The global recession of 2008-09 showed that correlations are dynamic and can change dramatically, particularly if affected by exogenous shocks. As a result, investors are increasingly seeking ways to diversify their portfolios as a means to maintain a relatively low level of correlation among the many asset classes to which they have exposure.
As an asset class, infrastructure has proved to be an attractive diversifier when paired with a number of traditional asset classes. In this piece we look at inter-market correlations, i.e., the return correlations of infrastructure to other asset class returns. This complements our May 2016 piece, Infrastructure debt pairwise asset correlations, where we analyzed intra-market correlations, i.e., the pairwise asset correlations within the infrastructure market.
Credit Suisse Asset Management (CSAM) did an infrastructure study in 2010. Their analysis showed that infrastructure equity return correlations stayed relatively low against most analyzed asset classes for the period July 2000 to March 2010, with the exception of non-US equities (see Exhibit 1). They examined correlations against two broad measures of the asset class: Firstly, the Macquarie Global Infrastructure Total Return Index, which covers publicly listed funds in 48 countries; and secondly, the correlations of a customized infrastructure mix of equally-weighted greenfield airport, port and energy returns, so as to illustrate what the results would likely be for an infrastructure strategy with a tilt toward greenfield economic infrastructure assets.(1)
Exhibit 1. Listed infrastructure equity returns have relatively low correlation with other asset classes
(July 2000 – March 2010)
CSAM used the Macquarie Global Infrastructure Total Return Index because of the benchmark’s long-term track record since July 31, 2000, and its broad coverage across 48 markets. The Macquarie Global Infrastructure Index (MGII) Series calculated by FTSE shows the stock price performance of companies worldwide within the infrastructure sector, principally those engaged in management, ownership and operation of infrastructure and utility assets.
Unlisted equity and infrastructure debt correlations significantly lower than listed equity
The average correlations for Global Infrastructure and Customized Infrastructure vs the other asset classes in Exhibit 1 are 0.52 and 0.49, respectively. Note that these correlations are for public investments, that is, listed shares of companies that operate in the infrastructure business as described above. The correlation of private infrastructure investments is much lower, however, because of the less efficient nature of the private market. A study in the Australian market found unlisted infrastructure fund returns to have correlations of only 0.06 to 0.26 vs other major asset classes such as equities, bonds and property.(2)
The correlations in Exhibit 1 are for infrastructure equity investments but the principle also holds true for infrastructure debt. See Equity betas for listed infrastructure funds, where we show that the betas between listed infrastructure debt funds and the broader equity market are very low at 0.079 on average and that almost all the of total risk of infrastructure is idiosyncratic risk and not market risk.
Portfolio effect – higher Sharpe Ratio and beneficial shift to efficient frontier
The practical benefits of adding infrastructure to a portfolio are an improvement in the risk-return profile (i.e., a higher Sharpe Ratio) and a beneficial shift in the efficient frontier, allowing for lower risk with little effect on return.(3)
We performed an analysis to investigate how including an allocation of infrastructure loans with an average duration of 7 years affects the efficient frontier on a Baa 5- and 10-year Euro-corporate bond portfolio.(4) Based on historical data from 2005 through to 2011, which is through the financial crisis, infrastructure loans demonstrated very little return correlation to corporate bonds and their inclusion in the portfolio resulted in approximately a 30% reduction in risk, with almost no reduction in returns in the tangency portfolio (see Exhibit 2). This analysis considers senior infrastructure loans. Based on our experience, by including mezzanine infrastructure the returns would be higher without a material increase in portfolio risk (i.e., the standard deviation of returns). Mezzanine loans are mostly in the private market, are relatively illiquid and still benefit from stable project cash flows.
