The great data debate
‘The idea that we can somehow measure “the thing that matters most” is quite absurd’
As he appeals to the British public to vote him in as prime minister, the leader of the opposition proposes collecting new data to provide a better picture of how the country is doing. “Wellbeing can’t be measured by money or traded in markets,” he says. He adds, “We measure all kinds of things but the only thing we don’t measure is the thing that matters most.”
All of the preceding paragraph is true, except for one detail: the first quotation is from David Cameron, then leader of the opposition, in 2006. The second is from Ed Miliband, the current leader of the opposition, a couple of weeks ago. Both men are united, it seems, by a feeling that the most familiar economic measuring stick, GDP (Gross Domestic Product), just isn’t up to the job. Cameron wanted to gather data on wellbeing or happiness; Miliband wants a “cost of living” index. Few reasonable people can object to gathering timely and authoritative economic and social statistics, yet Miliband and Cameron have managed the impressive feat of being cynical and naive at the same time.
The cynical motives in both cases are plain enough — as were, for example, Nicolas Sarkozy’s when, as French president, he commissioned some alternative economic measures that just happened to be more flattering to France. As the leader of a party with a reputation for liking free markets and low taxes, Cameron wanted to soften his image and suggest a broader, more caring perspective. Miliband is trying to replace a government that is presiding over a sudden uptick in GDP, so naturally he wishes to point the spotlight somewhere else.
The naivety requires more statistical digging to uncover, and it’s in three parts. The first point is that many of these data already exist. The Office for National Statistics asks questions about wellbeing as part of the Labour Force Survey. The ONS also publishes regular data on inflation, while wage data are in the Annual Survey of Hours and Earnings. Neither Cameron nor Miliband was really asking the statisticians at the ONS to do something new, just to do it more often or in more detail.
The second point is that no mainstream politician has ever regarded GDP (or its cousin Gross National Product) as the only worthwhile policy objective, although we are often invited to draw that conclusion. Robert Kennedy’s famous complaint that GNP counts “napalm” and “nuclear warheads” but not “the health of our children” or “the strength of our marriages” was wonderful rhetoric — but surely nobody believes that if only the statisticians had collected different data, divorce would be prevented and the Vietnam war would never have happened.
An acerbic comment in Nature last year complained that, “Despite the destruction wrought by the Deepwater Horizon oil spill in 2010 and Hurricane Sandy in 2012, both events boosted US GDP because they stimulated rebuilding.” But this is only a problem if the Deepwater Horizon spill was in some way caused by the collection of GDP data.
If politicians truly sought to maximise GDP they would immediately abolish all planning restrictions, all barriers to immigration and a good chunk of the welfare state. These ideas are political suicide, which proves that GDP is not the sole objective of public policy — it’s just a way to try to measure the size of the economy.
The deepest piece of naivety is the idea that — in Ed Miliband’s words — we can measure the one single “thing that matters most”. ONS data on median wages are a case in point. According to one measure, the median wage for people in full-time employment rose just 0.1 per cent in the past tax year — well below the rate of inflation. According to another way of calculating exactly the same number, median wages rose by 4.1 per cent, well above the rate of inflation. (The median is the wage earned by someone slap in the middle of the sample.)
How can that be? The lower measure is the median for the entire sample. The higher measure looks at the median wage of people who’ve been in the same job for the entire year — the vast majority. The two numbers would differ if — for example — some high-income people retired and some low-income people joined the labour force (school-leavers? immigrants?). It’s possible for most people to enjoy a decent pay rise while median wages stagnate, and that may be what is happening now. One rather narrow question — “how are things going for people in full-time employment in the middle of the income distribution?” — turns out to have two very different answers. Each one is perfectly justifiable.
We haven’t even got into questions of part-timers, the self-employed, the poorest, the richest, pensioners or benefit recipients. The idea that we can somehow measure “the thing that matters most” is quite absurd.
It’s the duty of our official statisticians to provide a range of timely and objective statistics that will lead to better decisions. That is why so many different types of data must be gathered, analysed and published. It is a hard job, which is why the ONS has better things to do than help our schoolboy politicians score points off each other.
Written for and first published af ft.com.