Working Paper: NBER ID: w28138
Authors: Victor Stango; Jonathan Zinman
Abstract: We examine how 17 behavioral biases relate to each other, to other decision inputs, and to decision outputs. Most consumers exhibit multiple biases in our nationally representative panel data. There is substantial heterogeneity across consumers, even within similar demographic/skill groups. Biases are positively correlated within person, especially after adjusting for measurement error, and less correlated with other inputs—risk aversion, patience, cognitive skills, and personality traits—with some expected exceptions. Accounting for this correlation structure, we reduce our 29 decision inputs to eight common factors. Seven common factors load on at least two biases, six on clusters of theoretically related biases, and two or three are distinctly behavioral. All but one common factor is distinct from cognitive skills. Factor scores strongly conditionally correlate with decisions and outcomes in various domains. We discuss several potential implications of this taxonomy for various approaches to modeling influences of behavioral biases on decision making.
Keywords: behavioral biases; consumer decision making; decision inputs; decision outputs
JEL Codes: C36; C81; D90; E70
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| biases (D91) | consumer decision making (D12) |
| biases are positively correlated (C10) | biases are interconnected (D91) |
| cognitive skills (G53) | biases (D91) |
| classical inputs (C67) | variance in biases (C46) |
| factor analysis (C38) | dimensions of decision inputs (D79) |