Attention FAE Customers:
Please be aware that NASBA credits are awarded based on whether the events are webcast or in-person, as well as on the number of CPE credits.
Please check the event registration page to see if NASBA credits are being awarded for the programs you select.

Former Bank of England Head Talks About the Limits of Certainty in Financial Predictions—Including the Ultimate Impact of COVID-19

Chris Gaetano
Published Date:
May 4, 2020
Lord Mervyn King

Lord Mervyn King served as the governor of the Bank of England from 2003 to 2013. His new book, Radical Uncertainty Decision-Making Beyond the Numbers, co-authored with economist John Kay, discusses the failure of policymakers to properly account for the unknown, and how, in their zeal to come up with definitive answers to unanswerable questions, they create a sense of false certainty that prevents us from seeing the world as it really is. He took the time to talk to The Trusted Professional about his book, as well as about current developments in an increasingly chaotic economy. The Q&A has been edited for length and clarity.

It seems that, overall, you have written a book about hubris—about misplaced faith in our models, calculations and even narratives. You heavily critique the axioms on which much economic and financial analysis rests, and urge a more empirically oriented approach. Why did you think this message was so important to convey, and why now?

I think, in a large part, this was because an awful lot of commentary on economic policy issues is based on the idea that we have models that are good at making predictions, but I think they misunderstand the value of models. We’re not against models at all, but rather, their misuse, which we think is very common and widespread.

Typically, models are useful in two ways. One is to generate valuable insight into what’s going on in the world. I’ll give an example: The efficient market hypothesis has been debated for years, and we basically assume it is right. [This hypothesis holds that share prices reflect all information and that stocks always trade at their fair value on exchanges.] But what value do we get from it? I’d say it’s the enormously important insight that, when making investment decisions, you should ask whether other people have already had this great idea and have already priced it into the market. But it is not a literal description of the world. There are many ways where the world differs from efficient markets. But that doesn’t undermine the value of the model, because it does give insights into the world, and many investors have made money being conscious of this insight.

Second, models tell you what information you need before making a decision. I think the COVID-19 crisis is a good example. On page 40 of our book—which we actually wrote last summer—we say we expected there to be an epidemic of an infectious disease resulting from a virus that did not yet exist. But that did not let us make predictions about where that would happen or when, or with what kind of virus. And so it also didn’t make sense to ask, “What is the probability of a virus coming out of Wuhan in December 2019?” because there was no way you could actually answer that question in terms of probability.

However, the epidemic models we do have are good at telling us how we might respond: We need to know how strong this virus is, which mean testing populations. We need to know how fast it’s spreading, we need to know who is spreading it to whom, and then, when it becomes too late to contain, what other approaches we can take, such as focusing on flattening the curve instead. This model gives us insight into how to tackle it, but not how to predict it, because you’d need to know the parameters for how a particular epidemic will spread, and we just don’t know that. Even now, there is still debate on the mortality, speed and spread of past epidemics we already know about.

So it’s a mistake to think of models as literal descriptions of the world versus something giving you deep insights you keep in mind as you go in the world and ask, “What is going on here?”

The points you make in your book about the inadequacy of many of our attempts to predict the future and account for all risk—were these lessons you yourself had to learn over the years? Did you once believe that everything could be reduced down to statistical probabilities? If so, what led to you change your views?

It’s something I learned over time. I started my life as an academic, and so I was doing theoretical and modeling work. But, really, during and after the financial crisis, I realized that these models were no use in making predictions. ... Most of the big problems in business and finance are unique problems, not ones that can be easily categorized like a coin toss we’ve seen 99 times before and have a pretty good idea of what the outcome will be. When businesses are making a big decision about whether or not to merge or take over another company or some other big decision, they’re not repeating something they’ve done hundreds of times before, so I think that the idea [that] you can use probabilities to capture the essence of uncertainty is deeply misleading. ... It was really after the crisis, when we sat down and talked about things, that we both independently came to the view that you can’t explain what’s going on in any of these areas in terms of a probabilistic calculus. This is a unique problem. How do we think about it?

You often speak of the futility of trying to calculate precise probabilities of complex phenomena in the larger world and say that we need to become more comfortable with uncertainty. Yet the trading algorithms that dominate Wall Street right now are run via these sorts of calculations, and what’s more, they have no context or awareness of the real-world implications of their actions, since they are merely lines of computer code. Yet firms still find these algorithms valuable because they make the firms lots of money. Are the algorithms just making the same mistakes as humans, only faster, or have they somehow found a way to deal with radical uncertainty in a way that humans have not?

I don’t think they found a way to deal with radical uncertainty. I think one of the big differences between computers using [artificial intelligence] and human beings is precisely that humans are very good at adapting and making big jumps in how they think about problems. Just look at how we’ve adapted to the virus and changed many things we do in terms of communications. This is a very good example of individuals deciding how best to use technology; it’s not computers saying to us, “This is how it will be.” We’ve taken that decision.

