Working Paper: NBER ID: w19269
Authors: B. Douglas Bernheim; Daniel Bjorkegren; Jeffrey Naecker; Antonio Rangel
Abstract: A central task in microeconomics is to predict choices in as-yet-unobserved situations (e.g., after some policy intervention). Standard approaches can prove problematic when sufficiently similar changes have not been observed or do not have observable exogenous causes. We explore an alternative approach that generates predictions based on relationships across decision problems between actual choice frequencies and non-choice subjective evaluations of the available options. In a laboratory experiment, we find that this method yields accurate estimates of price sensitivities for a collection of products under conditions that render standard methods either inapplicable or highly inaccurate.
Keywords: Nonchoice Evaluations; Behavioral Responses; Economic Conditions
JEL Codes: C91; D12; H31; Q51
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| nonchoice evaluations (C52) | actual choice behaviors (D01) |
| nonchoice evaluations (C52) | purchase frequencies (D19) |
| price changes (P22) | purchase frequencies (D19) |
| nonchoice evaluations (C52) | price sensitivities (D40) |
| nonchoice evaluations (C52) | predictive accuracy (C52) |