that it allows us to explicitly identify those individuals whose consumption, financing, and savings behavior can be classified as present biased. We then assess our findings by conducting an online experiment with the same people and checking whether an experimental measure of present bias actually predicts real behavior that is consistent with present-biased preferences.
The hereby proposed extension of the project aims at substantially improving this last step - namely the measurement of present bias using behavioral experiments by focusing on allocating real effort decisions rather than monetary payments over time. This serves two purposes: First, it represents a potentially substantial enhancement to our initial study, as a robust and precise measure of present bias is a fundamental requirement to assess its role in financial decision-making. Second, how to measure time preferences in general and present bias in particular is currently a very active debate in the literature. Hence, the possibility to compare different recently developed methods in its ability to predict relevant field behavior in itself represents an important contribution to this debate.
Concretely, we plan to run an additional series of online experiments in which we elicit time preferences and present bias of the clients in our sample. However, in contrast to the initial study, we will now use the measure by Augenblick et al. (2015), which relies on choices over allocating real effort tasks over time instead of monetary payments. Enhancing our existing data set with this new measure then allows us to systematically compare (i) the relation between different measures of time preferences and present bias among each other, (ii) the relative power of different measures in predicting present biased financial decision-making in transaction data, and (iii) the relative power of different measures in predicting bad financial health as visible in the extensive use of overdraft fees.