I think the algorithms are an example of people finding there were opportunities to front-run other traders, find misprices and spot short-term trading patterns that enable them to make money in a certain area. But there are no guarantees the algorithms that work today will work in the future. They need to be adapted and updated. They may work for a period in capturing something that isn’t easy to observe, but whether they surmise something from what’s actually happened in the last couple of months of enormous volatility remains to be seen.

One of the big problems with traditional statistical inference is that it depends critically on examining a stationary system [with fixed rules]. You could try to transform data to make them look like a stationary system, but that won’t change things. ... So let’s take growth rates. You find, for most of the postwar period, it looks like a stationary series, but when you come to the financial crisis and to COVID-19, you find that even growth rates aren’t stationary. You get jumps in how economies behave. People refer to this as shifts and shocks. The great thing about economic analysis of shifts is there’s no explanation behind it; they’re arbitrarily imposed and certainly don’t make sense, in terms of predictions.

Our world is nonstationary, which is why you can’t use past data to infer the likely frequency of future events. We start one of the chapters with the rocket fired to Mercury, and then, seven years later, it arrives exactly on schedule in the right place at the right time. NASA could do this because the laws of nature we have understood for centuries offer a very good description of how the world works. The rocket today is going to be obeying the same rules it would have 100 years ago, if they’d been able to make a rocket.

The laws governing the motion of the planets and rockets do not depend on what we believe about them or what we think will happen, but none of these are true about how an economy behaves. … Anything to do with business and finance, the rules aren’t unchanging; they change all the time. We don’t understand the system entirely—it’s so complex—but much of what happens does depend on what we believe about the future.

You say in your book that broad diversification in a portfolio, which will be robust and resilient to unpredictable events, is the best protection against radical uncertainty. With this in mind, do you view the rise of index funds as a good example of this principle?

Certainly, people have gravitated to index funds because they have come to realize that stock picking is a risky and dangerous game. Some of the most successful investors do not try to time the market. Warren Buffett doesn’t invest in anything he does not understand. He spends a lot of time thinking about his investments; he is not trading all the time. He’s thinking most of the time about whether this is a good company, whether he likes the people running it, whether he understands how it works. That makes a lot of sense. If you feel you don’t have that kind of expert information, though, there is a lot to be said for spreading your investment over a wider portfolio.

In your book, you repeatedly stress the importance of not getting too caught up in statistics and, instead, asking broadly, “What is going on here?” So, taking a page from your book, given the chaos that we are seeing in the global economy today, what is your answer to that question?

What we’re experiencing is something I don’t find helpful to call a recession or a depression because those words describe very different phenomena from the present. Those words describe situations where the private sector, businesses and families, are reluctant to spend and, because of that, demand falls, output falls and unemployment rises. That is totally different from where we are today. The government has decreed we’re not allowed to go to work, a mandated shutdown on the economy, and therefore, we should think of it in a very different way. ... What will happen to spending if and when we get back to normal is hard to judge, and it’s almost impossible to know how quickly we will get back, because there are no obvious precedents. This is a time we could reasonable say is different, and we don’t know how people will respond.

Suppose government relaxes the restrictions tomorrow. A lot of people, I think—including myself—will be very cautious about going out to eat in a crowded restaurant until we felt there was the possibility of some sort of treatment or vaccine.

But these are things for which there is no basis from past data to form judgments. So I don’t think economic models give us much guidance for the future. Whatever governments say or do, we just don’t know how nervous people will be, how reluctant they will be to go back to work. Many with low incomes will be desperate to do so because they have no other sources of income, but others being paid by their employers or receiving pensions may … still social distance themselves on their own initiatives, without guidance from the government. So these are the unknown things.

I think it’s important to get away from the idea that we can use traditional methods of understanding recessions and, instead, start to think of the consequences of the government intervening to freeze the operations of a market economy. What will happen to businesses? Can we afford to let them go bust? What businesses will be left to expand output and go back to work once restrictions have been lifted, if we do allow them to go bankrupt in the first place? These [questions] are what we need to be thinking about. Then, of course, there is the cost of the lockdown in terms of the well-being and health of the population. It’s a very serious issue, one to which, perhaps, not enough attention is being paid.

When last we spoke, in 2016, I asked about the effects of maintaining ultra-low-interest rates—or even negative-interest rates—for years on end. You said that “when you transfer spending from the future to the present, you dig a hole—time passes, and the future becomes today. So, now you cut interest rates again to bring even more spending forward, and that digs an even deeper hole. As time passes, that too becomes the present, and if you haven’t tackled the underlying problem, you create more and more of an incentive for central banks to cut rates further.” Right now, we are seeing the corporate bond market in shambles, as companies that had been able to take on debt at nearly zero cost are faced with the bill coming due. Do you think the risks you laid out during that  interview are finally coming to pass? Overall, how do the points you made in your previous book, The End of Alchemy: Money, Banking, and the Future of the Global Economy, apply to today’s economic crisis?

