The end of a calendar year and the start of a new one provides a chance for reflection.
For the UK’s economic establishment, particularly forecasters, introspection is more appropriate. Complete with their New Keynesian models, most have been proven utterly wrong so far that voting for Brexit would lead to a short-term downturn. As the most significant economic event since the financial crisis, this is some failure.
Much proverbial ink has been spilled in recent weeks about what this “failure to forecast” tells us about the underlying economics. Was it simply some faulty assumptions or fundamentally misguided models? Did modellers take for granted things that are as yet unknowable? Are their models better for short-term analysis but just bad for the long-term assessment in hand? Some of the country’s top economic commentators and forecasters are currently jousting on these questions via email. Sadly, those whose forecasts have proven so faulty have not offered up any rigorous assessment of why they failed.
I do not claim to have all the answers either. But the lack of humility has been astonishing. In a recent Times article defending the Treasury and other forecasting bodies, David Smith explained away the Treasury’s mistaken prediction of recession by claiming they had not appreciated the Bank of England might act, nor that Article 50 would not be triggered straight away.
This is dire sophistry. That these two things could swing an economy from immediate recession to sustained growth goes against most received knowledge about the time lags through which monetary policy operates and the forward-looking nature of consumers and investors.
If we really want to improve policymaking, far more self-critical analysis needs to be undertaken on both the assumptions of the analysis and the nature of these models in the first place.
The importance of this cannot be understated. As John Llewellyn outlined in an interesting Financial Times article, forecasting as a general concept is very important in facilitating planning. Get it wrong, and the results can be disastrous. And the truth is that economic forecasting, in seeking to predict the results of billions and billions of daily choices, actions, purchases and investments, is likely to be less easy than that in many scientific fields. This requires acknowledging uncertainties and the limits of knowledge.
That something is difficult does not mean one should not try, of course. But that something is difficult does not mean one can explain away large errors by repeating that the task is difficult.
Many of us pointed out on publication that the Treasury and others made very negative assumptions about the long-term policy choices made outside of the EU. The Treasury assumed leaving meant a Britain which politically decided to become a more closed economy. It assumed we’d raise tariffs and other trade barriers, and would not use any of our new-found freedoms to positively change regulation.
This long-term finding that Brexit could only have downsides, and make us poorer, was fed into shorter-term forecasts. Customers and investors, foreseeing a worse economic outlook, would (it was thought) cut investment and spending now. Add in the uncertainty of political negotiations, and a short-term downturn looked inevitable.
This was not forecasting per se, but scenario analysis fed into forecasts. At least one problem of the forecasts then was that forecasters implicitly made judgements on long-term political outcomes for which they had no particular expertise. They overreached, without acknowledging this clearly in public debate. Since the referendum, economic agents are not acting as if they expect Brexit to be economically disastrous, of course. Maybe the voters have more faith in our political processes developing better outcomes on trade and regulation than the experts envisaged!
But biased assumptions only get us so far in explaining away failures. These models also seem less adept at ever foreseeing anything out-of-the-ordinary occurring (see the financial crisis), or predicting downturns and recoveries (see 2010 to 2013).
Theses could be written on why. But as a starter, you might be surprised to learn that the basic models utilised by the alphabet soup of acronyms mostly do not even incorporate modelling of banking and money, an observation long-lamented by the economist professor Tim Congdon. If such components of modern economies are missing, then what hope have these models of assessing fundamental policy regime change, such as Brexit?
If this all seems like a lament without a solution, it is. Economies are complex organisms, and there’s not a top-down answer or perfect replacement model available. What we must hope for is that Brexit can catalyse the humility necessary for continual improvement. Forecasters should spell out their assumptions and model limitations more clearly. And we, particularly the media, must always be careful not to raise them to the pedestal of truth. How’s about that for two geeky New Year Resolutions?