ABSTRACT
We examine the link between uncertainty and the excess returns of direct commercial real estate markets using the well-established economic policy uncertainty index (EPU) and a novel measure of uncertainty, the world uncertainty index (WUI). Using a panel dataset of 17 European markets, we find a strong positive relation between the two uncertainty measures and excess returns in our univariate models and when we add real estate control variables. However, once macroeconomic control variables are included, only the WUI remains significant. We relate this finding to the more country-specific nature of the WUI, while the EPU potentially gives more weight to global news.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. The results further show that the variation in mean and variance when analysing the link between the EPU and housing returns needs to be accounted for.
2. The EIU is a sister company to The Economist newspaper. Established in 1946, it is a leading firm in the area of country intelligence that publishes country reports for 189 countries on a quarterly basis.
3. Also see, Ahir et al. (Citation2018), p. 8.
4. Ahir et al. (Citation2018) report a correlation between the two indices of 0.705 for a matching country dataset.
5. We include the following countries and cities: Austria (Vienna), Belgium (Brussels), the Czech Republic (Prague), Denmark (Copenhagen), Finland (Helsinki), France (Lyon, Paris), Germany (Berlin, Cologne, Duesseldorf, Frankfurt, Hamburg, Munich, Stuttgart), Greece (Athens), Hungary (Budapest), Ireland (Dublin), Italy (Milan, Rome), the Netherlands (Amsterdam), Norway (Oslo), Poland (Warsaw), Spain (Barcelona, Madrid), Sweden (Gothenburg, Malmö, Stockholm), and Switzerland (Geneva, Zurich).
6. co-star data are available at https://www.co-star.com.
7. According to the co-star, the average office segment only includes prime property data if sufficient data for the average segment is not available. This is only rarely the case.
8. See, also Lim et al. (Citation2013).
9. The excess return is of further interest because the choice between prime and average properties allows for diversification (Lim et al., Citation2013).
10. Both the WUI and EPU were retrieved from the website https://www.policyuncertainty.com.
11. OECD data are available at https:/stats.oecd.org, and FRED data are available at https://fred.stlouisfed.org.
12. Bloomberg data are available at https://www.bloomberg.com.
13. Since the number of observations is not the same for all the variables, our panel is unbalanced.
14. As analysing the relationship between real estate markets and uncertainty requires full time variation, we do not apply time fixed effects in our models (see, e.g. Barros et al., Citation2015). But nevertheless, we also tested a random effects model. The Hausman test (redundant fixed effects test) suggests that the fixed effects model is superior in our model specification.
15. The choice of the lag length k = 4 is also supported by the Akaike and Bayesian information criteria of a lag selection test within a VAR framework that we ran beforehand. We additionally checked for autocorrelation due to the long time span of our sample period. Thus, we have added an AR(1) component. However, since the AR(1) was not significant, we report all our models without the AR(1) correction.
16. The full models including all control variables and their lags are provided in the Appendix .
17. Su et al. (Citation2016) show that the EPU has no direct impact on real estate returns, and they document reverse causality from residential returns to the EPU for the German housing market. Chaney and Hoesli (Citation2015) and Das et al. (Citation2020) document similar linkages between housing markets and the broader economy. The latter study, however, finds that the same does not hold true for commercial properties. We thus used the Hausman test, which did not indicate endogeneity issues for our estimations. The results of the Hausman test are provided in Appendix .
18. Since the city fixed effects setup could potentially restrict some temporal dynamic in the model, we additionally tested a VECM to allow for short-term deviations in the time series variables. However, we only find significant values for the second and third lag for the natural logarithm of equities. The results are available for authors upon request.
19. Baker et al. (Citation2016) show a strong linkage between implied stock price volatility and their EPU measure.
Additional information
Notes on contributors
Elmar Lang
Elmar Lang is a PhD-student at EBS University. He has many years of experience in the private banking industry. Currently, he is working at PwC in Frankfurt as a Senior Associate in Real Estate M&A.
Ferdinand Mager
Ferdinand Mager is a Professor for Finance, Banking and Accounting at EBS University. He has vast experience in corporate finance, risk management and banking.
Kerstin Hennig
Kerstin Hennig is a Professor at EBS University. She has over 20 years of experience in the real estate industry and is specialized on real estate development and real estate marketing.