Four ways economists helped usher in the "post-fact" era ★

We are entering, it is said, a post-fact era. Populist politicians dismiss data and expertise out of hand, and the UK and US publics seem to nod along. Michael Gove, a UK MP campaigning for Brexit, summed this up simply: “people in this country have had enough of experts.”

This comes at a time when good-quality expertise and data have never been easier to access. Fact-checkers, explainer websites, data journalism and gigabytes of open data are all a click away.

And yet despite all this, in 2016 so far, numbers themselves seem to be under assault by a populist insurrection. We could blame populist anti-elitism, innumeracy, racism and xenophobia—and indeed we should. But economists (including economic journalists) must also consider our role.

Here are four ways in which we open ourselves to easy criticism.

1. We communicate uncertainty poorly #

There is an old joke: How do we know economists have a sense of humour? They put a decimal point in their forecasts. It’s funny because it contains more than a grain of truth. John Cassidy, writing in the New Yorker about the Brexit Remain campaign’s claims, makes the same point—only here it’s not so funny:

Almost all economists agree that the E.U. has been good to Britain. But the sixty-two-hundred-a-year figure [claim, for losses per X from Brexit] was so large, and so specific, that many people didn’t believe it.

Economists have to provide forecasts (who else will? how else to plan?) But too often we do it without adequately expressing the uncertainty, the likely interval of outcomes around a point estimate.

Government institutions have worked hard at this. The Bank of England pioneered the use of fan charts. Australia’s Treasury has been quite self-reflective about this topic in recent years. But still, often intervals estimates do not bubble up to the executive summary or the media release—much less the soundbite.

It’s time uncertainties—even simple interval estimates - became first-class citizens in economic prognostication, rather than being relegated to appendices. A point forecast sets an expectation of precision; it guarantees you’ll be precisely wrong every time. And if you’re wrong every time, people will dismiss you, even when you’re mostly right.

Economic journalists have a strong role to play here too.

2. We use quantities divorced from everyday experience #

Donald Trump regularly claims true unemployment is much higher than the official rate. ”A phony number,” he calls it. Look around you, he’s saying, do you believe unemployment is only 5%?

Of course, it’s easy for economists to dismiss his claims: the latest BLS release puts (seasonally adjusted) unemployment at 4.7% for the month of May. But while Trump may not have truth on his side, he has what Stephen Colbert called ‘truthiness’.

The way we define unemployment is bizarrely narrow compared with its everyday usage. You have to be actively searching for work. You have to have worked zero hours for pay that week.

We can stand by our definition, but regardless of what we say, a 50-something auto worker who got laid off in 2008 and has given up looking for work probably thinks he’s unemployed.

Late last year, Politifact evaluated one Trump claim of 42% unemployment, and concluded that you had to make extreme assumptions to support that rate, rating it Pants on Fire. They preferred instead 16.4% as

the highest alternative unemployment-rate measure we could come up with that had any credibility… only about one-third of the way to Trump’s 42 percent.

And yet that’s three times the official rate at the time. Trump supports can easily reassure themselves: sure, he’s prone to exaggeration, but the east coast liberal conspiracy that is the BLS is wrong too.

Interestingly, Bernie Sanders has been making precisely the same point, arguing that for African-Americans “the real unemployment rate for young people is 51 percent.” This claim was rated Mostly True by Politifact, with the caveat that “[h]is terminology was off”.

Perhaps its time to admit that it is our terminology, and not that of ordinary voters, that is off.

3. We over-generalize #

As economists, we far too often focus on quantities aggregated and netted out over the whole population—growth in median income, unemployment rate, etc—ignoring significant variation between different subgroups.

Just prior to the EU referendum, a large group of UK economists signed a letter to “speak out on the risks of Leaving and opportunities from Remaining in the EU.” It included the following text (emphasis in original):

Leave have also misled the public on EU immigration. The UK unemployment rate today is 5.1%, significantly less than it was before EU enlargement—**EU immigration hasn't taken jobs**.

But this is only true at the aggregate level. Undoubtedly, without question, EU immigration has taken jobs—at least some specific jobs, from some people. Economists may think of a job as an abstract, fungible quantity, but to a voter a job may be the job they applied for, but lost to an immigrant.

Moreover, the most credible summary of immigration impact I can find, from the Oxford Migration Observatory, says:

UK studies find that immigration has small impact on average wages but more significant impacts along the wage distribution: low-waged workers lose while medium and high-paid workers gain.

I’m pro-immigration. I used to be an immigrant in the UK. I believe the evidence suggests that immigration does benefit countries like the US, UK and Australia, in aggregate, in most cases. But that’s little consolation if it happens to hurt you, personally. National aggregates may be little relevance in a campaign where 37% is enough to win the vote.

Moreover, if I were an ordinary person leaning towards leave, and read the bold statement “EU immigration hasn’t taken jobs” I would stop reading the rest of the argument. Well obviously not your jobs, I would think, looking down the list of Lords, Sirs and Professors who put their names to this statement.

4. We’re more political than we admit #

Underlying all of the above is the pretence that economists are objective, apolitical animals: just telling you what theory predicts, what the data reveal.

But the reality is that economists come with a point of view. Most of us—myself included—are card-carrying members of the “liberal elite”. Regardless of partisan persuasion, most of us instinctively lined up against Brexit, and think Trump should stick to reality TV. Not just for economic reasons, but because illiberalism, anti-intellectualism and xenophobia are anathema in this circles in which we mix.

To pretend this doesn’t affect our economic judgments is foolish. Sometime we are loose with the evidence we cite. We may choose measures and modelling that agree with our point of view—even when others exist. We knowingly downplay uncertainty and disagreement, if they seem to diminish our argument. Why give the other side ammunition? They certainly don’t play fair.

And, it’s true, the populist Leave and Trump have campaigns have not just been loose with the truth, they’ve abandoned it altogether. They don’t play fair. But this is not a situation in which we can fight fire with fire. In a shouting contest, Donald Trump will win every time.

We must restore credibility in expertise through introspection and humility. More than ever, we must strive to acknowledge nuance, uncertainty, disagreement, our biases and the limits to our knowledge. If that means we must make smaller, more cautious, more qualified, better explained claims, so be it.

Note: Paul Johnson, the Director of the UK’s Institute for Fiscal Studies, has also just written on this topic. The IFS, which warned against Brexit (with, indeed, interval estimates), is traditionally one of the most respected independent voices in economic analysis, unmatched by any think tank in Australia or the US of which I’m aware. For them, this must be particularly galling. The only consolation is that their estimates will probably be proved accurate.