Exhibit 2. Efficient frontier with the inclusion of infrastructure loans in a corporate bond portfolio
The output of the distributions from the Sequoiaanalysis is shown in Exhibits 3 and 4:
Exhibit 3. Mean monthly returns, variances and standard deviations
| Infra | 5-yr corps | 10-yr corps | |
|---|---|---|---|
| Mean | 0.2140% | 0.2081% | 0.1922% |
| Variance | 0.0098% | 0.0178% | 0.0515% |
| Standard dev. | 0.9912% | 1.3328% | 2.2698% |
| Note: mean annual return | 2.57% | 2.50% | 2.31% |
As can been seen, over the 6-year period, the mean performance of the three asset classes has been remarkably close but infrastructure debt has the lowest return volatility by a material measure. The table below shows the covariance and correlation matrices for the three asset classes:
Exhibit 4. Covariance and correlation matrices
| Covariance Martix | Correlation Martix | |||||
|---|---|---|---|---|---|---|
| Infra | 5-yr corps | 10-yr corps | Infra | 5-yr corps | 10-yr corps | |
| Infra | 0.010% | 0.001% | 0.002% | 10% | 8% | 9% |
| 5-yr corps | 0.001% | 0.018% | 0.028% | 8% | 100% | 93% |
| 10-yr corps | 0.002% | 0.028% | 0.052% | 9% | 93% | 100% |
It is notable how low the return correlation is between infrastructure loans and corporate bonds, at less than 10%, as well as how low the corresponding covariances are. These correlations are in line with the study of private unlisted infrastructure funds in the Australian market discussed above.
Large benefit from diversification and lower tail risk
Infrastructure has low return correlations not only to other asset classes but there is also very low pairwise asset correlation within the infrastructure market from one project to another; significantly lower than within the broader credit markets, for example. This pairwise asset correlation drops when comparing infrastructure projects in different regions, countries and sectors.(5)
Because infrastructure has such low inter- and intra-market correlation, this asset class benefits from diversification more than other asset classes do. The most optimal way for most investors to benefit from this characteristic is to invest through a fund because it is easier to get exposure to assets diversified by region, country, sector and sub-sector. Low correlation also implies lower tail risk (i.e., a lower probability of a large unexpected loss), which has taken on heightened importance after the financial crisis of 2007-08.
Sequoia analyzed the effect on portfolio risk of incrementally adding infrastructure loans to a portfolio.(6) In the study, the portfolio loss distribution was driven by the number of loans in the portfolio and the correlation between the loans. We assumed equally sized loans and tested three correlation levels: normal (similar to general credit), low (similar to infrastructure) and independence.
Exhibit 5. Portfolio standard deviation of returns vs. number of infrastructure assets
Infrastructure debt – more scope to outperform than credit
Due to low infrastructure correlation, portfolio risk drops rapidly and tail risk is reduced as assets are added as shown in Exhibit 5. Our analysis shows that 17-18 infrastructure assets diversify away 70-80% of the idiosyncratic portfolio risk. By comparison, 70+ normal correlation assets (e.g., leveraged loans) would be needed to achieve the same amount of risk reduction. This is indeed the typical number of loans a CLO needs to achieve the diversity score required by the rating agencies. This means the manager would more or less have to ‘buy the market’ and therefore would have much less scope to find positive alpha assets.
By not needing as many assets to achieve the same level of diversity, an infrastructure manager can pick their exposures more carefully, looking for optimal risk-return opportunities.
(1) Infrastructure investments are often classified depending on their stage of development and are traditionally seen as “greenfield” or “brownfield” assets. Sometimes referred to as “growth infrastructure,” greenfield investments refer to new facilities that require design, financing, building and operating. Brownfield investments refer to existing assets and are sometimes referred to as “mature infrastructure.”
(2) Unlisted infrastructure funds based on infrastructure investments in Australia between 3Q1995 and 2Q2006. See Wen Peng, Hsu, and Newell, Graeme, University of Western Sydney, “The Significance of Infrastructure in Investment Portfolios,” Pacific Rim Real Estate Society Conference, Jan 2007.
(3) Sequoia, “Sharpe Ratios for infrastructure debt, infrastructure equity and corporate credit,” May 2016.
(4) Sequoia, “Effect of Including Infrastructure Debt on the Efficient Frontier of a Fixed Income Portfolio,” Jan 2012.
(5) “Moody’s Approach to Rating CDOs Backed by Project Finance and Infrastructure Assets,” Oct 2013.
(6) Sequoia, “Portfolio Risk – Diversification Effects,” July 2011.
