November 13, 2023

Bank of England review into forecast failings must focus on the people, not just the models

“Central banks have come under fire for missing their inflation targets, but at least as much because their inflation forecasts have really missed outturns.” This is a quote from a recent speech by Catherine Mann, Monetary Policy Committee (MPC) member. It is a position that seems to be shared across the Bank, given the terms of reference for the ongoing review of the MPC’s recent performance, which focuses on the technical aspects of their forecasts and processes.

This approach is misguided and will do nothing to prepare the Bank to perform better in the next seismic economic event. To suggest that the MPC’s failings in recent years are purely down to their forecasting models shows a lack of understanding of the purpose of using them.

As Andy Haldane, former BoE Chief Economist, stated in a 2021 speech about the economic impacts of the pandemic, “There are few if any, historical precedents to help judge the response of the economy to this scale of shock”. But that is exactly what forecasting models do. In simple terms, they use historical data inputted into some set of equations to predict the future.

Given the complexity of the global economy, every economic event will be unique in at least a nuanced way, which would lead to a forecast of it to be imprecise to some degree. It doesn’t help that the Bank of England admittedly only take into account the last “thirty years or so” in their forecast models, which misses the period of high inflation in the 70s[1].

It should have been obvious to the MPC at the time, therefore, that the forecasts would be highly likely to be largely inaccurate over this period. In this situation, a greater weight must be placed on the forecaster’s judgements as to what the outlook is.  

Judgements are, in fact, proven to be very useful in forecasting, with a plethora of studies proving that using judgements generally improves estimates[2]. This shouldn’t be surprising; while models are hindered by the historical data that is used, a good economist can think about a problem more theoretically and flexibly to navigate the uncertainty in a way that a model cannot.

However, as inflation was rising, the MPC’s judgements were wrong. They failed to react quickly enough when inflation kept on rising beyond what their forecasts were predicting. While some of this was down to unpredictable developments such as Russia’s invasion of Ukraine, the Bank generally were too slow to realise that their forecasts were not accurately predicting where or when inflation would peak. It is worth mentioning that Cebr’s judgements were much more accurate, which our CEO discusses in a recent Forecasting Eye[3].

That doesn’t mean the Bernanke Review is a waste of time. It is right in any case to constantly review and develop the tools at your disposal, given the ongoing development in both technology and our understanding of how the economy works. But to not consider the failings of the MPC’s judgements seems short-sighted.

This is not to say that we should expect the MPC to perfectly judge the economy at every point, but we should investigate why judgements turn out to be wrong and what can be done to support the MPC and other economists to learn and improve.

One such enhancement could be to look at the role of bias in judgement formation. An example of this, which could be of particular relevance for past few years, is recency bias. When faced with the highly uncertain picture in which it could have been reasonably argued that inflation will be much higher or much lower as a result of the pandemic, it is likely that their recent experiences played a part in their decisions.

Remember that, just before the pandemic, central banks across the world were dealing with interest rates close to zero. This gave them fewer weapons in their arsenal to support the economy in the early onset of the pandemic, when more stimulus was needed. The BoE were actually in the middle of a review around what creative new tools they could use when rates hit zero, at the point when inflation started to pick up.

Set in this context, it seems likely that recency bias played a role, perhaps only subconsciously, towards MPC members being overly dovish.

The Bernanke Review must look specifically at the role of judgements in the forecasting process. It should look at how biases may affect these and what can be done to mitigate the risk of bias. While everyone on earth has biases, it can’t help that the MPC deeply lacks diversity in terms of educational and career background and broad economic points of view.

This lesson must be learned fast to avoid the MPC making the same mistake twice in quick succession. In the current uncertainty of when or if to start monetary loosening, they risk again placing too much weight on recent history. Indeed, Catherine Mann specifies in her speech that she would rather monetary policy be overly tight rather than overly loose.

The Bank must not be biased by their recent experiences of overly high inflation and neglect, subconsciously or otherwise, the real downside risks that are present for the inflation outlook. Indeed, there is an argument that the natural interest rate (which is extremely difficult to measure) remains near zero, given ongoing long-term demographic changes such as the ageing population.

If the Bank does choke off inflation longer than is necessary, it might have to embark on an overly fast or aggressive monetary loosening campaign, adding in more unnecessary volatility in the economy and perhaps even landing itself in the situation it had previously where rates are dangerously low.

[1] House of Commons (available here)

[2] McNees, 2017 (available here); Wright et al., 1996 (available here)

[3] Cebr

For more information, please contact:

Christopher Breen, Head of Economic Insight
Email: cbreen@cebr.com, Phone: 020 7324 2866

Cebr is an independent London-based economic consultancy specialising in economic impact assessment, macroeconomic forecasting and thought leadership. For more information on this report, or if you are interested in commissioning research with Cebr, please contact us using our enquiries page.

The site uses cookies, as explained in our cookie policy. If you agree to our use of cookies, please close this message and continue to use this site.

Accept & Close