Using forecasts in building portfolios: A perilous undertaking
This article was first published in the Quarter 4 2020 edition of Consider this. Click here to download the complete edition.
- History has shown that it is dangerous to rely on forecasts for macro variables or the outcomes of important events to decide how to invest, as even the experts often get their calls wrong and cause large losses for their clients. We have seen innumerable examples of this in recent years.
- At Prudential, we do not rely on forecasts. Instead, we assign a valuation to an asset and compare it to its market price. The quantum of difference reflects the risk premium (or discount) assigned by the market. We are mindful of the future scenarios that markets are embedding, and to the extent that these become excessively focused on a single idea or outcome, then the potential for a future market surprise is elevated, as is that for sharp shifts in price. We can then position our portfolios to take advantage of such opportunities where valuation is on our side.
For some investors, forecasting the future is an integral part of their investment process. This is because the price of an asset is nothing more than the present value of the cashflows it is expected to generate in the future, and these future cashflows are influenced by myriad macro variables. As a consequence, investors spend an inordinate amount of time considering what that future might hold and imagining complex scenarios of how events are expected to unfold -- from which judgements are made about whether the current asset price offers value.
Humans really know little about the future
Yet at Prudential, we recognise that while we know the past and we are living in the present, there is little we can genuinely say about the events of tomorrow. Time and again markets are surprised by the circumstances presented to them. This suggests to us that it is difficult to base your investment process on one that seeks to forecast what will happen with any accuracy. Our starting point is not to ask: “what will tomorrow bring?” but rather, amid the uncertainty of the future, how much risk premium is embedded in current market prices that could protect a portfolio in the event of adverse outcomes. At Prudential, valuations are the critical input, and history has suggested that owning assets where those risk premiums are high adds risk protection by skewing the likely investment return distribution in your favour. In other words, if a future event or macroeconomic data turns out much worse than expected, the asset already priced with a high risk premium by the market should, in theory, fall less than other assets. Equally, if the event or data turns out better than expected, that asset should gain more than other assets.
Of course, there are no guarantees. Even a great hand can still lose at the blackjack table once in a while; but not every time and not over the long run. Consistently skewing the odds in your favour is key to superior returns. What we can’t say is what will happen to interest rates the next time the Federal Reserve meets, or whether Britain will manage to sign a trade deal with Europe, or if South Africa’s GDP will be -6% or -10% this year. We do know that these issues will influence markets, as investors integrate the news into prices as it emerges. However, the news flow is entirely unpredictable, and sometimes its effect on markets is unexpected, making forecasting a perilous business.
We also often see examples where changes in asset prices alone become market-moving factors: they can produce a spiral where risk-loving or risk-averse behaviour reacts solely to shifts in prices to drive further increases or declines.
History points to a low probability of getting it right
As such, we do not base our investment decisions on forecasts for any specific events or macro variables. This is a deliberate decision on our part, since time has shown that: 1) the probability of getting such a call correct is low; and 2) even if you do get a call correct, the market reaction isn’t consistently predictable. Together, the low probability of getting a macro call right and the low probability of predicting the correct resulting market move present investors with a high risk decision, one that we believe is too risky to invest clients’ money in.
Taking a look at past evidence, in January 1990 the US began its campaign to liberate Kuwait from the Iraqi forces that had invaded it. Between then and June that year, when victory was officially declared, the S&P 500 rose 20%. More recently, in the second half of 2008, when Lehman Brothers failed and the US government was forced to bail out AIG, guarantee money market funds and start quantitative easing, South African government bonds rallied 400 basis points. In the six months before that, as the US sub-prime debt market started to sour, few people, and certainly no one at the Federal Reserve or in government, suggested the worst global recession for 50 years was knocking at the door. Some of those that did call the Global Financial Crisis also lost out – once the market had fallen significantly, they were then stuck on the sidelines, so that as the stock market tripled between 2009 and 2013, they missed out on most of the gains. Even the most respected global analysts have been badly wrong, such as when Goldman Sachs in 2008 famously forecast the oil price to rise from $100 per gallon to as much as $200, only for it to halve over the following months.
At Prudential we didn’t know that China was about to embark in 2009 on unprecedented stimulus packages; we didn’t know that in 2011 Mario Draghi would say that he would ‘do whatever it takes’ to save the Eurozone; we didn’t know that global interest rates would end pinned at zero. However, we did recognise that by late 2008, equity valuations were extremely compelling: the Price/ Book value ratio on the JSE All Share Index fell to 1.0X, an all-time record low, offering great potential return upside since the potential distribution of outcomes was skewed in investors’ favour. At the same time, 10-year SA bond yields had reached 7%, close to an all-time low, and offered poor value. In 2009 equities began a multiyear rally which led them to ultimately more than double in value. Bonds had a miserable year, despite the SA Reserve Bank slashing short-term interest rates again and again. Today we are staring down the barrel of the US Presidential elections on 3 November. In light of the above, it shouldn’t be surprising that even though the majority of professional pollsters and experts are predicting a solid Biden victory, we are not explicitly factoring this specific outcome into our portfolio positioning. Just look at the outcome of the 2016 US election when Trump was elected, and the Brexit vote the same year. As is well known, most forecasts predicted a solid victory for Hillary Clinton, and a fall in the stock market if Trump won. But neither came to pass. Equally, Brexit passed when David Cameron had been convinced this could never happen, and as recently as February this year the IMF (with hundreds of the world’s most educated and respected economists) had suggested the world economy would post positive GDP growth.
Our valuation versus the market price
In building our Prudential portfolios, we analyse the value of an asset based on its own history and relative to other assets. We also assess the market’s pricing of that asset, and carefully consider how this might protect us against adverse outcomes. If our calculated value is different than the current market price, then the quantum of difference reflects the risk premium (or discount) assigned by the market. We are mindful of the future scenarios that markets are embedding, and to the extent that these become excessively focused on a single idea or outcome – for example, the US/China trade war that dominated trading during periods in 2019, or the daily Covid-19 infections or death rates earlier this year – then the potential for sharp shifts in prices is increased, and we can position our portfolios to take advantage of such opportunities where valuation is on our side. It is certainly not the case that “we know nothing” about the future. Nor is it simply saying that we know what valuations should be.
It is a combination of attempting to consider what the market is “thinking”, how much weight it is putting on single ideas, and how invested the market is in a particular outcome, given that we know particular outcomes rarely unfold as forecast. These conditions represent opportunities for us. This makes us contrarians in the market, and as such we must be very risk-aware in building our portfolios.
Risk control is therefore a key part of our process: we assess the risks associated with each asset, as well as a combined portfolio’s overall risk. The risk analyses we run take into account the portfolio’s exposure to macro variables like interest rate risk, currency risk, event risk, etc. We then decide if we have any unintended positions that could make the portfolio too risky or too skewed to a particular outcome. It can happen that the one scenario the market is betting on (and that implicitly we are positioned against by backing lower-valued assets) comes to pass. It can also happen that valuation signals are eroded by unexpected outcomes – for example, the destruction of property company earnings under Covid-19 such that an asset class that looked cheap was in fact expensive because the earnings were about to disappear.
In summary, at Prudential we will not take an explicit bet on the outcome of a future macro event (like the US election) or macro data because we believe that approach is too risky for our clients. Forecasts are very seldom correct, and the chance of loss is therefore too high. However, at the heart of our investment process we consider the current market valuation of an asset, which includes the risks (current and future) the market is attributing to it, and carefully assess how this might protect us against future adverse outcomes.