How did we do in 2008?
Whose performance is it anyway? When clients’ portfolios are genuinely customised to personal goals whose terms of reference they have jointly planned and taken responsibility for, it is reasonable to answer questions about performance with a question of our own. Yet the question posed is also a matter of fact we ought to be able to answer. In this article we provide the answer together with some guidance as to what, if anything, it means.
Goal-based portfolio management is to the retail investment scene what ‘Liability Driven Investment’ (LDI) is to the institutional markets. Goals are defined and quantified and the sources of uncertainty that characterise them are also identified. The solution to how these risks are then managed is some combination of complete hedging (either by insuring or matching the liability to an asset with the same nature) and mismatching, which is equivalent to a bet. LDI managers need to provide the inputs that help clients choose how much risk or which risks to lay off or retain but they no more make the choices than a casino croupier will tell you how to place your bets. So in a goal-based portfolio which reflects how clients chose generically, at a high level, to lay off some risks and retain others, whose performance is it?
Furthermore, if clients have different terms of reference that describe their goal (such as the rate of drawdown from the portfolio; required real terminal wealth at future dates; the dates themselves or the risk preferences), then it is possible no two clients will have the same performance.
For an asset gathering business used to marketing itself using the track record of its clients’ portfolios, dispersion of returns is anathema. For a business seeking to maximise economies of scale, customisation is itself anathema. The typical industry solution is to shoe horn every portfolio into one of a small number of discreet categories such as ‘conservative’ or ‘growth’ (HSBC Private Banking, for example). Goal-based investing means the return dispersion will not be explained by these categories.
Defining the portfolio sample Our goal-based portfolios (which we refer to as ‘Defined Outcome’) are significantly clustered around goals which share a common characteristic: their owners are either already drawing down from portfolios to meet spending or plan to draw in the future. The actual terms of reference could be as diverse as drawing subject to a constraint of not eroding the spending power of the capital at intervals of say 10 or 20 years (such as an endowment fund or family trust) or consuming all of the capital assigned to the goal by death subject to the constraint of not running out before death (a retirement plan).
These drawdown portfolios are subject to the normal rule that the dispersion of returns in short periods is significantly explained by the split between risks laid off and risks retained. Because most clients have goals defined by purchasing power rather than in today’s money (such as retirement spending), the asset that perfectly matches their liability or goal is an index linked gilt with the same duration as the spending. If equities and index linked gilts generate very different returns, the split between each type of holding will explain much of the performance achieved.
The split itself is explained by two independent attributes: the duration of the spending plan and the client’s risk preferences. These may overlap. A client with long-term liabilities but low risk tolerance may end up with more risk free index linked gilts than a client with shorter duration liabilities but a greater appetite for the upside potential of equity bets. Since risk tolerance clearly belongs to the client not to us, the best way to represent our performance is to select a group of client portfolios with the same risk tolerance, so that the performance dispersion explained by asset allocation differences is in turn explained by the differences in the goal terms of reference.
Ideally, these will be mainly about the plan duration: the profile of real drawdown targets and how long these targets last. The profile of drawdown (though typically in a retirement goal front-loaded to the earlier, fitter years) may vary over the life of the plan. Some plans may also terminate or taper at different ages, such as because of assumptions about housing equity contributing to spending in later years. Whatever these differences, they are captured by the notion of the plan duration.
However, there are some other return differences explained by a number of further client-determined preferences, such as the degree of bias to the UK as ‘home market’ or the extent to which the client chooses to back our expected equity returns over shorter periods than their longest time horizons in a plan (equivalent to betting more heavily on our estimates of market conditions). This will alter the mix of the four equity building blocks we use in these ‘Defined Outcome’ portfolios: the UK, Europe ex UK, the USA and Japan.
These caveats aside, we have taken three drawdown portfolios as representative. They cover a client with 10 years to go before retiring and a near-fully funded retirement spending goal and a projected drawdown duration of a further 35 years. The second client is within 4 years of starting draw during which he continues to earn but is no longer adding to capital and then plans to draw over the following 30 years. The third client is in drawdown and, since the spending targets are easily met but there is no competing need for the money, the plan duration is to age 100: 40 years. We have carefully avoided selecting clients with very low risk tolerance or much shorter time horizons as these will obviously look clever in a period of falling equity prices.
