Volatility Decomposition of Australian Housing Prices

1. Volatility Decomposition of Australian Housing Prices Chyi Lin Lee and Richard Reed The 17th European Real Estate Society Conference

2. Outlines • Introduction • Objectives • Data and Methodology • Results and Findings • Conclusions

3. Introduction • Australia- high homeownership -70% (IBISWorld, 2007). • Housing- 57% of the total value of Australian household assets (ABS, 2007). • The determinants of housing prices (first moment)- attention. • BUT the volatility patterns in housing prices- limited.

4. Introduction • Several studies: • Volatility clustering in housing prices (Dolde and Tirtiroglu, 1997; Crawford and Fratantoni, 2003; Wong et al., 2006). • The determinants of housing price volatility: • Miller and Peng (2006) -the home appreciation rate and GMP growth rate. • Hossain and Latif (2007)- GDP growth rate, house price appreciation rate and inflation

5. Introduction • Previous studies -the conditional volatility of a housing market. • The conditional volatility could be further decomposed into: (Pagan and Schwert, 1990; Nelson, 1991). • “permanent” component – persistent • “transitory” trend –strong impact • The transitory volatility is caused by noise trading (e.g. speculation activities and trading by irrational investors) • The permanent (fundamental) volatility is caused by the arrival of new information (Hwang and Satchell, 2000)

6. Introduction • Real estate literature: • A common (fundamental) component of volatility shared by direct properties and securitised real estate (Bond and Hwang, 2003). • The strong evidence of long-term memory volatility is also observed in most international real estate markets (Liow, 2009). • Significant differences between the “permanent” and “transitory” volatility movements (Liow and Ibrahim, 2010) .

7. Introduction • Previous studies - securitsed real estate • Exception - Fraser et al. (2010) - real house prices have a long-run relationship (permanent) with real income in the UK, the US and New Zealand.

8. Objective • To provide an insight into the pattern of housing price volatility by decomposing the volatility of housing price into permanent and transitory components.

9. Data and Methodology • Quarterly data of 8 capital cities for the period Q4:1987-Q3:2009, for a total of 88 observations were obtained from the ABS. • These capital cities are Sydney, Melbourne, Brisbane, Perth, Adelaide, Hobart, Canberra and Darwin, as well as the Australian housing market on aggregate. • Returns are calculated by the first difference of the natural logarithm of the quarterly indices.

10. Data and Methodology • Engle and Lee (1993,1999) and Liow and Ibrahim (2010)- Component-GARCH model.

11. Data and Methodology • C-GARCH Mean Equation: Variance Equations:

12. Results and Discussion Table 5: ARCH Tests Volatility Clustering

13. Results and Discussion Table 6: C-GARCH(1,1) Model

14. Results and Discussion Table 7: Specification Tests for the C-GARCH Model Correct specifications

15. Results and Discussion Table 8: “Permanent” Volatility Spillover

16. Results and Discussion Table 9: “Transitory” Volatility Spillover

17. Conclusion • Volatility Clustering • The volatility of housing price can be decomposed into permanent and transitory components  Differences between both components. • Both volatilities capture different sets of information and have different determinants.