Forecasting commercial property market performance: beyond the primary reliance on econometric models

Perera, T 2018, Forecasting commercial property market performance: beyond the primary reliance on econometric models, Doctor of Philosophy (PhD), Property, Construction and Project Management, RMIT University.


Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Perera.pdf Thesis application/pdf 10.50MB
Title Forecasting commercial property market performance: beyond the primary reliance on econometric models
Author(s) Perera, T
Year 2018
Abstract All commercial property stakeholders have a strong interest in forecasting as property market forecasts have become an integral part of the investment processes supporting asset allocation and property fund investment strategy. It is critical that a wide range of techniques is employed to forecast the performance of real estate assets (Ball, Lizieri & MacGregor 1998; Brooks & Tsolacos 2010) in terms of predicting reasonable estimates of key dependent property variables (demand, supply, rent, yield, vacancy, and net absorption rate). However, accurate forecasts can be conducted in a situation when ample quantitative data are available with a few uncertainties. For this type of problem, statistical methods in the standard risk analysis can be employed, with identified events and assigned probabilities (Aven 2015b; Bralver & Borge 2010). But surprises occur, and econometric results are unsettled by unknown risk factors. For this reason, well formulated theories of forecasting are still being conjectured that can be proven incorrect by one random event. Although risks can occur in either tail, downside risk associated with the lower tail is mainly focussed on risk management than the symmetric case of upside potential. To advance real estate decision making practice in this area, there is a need to improve forecasts through conceptualising downside risk integration in commercial property market forecasting. To do this, the study employed quantitative–qualitative sequential strategies (mixed method research design) aimed at developing a decision making model for improving the accuracy of the Australian commercial property market forecast.

In the quantitative phase, prime and secondary office market data and economic data for 2001-2011 were analysed statistically to test the forecast accuracy using various scale dependent/independent statistics, descriptive analysis and time series regression analysis. Secondary data was obtained from the premier Australian office markets. The quantitative evaluation was limited to office market data due to the sector’s good performance data coverage, compared to retail and industrial sectors. The 10-year timeframe is due to the discontinuation of the Australian Financial Review quarterly survey of economists, the source of reliable economic forecast data. The accuracy measurement revealed that rental movement and net absorption forecast errors were critical, illustrating limited predictive capacity in forecast models. Further, a scale-independent analysis that involves benchmarking, identified that naïve forecasting strategies outperformed the forecasters. In this study, naïve assumption is the last observed value six months ago. As per the outlier analysis, the timeframe around the Global Financial Crisis (GFC) has witnessed a significant inaccuracy, as with key manmade Black Swan events. For instance, January 2009 was identified as a worst-case scenario with a high over-estimation of prime office rental movements, where the 6-month forecast was 9.93% with an actual movement of -0.07%. Therefore, forecasting uncertainty has become a highly critical issue.

The quantitative study also revealed that property forecast errors deviate significantly from economic forecast errors without any consistency in the underlying relationships from mainstream economics. Hence, forces beyond economic factors need to be integrated into property forecast modelling. Apart from the effects of the GFC, the quantitative analysis lacks the explanation of causal factors for property forecast errors. The results needed further exploration. As such, a qualitative investigation (semi-structured interviews) was undertaken with 22 leading property and financial market experts to gain insights into the current Australian commercial property market (office, industrial and retail) forecasting practices. Thematic analysis was done with the view to identifying strategies to aid forecast accuracy for commercial property market performance.

The qualitative results show that 45% of experts interviewed are regularly over optimistic about their forecasts. Whilst forecasting with confidence is deemed important, forecast accuracy checks are underrated in the market due in part to unrealistic self-assurance about the forecast models. The analysis was triangulated with findings from literature and quantitative analysis, explaining that leading underlying macroeconomic determinants can be impacted by short-lived Black Swan events and long-term structural changes (key causes of forecast errors). The difficulty in understanding the timing and the magnitude of Black Swan events resulted in a willingness to accept the unexpected downside when it occurs, rather than spending on unrealistic and impractical modelling of one-off effects. With the due consideration of root causes and effects of the GFC, it was identified by interview respondents that the current property market is reaching the peak of the property cycle, reflecting the pre-crisis situation. Such similarities include low vacancy levels, strong face rents, strong construction pipeline and yield compression. Positively, the leverage is relatively low in the present market, but it has the potential to revert to a highly geared condition in a structurally low interest rate environment. Furthermore, the qualitative analysis involved identifying and mapping the potential impact of structural changes in Australian commercial property market performance. Key commercial property specific findings include rise in cross border capital flows, emerging outer suburb cities, property conversions, integrated property facilities and the ‘Give and Take’ effect (for example, retail market (give) to industrial market (take) by offsetting the effects of online retail).

The analysis shows that an accurate forecast requires both powerful algorithms and following best practices to improve the forecasting process. Therefore, this research proposes an innovative ‘Ten best practices’ empirical framework based on industry expert recommendations to improve commercial property market forecasting: (i) clear objective setting, (ii) collaborative approach to knowledge sharing, (iii) market data analysis to capture the changes in the determinants, (iv) atheoretical quantitative (data-driven) approach, (v) adhering to the parsimony principle, (vi) qualitative overlaying, (vii) eliminating biases to predict independent estimates, (viii) forecast validation, (ix) integrating an error positive culture, and (x) reviewing, redefining and remodelling. However, forecasting is challenged with the increasing level of downside risk. Thus, in addition to the ‘Ten best practices’, a protective layer of strategies to reduce forecast errors associated with downside risk is also proposed to retain forecast credibility. These strategies include historical referencing for errors and risks, stress testing, identifying warning flags, and comprehensive analysis of investment decision makers’ responses to risk. Finally, the overall findings were presented in an Applications > Determinants > Strategies > Validation (ADSV) decision making model, illustrating the recommended approaches and the deliverables for key steps in forecasting to improve forecast accuracy.

This research makes significant contributions to knowledge for both academia and industry practitioners by offering a methodological approach to how commercial property market forecasting can be improved. As deductive reasoning dominated in the real estate literature, this research predominantly employed a qualitative approach to capture the behavioural information. The models developed from the research will enhance academic theory. A better understanding of tail risk integration results in higher reliability of forecasts about the performance of commercial property market across different regions and sectors that in turn is beneficial for property investment managers in their investment decision making. Property developers are particularly interested in the market performance, when setting up feasibility studies. When making credit approvals, banks and other real estate financiers are benefitted from reliable market moves based on improved forecasts. Furthermore, identifying and mapping unknown risks are important in asset allocation strategies that are relevant for property fund managers and investors.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Property, Construction and Project Management
Subjects Built Environment and Design not elsewhere classified
Economic Models and Forecasting
Keyword(s) Australian commercial property market
Downside risk
Forecasting
Forecast errors
Forecast accuracy measurement
Structural changes
Black Swan events
Versions
Version Filter Type
Access Statistics: 160 Abstract Views, 107 File Downloads  -  Detailed Statistics
Created: Fri, 30 Nov 2018, 07:32:42 EST by Keely Chapman
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us