Investment risk modeling and forecasting of the property cycle

Abstract

In the last ten years the available information and used technology has massively increased in nancial markets. [1] Besides, stocks this trend has also infected the real estate market. In light of the nancial meltdown of 2007, cyclical research has become an important topic again, even though bubbles are somehow seen as normality of the system. 3 Hence, national banks have taken the monitoring of nancial markets much more seriously and expanded their repertoire of statistical variables. 4 Using the assumptions of cyclicality is often done passively by bankers and property developers in their investment decision, even though historical numbers hint these relationships. The reason for this is that more experienced market participant already have seen the patterns and recognize them compared to novices. However, besides having expectations on the future market behavior no further research is often done. However, as other studies show [2] investing or lending within the right timing can be an important determinant of success. This thesis, is split into two large parts. On the one hand, we will show chronologically the cyclical research, which has been done in the past and pointing out the methodology of each paper. Further, we will also investigate current trends in the property market and show the rise of technology. On the other hand, we will present a comprehensive methodology, which uses several time-series models to forecast the property cycle. Our research will focus on Austria. All inputs, will be selected by an extensive variable selection process, which will be outlined. As a second part we will view these forecasts as cyclical sensitivities and simulate each model result, on a synthetic real estate portfolio to show the eects of the cycle. The forecasts will be evaluated by their accuracy to assess the usefulness of our prediction models. Several accuracy measures have been picked to support a comparable view among all time-series models. Our results, suggest that forecasting the property cycle helps to identify valuable lending and investment opportunities for bankers and property developers. Our nal results suggest that the most accurate model results for our time series.