Abstract
This paper contends that returns generated by a transaction-based property price index provide biased estimates of the true underlying real estate returns. Two biases - temporal and distribution - are examined. Transaction data from 34 condominium developments in Singapore are used to test for the existence of these biases. In addition, the temporal and distribution biases provide testable hypotheses that suggest that the bias differ across market conditions. The empirical evidence supports the hypotheses. Finally, policy implications and measures to rectify the biases are highlighted.