The Role of Housing in a Mixed-Asset Portfolio: The Particular Case of Direct Housing Within the Greater Paris Area

Abstract

This article explores the role of physical residential real estate within the optimal multi-asset portfolio. Specifically, we consider direct housing within the Grand Paris metropolis between 1996 and 2017 as an asset class together with financial assets. Our findings bring several contributions to the residential market literature. First, directly held housing investment brings diversification benefits to the mixed-asset portfolio. Second, using hierarchical clustering technique, we divide the Greater Paris area into five homogenous groups of communes and compute the optimal weight of each commune as well as each group of communes in the tangency portfolio. Third, we check the weights’ stability through time and confirm that residential real estate always catches the highest weight in the optimal portfolio. Finally, we run additional tests to compare listed real estate performances in a mixed asset portfolio with those obtained by considering physical residential real estate. We conclude that listed real estate is not a substitute for direct housing.

Key Words:

• Direct housing
• “Grand Paris” metropolis
• inflation hedge
• diversification
• optimal portfolio

Notes

1 Direct housing investment is made through the purchase of a tangible residential property.

2 « Institut National de la Statistique et des Etudes Economiques » which is the French national institute for statistical and economic studies. 

3 The process of metropolization involves a higher concentration of the population within an expanding geographical area (largest cities and urban districts), as well as a growth in its functional weight, activities and wealth creation.

4 The largest transportation project in Europe.

5 This is partly due to the high share of owners in these countries, there are in addition cultural facts towards housing investments particularly in the United Kingdom according to Joaquim Montezuma in his study « Housing investment in an institutional portfolio context », 2003.

6 The metropolis of Greater Paris is a major center of higher education, which results in a high representation of the students and also the young workers. These two particular categories of population are emblematic of the current situation in the metropolis that we are seeking to demonstrate: high proportion of renters compared to owners, with a low home purchasing power, and a low vacancy rate due to the important demand.

7 In our case, the housing total return for each of the 127 communes is computed as follow:

$\begin{array}{cc}{\mathit{Housing} \mathit{total} \mathit{return}}_t& =R\mathit{ental} \mathit{return}+\mathit{capital} \mathit{return}\\ \iff {\mathit{HTR}}_t& =\frac{R_t}{P_t}+\frac{P_t-P_{t-1}}{P_t}\end{array}$

Where $R_t$ is the average level of rent per square meter in a commune during year t, and $P_t$ and $P_{t-1}$ are the average transaction prices per square meter in a commune during the years t and t-1.

8 This is done for example in the study of Hoesli and Hamelink (1997). For operating expenses, they subtracted 150 basis points from the rental return in two swiss housing markets.

9 The series is initially provided by Macrobond, and the computation of total return series is provided by BNP Paribas Real Estate.

10 We weighted each commune’s total return by a size factor that indicates its weight within its group. The size factor selected is the stock in the commune, such that a total return of a group in year t is computed as follow:

${\mathit{Housing}\mathrm{ }\mathit{group}\mathrm{ }\mathit{total}\mathrm{ }\mathit{return}}_t=\sum _i^n{\mathit{HTR}}_{it}\times \frac{{\mathit{Stock}}_{it}}{{\mathit{Stock}}_t\mathrm{ }}$

Where ${\mathit{HTR}}_{it}$ is the housing total return in a commune i during the year t, ${\mathit{Stock}}_{it}$ is the total number of dwellings (collective housing) in a commune i during the year t, and ${\mathit{Stock}}_t$ is the total number of dwellings in the group of n communes.

11 « Schéma Directeur de la Région d’Île-de-France »

12 Rolling correlation using moving window also corroborates the diversification ability of housing in a mixed-asset portfolio. Graphical representations of rolling correlations is available upon request.

13 Limitations are linked to some simplistic assumptions such as liquidity is infinite, all investors are rational and risk averse, investors control risk only by the diversification of their holdings, etc.

14 The risk-free rate considered in the study is the average OAT 10-year French government bond over the 2000-2017 period. In this case, the accrued coupon amount is not included in the computation. It is commonly used in the real estate industry as the benchmark to which we compare real estate returns (Crédit Foncier, BNP Real Estate, etc.).

15 Only stocks, bonds and housing are represented because cash did not appear in the tangency portfolio. This could be explained by the characteristics of the proxy used for the cash; 3-month Euribor. As seen before, it provides the weakest return and it is the most positively correlated asset with housing one. In contrast, due to its lowest volatility among other assets, it captures the highest share of the minimum variance portfolio, while no housing asset appears in this portfolio.

16 Besides, for easiness of visual representation, we chose to classify the existing communes into 5 homogenous groups.

17 Detailed results per commune appear in Figure 7.

18 Detailed results are available upon request.

19 Available in the supplemental file.