In Part one of this article, we analysed and understood the historical trend and performance of real estate assets relative to the traditional asset classes. But the big question still requires an answer: How much real estate should investors hold in order to maximise the return of their portfolios?
Of course, there is no single answer. At any point in time, the optimum amount of real estate investment changes according to the industry trend and its performance relative to other investment opportunities, i.e. opportunity cost. So, our aim can only be to understand how the optimum fluctuates and what causes these changes.
In order to gain this insight, let’s consider the impact of adding direct real estate assets to an initial portfolio that invests 60% of its capital in shares, 35% in bonds and the remaining 5% in UK Treasuries. We will then gradually add more real estate to the portfolio, reducing the amount of other assets. Figure 1 below shows the results for the period 1950-1980. Since the line is noticeably straight, it can be deduced that the addition of real estate assets reduces risk more than return, with an aggregate positive effect over the Sharpe ratio. This becomes clear when looking at Table 1: the red value represents the maximum Sharpe value, and is obtained with a 90% real estate portfolio. While it is extreme, this result is consistent with the previous findings regarding returns and risk: during the 1960s, the real estate sector outperformed both shares and bonds. It was also characterised by lower risk, so it comes as no surprise that the most efficient portfolio invests 90% into real estate. Additionally, the same combination of assets will yield the minimum-variance portfolio.
Figure 1. Portfolio Optimisation before 1980
Table 1 confirms our previous findings: during the period 1960-1990, the real estate market outperformed all the other asset classes in terms of risk-adjusted return, and therefore it should have constituted a significant proportion of every investor’s portfolio, regardless of the risk appetite. However, things look much different after 1980 up to 2000, as we can see in Figure 2. Adopting the same technique as before, let’s consider a hypothetical portfolio that invests 60% in equity, 35% in government bonds and 5% in T-bills. Now, the frontier of risk-return opportunities is C-shaped, which implies that above 40% real estate, the resulting portfolio will be dominated, i.e. there will be another portfolio yielding higher return for the same risk; also, the maximum Sharpe ratio is obtained without investing in real estate.
Figure 2: Portfolio Optimisation from 1980 to 2000
Despite carrying the same risk, the average performance of the real estate market dropped below bonds during the period in question; also, the risk-adjusted return is below equity. Given the low level of the Sharpe ratio, investors’ preferences move towards more rewarding assets, and, starting from 1980, an increasing proportion of portfolios were constituted by company shares, so that investors could exploit high growth rates in the stock market.
Next to determine the impact from 2000 to the Global Financial Crisis (GFC). The stock crash in 2001 reversed the trend of the previous decades, increasing the amount of real estate required in order to optimise portfolios: as shown in Figure 3 below, the maximum Sharpe ratio of 0.45 is yielded by a 100% real estate portfolio.
Also, it is important to notice that the pattern followed by real estate assets is completely different to before: adding real estate to the initial portfolio increases risk and return, instead of reducing them. In other words, the chart is “flipped” and the 100% real estate portfolio is located in the top right corner, as the maximum risk-return bet. This can also be seen in Table 3, which highlights that the real estate market yields much larger returns than equities, with less risk.
Figure 3. Portfolio Optimisation from 2000 to 2007
Also, Table 3 provides an explanation for these results: after the dot-com crisis, shares yielded, on average, lower returns than risk free assets, thereby granting negative Sharpe ratio to their owner. Conversely, the risk embedded in holding real estate reduced slightly with respect to the previous decade, while returns remained stable, leading to a significant increase in the acquisition of these type of assets.
Unfortunately, the high-scale speculative strategies used by investors had a negative impact over the debt market, and LTV skyrocketed in a few years leading to the infamous GFC. Figure 4 and Table 4 show how the real estate market reacted to the drastic consequences of the worst financial crisis since the 1929 Great Depression.
Figure 4. Portfolio Optimisation from 2009 to 2016
Table 3 depicts that real estate returns dropped while risk rose consistently, with a consequent negative impact over the Sharpe ratio. The major consequence of this can be seen in the following graph: adding real estate to the initial portfolio only increases standard deviation, without any benefit for returns. The best risk-adjusted return is therefore obtained without investing in the real estate market, as in the period 1980-2000. All the previous findings are in line with the following graph, which shows how the Sharpe ratio behaved through the years. Particular attention should be paid to the GFC effect, following which real estate Sharpe plummeted and has not yet recovered.
Everything considered, the benefit that investors might gain by adding real estate to their portfolio is much more limited now than in the past. However, there are some significant factors to consider regarding the nature of the dataset. It is important to remember that the figures refer to direct aggregated real estate investments, i.e. there is no distinction whatever between sub-markets. This is important since each of them has its own specific characteristics. Anyway, this should not refrain us from advancing a hypothesis regarding the behaviour of the largest real estate sub-market: on the one hand, the housing market is supposedly correlated with government bonds because of the strict connection to mortgage rates; on the other, it is said to have zero or slightly negative correlation with the stock market due to the non-cyclical behaviour of the demand for houses. The main consequences of these hypotheses being true are:
(1-5) Scott The Real estate Masters to 1970, MSCI / IPD, Barclays Capital Equity-Gilt Study
(1-4) Scott The Real estate Masters to 1970, MSCI / IPD, Barclays Capital Equity-Gilt Study
© Treex 2020