Abstract:
Quasi-hyperbolic discounting is one of the most well-known and widely-used models to capture self-control problems in the economics literature. The underlying assumption of this model is that agents have a "present bias" toward current consumption such that all future rewards are downweighed relative to rewards in the present (in addition to standard exponential discounting for the length of delay). We report a meta-analytic dataset of estimates of the present bias parameter β based on searches of all major research databases (62 papers with 81 estimates in total). We find that the literature shows that people are on average present biased for both monetary rewards (β = 0.82, 95% confidence interval of [0.74, 0.90]) and nonmonetary rewards (β = 0.66, 95% confidence interval of [0.51, 0.85]) but that substantial heterogeneity exists across studies. The source of this heterogeneity comes from the subject pool, elicitation methodology, geographical location, payment method, mode of data collection (e.g. laboratory or field), and reward type. There is evidence of selective reporting and publication bias in the direction of overestimating the strength of present-bias (making β estimates smaller), but present bias still exists after correcting for these issues (for money β = 0.87 with 95% confidence interval of [0.82, 0.92] after correcting for selective reporting).
