Using stochastic frontier models to mitigate omitted variable bias in hedonic pricing models: A case study for air quality in Bogotá, Colombia

- Failure to account for unobserved housing quality (UQ) in hedonic regressions will bias parameter estimates.
- We argue that variation in UQ will tend to be asymmetric and correlated with environmental quality.
- A frontier model with asymmetric random error can mitigate omitted variable bias from unmeasured housing quality.
- We found a smaller estimated elasticity for air pollution when using a frontier model than when using a conventional OLS.
- The previous finding is consistent with the hypothesis that unmeasured variation in UQ biases conventional results.

Hedonic pricing models use property value differentials to value changes in environmental quality. If unmeasured quality attributes of residential properties are correlated with an environmental quality measure of interest, conventional methods for estimating implicit prices will be biased. Because many unmeasured quality measures tend to be asymmetrically distributed across properties, it may be possible to mitigate this bias by estimating a heteroskedastic frontier regression model. This approach is demonstrated for a hedonic price function that values air quality in Bogotá, Colombia.