Paul Diggle

Hello and welcome to Macro Bytes the economics and politics podcast from abrdn. My 
name is Paul Diggle

Luke Bartholomew 
And I’m Luke Bartholomew. 

Paul Diggle

And this week we want to talk a bit about the nature of economic forecasting and scenarios. How we think market participants can navigate a world filled with uncertainty by thinking through alternative scenarios and preparing for lots of different possible outcomes and the trigger for that conversation really is twofold.
First, Ben Bernanke, the former chair of the US Federal Reserve, has just authored a review of the Bank of England's forecasting record, particularly around inflation, and he's given a series of recommendations for how it could do better in future and thinking deeply about alternate scenarios is part of his list of recommendations. And then secondly, because the world of macro markets at the moment is just currently very full of risks and uncertainties.

Now it's definitely a well-worn cliche that the world is more uncertain than ever. Almost every forecasting house’s year ahead outlook document has said that for years, if not decades. And of course, it can't always be true. It's a form of recency bias to always think the world is more uncertain than ever. Nonetheless, two particularly prominent uncertainty stand out at the moment.

One is inflation. Following a series of stronger than expected inflation data, particularly in the US market, and central banks are grappling with quite how difficult the process of lowering inflation back target is going to prove, how long interest rates will need to remain elevated to bring that about. And lots of our recent podcast episodes have been discussing exactly that sort of question.

And then secondly, Middle East security and geopolitics. We're recording this in the days after Iranian attacks directly on Israel. It's clearly a critical and worrying juncture for the whole world but including for financial markets. And we don't know quite how the situation would have evolved by the time you're listening to this. But the way we're trying to think about it right now is through a scenarios lens. There are different alternative scenarios of escalation, de-escalate and so on. So the question we want to answer is how should we think about and navigate macro uncertainty? 

Luke Bartholomew 

And I think the best place to start then is with the Bernanke review itself. And Paul, just to credential Doctor Bernanke a bit further, not only was he Chair of the Federal Reserve, he's also a Nobel Prize winner in economics for his work with co-authors on banking and financial crises, so extremely well qualified to come at that review. And I think the review is interesting for several reasons. First, just in its own terms, of course, we discussed with Michael Saunders on this podcast back in August aspects of the review and what we thought was likely. But then also because I think thinking about the review opens up some interesting avenues for further thoughts around uncertainty and forecasting as well.

But I guess the context for the Bank of England review is this sense that it had made at least one, or maybe several, big forecasting errors over the last year. The big one, of course, being around inflation, which is the bank's responsibility. It has an inflation targeting mandate. But then also as well, it had high profile forecasts for a deep recession about 18 months ago around the energy price shock that never occurred as well. And I think the Bank internally was criticizing itself a lot for these forecasts, but also faced quite a lot of political flak as well. And I think announcing this review has being helpful on both those fronts. And in particular, I think what the Bank has felt is its communication strategy during this time has been what's been most challenging, because in some ways its forecasting record is no worse than a lot of other private sector forecasts or indeed peer central banks.

But the bank's credibility seems to have been more damaged around this, and that's showing up in inflation expectations a bit higher in the UK than elsewhere. And of course, credibility is a key aspect of modern central banking. So, I think the Bank of England thinks if it can communicate better, then it has a better chance of maintaining its credibility with the public. And in particular what it has used is these fan chart-based forecasting methods. Previously in which the bank forecast a modal view, its most likely view on how inflation and other variables will behave if the market implied path for policy rates is followed and then has a fan chart around that. So quantifies the degree of uncertainty with a distribution of potential outcomes around that modal baseline. And over time the fan grows wider with the idea, you know, that there's more uncertainty in the future relative today. It’s hard as know where inflation will be in three years’ time compared to, you know, the next month or so. 

So that was the that was the context going into the review, and I guess the key thing that the Bernanke review suggested was a scrapping of these fan charts, that it led to quite cack-handed communication techniques that weren't really well understood by the general public. It was difficult to interpret quite what the bank was trying to say about the market implied path of policy and whether the Bank of England agreed with it or not.

It required some understanding of probability distribution functions to work out exactly how the fan chart was being constructed. So that was a key aspect of the report. Another was just to put in more resources into forecasting - both human and capital investment. So, more money into the software that's behind forecasting, upgrading skills and expertise of the staff and trying to integrate different forecasting models.

And then with this upgrade in forecasting infrastructure, another key aspect of the review was for the Bank of England to make use of explicit scenario analysis as part of its communication techniques - and not just the base case, but also other scenarios around that base case as to how the world might evolve. 

Paul Diggle

And you're a keen Bank of England watcher yourself, Luke. Tell us how exactly the Bank in the past it has communicated its own expectations for the course of interest rates by conditioning on the market course. Because that's confusing even for people deeply involved in this to understand, let alone when they communicate with the general public. 

Luke Bartholomew 

So, the Bank of England has a mandate to have inflation at 2% over the medium term. So, there's a sense in which if it's fulfilling its mandate, inflation over two- or three-years’ time should be at 2%. So, when the Bank of England is not forecasting inflation at 2%, either it's higher or lower than 2% over that horizon, that implies the Bank of England has some disagreement with the market path that these forecasts have been conditioned on. So, if inflation is higher than 2% over several years, that implies the Bank of England thinks that policy is going to be tighter than the market has priced in. And equally, if inflation is below that 2%, that implies the Bank of England thinks policy is going to be more accommodative than the market has priced in. So, as I say, it's a somewhat cack-handed way of getting at this. It never explicitly says, it just sort of says, well, we're not going to be meeting our mandate on the basis of what you the market have priced in, so something somewhere has got to change for us to meet the mandate and that is on you, Mr. Market, to update accordingly. 

Paul Diggle

Yeah, it's a pretty roundabout or hall of mirrors type of way of communicating their expectations, but of course Bernanke, in his review, stopped short of recommending something like the Fed’s dot plots where Bank of England officials would be explicitly saying what their own expectations were.

He thought that was perhaps too big a change, at least in the near term. But clearly this recommendation of using alternative economic scenarios is a big part of the review. It's a big part of what the Bank of England could start doing before too long. And I think it's important Luke to make clear quite how difficult forecasting is. I mean, there is this  saying ‘forecasting is very difficult, especially about the future’ variously attributed to Mark Twain, Yogi Berra, a variety of other people who always have these kind of humorous quips. And in fact, what I really love about that quip is that economists do also make forecasts about the present and the past. These are things like nowcasts and backcasts, used to figure out what the economy is doing right now or how it's performed  in the recent past. There's a big business in providing estimates of what economic data releases relating to months just gone are going to show, and they're often wrong. So even forecasts about the past and about the future are difficult. And I also think there's something in this idea that forecasting GDP, inflation, interest rates is the most dismal end of the so-called ‘dismal science’ of economics.

It's actually not what most economists do. They're primary doing a mixture of economic history, about trying to describe the functioning of the economy, making policy prescriptions, whether macro when it comes to fiscal, central banking or micro, things like competition policy. Now, of course, policy prescriptions are a form of prediction, but it's not quarter-to-quarter numerical forecasts of GDP and inflation and the like.

That said, before I get too far in talking down the role, the importance of, market economists, what can we do, what can investors do, in the presence of uncertainty? What can the Bank of England do? And I agree that with the Bernanke review, scenarios must be the answer to that. No one really seriously expects, you know, one two decimal place baseline forecasts to be exactly correct and nor are baseline forecasts all that we care about. Markets price a probability distribution of the future rather than a single outcome. And often large moves in markets are about shifting relative weights to alternative scenarios. 

Luke Bartholomew 

Now, I mean, for all we are very big fans of scenarios, Paul. I mean, it is probably also worth saying that maybe we're a little bit less down on the fan charts than seems to be fashionable at the moment. In some ways they are quite nice numerical ways of capturing the degree of plausible uncertainty around the base case and make use of the way in which past forecasts went wrong to say something about sort of uncertainty in the world. So, I think when these fan charts were first introduced, they were quite revolutionary as communication and forecasting devices. But yeah, clearly the Bernanke review thinks that they are now past their sell-by-date, and that's partly, you know, for those as described somewhat cack-handed communication aspects of the fan charts. Also, because I think he feels that too often, for all that these are meant to be relatively rigorous quantitative assessments of uncertainty, there tends to be a degree of sort of qualitative override that policymakers will sometimes do and just skew the distribution a little bit just on the basis of their perceptions of the risk - rather than what the model is spitting out. And it sort of sometimes leads to the worst of both worlds if you have that approach. 

