Credit Constraints and the Valuation of Public Amenities
John Heilbron, Finance PhD student
Using house prices to estimate the value of non-traded amenities is a technique dating back to Rosen (1974). Later work improved estimates by identifying the role of “polluting variables”, like wages, that might bias estimates of amenity values (Roback, 1982), exploiting clean cross-sectional or time-series variation (Black, 1999; Chay and Greenstone, 2005) and correcting for selection across boundaries (Bayer et al., 2007). Hedonic price regressions and the hedonic approach continue to be used (Kulka, 2019; Diamond and Mcquade, 2019). Though Rosen considered a static household tradeoff between non-housing consumption and housing amenities captured in rental rates, empirical applications commonly use house prices instead. Ouazad and Rancière (2019) write a dynamic stochastic matching model that considers ways in which financial frictions might distort the relationship between the price schedule and willingness to pay for amenities. This paper relies on a considerably simpler intuition and shows how it is possible to exploit a central feature of the mortgage market, the g-fee and PMI requirements of the GSE market, to empirically assess the degree of bias in hedonic estimators. I propose to demonstrate this by reproducing results from Currie et al. (2015), a recent application of hedonics, and correcting them for bias due to credit constraints.