When Formulas Fail: On the Variability of the Exponential Growth Bias

This paper analyzes whether individuals who have been well trained to apply exponential discounting and compound interest calculations in simple savings scenarios are also successful in dealing with exponential effects in more complicated debt scenarios. In an experimental study, involving 251 German undergraduate business students, we find strong evidence of an “amortization bias”, a tendency to linearly estimate the remaining balance on a loan at various points of a debt payoff schedule. Interestingly, the same group of subjects shows almost no exponential growth bias for simple savings questions. We develop a measure for the exponential growth bias that naturally extends over different tasks and parameter settings and provide first within-subject evidence for the variability of the bias in different areas of application. Our findings indicate that teaching individuals a formula for calculating compound interest does not help to develop a general or intuitive grasp of exponential effects in more complicated settings. This discouraging insight adds complexity to the discussion about effective ways to support individuals in overcoming the bias. JEL Classification: D14