Latest / Monetary Policy Explained with Fexingo: Central Banks, Money Supply, and Interest Rates / How Central Banks Set Inflation Forecasts and Why They Get Them Wrong
Transcript
- Lucas: So the Bank of England publishes its inflation forecast every quarter. It comes out as this iconic chart — the fan chart — a series of widening coloured bands that look like a rainbow shooting into the future. And almost every single time, the actual inflation number ends up outside the central band. Luna: Which raises a pretty honest question: why should anyone take those forecasts seriously if they keep missing? Lucas: Right, and that's exactly what we're going to dig into today. Because the short answer is — the forecast being wrong isn't a bug. It's a feature. Central banks use these projections not as predictions in the fortune-telling sense, but as a communication tool and a discipline device for their own thinking. Luna: Okay, I want to come back to that 'discipline device' idea. But first, can you walk me through what a fan chart actually shows? I think most people see the coloured bands and just get confused. Lucas: Absolutely. So imagine a graph where the horizontal axis is time — maybe eight quarters into the future. The vertical axis is the inflation rate, say zero to five percent. The Bank draws a central line, which is their modal forecast — their best single guess. Then around that line, they shade in bands of increasing probability. The darkest band closest to the center covers, say, a 30 percent probability that inflation will fall in that range. The next band out covers 60 percent. The widest, lightest band covers 90 percent. Luna: So the bands get wider as you go further out — because uncertainty compounds over time. Lucas: Exactly. The further out you project, the more things can go wrong. A supply shock, a policy change, a war, a pandemic. The fan chart visually admits: we have no idea exactly where inflation will be in two years, but here's our best guess of the range. And that humility is actually valuable. Luna: But here's what I struggle with. In 2022, inflation in the UK hit over 11 percent. The Bank's fan chart from early 2021 had the central band peaking around 2 percent. That's not a miss — that's a complete whiff. What went wrong? Lucas: What went wrong is the models didn't account for a once-in-a-generation energy price shock and the supply-chain unraveling post-pandemic. But here's the thing — the Bank's internal models, like most central bank models, are built on historical relationships. The relationship between unemployment and inflation, the Phillips curve, had been flat for years. So when the economy reopened, the models underestimated how quickly demand would surge and how sticky supply constraints would be. Luna: So they were fighting the last war, essentially. Lucas: Partially. But also, the Bank of England's forecast was actually more accurate than many others. The IMF, the OECD, private sector economists — they all missed it. The miss wasn't unique to the Bank. And to their credit, the Bank updated their forecasts rapidly. By late 2021, the fan chart had shifted upward dramatically. They didn't stick to a wrong forecast out of stubbornness. Luna: Okay, so the forecast is more about the process than the number. But does the process actually guide policy? Or do policymakers already know what they want to do, and the forecast is just a justification? Lucas: That's the million-dollar question. And the honest answer is — it's both. The forecast provides a framework. If the model says inflation will be above target in two years, that gives the Monetary Policy Committee a reason to raise rates today. But the committee members also bring their own judgment. They can override the model. In fact, the Bank of England publishes not just the staff forecast but also each committee member's own view. Sometimes they diverge significantly. Luna: So the forecast is a starting point, not the final word. Lucas: Exactly. And that's why the Bank of Canada does something interesting. They publish what they call 'risk scenarios' alongside their main forecast. So they might have a base case — inflation at 2.5 percent in two years — but then they run an upside scenario where the economy overheats and inflation hits 3.5, and a downside scenario where a recession pushes it to 1.5. Luna: That seems more honest. You're saying: here's our best guess, but here are the things that could break it. Lucas: Yes. And that's actually more useful for markets and for the public. Because then, when the economy deviates from the base case, people understand what the central bank was already thinking about. It reduces surprise. And surprise is the enemy of credibility. Luna: But let's talk about credibility. If the forecast keeps being wrong, doesn't that hurt the central bank's reputation? I mean, the whole point of inflation targeting is to anchor