Access all areas

Portfolio construction is an essential component of any investment strategy, helping investors diversify asset- or sector-specific risks to boost risk-adjusted returns.

Nobel Prize-winning economist Harry Markowitz famously described diversification as “the only free lunch” in investing. However, within private markets, specialisation can give investors an edge, through informational and deal-sourcing advantages that can contribute to outperformance. A true understanding of portfolio risk and sector-level correlations helps investors balance the trade-off between specialisation and diversification to maximise alpha generation without taking undue concentration risk.

The challenge of portfolio construction in private markets

Quantifying portfolio risk and the diversification benefit within private markets is challenging. Unlike public markets, historical performance data is often unavailable, incomplete or based on smoothed valuations that underestimate the volatility of transaction-level pricing. Moreover, differing methodologies can lead to time lags between changes in transaction pricing and valuations. This can create an illusion of diversification, which may not exist if assets need to be sold.

Further complicating matters, infrastructure and real estate markets are constantly evolving. New digital and carbon-transition sectors now represent a larger proportion of infrastructure investment. Within real estate, there has been a drastic shift in the composition of investor portfolios over the last two decades, with alternative sectors seeing rising allocations, whilst retail has been declining and industrial assets growing (Figures 1 and 2). Some of these sectors are highly heterogenous, with different levels and types of risk.

As a result, most investors have historically sought to reduce portfolio risk through a “naive diversification” approach; allocating to multiple sectors, geographies and asset classes in a logical manner, as opposed to using more sophisticated quantitative methods.

The weakness of simplistic approaches is that they rely on implicit diversification among sectors, potentially underestimating portfolio risk or, conversely, investors may under-allocate to attractive opportunities. Moreover, allocations are usually considered on a capital rather than risk basis. Diversification is generally maximised by allocating equal risk weightings, which can be very different to capital weightings in a portfolio consisting of sectors with different levels of total risk (Figure 3).

Generalisations are often made at the asset class level based on historical metrics and relationships. In our view, a more granular, forward-looking approach is required, considering sub-sector allocations and risks; for instance, the risk profile of a subsidised solar farm is different from an airport, although both are infrastructure assets.

Additionally, we believe investors should be most concerned about portfolio risk and diversification over multi-year horizons, matching the hold period of their investments, as opposed to diversification of annual measured performance. We have built a portfolio construction tool based on these premises (see Relative value).

Diversification in real assets

The search for higher returns and diversification from public equity and credit markets has led many institutional investors to increase their allocations to real assets. Globally, private market fundraising increased by 139 per cent from the five years preceding the global financial crisis to the five years to 2021.¹ With growing allocations, diversification has become increasingly important. In our 2023 Real Assets Study, 57 per cent of the investors surveyed cited diversification as their primary driver for investing in real assets.²

Real assets offer meaningful potential for diversification. Real estate is cyclical, but local supply/demand factors, sector, asset quality, lease length and tenant characteristics all lead to differentials in asset-level risk profiles (as we discussed in How to benefit from specialisation and diversification in real estate).³

Infrastructure is even more heterogenous, with risks often sector-specific, including regulatory risk, toll-road traffic, fibre broadband take up or power prices. These risks can have a degree of economic linkage, but the essential nature of infrastructure makes others largely idiosyncratic.

Even within sectors, risk can differ depending on the revenue mechanism and contractual structure, which can alter exposure to price and volume risk. For example, a wind farm with revenue entirely derived from government subsidies has a different risk profile to one that sells power at the wholesale power price. Figure 4 shows our assessment of the risk breakdown for certain sectors.

Despite the variety of cashflow risk drivers within real estate and infrastructure, the grouping of these sectors can contribute to heightened correlations of market pricing within an asset class. The same pool of investors tends to look across sectors within an asset class; when fund dry powder and liquidity are high, this tends to simultaneously increase the pricing of sectors that may otherwise be unrelated.

It is by combining multiple real asset classes that the most powerful diversification can be achieved. Utilising our bespoke relative-value methodology, we can quantify forward-looking portfolio risk. Figure 5 shows our assessment of the extent to which diversification can reduce risk within a real estate and infrastructure portfolio: a 45 per cent reduction can be achieved relative to the weighted average risk of each sector individually.

Although quantitative tools can be helpful, it is important to keep in mind that correlations oscillate over time; left-tail risks may not have occurred in historical datasets or could be inconceivable until the moment they happen.

