Control function quantile hedonic pricing
本文提出控制函数分位数回归框架，解决特征定价中营销时间内生性和属性估值异质性问题，发现高端房产营销时间每100天溢价2.5%，且传统工具变量法低估该效应。
Abstract This paper addresses two persistent challenges in hedonic price analysis: endogeneity of marketing time and heterogeneity in attribute valuations, using a novel control function quantile regression framework. By exploiting characteristic-induced heteroskedasticity for identification, we address simultaneity bias without exclusion restrictions and recover heterogeneous effects across the price distribution. We find a U-shaped relationship between marketing duration and sale prices in Multiple Listing Service data. High-end properties at the 95th price percentile show 2.5% price premiums per 100 days of marketing time, three times greater than the effect at the median price level. Control function adjustments show that conventional instrumental variable approaches underestimate premiums, with bias greater in upper quantiles. The estimator accommodates both conditional and unconditional quantile specifications and remains robust to parametric variance assumptions using series estimation. Our results highlight the role of distributional analysis in housing markets, where seller patience and buyer bargaining power interact asymmetrically across price segments.