• Stuart Fowler

'Break-even Year' for equities

It’s 2023, just 7 years away. It’s exceptionally close. It's a measure of the QE-induced equity risk premium. It matters to valuation, asset allocation and funding.

One of the outputs of our modelling of horizon-specific real equity returns is a number we call the Break-even Year. It serves as a headline marker of the size of the expected risk premium, or 'excess return', available to investors willing and able to bear equity risk. Our quantitative asset-allocation and funding model is systematically responsive to available risky returns relative to the risk free return, as are many portfolio-construction approaches. A very close break-even time horizon for risk taking is normally only available in equity bear markets.

What exactly is it?

The number is the time horizon at which the 95th percentile (think of it like a worst-case scenario) in our calculated distribution of probable real equity total returns for an optimised combination of global equity markets (for this purpose, without any rebalancing) exceeds the risk free yield given by Index Linked Gilts (ILGs) with the same duration. In other words:

  • we are as confident as we realistically can be that, at and beyond the Break-even Year, essentially all of the probable real returns from equities exceed the risk free rate;

  • it requires extreme risk aversion to rationalise a preference for the risk free alternative to equities for objectives with longer consumption horizons than this.

There may be some investors who fund their objectives at higher than 95% confidence but they are likely to be responding to regulation or legislation and are not truly part of the population of risk bearers. Our clients have purchasing-power goals and fear inflation risk as much as capital risk. They always have the freedom to select between a hedging asset that is risk free in terms of the nature, date and quantum of their goal, and an optimal portfolio of risky assets. How they select between them, or combine them, is dependent dynamically on market conditions and their risk tolerance. (In modern portfolio management this is known as the 'portfolio separation theorem' or 'two-asset portfolio' and, though not as common as relying solely on diversification of different sources of risk, it is increasingly being adopted as the best way to manage risk.)

We’re used to seeing the Break-even Year fluctuate with ILG yields and equity prices but it’s very rare for it to be this low. It's extremely low compared with data long before inflation-linked bonds existed, when ex post deflated nominal bond yields can serve as a proxy for real yields. It is about QE but it’s only indirectly about Brexit. Because the equity expected return has a currency component, the combined actual movements of equity markets and sterling mean that the Breakeven Year altered little before and after the Referendum, although Brexit did cause risk free rates to fall to a new low.

Why so unusual?

To put 7 years into perspective:

  • The Break-even Year is normally 25 years out. ‘Normal’ assumes i) ILG yields of 1% (an estimate of the pure time value of money with no risk premia and no time preferences); ii) all equity markets are in line with their own long-term regression trend for real total returns; iii) sterling is in line with its purchasing power parity with other currencies. It's a theoretical norm, or 'normalised', rather than a state that is usual or frequent.

  • If we substitute normal with actual ILG yields of -1.4% pa but keep sterling-adjusted equity expected returns normal, the Break-even Year is 11 years out.

  • Currently, we think expected sterling equity real returns are higher than normal, assuming the long-term trend will hold. Actual market levels vary between 52 and 98% of their long-term real-return trends (Japan and Emerging Markets at the low end, other developed markets at the high end). Real exchange rates are close to parity. If we notionally combine these equity conditions and the actual current risk free rate, we get the 7year horizon, which means that risky-asset valuation explains only 4 years’ difference and the risk free rate explains 14 years'.

  • Seven happens to be about the holding period at which real returns from equities appear to be log-normally distributed and our model assumes both a normal distribution and long-term mean reversion. If the Break-even got any shorter than this we would have to question whether 95% was really ‘extreme'.

What does it imply for investors?

For investors generally this situation is extraordinarily good news. Risk is being rewarded in a manner that would otherwise require a deep bear market in equities. The actual implications for different classes of investor are a little more nuanced.

  • For individual investors, the conventional approach to suitable advice, which evolved with custom but has been reinforced by lazy regulation, implies a high opportunity cost. Once their attitude to risk (a constant, maybe even a personality trait) has been defined by reference largely to volatility, their attitude to volatility has to be shown to change if their actual bond:equity mix is to respond to large changes in market risk premia rather than remain fairly constant. Responding to horizon-specific risk premie, which is likely to lead to better outcomes, assumes the risk approach is constant but the risk level can change.

  • For equity investors with very long horizons and normal risk aversion, such as some sovereign wealth funds, endowments and intergenerational family trusts, it’s all a bit academic as they would rarely anyway hold risk free assets.

  • For Defined Benefit pension schemes, who are punished for variance in the funding position calculated by comparing portfolio values with a notional value of liabilities discounted by bond yields, the deficit goes up as their own liability-equivalent break-even horizon goes down. Only by agreement with the Pension Regulator will the period allowed for spreading deficit-recovery contributions increase to compensate. The implication of QE-induced market distortions is that it is not enough for the recovery period to be stretched out: the recovery contribution level also needs to fall.

  • It would be a disaster too for individuals retiring now had they, without the Pension Freedoms, felt forced to buy annuities or bought them by default. A level annuity without inflation protection never was risk free for a retirement spending goal, so the harm would have taken the form of giving up the loose inflation indexation in equities and property in favour of a bet that inflation would not exceed market expectations embedded in nominal bond prices. Exchanging zero opportunity-cost equity upside for naked inflation exposure on a meagre income is not rational. But by the same token, a fully-indexed annuity (equivalent to the benefits delivered by a public-sector pension) always was beyond reach for all but the wealthiest savers who wouldn’t on these terms choose it anyway.

In fact, individuals are perhaps the only investors left in the game who truly can respond without unavoidable external constraints to the Governments’ efforts to steepen the slope of the risk premium (for that’s what QE does). This is remarkable considering the merit of the strategy necessarily assumes an unconstrained rational response by agents in the economy.

What does it imply about expectations?

The presence of so many constrained investors perhaps explains why a high ex ante risk premium has not been arbitraged down, in spite of fairly good equity returns. The other explanation is that investors generally are assuming that historical real returns will never revert to historical means and are shifting the whole distribution of expected returns down, rather than assuming that there will be a long period of below-trend returns (as has occurred in the past) before eventually recovering.

History teaches that it is better to assume the long-term trend persists (capitalism adapts; markets are equilibrium systems; indices are Darwinian reflections of the system) and make realistic allowance, based on data evidence, for the time-dependent risk of deviations from trend. When much lower sustainable or ‘true’ real return trends from global equities (no more than 3.5% pa, or nearly half the historic trend) are required to make an apparent 7-year breakeven a true, normal 20-year breakeven, it pays to bet against those implied expectations of a system change, even when you think the change is plausible.


© 2016-2020 Fowler Drew Limited

  • Black Facebook Icon
  • Black Twitter Icon
  • Black LinkedIn Icon

Authorised and regulated by the

Financial Conduct Authority