For comparison we show the two indispensable building blocks for hedging or retaining risk: index linked gilts (represented by the FTSE 5-15 year index) and UK equities (represented by the Legal & General FTSE All Share tracker institutional units, so after expenses). 2008 was one of those periods in which risky and risk free returns diverged dramatically, with ILGs up slightly and the All Share down nearly 30%.
The average equity exposure over the year was 96% for the youngest client still accumulating, 77% for the client close to the start of draw and 64% for the client already drawing.
Considering these are relatively high equity exposures, we have obviously done well to have diversified the equity risks geographically. As noted in our article on diversification, this is not a risk control but is an essential part of maximising the risk-adjusted expected returns of the risky portfolio. In general, our portfolios use international equity markets more freely than most wealth managers and particularly so when the horizon-specific risk-adjusted expected returns are higher than in the UK, as was the case throughout 2008.
Currency was a significant factor in the short-term payoffs to the international exposure. In our return generating model, currency is a source of additional uncertainty in the horizon-specific real returns. Because this is a real return model focused on real wealth outcomes, the relevant source of currency risk is the variance in the exchange rate not explained by relative inflation, in other words the variance in the real exchange rate. In 2008 currency movements far exceeded inflation differences. As currency exposure lifted the portfolio returns, and therefore actual exposures relative to the model target exposures, we have gradually adjusted the weights, both within the three international markets and back to the UK. We anticipate there may be a greater move back to the UK in 2009.
What about these rebalancing dynamics themselves? The return-generating model assumes gradual or weak ‘mean reversion’ in equity returns and so we tend to be naturally contrarian. However, we are also gradualists, rebalancing to new targets roughly monthly and usually in small increments. Hence in 2008 we gained from the relative returns between markets but not as much as if we had not gradually moved against the trend. But the flip side of that is that some of the position taking that we benefited from in 2008 stemmed from foundations laid in earlier years when return differences (and overall portfolio performance effects) were much smaller.
Looking forward instead of back The messages we give to clients are less about the actual returns in a single period as short as one year and more about the implications of the market returns for the chances of achieving the goal targets with the original agreed confidence. We look back at past performance as auditors but we look forward to the goal outcomes as investment planners.
The audit looks at just one single-step scenario in a plan with multiple steps and many possible returns for each step. It is nonetheless important if we found the actual return path in any of these steps was outside the model parameters. Bad as they are, the 2008 equity returns in local currency are within our model.
What singles us out from most (if not all) private client advisers is our ability to quantify the progress towards their defined outcomes. In a perfectly hedged portfolio, market volatility affects the path of the valuation (per the audit report) but has zero impact on the real outcomes (per the planning report). But equity bets have a different effect which needs to be continuously monitored and explained.
Very short horizons (usually up to three years) are matched by cash but longer horizons need an inflation hedge as well as a capital hedge, hence the need for index linked gilts. The fully hedged horizons immune to volatility may extend up to 10 years. Long horizons (over 15 years), which at most levels of risk tolerance are backed by equity bets, have time to make good interim losses through mean reversion. The range of probable outcomes for the payoffs of these equity bets is also therefore only slightly affected by volatility. It is the ‘middle’ years, where some combination of equities and index linked gilts is held against the drawdown ‘liability’, that get pushed away from their targets when the equity return path is negative. However, the impact is not the full extent of the drop in actual equity values.
What did we say after 2008? Our monitoring of goal progress refers back to simulations of plan paths and outcomes using the same assumptions as the model. This suggests that the ‘interim’ shortfalls emerging in 2008 in the resources required to meet the original goal outcomes, at about 10%, are small enough, given their volatility, for us to be able to reassure clients that they do not need to assign more resources to their drawdown plan or lower their spending expectations.
These simulations and the original model runs assume, by the way, that a worse case scenario is similar to Japan’s ‘lost decade’ but experienced in all equity markets. Big equity falls and slow recovery are a killer for drawdown plans. When explaining to new clients in the planning process how bad equities can be, Japan was the example we used. In 2008 such a possibility looked frighteningly real where before it might have seemed theoretical. Because it was allowed for, it looks a lot less frightening to our clients. Has it been allowed for in your planning and do you know what it implies for your outcomes? Or do you only know what your portfolio earned last year?