Navigating small datasets: The efficacy of the factorial index over the repeat sales index

Abstract Repeat sales index methodologies are prevalent for computing indices, largely owing to their simplicity, minimal data requirements, and reliable approach to addressing real estate heterogeneity. However, this method is not without flaws, being restricted to properties sold at least twice within the sample period and demonstrating high sensitivity to the size of the dataset. In this paper, we introduce a stable and robust factorial approach that holds firm even when the data quantity is sparse due to factors such as frequent publications, thin markets, data unavailability, or a drastic drop in the number of transactions. We demonstrate that trend and volatility estimates tend to be biased with the standard repeat sales approach, but remain unbiased with our factorial method, regardless of the size of the dataset. The factorial approach exhibits greater robustness in relation to transaction numbers.