In the third article in a new series, Pascal Blanqué and Amin Rajan argue that the ex post returns no longer match the ex ante promises
Since the 2002 bear market, pension plans have learnt an enduring lesson: seeing how a given asset allocation might work on paper is one thing, realising what it delivers in practice is quite another.
The gap is due to implementation leakage, caused by untoward external and internal
factors that undermine optimal execution, according to the Amundi Asset Management/CREATE-Research 2014 survey ‘Alpha Behind Alpha: Rebooting the Pension Business Models’.
The external ones have included fees, commissions, spreads, market impact, portfolio drift and the opportunity cost of trading. Additionally, simulation models have overly influenced asset choices, without taking into account the fact that investors’ own actions and reactions introduce new risks, as do the reactions of other investors as they weigh up every situation. Back-tested performance often ignores the market impact of implementation. The past may be the best guide to the future but it is still a very imperfect one.
There are internal factors too. To tackle them, plans have been taking action to minimise leakage while rebooting their business models (see figure) in line with the old military dictum: plans are nothing, but planning is everything.
As well as improving their governance practices to develop strong investment beliefs and the necessary prime-mover advantages, pension plans are making more judicious use of investment vehicles to get the best out of passive investing (covering cap-weighted indices, ETFs, smart beta) and life-cycle products. The winds of change are evident.
A loser’s game
If governance is about setting goals, and asset allocation is about the strategy needed to achieve them, execution is about its delivery. Together, they are like a three-legged stool, each leg needing the others to work. Before the 2008 crisis, 80% of a portfolio’s returns reportedly came from intelligent asset allocation. Now that is reduced to 50%: the rest comes from savvy execution. Rickety markets have ensured that poor execution delivers poor results.
Indeed, many pension plans now see investing as a loser’s game, like tennis – a game in which success goes not to the player with the best strategy, but to the one making the fewest mistakes. Hence, apart from risk management, three other areas have attracted significant attention.
Far and away, the biggest change relates to risk management models. With time-varying risk premia, risk is seen as a dynamic concept. It is therefore vital that: its causes are separated from their consequences; its management is separated from its measurement; its impact within a period is separated from its impact at the end of the period; its time dependency is separated from its randomness; its new focus on maximum tolerance for a drawdown is separated from the old focus on volatility. Hence, the emerging risk models are used to frame the questions, not deliver the answers. Effective risk management relies on human judgement. That’s why governance is the alpha behind alpha.
The second biggest change is the search for a value-for-money fee structure. The heads-I-win, tails-you-lose fee structure is gradually withering on the vine. There is strong resistance to paying alpha fees for beta performance. Fees are seen as a key source of outperformance when compounded over time. Once just a mirage, fee clawbacks are now on the agenda.
Investment vehicles are the third area attracting a lot of attention. Having done their asset allocation, pension plans used to invest directly into their chosen asset classes via active management. Now, the trend is to seek advice-embedded vehicles that are cost-effective, liquid and transparent. In the DB space cap-weighted indices, ETFs, smart beta and multi-asset class products have been the most popular vehicles. In the DC space, life-cycle funds and diversified growth funds have led the line up. This distinction between asset classes and their vehicles is driven partly by cost and partly by rising interest in outcome-oriented investing.
The final area attracting interest is financial engineering. There is less resistance to using devices such as shorting, leverage and derivatives to extract extra value and/or downside protection. Interest in long-short funds, risk parity, CDS and credit and equity derivatives has ballooned over the past three years.
Case study of a Swiss pension plan
“In the past, our risk models relied on the notions of risk-free assets, static asset class correlations and stable risk premia. Volatility was the key measure of risk, backed by VAR. Since the Lehman collapse, extreme spikes in volatility and asset class correlations have been common. Dominated by the euro crisis, 2011 was a nerve-shredding year: the pool of AAA-rated government paper contracted by 65% on account of sovereign downgrades on both sides of the Atlantic.
“Consequently, the old risk models became obsolete. They had relied heavily on technology only to realise that it improves the measurement of risk, but not our understanding of it. Hence, we have implemented three changes.
“First, our key measure of risk is the maximum drawdown our board is willing to tolerate, taking into account our sponsors’ unwillingness to make additional recovery contributions. The limits are reviewed annually.
“Second, we allow for path dependency. We look at risk in a multi-period context, to allow for the fact that returns in any one period can be heavily influenced by returns in previous periods – especially when momentum is working.
“Finally, our diversification is increasingly based on risk factors that allow us to understand the changing asset class correlations in different market regimes and the real sources of risk for all asset classes.”
Pascal Blanqué is deputy CEO and CIO of Amundi Asset Management and Amin Rajan is CEO of CREATE-Research