And then finally, and this gets to part of the reason that we like scenarios, but also what other people I think, when they think about risks like to do, is think about very specific states of the world, plausible actual outcomes rather than statistical distributions and what scenarios are in that sense then are sort of the internally consistent set of events or the causal path that led to those specific quantitative outcomes rather than some disembodied number that fits somewhere within a statistical distribution - but it's not clear what it is about the economy that has generated that outcome within the distribution. Why it's that number rather than any other you can't tell from the fan chart, but you can tell from a scenario. It has a clear causal path that's got you there. And the scenarios can then have associated with them waymarks or indicators, some way of assessing whether that scenario has become more or less likely over time.

Paul Diggle

Yeah, at any one time, we are running and maintaining five, six or so all-encompassing global macroeconomic scenarios. So right now, we have a baseline scenario which is something like a gradual slowing in the global economy, a gradual moderation in inflation and an eventual rate cutting cycle to really boil it down - as a baseline, but we've also got scenarios like the so-called ‘no landing’. This term is a part of the market parlance at the moment, meaning a continued period of very strong growth, particularly in the US, inflation that remains well above target and within that context, perhaps very limited, if any, scope for an interest rate cutting cycle. So, we've got a ‘no landing’, we've got a ‘Middle East war’ – so a full-blown conflict there, leading to an oil price shock and associated upward inflation and downward growth impacts, a scenario in which US and indeed global productivity growth is very strong. It has been very strong over the past year. Perhaps that's an early gain from the AI revolution. Perhaps it's very strong inward migration and other changes in the labour force. It would allow very strong growth to exist alongside lower inflation. It would mean long run equilibrium interest rates are higher. We've got a scenario which involves past monetary policy tightening, tipping the US economy into a recession, and then we have a scenario in which China suffers something like a balance sheet recession. There are big headwinds to growth there. So all of these are trying to describe ways in which key judgments underlying the baseline view of the world could be incorrect. And a thing we do alongside these global macro scenarios, is run a whole set of ad hoc scenarios framed around event risk, often political event risk. So alternative ways in which volatility in the Middle East could play out would be an example, or alternative ways in which the US election could play out -different combinations of who wins the White House and who wins the House and the Senate. And the baseline then it's just another scenario within that kind of distribution. It’s the modal. It's the most likely one. But importantly, it's not the be all and end all of  an economic view. And I should also note that a very common approach that a lot of investors have is an up, down, and base case. So, a three-scenario type of approach. I do wonder if that sometimes misses the nuances in quite how the global macro environment could evolve along several different dimensions. For example, if one has a baseline forecast for inflation and then a high inflation and a low inflation scenario. That could be the sort of thing the Bank of England eventually does. While it matters if the source of higher inflation is, say, a geopolitical oil price shock that would weigh on economic growth, or a domestic growth shock, stronger domestic growth, that would push upwards on activity growth, or it matters if the low inflation scenario is a negative demand shock -  so something like a recession, or a positive supply shock, something like the US experience last year. So, I think actually a multi-scenario approach is the way to go to tease out quite how complicated some of the key judgments about the global economy are. 

Luke Bartholomew 

And so, I guess with those multiple scenarios Paul, what you might think that you are modelling, in the language of Donald Rumsfeld, is something like the ‘known unknowns’. What scenarios can help you do is understand sort of exogenous shocks that you can then model. They're not what you think is going to happen in your baseline, but they are plausible risks out there that you’ve identify, right? You've thought through what the consequences are. And I think scenario analysis is very good at that. 

But then to extend the Rumsfeldian language, there is, of course, also famously ‘unknown unknowns’. And these are risks out there that, you know, you don't think to model – ‘black swans’ as they're sometimes called. And, you know, the nature of scenario analysis is that you are never going to be able to figure out all the potential risks out there, that there will be aspects of the distribution that you don't cover. Black swans do exist, and that sort of one source of uncertainty that doesn't go away even as you bring scenario analysis in. 

And then the other source of uncertainty in your forecasting, I guess, is sort of the model uncertainty itself - the process that you use to think about how the economy works, that you think through causal chains could be faulty in some way, and whether that sort of the intuitive model that you carry around in your head or some more explicit model that occupies spreadsheets, lines of code or whatever it might be, you know, the kind of thing that the Bernanke review suggested that the Bank of England needed to invest more in. A classic example of where that kind of model uncertainty could exist in the economy, and perhaps it's been a big part of the inflation experience over the last couple of years, is its relationship, called the Phillips Curve, which is the connection between the degree of slack in the economy and inflation and the kind of models that many forecasters were using implied that the Phillips Curve was relatively flat. That is to say that even quite big changes in slack wouldn't lead to a big change in inflation. As is,  it’s seemingly being the case that the Phillips Curve is very steep, that we've got quite big changes in inflation as a consequence of, you know, smaller swings in activity than you think would be necessary to generate those kind of inflation outcomes.

