Taking risk budgeting a step further
An analytical framework at the fund selection stage can help spare DC fund participants the pitfalls of a more advanced approach to diversification, writes Thierry Roncalli
One of the main risks faced by defined contribution pension plan participants is rooted in the implementation of rather shallow, or ‘naïve’ diversification strategies by the participants themselves, according to Shlomo Benartzi and Richard H Thaler in their 2001 article ‘Naive Diversification Strategies in Defined Contribution Saving Plans’(1),
In their study, Benartzi and Thaler found that DC participants tend to follow a ‘1/n’ diversification strategy, simply allocating equally their assets among the funds offered by their scheme. Regardless of their actual investment objectives, participants tend to mechanically give more weight to whichever asset class the DC plan fund offering is slanted towards. The outcome may be rather benign when the fund offering is half in bonds and half in equities, but might have more dire consequences in terms of risk concentration where 80% of the funds offering is in equity and the rest in bonds. Yet, the strategic allocation phase is essential for DC participants, who are unlikely to review their allocation past their initial investment.
If a DC participant were to reason in terms of risk, could a risk parity approach be of any help in implementing a more robust and adequate asset allocation? Or would a risk parity allocation be equally naive? Whereas it would certainly not be as naive as the first approach – notably avoiding risk concentration – we find that a straightforward risk parity approach might not be adapted to the objective of a DC pension participant with a long-term investment horizon that can extend over various decades. A more refined risk-budgeting approach based on the analysis of the risk factors facing these investors – among others inflation and growth – might provide an optimal way to allocate these assets and even to select specific investment strategies.
The principle underlying risk budgeting-based allocation is simple: the asset allocation is established on the basis of the risk contribution of each portfolio component, rather than their expected returns. In the case of a portfolio comprising two assets managed using the equally-weighted risks, or ‘risk parity’, approach, each asset makes an equal contribution to total risk. By way of symmetry, and under the efficient market hypothesis, each component will generate the same proportion of the total performance of the portfolio. To achieve this balance, exposure to the riskier assets is reduced, and vice versa. Traditional diversification via market cap, or by any means ‘naïve’, weightings leads to markedly different results. For example, in a balanced portfolio composed of 50% equities and 50% bonds, the equity component generally accounts for over 90% of the portfolio’s volatility.
Whereas reasoning in terms of risk parity would certainly avoid the pitfall of risk concentration, it may not be adapted to DC participants. Without the possibility to resort to leverage, the risk parity returns might be insufficient over a long period. And even if leveraging a portfolio were possible, a major challenge would still to be addressed – namely, defining a risk allocation in line with investors’ anticipations, beyond simply achieving an intuitively better diversification than the market cap approach.
Beyond asset classes and into risk factors
Even if a portfolio’s strategic allocation might appear to be neutrally diversified using the asset class-based risk-parity approach, it may still harbour hidden risk concentration. Different asset classes are affected by various risk factors, which themselves might affect more than one asset class at a time. Simply splitting risk budget equally may result in a concentration on a limited number of factors.
Long-term investors such as DC participants may want to go beyond risk budgeting based on simply asset clases by no longer considering only asset classes (equities, bonds, commodities, and so on) but as well economic risk factors such as economic activity (GDP, industrial production), inflation (commodity and consumer prices), interest rates (real interest rates, steepening and convexity of the yield curve) and the effective exchange rate.
Under such an analytical framework, the portfolio is constructed around the budgeting of risk factors. At first glance, this might seem easily achieved by simply linking an asset class with a particular type of risk – eg, bonds with interest rate risk – and applying the traditional risk budgeting method. But remember that single risk factors will affect various asset classes at a time. For instance, while equities are affected by economic growth and industrial production, they are also affected by interest rates and inflation. Therefore, the sensitivity of each asset class to certain risk factors must be identified in order to establish an asset allocation.
Consistent strategy selection
Risk-factor analysis does not only enable a DC investor to allocate a portfolio according to strategic anticipations, it may as well be used at the strategy selection stage to ensure full adequacy with these anticipations. As a matter of fact, analysing apparently similar strategies through the risk-factor lens gives investors a clearer, and sometimes suprising, picture of the variables that drive their performance.
By way of example, an increasing number of investors are opting for ‘alternative’ or ‘smart beta’ indices to supplement their passive management activities. Several competing methods currently exist. They include the equally weighted, minimum variance, most diversified, and risk parity strategies, among others.
Applied to the S&P 500, strategies such as the minimum variance portfolio or the most diversified portolio give a fixed income flavour to an equity portfolio. According to Lyxor’s analysis, the sensitivity to interest rate of such strategies are respectively of 73% and 49%, versus only 6% for the original capital-weighted strategy. This can be explained by the very equities that make up the lion’s share of such low-volatilty strategies, which tend to have bond-like (ie, low volatility) characteristics.
In turn, implementing a minimum variance component in the equity component introduces bond risk that may run counter to the intial strategic allocation. In contrast, other ‘smart beta’ strategies such as the equally weighted or the risk parity portfolio retain their strong equity tone – which may be closer to an investor’s original intentions. From a DC investor’s perspective, using such analytical framework at both the strategic allocation and fund selection stage may help to secure consistency at each stage of the portfolio construction process.
1 Thaler, R and Benartzi, S (2001), Naive Diversification Strategies in Defined Contribution Saving Plans, American Economic Review, vol 9, issue 1, pages 79-98
Thierry Roncalli is head of quantitative research at Lyxor Asset Management