Although far from straightforward, global diversification can result in exposure to higher GDP growth combined with lower volatility, argue Thomas Au, Simon Redman, Kim Politzer and Mark Roberts

The benefits of a global approach to diversification are threefold. First, from the evidence provided by the real estate investment trust (REIT) market, the correlations of global real estate markets are significantly lower than either the stock or the bond market. The average of the correlations of markets, which comprise the global real estate securities index, is 0.30 for the 17-year period from 1990 to 2007.

The correlation of the global stock and bond markets over the same period is 0.56 and 0.52, respectively. We believe the key reason for the difference in correlations among asset classes is that local property market fundamentals play a greater role in driving performance and may be less affected by global capital flows.

Second, the US share of global GDP is stabilising, while the share of Europe and Asia GDP is increasing. This shift reflects the expansion of Europe into Central Europe and the impact of a single currency across most of Europe. In Asia, the growing economy of China provides the greatest fundamental shift.

Historically, Asia has grown on par with the US with similar volatility, while growth in Europe has trailed but with significantly lower volatility. When these factors are combined, a globally diversified portfolio constructed in an optimal manner can create exposure to slightly higher GDP growth with lower volatility.

Finally, non-listed real estate performance databases are expanding globally and fundamental supply and demand drivers for real estate are becoming more transparent. While differences do exist across the globe, the preliminary analysis we outline below highlights the diversification benefits of global real estate investing.

To consider global diversification, it is a useful exercise to review the benefits that are available from doing this on a regional level. This will necessarily impact the investment views and actions of domestic or regional investors. The key question is, are there compelling reasons for investing globally versus regionally?

As a starting point, we have reviewed the major global office markets. This is partly driven by the availability of reasonable performance data over a meaningful time period in order to develop a basis for comparison. It is important to highlight the difference in performance data we are using for these markets. First, for the US, we are utilising total returns from the NCREIF Index.

In both Europe and Asia, we are utilising prime data. The data from the US reflects the performance on assets owned by institutional investors. According to NCREIF, these assets are reappraised quarterly. In both Europe and Asia, we are utilising data which reflects the practices used in these regions. In both regions, local agents established a net effective rental value for "prime" space.

These rents are capitalised at an appropriate income yield to determine capital value. Total returns are produced based upon the changes in capital values and the initial income yields. For all markets, we utilised returns produced in local currency terms. In other words, the analysis assumes investors have not hedged currency risk.

This is an appropriate assumption for some investors as one of the features provided by global diversification is to capture changes in local currency terms in an effort to reduce home country currency risk. The markets we have selected are shown in table 1.
These markets have been adopted because they all provide a substantial focus for domestic investment activity and represent an appropriate geographically diverse mix of investment locations.

For each region we have plotted the historical return information below:
It is clear from this analysis that office markets in the US are highly correlated even though they are geographically disparate. In contrast, markets in Asia and in Europe show lower correlation and comparatively higher diversification benefits. This is further supported by Table 2, which shows the average correlation of markets within their respective region.

The average correlations clearly suggest that regional investors within Europe and Asia are able to achieve diversification benefits by adopting a regional strategy. Even in the US, where the correlation is relatively high compared to markets in Europe and Asia, investors can achieve regional diversification benefits since individual markets have unique embedded characteristics, which drive capital market and tenant demand for real estate.

On the face of this there may be no benefit to adopting a global approach. However, this assertion ignores the benefits of the relative correlation of each market within a region to those markets outside the home country region. To show this, we grouped the correlations of markets across two regions and then computed the average of the correlations between all the markets in one region with those in another region. The results are shown in table 3.

As is evident in the table, there appear to be significant diversification benefits for investors who look beyond their regional borders. For example, for European investors, while there appear to be significant diversification benefits within Europe as shown above in Table 2, further diversification benefits may be achieved by investing in Asia and the US.

In particular, since the average correlation between the European and Asian markets depicted is only 0.19, both Asia and European investors benefit through cross-regional investment strategies. Similarly, for US investors, the relatively low correlations with Europe (0.50) and Asia (0.36) indicate the markets depicted suggest greater diversification benefits can be achieved with a global investment strategy.

Given the events of the last several months, few would argue against the globalisation of markets. Yet, while we have witnessed greater synchronisation of the main global economies and their capital markets, real estate markets benefit from significant supply and demand side differences such as planning regulations, geographic differences and occupier preferences, which lead to the correlation differences highlighted above.

These differences emanate from the embedded characteristics unique to each market, which partially offset the impact of increasing capital market convergence. To show this, we plotted the peak and trough returns during the last 10 years (2Q 1998 to 2Q 2008) along with the average of total returns which occurred over the period. In addition, we ordered the markets from left to right with those on the left producing the narrowest range of returns, while those on the right produced the widest range of returns.

As seen in figure 4, the range of total returns varies considerably across the globe. Generally speaking, a number of US markets, such as Washington DC or Chicago, have produced a narrow range of returns while a few Asia-Pacific markets, such as Tokyo or Singapore, have produced a broader range of returns. Clearly, some of the performance differences can be explained by the return methodology utilised across regions.

