New Year, Many Futures
Every January the sell-side bombards investors with seemingly precise forecasts for the coming year (ALSI up 10% by Christmas etc). Such ‘single point’ forecasts aren’t very helpful at the best of times; in the current environment of heightened uncertainty they are of little value.
- In STANLIB’s Multi-Strategy team we recognise that the future is best understood as a range of outcomes and is constantly evolving. Multi-asset portfolios must be constructed with a particular set of market and macro-economic outcomes in mind, but to think of the future in terms of single point forecasts of growth, inflation and rates creates the risk of being blindsided.
Embracing uncertainty, we define plausible economic scenarios, assign them probabilities and then devise the optimum asset mix for that ‘risk-weighted’ macro outcome. We debate our scenario probabilities on an ongoing basis, fine-tuning the portfolio as we go. We believe that a culture of intellectual diversity and open debate improves risk management and delivers better portfolio outcomes for our clients.
There is an apocryphal story about how a horse racing tipster convinced gamblers that he had a crystal ball. He would begin by writing to many of them with his prediction for the next big race. In reality he would tip a different horse to each group of recipients; the next week he would only selectively write to those who had received the last winning tip. After a couple of iterations, the small group of punters who had received only winning tips would be convinced of his genius.
Unlike our tipster, investment banks’ analysts and strategists don’t have the luxury of hedging their bets: every January they must pick one forecast which is then broadcast far and wide. This new year has brought the usual logjam of ‘market outlooks’ and the likelihood that they will all be wrong (though some more than others).
Sell-side strategists face a conundrum: any sensible person knows that the future is uncertain, but industry practice is to pick a single point. Saying ‘GDP +5%’ seems more authoritative than ‘GDP growth somewhere in the range of 3%-7%’, but the apparent conviction of the former is deceptive: there is more information in the latter because it gives the reader a sense of the range of likely outcomes. This matters, because preparing for unlikely but possible scenarios is the essence of risk management: knowing ‘how bad things might get’ is therefore much more valuable than knowing an economist’s base case, which is in any case a risk-weighted average of the various outcomes they can envisage. The reality that a single point forecast isn’t the whole picture can be illustrated in terms of planning your holiday this year. Knowing that the average daily temperature in Mauritius in March is 29.8 degrees Centigrade tells you to bring plenty of sunblock; learning that there is also a small risk of a bad tropical storm this season gives you a chance to pay up for decent travel insurance.
We understand that a successful investor must be prepared to contemplate multiple possible futures and we have built our investment process accordingly. We define a range of possible and distinct macro and market outcomes over the next 12 months and blend them into a probability-weighted scenario to guide our portfolio construction. By thinking in terms of ranges of outcomes, we aim to minimise common behavioural biases such as false anchoring (the status quo or scenario you first thought of dominates your ongoing thinking) and confirmation bias (only looking for data points which support your initial thesis). Team members independently assess the probabilities of these scenarios, and we “score” them as a team on a regular basis. This readies us for change, sharpens our debate and keeps us nimble. We believe scenario planning keeps portfolio thinking more agile, resulting in better risk-adjusted returns over time for our clients.
New Year, New Forecasts
The thump of mail hitting the doormat may have been replaced with the pinging of your laptop as emails arrive, but January still heralds a swathe of investment banks’ predictions for the coming year. Many may be well-articulated and based on solid data, but most will still be wrong.
Sell-side forecasters will empathise with Harold Macmillan, the UK’s prime minister between 1957 and 1963. When asked what the greatest challenge for a statesman was, he replied: “Events, dear boy, events”. Reflect how growth forecasts were put in intensive care by Covid-19 in 2020 and how inflation expectations were blown sky-high by Russia’s invasion of Ukraine in 2022.
As we enter 2023, market watchers must try to predict the complex interactions between inflation, interest rates and growth. Table 1 shows how some of the largest institutions assess the risk of a US recession this calendar year.
