A large public plan sponsor recently asked us at State Street Global Advisors to take a look at its equity program. The investment team was concerned and also a little perplexed about recent performance. The program had marginally trailed its benchmark for several years, yet it was not immediately clear why. Like a lot of institutions of its size, the plan had constructed its portfolio in classic core-satellite fashion, with the lions share of the assets passively managed against a standard cap-weighted index and the rest spread among more than a dozen top active managers.
It was only as we dug deeper that a pattern emerged. Whereas each manager delivered a high degree of active risk on its own, together they delivered almost none. Given the capacity constraints of each manager and the plans understandable desire to diversify its holdings, it had added managers to the point at which they were canceling each another out. For all its efforts to construct a well-balanced and high-active-share program, it had essentially built the equivalent of a very expensive index fund.
Here, weve run simulations similar to the analysis we conducted for the plan. We have designed a universe of ten active large-cap U.S. managers by randomly selecting from the top institutional managers over the past five years in the large-cap-blend Morningstar U.S. mutual fund database. (This emulates the behavior of many institutions, which tend to focus on the top managers when beginning an active search.)
We then set out to design a classic active core-satellite program, with 50 percent allocated to the S&P 500 index and 50 percent to active managers. For the active portion, we first allocate to the manager with the highest active risk and then make additional allocations to managers in descending order of active risk. We utilize an Axioma risk model to estimate a holdings-based active risk level as of year-end 2014. In each case, we equally weight our active managers. So, if we have two managers, each gets 25 percent, and if we have ten, each gets 5 percent.
If we increase the number of active managers while still keeping the overall 50-50 passive-active breakdown the same, we quickly see a material decline in active risk and active share. With one active manager, the equity program achieves an active risk of 1.7 percent and a high degree of active share at 46.3 percent. After five managers, our active risk falls to 0.65 percent, and it tails off to 0.43 percent at ten.
We see a similar erasing on asset-specific and factor risk. For example, the risk model estimates a significant small-cap bias from the first manager of 0.34 standard deviations from the index, but this bias collapses to 0.06 at ten managers.
Does this mean that investors should give up on active management? Hardly. But it may mean we all need to seriously revisit the traditional core-satellite paradigm. Lowered return expectations have left a growing number of institutions anchored on policy benchmarks that fall well short of their targets. Yet they may be making a mistake if they think they can make up lost ground by building a sufficiently robust program of active managers. A better approach, we think, is to start reconceiving their core in terms of the specific market factors they can harvest to boost returns and mitigate risk. Indeed, were currently engaged with the large plan sponsor to identify which combination of factors size, low volatility, value, quality and momentum is most appropriate for its needs.
Active management still plays a role in the new setup, but with the core passively managed to capture the markets recognized premiums, the number of active mandates can come down. Although the size of these mandates may increase, in many cases overall fees will shrink. Active share, ironically, will go up. And most important, institutions will be able to build portfolios more closely aligned with what they are actually trying to achieve.
Ric Thomas is the global head of strategy and research for State Street Global Advisors investment solutions group in Boston.
Get more on indexing.