 I think the arguments I put forward in that book—and put forward again in a speech to the [International Monetary Fund] last October—, made clear that the world economy has not recovered because we’re still in a basic disequilibrium, and there was no way central banks could get us out of it. It requires a bigger adjustment in the world economy. One way it could come about would be through a further crisis caused by a significant default on debt. I stick to that. I think we’re beginning to see that this virus is the trigger that is likely to lead to significant defaults on debt.

One of the problems of dealing with proposals for debt forgiveness or suspension on debt repayments is that it makes it much more likely that those periods of debt forgiveness or suspension are turned essentially into defaults or genuine debt write-offs, and I think that will have major repercussions on all pillars of the economy, especially the banking sector, which we thought was in much better health. But at the end of this process, a lot of the defaults will fall on the banking system, and the need for capital will be larger than the banks are prepared to admit.

The Federal Reserve’s main strategy, it seems, is to do what it can to keep borrowing costs low so that credit can continue flowing, so that bonds can continue to be issued, so that firms can keep taking on debt. To what degree do you think this is sustainable? Are we just pushing the reckoning back a few years? Or is there something going on that is addressing the core problem?

I think it’s important to divide what’s going on in two columns: The first is that, if you mandate a shutdown of the economy, the government has a responsibility to ensure lending is carried out with government guarantees to make sure businesses, big and small, keep going. They can’t sell their goods now, but we need them to still be there when we start the path back to normalcy. I think a massive loan program is the only answer to that.

But looking further down the track, if that’s all we do, in a year or two’s time, we’re going to be in the same position we were before the virus hit, and we still will have disequilibrium and too much debt, and the potential for another crisis will still be there.

It may be that these things can be conflated. As we get through this crisis, a lot of debt will just be written down, and it would be a long-run benefit if we could find a way to write down a lot of that debt. But it’s going to be extraordinarily difficult to navigate both the short-term problem of massive loan programs backed by the government and the long-term problem of finding a way to reduce the debt burdens of the private sector, which they’ve inherited from the past. Debt was higher relative to GDP [gross domestic product] at the beginning of this virus problem than it was in 2007, and that is not a happy position to be in.

The massive aid package passed in the United States will add trillions of dollars of debt to the U.S. government. However, the Federal Reserve also said it plans to buy effectively limitless amounts of U.S. Treasury bonds, the proceeds of which will eventually be sent back to Treasury. As a result, it would seem that these bonds never existed in the first place. Is the Fed just magicking this money into existence? And, if so, how sustainable is this?

If it does this indefinitely and to an unlimited extent, we will all end up with significant inflation. But I think what’s happening is that governments know their expenses will rise sharply over the next three to six months, and they want to smooth out the timing of the issuance of this government debt, ... allowing the government more time to issue debt to the private sector, and, if necessary, the central bank can buy all [this debt] back.

The key here is not to worry about the mechanism by which money is created. When central banks buy government debt, they create money. It’s not the mechanism that matters, though, but who is making the decision, and if the central bank is doing it and can determine the amount of money it wants to create to ensure there is continued growth, reasonable steady and with low inflation, there is no problem.

If, on the other hand, in a year or two, the government puts so much pressure on the central bank to print money to buy whatever the government wants through purchasing government debt, then it would lead to inflation down the road. The question of whether this is going to happen is hard to judge now, but at present, there is no reason to assume the independence of the Fed will be challenged. The government, of course, can pressure individuals and nominate the wrong kind of people to the Federal Open Market Committee to try to influence what the Fed does. The key thing is that there is a high degree of political consensus to allow central banks to determine how much money they print to keep inflation under control.

I think there are no new monetary policy instruments there. This is not about the mechanism of printing money. This is about who makes the decision.

Much as in the last crisis, it seems that, in this one, we are seeing central banks take unprecedented actions to shore up the economy. With the Fed’s current plans, through undertakings such as the Main Street Loan Facility and its corporate bond-buying program, is the line between fiscal and monetary policy getting blurred here? And if so, what do you make of such blurring?

It is being blurred, and I think this is, of course, a concern. As long as central banks buy government debt and bonds, I wouldn’t be concerned. But as soon as the central bank starts buying private-sector instruments or instruments issued by some localities but not others, then it is getting into the business of something that is really fiscal policy.

That kind of operation may be very sensible to do, but it ought to be something decided by governments, with the central bank as an agent. That’s why it’s important for central banks and governments to work closely together in situations like this, because the central bank can be the agent in implementing something the government needs ... So if the central bank is seen making judgment decisions on which private-sector securities to buy, which companies to help and which not to help, I think that is a danger, because once the episode is over, politicians will say, “Well, that’s interesting. You made judgments favoring one sector over another, one company over another. So we want to have more control over you, the central bank.” That worries me.

Click here to see more of the latest news from the NYSSCPA.