Shallow risk, deep risk
An article in FT Money last weekend caught our eye. Former Russell Group pension consultant Don Ezra was quoting US investment writer William Bernstein, whose most recent book introduces the concept of shallow risk and deep risk. It caught our eye because the two types of risk fit almost exactly the two that we differentiate as ‘path risk’ (captured by nominal return volatility of a portfolio, such as the standard deviation of daily or monthly returns) and ‘outcome risk’ (the size of the uncertainty about long-term real outcomes of a long-lived plan). It fits, but it’s also a lot more catchy.
Deep risk explains why two different generations of savers, both benefiting from averaging the ever-changing market conditions over long phases of accumulation and decumulation (possibly even with the same observed standard deviation of returns), may nonetheless emerge from the entire plan with vastly different real outcomes. Plan outcomes here may be measured as lifetime capital drawn and spent or capital retained and passed on to the next generation.
A ‘lucky’ generation will benefit from average market conditions being poor whilst accumulating, as long as returns are high in decumulation – somewhat counterintuitive yet perfectly logical. Imagine, for instance, the potential benefits of being Japanese and making regular contributions into Japanese equities from the 1990s. As long as markets do eventually ‘revert to the mean’ so that the capital is drawn down from much higher average levels, the outcomes will be much better than if markets had done very well in the first phase only to do worse in the second phase. The trend of real total returns (with income reinvested, adjusted for Japan’s inflation or deflation) for Japanese equities slid from over 8% pa in the 1950s through the 1980s to close to zero in the next two decades, leaving the entire 6-decade period with a trend of 4%, about 2% pa less than other developed markets. Being twenty or forty years older will reflect that difference.
Diversifying equity risk
It could also have been dampened for all generations as long as they had diversified between Japanese and foreign equity markets, as the deep risk that hit Japan so hard was largely peculiar to Japan. Diversification is one of the best ways to manage deep risk, according to Bernstein. But its actual motivation has probably been to manage shallow risk, as it also helps smooth the short-term path of returns as long as volatile markets are also not very highly correlated. But correlations usually being inconveniently high, diversification is more important as a way of mitigating the damage of political and economic deep risks.
Confiscation and devastation are two of the geopolitical risks that history reminds us we should not ignore. Confiscation could mean outright sequestration of assets (perhaps limited to foreigners’ holdings), nationalisation without proper compensation or penal taxation. Devastation from wars or revolution can be a form of confiscation but may also be associated with currency debasement. Losing wars is ironically not necessarily as destructive but is also often associated with currency debasement through hyperinflation and that can destroy wealth for a generation.
Since these historical instances of deep risk tend to be country-specific, diversification is the key management tool. The risk Ezra thinks people are most likely to manage poorly is inflation. More particularly, inflation or deflation will make a big difference to long-term outcomes if businesses are not able to adjust to it, and so the normal means by which equities compensate for general price changes in the economy breaks down.
Inflation risk and debt instruments
Ezra’s reasoning fits perfectly what makes the most difference to a portfolio constructed to optimise volatility and one that aims at optimising long-term real outcomes. The first relies heavily on fixed income investments, as these have the most reliably low (often negative) correlation with equities. It is easy to see that a range of products or indices ranked by their expected volatility are differentiated largely by the exposure to bonds rather than the mix of other assets that are more closely correlated, such as property, private equity and hedge funds. The higher the bond exposure, the lower the risk.
What happens when we judge this approach against deep risk? Assets with a reasonable but uncertain chance of keeping up with inflation are substituted with ones that will inflict a real loss if inflation turns out to be higher than expected. They may also gain more if inflation comes in lower, as clearly happened after the Great Moderation, although outright deflation could have a different effect, wiping out businesses and households for whom the real burden of servicing and repaying debts becomes unbearable.
The Great Moderation was a very recent (current, even) generation-long difference that made as many people extraordinarily lucky as were made desperately unlucky by depending on fixed nominal incomes during the high-inflation 60s and 70s. Any factor in your portfolio that can make such a big difference to real outcomes has to be considered very risky, not low risk.
Long-period cumulative inflation is probably not symmetrical. The chances of high inflation are greater than those of deflation simply because governments can control the creation of money. So government-regulated inflation represents a vicious form of deep risk. Unlike equity risk, where real returns have historically provided a positive albeit uncertain premium for living with it, real bond returns show no systematic compensation as an incentive for bearing the inflation or currency-debasement risk.
How lucky or unlucky could you be?
When all possible future real outcomes are modelled using historical experience of deep risk, how good or bad could your outcomes be?
If, for example, you were starting retirement now, aged 65, and needed to make your capital last (but in this example without valuing inheritable surpluses as highly as you value your own lifetime spending), we can calculate the range of possible outcomes in terms either of when you might run out (spending at a fixed real rate) or the average level of spending. We will assume that death comes late and so will plan for the capital to last at least till age 100. (It would not make sense to plan how to survive deep risks yet not deal as prudently with the uncertainty of your own mortality.) The underlying capital is invested in a two-asset portfolio: globally-diversified equities (with no bias to the UK as ‘home market’) to fund long-term cash flows and cash and index linked gilts to fund the nearest years of spending. The projections assume that the balance will change with the shortening duration of the plan (the older you are, the more of your cash flows are short term, so less in equities) but around that essential risk discipline, necessary to make the plan resistant to deep risk, there is scope to vary the equity exposure with expected real returns depending on whether equities are above or below their long-term historical real return trend. Risk tolerance has been assumed as low.
For £1m of capital today, the rate of real draw that is safely sustainable in real terms is £41,000. ‘Safely sustainable’ means with 99% confidence, so excluding only a historically unprecedented and unsurvivable collapse of the capitalist, market system worldwide. If market conditions were average, sustainable gross draw could be as high as £57,000. The best outcome (so with only a 1% chance) is £78,000, although much of this additional spending would probably arise late in the plan when it is least valued – and indeed almost assumes a wish to share the good fortune with the next generation.
If you intended always drawing at the average rate, equivalent to planning on 50% confidence of coping with deep risk, you would not run out of capital until age 86 at the earliest, with a 1 in 20 chance it may not last beyond 87.
A range of possible real outcomes equivalent to nearly twice as lucky or half as unlucky is a much more realistic expression of the true uncertainty in a long-lived investment plan, even when shallow risk is automatically reduced by drawing over long periods in different market conditions and when diversifying deep equity risk.
Tax risk for UK-based investors
Even these illustrations ignore the uncertainty about the rate at which gross draw would translate into after-tax spending: a possible deep risk that is not diversifiable Allowing for the risk of excessive taxation for long periods before it finally becomes seen as counterproductive (think from Wilson to Thatcher), the true range of spending outcomes is probably over twice.
Where do the numbers come from?
These projections come from the Lambda Investment Technology long-term dynamic asset model that Chris Drew and I first developed in 1999 and has been tested now in 16 years of actual experience. The Lambda model is the basis of all our clients’ decisions about a goal-based plan they are willing to live with, partly because it allows for the true deep risk uncertainty, and our decisions about how to manage the goal-based portfolio that will deliver the planned outcomes within the agreed tolerances.