A Risk-Control, Risk-Asset Framework for Asset Allocation

October 11, 2024

By Peter Mladina, Executive Director of Portfolio Research, Wealth Management
Charles Grant, CFA, Director of Asset Allocation, Wealth Management

We demonstrated in Compensated Portfolio Risk that term and market risk factors are the systematic risks that almost entirely explain the compensated return and risk of multi-asset class portfolios. Conventional asset classes have overlapping exposures to these two primary risk factors. For portfolio construction and risk management purposes, we categorize conventional asset classes into “risk-control” or “risk-asset” subportfolios based on their sensitivities to term and market risk factors. The term factor represents maturity-related interest rate risk, and it is commonly defined as the return of Treasury bonds minus the return of Treasury bills. The market factor represents equity and equity-like default risks, and it is commonly defined as the return of stocks minus the return of Treasury bills.

Standard deviations (return volatility) and correlations are the risk characteristics that matter for asset allocation. As independent risk premiums, term and market factors have different volatilities and they are largely uncorrelated with each other. Exhibits 1 and 2 show historical risk characteristics for Treasury bonds and stocks.¹ Exhibit 1 shows that over the full history Treasury bonds have had a standard deviation of just 4%, which compares to 19% for stocks. When viewing short time periods (rolling 3-year windows), the standard deviation of Treasury bonds is always materially lower than the standard deviation of stocks. The standard deviation of Treasury bonds peaked at 12% in a 3-year period that occurred during the interest rate spike of the early 1980s. In contrast, the standard deviation of stocks peaked at 57% during a 3-year period that occurred during the Great Depression.

Correlations measure the degree to which assets comove from their respected average returns, capturing the diversification benefit. True risk factors should be lowly correlated. Exhibit 2 shows that over the full history the correlation between Treasury bonds and stocks is close to zero (0.06), confirming that they are different sources of risk (and return). The correlation between Treasury bonds and stocks varies over the short run (rolling 3-year windows), from 0.68 to -0.62 in Exhibit 2. But it is never highly correlated and it always reverts back toward zero in the long run. These risk characteristics show that the term factor is reliably less risky than the market factor, and it diversifies the market factor.

Next we show that term and market factors explain multi-asset class portfolio risk.² Northern Trust’s standard Moderate portfolio is comprised of 41% risk-control assets and 59% risk assets. Its risk-control subportfolio includes cash, investment-grade bonds, and inflation-protected bonds. Its risk-asset subportfolio includes high-yield bonds, U.S. equity, developed international equity, emerging-market equity, natural resources, real estate and infrastructure. Since the common inception (December 2002) of indices that proxy for these asset classes to August 2024,³ the standard deviations for the risk-control and risk-asset subportfolios within the Moderate portfolio are 3.9% and 14.8%, respectively, and their correlation is 0.29. These risk characteristics are consistent with the history in Exhibits 1 and 2. They combine to produce a standard deviation for the Moderate portfolio of 9.4%.

Exhibit 3 shows the risk attribution by factor since 2002. Although the Moderate portfolio has 10 asset classes, term and market factors explain 98% of its total risk. The betas measure the exposure or sensitivity to each factor. Notably the market factor contributes nearly all portfolio risk (97%) due to its higher standard deviation (and higher beta). In contrast, the term factor contributes very little to total portfolio risk (1%) due to both its lower standard deviation and low (diversifying) correlation. Idiosyncratic risk is the unexplained portion of total portfolio risk. It is small (2%) and generally uncompensated with an associated return premium — unlike term and market factors.

Exhibit 3 - Risk Attribution Since 2002

Next we evaluate the 3-year period ending August 2024, which includes the unexpected spike in inflation and bond yields. Over this difficult period for bonds, the standard deviations for risk-control and risk-asset subportfolios are 6.6% and 16.0%, respectively, which are higher than the prior example (since 2002) but still consistent with the history in Exhibit 1. However, their correlation is 0.78, which is high relative to the prior example and the history in Exhibit 2. These risk characteristics result in a standard deviation for the Moderate portfolio of 11.7%.

Exhibit 4 shows the risk attribution by factor over this 3-year period. Term and market factors explain 99% of the Moderate portfolio’s total risk. The market factor still contributes the vast majority of total portfolio risk (84%) due to its higher standard deviation (and higher beta). Despite the risk environment for bonds, the term factor contributes only a small portion of total portfolio risk (15%) due to its lower standard deviation and elevated but imperfect correlation.

Exhibit 4 - Risk Attribution Over Last 3 Years

Investors are more likely to succeed when they ignore short-term noise and focus on their long-term goals. For asset-only investors, the overall results show that risk-control assets (term factor) control total portfolio risk — which is dominated by the market factor — through both a lower standard deviation and generally low correlation. Building portfolios from risk-control, risk-asset building blocks is a far more reliable approach to risk management than building them solely from conventional asset classes or tactical market timing.

There are additional benefits for goals-based investors. The risk-control subportfolio is transformed into a goal hedge by employing high-grade, duration-matched bonds. The goal hedge can fully or partially secure an investor’s lifetime goals with minimal goal-relative risk, while the optimal risk-asset subportfolio can fund any unhedged portion of goals (and hold surplus assets). Perhaps ironically, a high correlation between the goal hedge and risk assets is actually beneficial because risk assets will then comove with goals just as the goal hedge does, further reducing goal-relative risk.

Asset allocation should rely on a sound theoretical foundation that is empirically valid and robust in practice. The risk-control, risk-asset framework represents the functional implementation of ICAPM portfolio theory.⁴ It provides a unified asset allocation framework for both asset-only and goals-based investors, with benefits that include better control over total portfolio risk, more robust diversification, and more reliable goal funding.