Quantifying the Reversibility Phenomenon for the Repeat-Sales Index

Abstract

The reversibility phenomenon for the repeat-sales index (RSI) is a serious obstacle for the derivatives products; it could hinder their introduction or their success. It is also an undesirable characteristic for the management of the real estate risk. This article provides a general solution for this problem, using an informational reformulation of the RSI framework. We present first a theoretical formula, easy to interpret and easy to handle, before implementing it. Our methodology is robust in the sense that its conclusions are not conditioned by any specific dataset; moreover, the numerical estimations of the reversibility percentages are reliable. For the derivatives our technique has strong implications for the choice of the underlying index. Indeed, even if the reversibility of the RSI is probably higher compared to the hedonic one, this index remains a challenger because of the predictability and the quantifiability of its revisions.

Keywords: Reversibility, Quantification, Information, Monte Carlo simulations, Markovian process