AIQ speaks to economist and author John Kay about risk, uncertainty and the longer-term implications of the coronavirus pandemic.
John Kay is one of Britain’s foremost economists. His long and varied career has spanned academia, policy and the corporate world. He has served as director of the Institute for Fiscal Studies and the Saïd Business School at Oxford University. In 2012 he produced an influential report for the UK government that drew attention to the damaging effects of short-termism in the stock market.
Kay has communicated with a wider audience through his long-running column in the Financial Times and a series of books, including the prize-winning Other People’s Money (2015), which examined the relationship between business, finance and the wider economy.
Kay’s latest work, co-authored with former Bank of England governor Mervyn King, is Radical Uncertainty: Decision-making for an unknowable future. The book was hailed as eerily prescient when it was published in March 2020, just as COVID-19 spread across the globe. The pandemic supports Kay and King’s thesis that the world is too unpredictable to model precisely; probabilistic risk management techniques are no match for the topsy-turvy contingencies of real life.
In this Q&A, Kay discusses the thinking behind the book, the longer-term implications of the coronavirus crisis for business and finance, and how companies can adapt to conditions of radical uncertainty.
Your book draws a distinction between risk and uncertainty – why is it important to distinguish the two concepts?
In 1921, two large books were published on this subject by John Maynard Keynes and Frank Knight. What they meant by risk were things that could be described probabilistically, whereas uncertainty referred to things that couldn’t be defined probabilistically.
What’s happened since then – and the financial sector is the extreme example – is that the distinction has effectively been elided, and the historic definition of risk and uncertainty is no longer seen as relevant.
Risk is when something bad materialises. Uncertainty, on the other hand, can be good or bad
We strongly dispute the contention that all uncertainty can be described probabilistically. We choose to define risk and uncertainty in the way ordinary people do. Risk is when something bad materialises. Uncertainty, on the other hand, can be good or bad; when you go on holiday and try a new restaurant, or meet new people, you don’t know what’s going to happen. It might be pleasant, or it might not. Risk arises when something jeopardises your reference narrative; the way you thought you were going to live your life.
Your book focuses on “radical uncertainty… a world of uncertain futures and unpredictable consequences, about which there is necessary speculation and inevitable disagreement”. Is COVID-19 an example of this?
COVID-19 is absolutely an example of radical uncertainty. The pandemic is not what Nassim Taleb calls a Black Swan, an event you can’t anticipate because you can’t imagine the event. You definitely could imagine a pandemic; indeed, we somewhat presciently wrote in the book that a pandemic would happen. But we didn’t know when or where, and we certainly couldn’t have sensibly responded to the question: “What’s the probability a global pandemic will start in Wuhan in December 2019.”
Are you confident policymakers will be able to deal with the economic fallout from COVID-19, even after the health risks have been addressed?
No. The health and economic risks are bound together. There seems to be a widespread belief that, before long, we will be able to announce the health risk is over and we can get back to normal. It’s not going to be like that; the most likely scenario is that this continues in one way or another for the next one to two years. We can’t be confident in policymakers’ responses, as we simply don’t know how this virus will evolve and what the economic consequences are going to be. Our argument is that we should stop pretending to have more knowledge about the world than we actually do.
You write that at times of radical uncertainty, decision-makers should ask the question: “What is going on here?” Why is this question so important, and how would we go about answering it under the current circumstances?
In business, politics and finance, you’re repeatedly confronted with unique situations. Even if the existence of a pandemic is not a unique situation – and it isn’t; it is something that has happened before and will happen again – this pandemic has unique features. You need to recognise that, and by asking “what is going on here” you can address the whole context of what is happening.
The big failure so far has been the failure to gather the information we need to make sensible decisions
To my mind, the big failure so far has been the failure to gather the information we need to make sensible decisions. We haven’t had the testing capacity to provide that information. There is now a discussion around random testing, and that is terribly important because we don’t know how many people have the virus and how many people who do contract it suffer serious consequences as a result.
Would you point to an example of an institution or industry that is managing risk and uncertainty well?