Another difference may be explained by the initial yields. In the case of the US, initial yields have been higher while those in the Asia-Pacific region have been lower. When initial yields are lower, small changes in rents cap rates can have a greater impact on total returns as compared with those markets which have higher initial yields. Therefore, the amplitude of returns can vary based upon initial yields. This is both a risk and a benefit.

On the face of it, one might question why they should invest in markets which show such a wide range of returns. Certainly, any return that deviates from its long-term average return can be considered "less predictable". As a result, some investors may decide to avoid investing in such markets. However, this overlooks whether or not deviation from the long-run average produced positive or negative results. Statisticians use the term "skewness" to describe whether or not the changes from the long-run average were positive "outperformance" or negative "surprise", which resulted in underperformance.

Going back to figure 4 and considering Tokyo as an example, when returns deviated from their average, the market generally produced more positive than negative "surprises". The peak returns approached 80%, while the downside in returns was lower at minus 20%. As a result of more positive "surprises" versus negative "surprises", Tokyo offices produced the second highest average returns of the markets in our sample.

Such embedded characteristics are important to consider when constructing a global portfolio. To explain, "risk", or the standard deviation of returns, is a key input into an optimisation model. In an optimisation model, the assumption is made that the negative surprises are equal to the positive surprises. Therefore, any market which has a greater risk value can be interpreted as "undesirable".

However, as we noted in our example of Tokyo, sometimes a higher standard deviation can produce an increased chance of outperformance. In addition, despite the fact that some markets may have lower relative return, it may be worth including those markets in a portfolio because their correlations are low relative to other markets. To show some of these factors, we plotted the markets along the return and risk spectrum as highlighted in figure 5.

As seen in the table, the risk and return attributes vary considerably across the globe. Over the last 10 years, the US markets tend to lean towards the lower-risk, higher-return spectrum while the Asian markets tend to show a higher degree of variability, with the notable exceptions being Seoul, Beijing and Shanghai.

The European markets are mixed. For example, Madrid has experienced significant growth, which has resulted in a higher degree of variability. In comparison, Amsterdam has produced higher risk-adjusted returns due to the stability of the local property markets. Finally, some caution is warranted when looking at past returns as an indicator of future performance.

In the US, low risk premiums and ample debt capital drove returns higher over the last few years and total returns are expected to revert to more normal levels. In the case of London City offices, it has been more adversely impacted by the recent credit crisis earlier than what may be eventually reflected in other markets. 

Still, taking a longer-term perspective, because the correlations are low and the risk and return attributes are different, combining different markets together can produce a more optimal portfolio. To investigate this we used an "optimiser" to construct a diversified portfolio of markets across the globe. In doing so, it was possible to analyse the composition of the portfolio across a range of expected return and risk. To simplify our analysis, we also constrained the weight to any one market to 10% to avoid undue concentration. Figure 6 plots the composition of the portfolio across a range of risk levels.

The lower-risk, lower-return portfolio lies to the left of the chart while the higher-return, higher-risk portfolio lies to the right of the chart. The low-risk portfolio produced a total return of 15% and a standard deviation, or risk level, of 6.5% over the last 10 years. This level of risk is lower than could be achieved by investing in any single market and highlights a benefit of global diversification. This portfolio would have exposure to London, Amsterdam, Shanghai, Beijing, Seoul, Boston, Chicago, Los Angeles, New York and Washington DC.

The portfolio at the opposite end of the spectrum returned 17.5% with a risk level of 11%. Had we not constrained the portfolio to 10% per market, we could have reduced the risk level. What is interesting to note though is how the composition of the portfolio changes across the risk spectrum. For this portfolio, Madrid, Paris, Tokyo and San Francisco replaced London, Amsterdam, Beijing and Chicago. The markets which remained constant across the risk spectrum were Shanghai, Seoul, Boston, Los Angeles, New York and Washington DC.

The case for global investment is clear. Because the risk and return characteristics of global real estate markets are sufficiently diverse and the correlations are low, investing globally offers the potential to add value. However, the practical execution of a global investment strategy is not straightforward. Investing in a global portfolio of REITs provides many of the advantages outlined in this article.

Unlike direct real estate, though, it suffers from short-term volatility associated with general equities markets. Therefore, private real estate can provide the purest form of real estate diversification as outlined for those investors who are of sufficient size to access real estate directly.

Global direct investing involves a few issues that domestic investment does not require. These include tax structuring in order to actually receive the expected returns, obtaining sufficient knowledge to have a coherent view of market opportunities and managing currency and cultural differences across multiple time zones.

If globalisation of markets and investor appetite is to continue, it is incumbent on investment managers to have the right skills and international capabilities to be able to make the investment choice between one region and another and to be able to distinguish between good and poor investments within a region.

Moreover, for significantly large investments, it may also require managers to have the ability to structure investments so that, in some cases, investors can pool capital in order to minimise specific asset risks and, in all cases, so that domestic taxes are minimised and both income and sales proceeds can be maximised.

While these issues may provide some impediments, the benefits of global diversification as outlined can serve to offset these issues and enhance investor access to a broader array of investment opportunities.Mark Roberts, head of US investment research; Kim Politzer, european research manager; Thomas Au, head of Asian investment research and Simon Redman, head of business development at Invesco Real Estate