Price and spatial distribution of office rental in Madrid: a decision tree analysis

Abstract

In this paper, we assess the drivers of office rental prices in the municipality of Madrid with a sample of 4,721 offices in March, 2020. The estimation was performed using the decision tree approach, which was built with a random forest algorithm. This technique allows us to capture the strong nonlinear component in the relation between price and its drivers, mainly geospatial location. Through a stratified analysis, we find that the willingness to pay high rent in the center of Madrid is a feature of particular relevance to medium-sized offices. For different reasons, we also find some office clusters located far from the city center with high rent for both large and small offices.