Working Paper: NBER ID: w27209
Authors: Peter Bergman; Eric W. Chan; Adam Kapor
Abstract: We randomized school quality information onto the listings of a nationwide housing website for low-income families. We use this variation and data on families' search and location choices to estimate a model of housing search and neighborhood choice that incorporates imperfect information and potentially biased beliefs. We find that imperfect information and biased beliefs cause families to live in neighborhoods with lower-performing, more segregated schools. Families underestimate school quality conditional on neighborhood characteristics. If we had ignored this information problem, we would have estimated that families value school quality relative to their commute downtown by half that of the truth.
Keywords: housing search; school quality; low-income families; randomized controlled trial; neighborhood choice
JEL Codes: I0; I21; I24; I3; R0; R21; R31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Providing information about zoned schools and their performance (I21) | Households search for and move to locations with higher-performing zoned schools (R23) |
| Households search for and move to locations with higher-performing zoned schools (R23) | Families live in areas assigned to schools that have a 0.10 standard deviation higher rating (I24) |
| Providing information about zoned schools and their performance (I21) | Families live in areas assigned to schools that have a 0.10 standard deviation higher rating (I24) |
| Families would trade an additional 54 minutes of commute (R41) | A 10 percentile point increase in school quality (I21) |
| Ignoring information frictions (G14) | Understatement of valuation of school quality by more than 50% (I21) |
| Observable neighborhood characteristics (R23) | Predict school quality (I21) |
| Beliefs about school quality (I21) | Systematic bias in predicting school quality (I24) |