Annual sector indices are susceptible to price differences provoked by major environmental changes. Therefore, it is really important to take into consideration different situations and apply an appropriate method for constructing WLS estimator. The objective of our research is to compare several well-known methods (in the framework of the repeat sales model) used for constructing house price indices based on unbiased estimators. We intend to show that a weighted least squares estimator (WLS) for the log price index can be improved due to the introduction of time dependency. We tend even to downweight housing units that have been held for a long time more than theory suggests. We suggest that there exist a pragmatic reason for doing this ñ long hold prices are differentially risky because of unknown structure changes.