The food retailing sector has responded fairly robustly and effectively to the current crisis, whereas a lot of other business sectors have been shown to have supply chains for which even a slight disruption creates problems.
There is a big set of issues there. To protect yourself against this kind of disruption, you have to find a structure that is robust and resilient, and for that you need to have what engineers would think of as “modularity” – i.e. a system built in such a way that when one part fails it doesn’t bring down the whole system. You also need redundancy, which means not trying to run things with the minimum margins of safety you can get away with.
In most of business – particularly in finance – we’ve tended to regard these kinds of things as signs of inefficiency. The siloing of financial services activities was effectively abolished in the 1980s. Since then you’ve had banks and insurers talking about “surplus capital”, as if it’s possible for financial businesses to have too much money.
Could the crisis lead to positive change, if companies are motivated to shift from “just in time” supply chains to “just in case” models? Or will we revert to business as usual as the pandemic recedes?
It will be somewhere in-between. People have these sorts of conversations, but the short-term pressures on management will be the same after the crisis. If a chief executive’s tenure is five years, they may think, “Is something going to happen on my watch? Probably not.”
Your book draws attention to bogus probabilities and flawed algorithms. Why are computer-based models ill-suited to conditions of radical uncertainty?
Because the models are constructed by people who assume they have knowledge that they don’t have and couldn’t possibly have. We talk in the book about the failure of risk management in the financial sector. During the financial crisis, [former Goldman Sachs CFO] David Viniar famously said: “We were seeing things that were 25-standard deviation moves several days in a row.” Which of course isn’t what happened: what he meant, or should have meant, was that this series of events looked impossible based on the Goldman Sachs model.
To derive a probability about the world you would have needed to take a probability based on the Goldman Sachs model, then multiply it by the probability the Goldman Sachs model was in some sense true; you couldn’t have known what the latter probability was, and it was clearly very low.
You cannot derive a probability about the world from a probability that’s developed in a model
The lesson is that you cannot derive a probability about the world from a probability that’s developed in a model. The database with which Goldman Sachs built its model came from a period in which banks didn’t go bust.
How about the implications for investors today? How can they ensure their portfolios are resilient?
Robustness and resilience are about diversification in large part, particularly in an investment portfolio. In any crisis, there are always some things that do well, as well as some that do badly. If you simply look at the dispersion of individual stock movements over the last few months, as against the average movement of the whole, you can see that: an investment in Amazon has worked out pretty well, for example; an investment in airlines has not. But you couldn’t have anticipated the precise nature of this crisis, and a financial crisis would have had very different effects.
Investment firms face a dilemma: they have to maximise returns for clients while allowing companies the space to build resilience against uncertain events, perhaps through the kind of investments that won’t show up on a quarterly earnings statement. How can they strike the right balance?
Investment intermediaries, asset managers, have the problem of being accountable to financial advisors and investment consultants who are constantly engaging in these kinds of very short-term comparisons, which will not demonstrate the advantages of widespread diversification. And widespread diversification is not something you approach by calculating betas in the way portfolio models typically do, but by asking the “what is going on here” question, and by understanding the underlying determinants of asset price returns.
One lesson is to understand that reducing risk is not the same thing as achieving certainty, and that has huge implications for portfolio management and planning. I sometimes say that someone who knows he is going to be hanged tomorrow has certainty but not security. And that may sound like a joke, but when you look at pension funds that, either collectively or on behalf of individuals, are largely invested in bonds, you see that’s more or less precisely what they are doing – offering the certainty of a low standard of living in retirement. That’s not risk management.
Could any of the painful lessons learned during this crisis be applied to avert other global threats – climate change, for example?
Only if we look at this crisis in a way that generates general lessons rather than specific lessons. If you take the example of climate change, there is quite an exaggerated faith placed in climate models that have all the characteristics of the bad models that I’ve described. The best approach is to recognise we don’t really know what’s going to happen, and therefore we need to have strategies that are robust and resilient. We basically need to be doing the equivalent of buying options, which is a matter of looking at fundamental technologies. We shouldn’t be paying attention to people who claim without foundation that they know what’s going to happen to the climate.