The role of sentiment indicators for real estate market forecasting

The purpose of this paper is to assess the behaviour of economic sentiment indicators at rent‐growth turning points and indicators' ability to forecast such turning points. More specifically, the paper looks at whether early signals are generated for forthcoming periods of negative and positive office rent growth. The analysis aims to complement structural model forecasting in the real estate market with short‐term forecasting techniques designed to predict turning points.

The objective of this study is achieved by deploying a probit model to examine the ability of economic sentiment indicator series to signal the direction of office rents and the strength of movement in this direction. The main advantage of this approach is that it is geared towards predicting turning points. Probit models are non‐linear in nature, and as such they can capture more effectively the likely asymmetric adjustments when turning points occur than linear methodologies would. The analysis is applied to three major office centres – La Défense, London City, and Frankfurt – to examine whether the results will differ by geography.

The findings reveal that the probit methodology utilising information from economic sentiment indicators generates advance signals for periods of contraction and expansion in office rents across all three markets: La Défense, London City, and Frankfurt. The lead times for La Défense and Frankfurt are longer than those for London City and range between three and nine months. The evidence in this paper clearly supports the appeal of sentiment indicators and probit analysis to inform forecasting and risk assessment processes.

Acknowledging the limitations of structural models and related methodologies and the lack of adequate research on turning‐point prediction in the real estate market, this study forecasts episodes of negative and positive office rent growth applying appropriate techniques and data that lead economic activity, are of monthly frequency, and are not revised historically. The paper raises awareness of a forecasting approach that should complement structural models and judgmental forecasting, given its suitability for short‐term forecasting and for signalling turning points in advance.