Quantitative portfolio construction tools generally only account for “known unknowns”, whilst many sectors have their own left-tail risks (such as regulatory risk), which can result in a permanent loss of capital. Therefore, portfolio construction must include an element of human judgement. In our view, it is always prudent to increase the level of diversification suggested by quantitative tools to protect against “unknown unknowns”.

Real assets are traded privately, with every deal bespoke and dependent on managers with specialist capabilities and networks. The opaque nature of these markets can create inefficiencies, but this also creates opportunities to exploit mispricings and gives an advantage to first movers who can deploy funds effectively and efficiently.

New sectors are particularly ripe for exploiting a first-mover advantage. Funds normally have a deployment period of one to three years. If you need to set up a new fund or mandate when a new opportunity arises, the time spent doing so can lead to missed opportunities. However, a multi-asset manager can swiftly enter a nascent market without needing to set up a new segregated mandate.

In terms of our own investments, a multi-asset approach means we can switch between asset types when we see value and go underweight on asset classes that either don’t provide value or are not available. For example, the flow of mid-market loans all but stopped during 2020.

A multi-asset manager can more easily switch to assets still trading and offering good relative value when others aren’t. This can be seen in Figure 6. In 2020, for example, structured finance trades offered good relative value and were in rich supply. This was not the case in 2021, so more traded in the previous year. In 2022, private corporate debt trades provided better relative value over other asset classes, giving multi-asset managers the opportunity to rebalance portfolios with these, making up for 2020 when few were available.

For a manager using a single asset class approach, these drastic changes in deal flow from year-to-year can make the pace of deployment highly uncertain.

While these assets are less liquid, growing demand is making them more readily tradable. This, combined with uncertainty over the inflation outlook and cost of traditional liquid market hedging instruments, offers a strong rationale for adding real assets to portfolios.

As mentioned, the real assets universe is varied with sectors and asset classes following their own unique cycles. Having a broad opportunity set can help investors take advantage of mispricings between and within asset classes to allocate capital to the most attractive areas.

Figure 7 shows the illiquidity premia versus equivalent public debt we have observed through cycles for infrastructure debt, real estate debt and private corporate debt. From 2012 to 2015, infrastructure debt offered the highest illiquidity premium, potentially a reflection that the capital cost to banks to hold long-dated liabilities was high following the financial crisis, with spreads widening as a result.

Conversely, in recent years, liquidity and competition for infrastructure debt assets have been significant, resulting in a lower illiquidity premium compared to real estate and private corporate debt.

Figure 7: Illiquidity premia by asset class vary through the cycle (basis points)

Source: ICE Bank of America Merrill Lynch, Bloomberg, Aviva Investors. Data as of December 30, 2022

As well as inherent differences to sub sectors, different valuation methodologies, fund flows to asset class-specific funds and regulation can contribute to differing market cycle timings (for example, there is normally a delayed impact of rates on real estate long income). But this is challenging to assess. A multi-asset opportunity set allows investors to rapidly switch between sectors or asset classes to find the best relative value.

Arbitrage opportunities can arise from different structures or different parts of the capital structure, which can have similar risks but offer different returns – for example, income strips versus real estate debt, as depicted in Figure 8.

In 2020, income strips offered a higher return, despite a wider potential distribution of returns, at the end of the holding period compared to real estate debt, but that has reversed in the last two years, implying real estate debt offers better relative value.

Figure 8: Expected returns over government bonds, income strips versus real estate debt (per cent)

Source: Aviva Investors, September 2022

Equity, subordinated debt and senior debt can also have different market cycles, even when financing similar assets.

As seen in real estate in Figure 9, depending on the extent of the change in property values, the attractiveness of positions in the capital structure will change. 

• There is scope for material diversification within real assets due to the varied return drivers within asset classes, enabling stronger risk-adjusted returns. A multi-asset portfolio allows for more effective portfolio construction, taking into consideration the facets of each individual investment, not just sectors overall.
• Areas of the real assets market can be subject to large fluctuations in deal flow. Being able to access a broader range of opportunities should increase the pace of capital deployment.
• The real assets universe is varied with sectors and asset classes following their own unique cycles. Having a broad opportunity set can allow investors to take advantage of mispricings among asset classes to allocate to the most attractive areas.