Illiquid Assets, Risk Budgets, and Factor-Based Thinking

Most investment portfolios are built from liquid, publicly traded assets: equities, bonds, and cash. But large investors increasingly allocate to private equity, direct real estate, and infrastructure. These asset classes offer potential return premiums but create real challenges for standard portfolio construction. Separately, factor-based allocation offers an alternative way of thinking about what drives returns. And risk budgeting provides the tool to check whether risk is being used as efficiently as possible. This article covers all three.

The Promise and Problem of Illiquid Assets

Illiquidity premium: The extra return an investor expects to earn as compensation for locking money into an investment that cannot be easily sold

Private equity, direct real estate, and infrastructure can offer higher expected returns than comparable public market investments. The premium compensates investors for accepting three things: the inability to sell quickly, the difficulty of precisely valuing the investment between transactions, and the higher concentration of company-specific risk compared to a diversified public index.

The problems start when you try to include these assets in an MVO. For liquid assets like global equities, high-quality index data stretches back decades, covers millions of securities, and can be tracked cheaply through index funds. For private equity, the situation is the opposite on every dimension.

The two-part problem with illiquid assets: First, the return and risk data for illiquid asset classes is unreliable. Private equity funds only value their holdings periodically and often use smoothed valuations. This makes the measured volatility of private equity look much lower than it truly is. Second, even if you had perfect data on the asset class, you cannot invest passively in private equity. Every actual investment is a specific fund with its own manager, strategy, and idiosyncratic risks. The fund's return profile is very different from the theoretical asset class average.

These problems lead to three practical approaches. The first approach is to leave illiquid assets out of the formal asset allocation entirely and treat any exposure as an implementation decision. Once the liquid portfolio is built, the investor adds private equity or real estate funds alongside it, judging them on their own merits rather than forcing them into an optimiser with unreliable inputs.

The second approach is to include illiquid assets in the optimisation but calibrate the inputs to reflect the actual fund characteristics, not the theoretical asset class. If the investor will realistically hold 5 or 10 private equity funds rather than the entire asset class, the risk and return estimates should reflect that concentration.

The third approach is to include illiquid assets using listed proxies. Global REITs can stand in for direct real estate. Listed infrastructure stocks can stand in for infrastructure. The advantage is that reliable data exists. The disadvantage is that listed proxies are more correlated with equities than the private assets they represent, which overstates portfolio correlations and increases input sensitivity.

Cash and the Risk-Free Asset: A Practical Note

Cash is sometimes treated as the risk-free asset and sometimes as just another risky asset in the optimisation. Neither approach is perfectly right. The return on cash is known in the short term but uncertain over longer periods. If an investor's time horizon is 10 years, the true risk-free asset for their purposes is not 90-day Treasury bills but a 10-year government bond.

In practice, many wealth managers include cash in the optimisation and let the optimiser decide the allocation. Others exclude cash and separately determine how much to hold based on liquidity needs, the common guidance being around six months of expenses. Both approaches are defensible.

Risk Budgeting: Making Every Unit of Risk Count

Marginal contribution to total risk (MCTR): How much total portfolio volatility increases when the weight in a specific asset rises by a small amount. Calculated as the asset's beta relative to the portfolio, multiplied by the portfolio's total volatility

Absolute contribution to total risk (ACTR): The actual amount of volatility an asset contributes to the portfolio. Calculated by multiplying the asset's weight by its MCTR

Risk budgeting answers a question that portfolio weights alone cannot: where is the risk actually coming from? Two portfolios can have identical asset weights but very different risk profiles if the volatilities and correlations of their assets differ.

The risk budget treats total portfolio volatility as a resource to be allocated wisely. Some assets have high MCTR (they are risky relative to the portfolio). Others have low MCTR (they are stabilising). Risk budgeting makes this visible so the portfolio manager can judge whether the risk allocation makes sense.

An asset allocation is optimal from a risk budgeting perspective when the ratio of excess return (above the risk-free rate) to MCTR is identical for every asset in the portfolio. If one asset has a higher ratio, more weight should be shifted toward it. If one has a lower ratio, weight should be reduced. When all ratios are equal, no trade can improve the risk-return trade-off.

Factor-Based Asset Allocation: A Different Way of Thinking

Investment factor: A systematic source of return that explains why certain groups of assets earn more than others over time. Common factors include market risk, company size, valuation, momentum, credit, and duration

Standard MVO allocates to asset classes: equities, bonds, real estate, and so on. Factor-based allocation allocates to the underlying return drivers instead. Rather than owning UK small-cap equities as a category, a factor investor explicitly holds exposure to the size factor (small companies tend to outperform large companies over long periods) and the market factor.

Factors are typically constructed as zero-cost, self-financing strategies: go long the better-performing attribute and short the worse-performing one. The size factor, for example, is the return from being long small-cap stocks and short large-cap stocks. Because short positions offset long positions, the factor has little market exposure and tends to have low correlations with other factors.

The key insight from research is that when asset class and factor opportunity sets are constructed to cover the same exposures, neither approach is inherently superior. The efficient frontiers they produce are very similar. The choice between asset-class and factor-based allocation should be driven by which space the investor can most reliably form capital market assumptions.