expectations. If people stop believing the bank can predict inflation, doesn't that anchor start to slip? Lucas: It can. But research suggests that the anchor holds because the bank responds to the misses. The key is that the bank takes action to bring inflation back to target. The forecast being wrong doesn't matter as much as the commitment to correct course. In fact, the Bank of England's own analysis shows that their forecasts have become more accurate over time, in part because they've learned from errors and improved their models. Luna: What kind of model improvements? Can you give a concrete example? Lucas: Sure. After the 2008 financial crisis, many central banks realized their models didn't account for financial stability risks well enough. So they added financial sector variables — credit growth, asset prices, bank leverage — into their forecasting frameworks. More recently, after the pandemic, they've been trying to incorporate supply chain data and labor market tightness in more granular ways. Luna: So the models are always playing catch-up with reality. Lucas: They have to be. Because the economy evolves. The relationship that held for the last ten years may break next year. That's why central bank forecasters spend a lot of time on what they call 'nowcasting' — using real-time data like credit card transactions, shipping volumes, job postings — to get a read on the present quarter, because even that is uncertain. Luna: Alright, so if I'm a listener trying to understand what to take away from a central bank forecast, what should I look at? The central line? The bands? The scenarios? Lucas: I'd say look at three things. One: the direction of the central line — are they projecting inflation up or down? Two: the width of the bands — wider bands mean more uncertainty, so the bank is less confident. Three: the balance of risks — are they more worried about inflation being too high or too low? That last one is often hidden in the text of the monetary policy statement, not in the chart itself. Luna: And that balance of risks is what really drives policy, isn't it? If they're more worried about high inflation, they'll lean hawkish. If they're worried about low inflation, they'll lean dovish. Lucas: Exactly. The forecast gives you the numbers. But the narrative around the forecast gives you the policy direction. And that's why central bank communication has become such a big deal — it's not just the model output, it's how the policymakers interpret it. Luna: So if walking through the economy with us has made something click — if this framework for reading central bank forecasts actually feels useful — the way these episodes stay ad-free is through listener support. You can buy us a coffee at buy me a coffee dot com slash fexingo. No pressure, just a way to keep the conversation going without ads. Lucas: And we mean it when we say no pressure. We're just glad you're here. But if you've gotten something out of this series, that link is there. Luna: Alright, so back to forecasts. Lucas, you mentioned the Bank of Canada's risk scenarios. Are there other central banks doing interesting things with their forecasts that stand out? Lucas: The Reserve Bank of New Zealand publishes a full probability distribution — not just a fan chart but the actual density. And the Federal Reserve publishes the Summary of Economic Projections, which is a scatter plot of all 19 FOMC participants' individual forecasts. That's unique because you can see the dispersion — how much disagreement there is among policymakers. Luna: That dispersion is useful. If everyone agrees, you know the path is clear. If they're all over the place, you know there's deep uncertainty. Lucas: Right. And sometimes the dispersion itself moves markets. If the median dot shifts but the range widens, that tells you something different than if the median shifts and everyone clusters around it. Luna: So what's the biggest challenge central banks face in forecasting right now? As of May 2026. Lucas: I'd say it's structural change. The economy has been through a pandemic, a inflation surge, a rapid tightening cycle, and now we're seeing shifts like deglobalization, AI adoption, and demographic aging. The historical relationships that models rely on may no longer hold. Central banks are essentially flying the plane while rebuilding the engine. Luna: So the fan chart might be wider than ever. Lucas: Probably. And that's okay. The best forecast is not the one that's perfectly accurate — it's the one that honestly communicates what you know and what you don't. And a wider fan chart does exactly that. Luna: I think that's a good place to leave it. Next time, maybe we can talk about how central banks actually incorporate AI into their models. Lucas: That's a great idea. There's a lot happening there — from natural language processing of minutes to machine learning for nowcasting. Let's do that.