N° 19-32: Robust Desmoothed Real Estate Returns

This research starts from the observation that common desmoothing models are likely to generate some extreme returns. Such returns will distort risk measurement and hence can lead to investment decisions that are suboptimal relative to those that would be made if a transaction based index were available. Thus, we propose to improve the desmoothing models by incorporating a robust filter into the procedure. We report that in addition to properly treating for smoothing, the method prevents the occurrence of extreme values. As shown with U.S. data, our method leads to desmoothed series whose characteristics are akin to those of transaction-based indices.