And famously, the Phillips Curve has been an example in the economic literature of precisely this kind of model uncertainty for at least 50 years or so. It's what Robert Lucas, another Nobel Prize winning economist, had in mind when he came out with his famous critique around policy making and modelling and it’s been behind how economists have tried to structure their models, as I say, for the last 50 years, and the fact that we're still grappling with this sort of modelling error, perhaps tells you how difficult it is to get away from these kind of concerns. 

But that's only one source of potential model error. When we were talking with Michael Saunders previously, he suggested, you know, maybe there have been changes to the household transmission in the way in which monetary policy affects spending as the shift in mortgages away from variable rates towards fixed rates that have changed how long it takes, the shifts in interest rates to affect spending. And that wasn't captured appropriately in the Bank's model. Or maybe the model had assumed too much that inflation expectations were well anchored, and as that turned out not to be the case, suddenly the Bank's forecasts went very wrong. So again, there was just something wrong with the structure of the model itself. 

And then we can think of ways in which, you know, in the future, the structure of the economy might be changing that makes models difficult. The big fashionable thing to do, post-financial crisis, was to integrate a financial sector into models. And arguably we have only made limited progress in that respect. There's also been a lot of work in trying to think about distributional questions and how changes in economic distribution, inequality, both of income and wealth, might have affected the way in which monetary policy works, and that is still very much a work in progress. Supply chains have, of course, become a big focus as well, and again, perhaps part of the difficulty in the last couple of years has been that economic models have lacked an explicit way of thinking about how disruptions in some sectors of the economy can affect others through interlinked supply chains. And finally, climate change as well might be doing things to change the structure of the economy such that the way in which our models are written down need to change beyond just doing scenario analysis around those base case outcomes.

Paul Diggle

Yeah, so many potential sources of uncertainty. But if you're going to use scenarios to describe possible future states of the world, something that comes along with that is needing to attach probabilities or send some kind of signal about the relative likelihood of the baseline and alternative scenarios. And again, there are multiple ways to do that. 
Economists can use quantitative modelling to help inform scenario probability. So things like the slope of the yield curve or a broad range of leading economic indicators can give you model-based quantitative recession probabilities. They're only one guess, but they are a kind of quantitative approach to doing it. Economists can also draw on historical experience when informing the balance of probabilities for different scenarios. So, you might look at past oil price shocks, the broad inflation environment of the 1970s as a comparison to certain risks that exist in the global economy today. But equally sometimes historical experience can lead you astray. For example, it might be that the experience of low inflation anchored inflation expectations of flat Phillips Curves after the financial crisis, as you were saying Luke, left economists complacent to the risks of an inflation breakout. It could be that we had limited past historical evidence for the effect of pandemics in modern economies. In the end, and what I suspect is perhaps widely underappreciated, is that navigating macro uncertainty always requires a big dollop of judgment. And that's why applied economics is as much an art as it is a science. 
And then the final thing perhaps as a parting thought, Luke, is that learning from past mistakes, reviewing forecast errors, is a critical part of improving the way one thinks about alternative scenarios. But perhaps I can end on another quip, which is from the economist Stanley Fischer. He was governor of the Bank of England, Bank of Israel. He was Deputy Governor of the US Fed under Janet Yellen. And he once said ‘Never look back at your forecasts. You may lose your nerve’. So, it's certainly not an easy task to  do forecasting. The kind of errors the Bank of England made were understandable. But having an alternative scenarios approach must be a good response to that sort of uncertainty.
 
Luke Bartholomew
 
Well, yeah, just to take this conversation full circle, I think Stan Fischer was in fact Ben Bernanke's teacher at MIT as well back in the day. So, it is all connected, but I guess that is as good a point as any to leave it. So as ever, let me please remind you to like and subscribe wherever it is that you get your podcasts. And all that remains is to thank you all for listening. So thanks very much and speak again soon.


This podcast is provided for general information only and assumes a certain level of knowledge of financial markets. It is provided for information purposes only and should not be considered as an offer investment, recommendation, or solicitation to deal in any of the investments or products mentioned herein and does not constitute investment research. The views in this podcast are those of the contributors at the time of publication and do not necessarily reflect those of abrdn. The value of investments and the income from them can go down as well as up and investors may get back less than the amount invested. Past performance is not a guide to future returns, return projections or estimates and provides no guarantee of future results.