At the start of this year Andrew Haldane, the Bank of England’s chief economist, declared that the economics profession was ‘to some degree in crisis’. He highlighted its ‘Michael Fish’ moment before the 2008-9 financial crash when most economists were as off the mark as the infamous BBC weather forecaster in anticipating an impending hurricane in 1987. Haldane went on to admit – ‘a fair cop’ – that economic experts at his own Bank, as at the IMF, the OECD and the Treasury, had seen their gloomy predictions of the effects of a vote for Brexit confounded by what actually happened. The economy continuing to run along pretty much as it had before the referendum, even ticking up a little as 2016 went on.
The clarity of Haldane’s mea culpa about economists getting their predictions wrong wasn’t matched by his explanation for why. In fact he thought their models generally work well but were ‘rather narrow and rather fragile’ to cope at times of change. The solution he suggested was more data, claiming that just as weather forecasting had improved since 1987 – that’s his perception at least – so could economic forecasts learn from mistakes and improve too.
This misses the inherent defect with economic models – however much data they include – which is that they are driven solely by the assumptions that go into them, not by economic reality. Their results therefore only substantiate these prior suppositions. They are intrinsically circular in that they show what the modelers believe. They illustrate their starting premises. If these are gloomy assumptions, the estimates are gloomy, and the same applies with optimistic presumptions. Models can’t ‘prove’ anything about the future.
For example, economists from the Centre for Business Research at the University of Cambridge explain that the model used by the Office for Budget Responsibility, the government’s official ‘independent’ forecasting body, makes certain assumptions about productive capacity. The OBR assumes labour productivity growth of about 2 percent a year. Hence ever since its creation in 2010 the OBR has been projecting a return to productivity growth of about 2 percent. Given the actual stagnation of productivity, every year this has been wrong. ‘In the OBR’s world, next year will always be better.’
This self-fulfilling defect explains Haldane’s recognition that models fail most at tipping points. They can’t explain economic change. Models extrapolate from the current presumptions. When the world deviates from these presumptions – as it tends to do, quite frequently – the models inevitably fail to predict the changes.
The official referendum models failed for similar reasons. In this instance they assumed a change that didn’t happen. Built into them was the presumption that a vote for Brexit would lead to economic activity stalling. Businesses would stop investing and people would stop spending. Hey presto, models which assumed lowered activity across the economy forecast lower economic activity. As the premise proved faulty in the months following the referendum, so the forecasts were fallacious. As Haldane noted, people hadn’t acted as their models presumed they would in the circumstances of post-vote uncertainty, by reducing their spending. In fact people carried on spending after 23 June pretty much as before.
The basic difficulty with spreadsheet forecasts is in knowing what an uncertain future will bring. This is another way of saying forecasts based on assumptions that extrapolate from present conditions can never reliably predict a future that is pregnant with change. Which rather defeats their supposed purpose. The whole concept of economic forecasting is therefore flawed.
Aware of this inherent limitation another discipline, demographers, often stress that their estimates of future population trends are projections, not forecasts. They project what will happen by rolling forward the film on present assumptions. They don’t anticipate changes or, to be more precise, they don’t anticipate changes in the rate of change. A jump in mortality rates, or a sharp fall in birth rates, will lead the actual population numbers to diverge from the ones projected. It is a shame that most economists fail to highlight the same health warning.
However, the main real world problem we face with economists today is not the intrinsic limitation of their models. Nor is it groupthink. Nor is it even the dominance of shaky and superficial theories, which at best only describe some economic phenomena but fail to explain underlying developments. The bigger contemporary problem is the enhanced role given to economists by politicians.
Increasingly since the 1980s, politicians of all stripes have outsourced their responsibilities for the economy to others, to non-elected technocrats and to various ‘experts’. For example, Haldane’s employer, the Bank of England, was in 1997 given its ‘independence’ and responsibility for running the country’s monetary policy by the New Labour government, with opposition support. Such policy had previously been a political, and therefore a democratically accountable responsibility. Ever since politicians could buck pass, as Theresa May recently did in blaming the Bank for some of the sided-effects of ultra low interest rates.
As part of this depoliticisation of economic life, politicians have deferred to professional economists for advice and making prognostications. Of course some economists are more arrogant and some are more humble about their capabilities, in forecasting and, more importantly, analytically. But when it comes to policy-making, give them a break. It is the politicians that have put the economists on a pedestal that we should be giving the tough time to. They are the ones we elect. We should not allow them to hide behind the economics profession, both for their failures in understanding what’s happening in the moribund economy, and also for their abdication from acting to revive it.