Table 1: US 2023 Recession timing dispersion of view
|Bank||2023 Recession Forecast|
|Morgan Stanley||No recession|
|Goldman Sachs||US will “narrowly avoid” a recession|
|PNC||Mild recession starting in the spring|
|Barclays||“Shallow” recession beginning in the spring|
|Bank of America||Mild recession likely beginning in the first half of the year|
|Wells Fargo||“Modest” recession beginning midway through year|
|Deutsche Bank||Recession in second half of the year|
|JP Morgan||Mild recession beginning late in the year|
Source: Washington Post. Data as at 9 January 2023.
Why is it Hard?
Given that every one of the institutions in the list can (and does) pay top dollar for their brainpower, and that all of them are working with the same data, how can there be such a diversity of views among them of what will happen over 12 months?
Macro and market forecasting is hard because economies are complex systems. The variables are interrelated, and information is only known with a lag. The direction of economies and markets is the aggregate of decisions made by millions of people. Market observers also select and weigh the information inputs that go into building these forecasts quite differently based on their prior beliefs and experience.
We have discussed in the past (see the noisy data piece) how forecast error since Covid-19 has risen given the large shock to the global economy and the extraordinary stimulus that followed. The future seems much less certain to follow a well-trodden path.
Figure 1 shows how forecast error dispersion has increased in this decade, reflecting the extraordinary dislocations that the global economy has suffered: the pandemic, the extraordinary stimulus that followed and the inflationary pulse of the war in Ukraine. We see a precedent for this in the recession triggered by the Global Financial Crisis (GFC) in 2008, so this forecast error may be exacerbated if we are indeed heading into recession.
Figure 1: Tougher times – widening range of forecast error around recessions and post-COVID
Source: Bloomberg. Data as at 26 January 2023. Forecasts are Bloomberg Consensus 3 months before year starts. (Nearest date from this point in first two years where data is sparse.)
Besting your Biases
Strategists may disagree about the destination, but they all start at the same point (the reality today); the path of least resistance for most forecasters is therefore to extrapolate the prevailing status quo.
This isn’t the only cognitive weakness that forecasters share:
Recency Bias: the tendency to rely too heavily on recent events when constructing a model of the likely future.
Confirmation Bias: our penchant for ignoring inconvenient data points that challenge our existing view, preferring rather to highlight any evidence that we are right. No one likes to be wrong, and reworking your calculations takes effort.
Anchoring: the tendency to inadvertently select an inappropriate or irrelevant ‘starting point’ in our mental process against which subsequent data is judged.
Herding: the preference for staying with the group even if wrong rather than being an outlier. This may be rational in a profession where the risks of being conspicuously wrong (lose your job) are greater than the rewards (small bonus) of being a lone winner.
You can see that these cognitive biases can make a complex problem even harder for a forecaster: human beings are wired to start at the wrong point, reject contrary data points, stay with the herd and stick to their guns long after they should have capitulated. A classic example of this was how long it took the Fed to drop the notion of transitory inflation.
We appreciate the range of forecasts that we see in the market. It teaches humility that combats overconfidence of reliance on a single forecasted view.
Events out of the blue
The impact of geopolitical events like pandemics and wars can be nigh impossible to determine with accuracy, even if anticipated. Even if you have a grip on the factors which might cause such an event, the difficulty of predicting its impact on markets is compounded by the uncertainty of how politicians and central bankers will react. Covid-19 was an object lesson in compound uncertainty: one could have imagined the impact on office property, but the dislocation in automobile supply chains and skyrocketing used car prices surprised many.
Strong Hand not a Single Ace
We have found scenario planning to be a valuable framework for our multi-asset thinking. Scenario planning describes our process of establishing plausible paths for markets over the next 12 to 18 months. This forces one to think of potential future paths and not a single point as shown in the hypothetical illustration in Figure 2.
Figure 2: Diagram of forecasts versus scenario planning
By scenario planning we are not talking about the typical three case approach, in which a most likely ‘base’ case is contextualised with a bull (optimistic) and bear (pessimistic) case. That is just a basic robustness test which involves tweaking a variable or two to get some variation around the main base case. We create four or five distinct possible market outcomes that are plausible over the time horizon. Behaviourally, a three-scenario approach creates a path of least resistance to the middle option.
We create narratives to describe these different scenarios to help provide colour, relate them to economic and market variables, and to imagine the asset class behaviours that we would expect in each. For each scenario we establish which asset classes would be core, which could add alpha and which would work as hedges. We think this approach has a number of benefits:
- Being prepared. This practice helps us to be mentally prepared for discontinuity and then react to it: having ‘done our homework’ we have allocation strategies in hand as different scenarios take centre stage.
- Open mind. Thinking in terms of a range of outcomes widens your peripheral vision and keeps your mind open. This helps to avoid anchoring, as one is thinking probabilistically across the range and not locking on to a single base case point. Not committing to a single scenario helps us avoid confirmation bias.
- Testing your view. We score the probability of these scenarios independently as a team at regular intervals. This helps us to keep testing our assumptions while keeping our eyes open for new potential scenarios which may emerge. As John Maynard Keynes is famously claimed to have said, “When the facts change, I change my mind. What do you do?”.
- Wisdom of crowds. We risk-weight the possible scenarios and summation of individual views to get a team view. The result of this can be seen in Figure 3: Multi-Strategy Team Scenario Probabilities over time. By drawing on all members, we hope to get a more robust outcome; just as diversification improves risk-adjusted investment returns, an intellectual democracy like ours dilutes the aggregate impact of each individual’s behavioural biases.
- “Manage Disagreement as an Asset”. The smoothness of the team line “hides” the robust internal contrasting debates of the team at the individual level. This process makes disagreement a virtue, as it provides value by unearthing marginal views to test one’s own opinion. Often it is the discussion around marginal changes at the individual level that can trigger ideas for portfolio repositioning, as well as consideration of protecting from adverse outcomes to the view (either way).
- Better risk management. We think of risk not only in terms of drawdowns, but also of not meeting our portfolio objectives. Scenarios are medium-term but they can help steer tactical thinking. For example, current portfolio positioning can be too bearish or bullish relative to the weighted forecast outcome for various asset classes. This can force us to think about protection on the downside and/or taking upside optionality if deemed necessary. How we apply this will be a function of positioning and mandated client objectives.
Figure 3: Multi-Strategy Team scenario probabilities over time
Source: STANLIB Multi-Strategy. Scenarios correct as at 20 January 2023
The practical application of scenarios and the impact on assets can be illustrated by two of our current four scenarios. A recession (‘Hard Landing’ in the chart) would result from the Fed keeping rates higher for longer than currently priced into markets: this could lead to equity markets buckling as margins subside, earnings fall and multiples contract. In the ‘Slowflation’ scenario, growth would remain positive (but below trend) while inflation persists at levels above central banks’ targets for longer. Here local markets could do well, buoyed by commodity stocks and China reopening, but global markets are likely to remain range-bound: big swings, in either direction, are possible but we would not expect to see any big trending moves down such as a hard landing might deliver.
It is important to emphasise that scenario planning is just one input into our process. Our scenarios are designed to look at least 12 months out, whereas we are managing our portfolios daily. We stress that “path dependency” is critical in this journey. Other inputs like our ‘market lenses’ can be useful over shorter time horizons (like 3-6 months) to manage downside, as well as upside, risks and exposures as scenarios develop and change ranking.
Nevertheless, scenarios are incredibly helpful if only by making it harder for us to slip into the behavioural traps that await. They remind us that the only certainty is change, that hubris is a constant risk and that events will eventually make fools of all single-point forecasters. They keep us more agile, which we believe provides better risk-adjusted returns over time for our clients. They also put the exchange of ideas at the heart of our team’s daily culture, making this not just a more interesting place to work, but, we believe, making us better stewards of our clients’ capital.
 We discussed the long lag for NBER to announce a US recession in https://stanlib.com/2022/06/14/market-behaviours-and-recessions/. We may get to December this year and still be unable to confirm if a recession had ”officially” occurred.