|
Abstract:
|
This study examines New Jersey's rehabilitation code and uses a difference-in-difference estimator to measure the effect of the code on housing prices relative to two control states. The primary results indicate that the rehabilitation code significantly raises home value. However, these results are not reliable measures of the effect of the rehabilitation code. Two forms of model misspecification have likely led to biased estimates of the rehabilitation code coefficient. I was unable to find the proper functional form relating home value to the independent variables, and I could not control for certain factors that likely influence housing prices. When my model included two extra controls, the association between the rehabilitation code and home value became negative, which suggests that the original estimates are upwardly biased. The variation in the treatment coefficient also suggests that determinants of housing price were varying differently in New Jersey than in the control states during the study period. This makes it difficult to reliably measure an unbiased treatment effect using a difference-in-difference strategy and raises doubts about both the primary and secondary estimates. |