Friday 29 November 2019

Bad predictions as political weapons

When Iain Duncan Smith’s office was vandalised this week, certain apologists from the far-left argued that the crime was partially, if not wholly, mitigated by the Tories’ alleged murder of 120,000 people. The murder weapon was ‘austerity’ and the evidence is in a paper from BMJ Open in 2017. The idea that austerity took 120,000 lives has become a proven fact in some quarters. The evidence itself is less compelling. The authors of the BMJ Open study note that improvements in life expectancy slowed between 2012 and 2014, at a time when increases in healthcare spending also slowed. The estimate of 120,000 lives comes from their extrapolation of how many fewer deaths there would have been between 2010 and 2017 had mortality continued to decline at the earlier rate, and assumes that the slowing of life expectancy (which was observed in many other countries at the same time) was due to spending constraints rather than other factors.   

As various fact checking experts have concluded, this is speculative stuff. The claim that austerity has killed 120,000 people strikes me as a classic example of a political factoid. It falls somewhere between the truth and a lie. It cannot be called a lie because it is impossible to disprove, but it would be a lie to call it a fact because it cannot be proven.

Remainers rightly complain about the £350 million a week for the NHS that was famously promoted on the side of a bus during the referendum. For them, it is the flagship example of a political lie. Leavers, meanwhile, complain about the government’s prediction of an immediate recession and 800,000 people out of work in the event of a vote to depart the EU. The former is considered a lie because it could not mathematically be true. The latter also turned out to be untrue but is not considered to be a lie because while the prediction may not have been made in good faith, a deliberate attempt to deceive cannot be proven.

I suspect that the Treasury’s forecast of imminent recession had more impact on voters than the bus (which could have used the more defensible figure of £280 million and had the same effect). Bad predictions might be more honest than premeditated lies, but they are no less misleading.

The last decade has seen an epidemic of terrible economic predictions. The Resolution Foundation (RF) is perhaps the worst offender. Relentlessly pessimistic about incomes, wages, poverty and inequality, they have spent a decade making Eeyore-ish predictions which grab headlines but never come true. In 2012, it claimed that ‘even on optimistic growth assumptions’, poorer households would have incomes 15 per cent lower than in 2008. In reality, the income of the bottom fifth never fell below 2008 levels and was 15 per cent higher within five years of the RF prediction.

Last July, the think tank’s CEO said that ‘child poverty is likely to have increased last year’. Office for National Statistics data show that it didn’t. In February, RF produced a ‘nowcast’ claiming that there had been ‘zero growth’ in the incomes of working people between 2017/18 and 2018/19. Official data published a few months later showed that there had been income growth of 1.3 per cent - £400 a year - with the Office for National Statistics noting that this was the sixth annual increase in a row.

Child poverty is a particular favourite. Usually measured in relative terms, a child is considered in poverty if they live in a household which has an income below 60 per cent of the average after housing costs are paid. The rate was 31 per cent before the financial crisis, fell during the recession as average incomes declined, and has been at 30 per cent for the last three years.

According to RF, child poverty is always on the brink of spiralling. In February, they projected it to rise to a record 37 per cent by 2023/24. Today, with an election approaching, they have made the front page of Guardian with the claim that child poverty will hit a '60-year high under Tories’. Their estimate for 2023/24 is now 34.5 per cent, lower than the previous figure but nobody at the newspaper seems to have asked why. A record high is a record high, I guess, and even 37 per cent looks rosy compared with the 41 per cent projected by 2021/22 in a report for the Equality and Human Rights Commission.

It is too early to say definitively that these forecasts are wrong. Suffice to say that current trends in child poverty are better news for children than they are for forecasters. And we know for a fact that predictions of spiralling poverty and inequality made by the Institute of Fiscal Studies and the Joseph Rowntree Foundation towards the start of this decade were way off.

Predicting the future is difficult and projections are almost bound to be wrong to a greater or lesser extent. The telling aspect of the forecasts above - and I could cite many more like them - is that they are all wrong in the same direction. Put simply, they all predict that things will get worse.

Obesity forecasts are no different. Obesity rates have been rising in Britain for the last twenty years, but far more gradually than ‘public health’ experts predicted. In 2007, when the male obesity rate was 24 per cent, we were confidently told by leading experts that 36 per cent of men would be obese by 2015. The current figure is 27 per cent. In 2010, we were told that 80 per cent of men would be overweight by 2020. The rate was 68 per cent at the time. It is currently 67 per cent.

Rates of obesity are high and the economy has had a terrible decade, so why the need for further exaggeration? The suspicion is that predictions of this kind are not designed to guide us to the future but to shape the present. The pressure groups and academics who produce forecasts of spiralling obesity, poverty and other evils do not present their forecasts as inevitable, but rather as what will happen if the government does not change its policies. Whether it’s welfare reform or sugar taxes, the underlying message is political. At it happens, the government generally doesn’t change its policies and the scary outcomes still fail to materialise, but everybody has forgotten about the predictions by then, not least because a new set of predictions - accompanied by political demands - has taken their place.

What should we call such factoids? ‘Fake news’ seems too harsh since the forecaster cannot know for sure that their forecast is wrong, and there is always a chance that they are making an honest effort to read the runes. If we are going to borrow a phrase from the Trump universe, ‘alternative facts’ is a better description. Wild estimates become a substitute for official statistics, with the former getting acres of news coverage while the latter are barely reported at all. ‘Falsehood flies,’ wrote Jonathan Swift, ‘and truth comes limping after it, so that when men come to be undeceived, it is too late; the jest is over, and the tale hath had its effect.’Vanishingly little has changed since then, except that falsehood flies even faster today.

[Previously published on the Telegraph website]

No comments: