Asymptotic properties of imputed hedonic price indices

Hedonic price indices are currently considered to be the state-of-the-art approach to computing constant-quality price indices. In particular, hedonic price indices based on imputed prices have become popular both among practitioners and researchers to analyze price changes at an aggregate level. Although widely employed, little research has been conducted to investigate their asymptotic properties and the influence of the econometric model on the parameters estimated by these price indices. The present paper therefore tries to fill the actual knowledge gap by analyzing the asymptotic properties of the most commonly used imputed hedonic price indices in the case of linear and linearizable models. The obtained results are used to gauge the impact of bias adjusted predictions on hedonic imputed indices in the case of log-linear hedonic functions with normal distributed errors.