House price indexes and cyclical behavior

Purpose – The purpose of this paper is to use data on new apartment offerings in the municipality of Sao Paulo, Brazil to illustrate its main claim that the hedonic direct method using time dummies as well as the simple average method include cyclical behavior of observables and non‐observables in a house price index that may overestimate or underestimate the actual change in house prices, well beyond the composition effects. Design/methodology/approach – The paper proposes the use of alternative characteristics hedonic functions to compute alternative Laspeyres house price indexes that differentiate the sources of observable shocks in the index. The decomposition allows for the inclusion of level and cyclical behavior of sets of aggregate variables into the index. Findings – The appropriate house price index should filter out real shocks that potentially affect the real estate sector. An index should capture nominal variation and incorporating real variation biases the measurement. Thus, the index is intended and able to buffer the bias spillover into the rest of economy. In the limited sample from the city of Sao Paulo, Brazil, the main finding is that real shocks and US foreign shocks give an upward bias in the house price index, while nominal shocks give mostly a downward bias. Real shocks make the index incorporate gains that should not be incorporated into the index, thus providing a noisy picture of the nominal variation in house prices. Originality/value – The key contribution of this paper is to provide a framework for the construction of a house price index that filters out real shocks that potentially affect the real estate sector. An index should capture nominal variation and incorporating real variation biases the measurement. Those biases can spillover to the rest of the economy is a detrimental way. Thus, the index is intended and able to buffer the bias spillover into the rest of economy.