Monday 19 January 2015

Great extrapolations in wealth and obesity

"If current trends continue..."

Oxfam have been given lots of news coverage today thanks to a report which claims that the richest 1% will 'control' more than half the world's wealth by 2016.

The prediction is based on this graph (from page 2 of their report).

The weaknesses of this extrapolation should be obvious, although they seem to have eluded the BBC, The Guardian et al. Oxfam cherry-picked four years of data and drew a straight line extending out from them. And, just to be sure of getting the right result, they ignored the most recent year, presumably because it suggests that the trend is flattening out.

The idea that the 1%'s share of wealth rises in a straight and linear fashion (because of Global Capitalism - boo!) is a fiction. One only needs to look at the other graph on page 2 to see this.

Here we see that the 1%'s share of wealth was declining when the global economy was booming, rose after the financial crisis began and seems to now be levelling off. We also see that the 1%'s share of wealth is smaller than it was 15 years ago—a fact that Oxfam chooses not to mention, perhaps because it doesn't suit their claim that 'global wealth is increasingly being concentrated in the hands of a small wealthy elite.'

The 1%'s share of wealth may now be about to fall again, or it may stay flat, or it may rise. Without an understanding of why it fluctuates in the first place (which Oxfam, being economically illiterate, does not have), it would be foolish to speculate. But what you really should not do is cherry-pick a handful of years when the share of the 1% rose and assume that this trend will continue inexorably.

Seeing Oxfam's graph I was immediately reminded of the various obesity predictions which use the same cretionously stupid methodology. Take the ironically titled Foresight report of 2007, for instance. This highly influential report told us:

The extrapolation of current trends, which underpins the microsimulation, indicates that, by 2015, 36% of males and 28% of females will be obese. By 2025, these figures are estimated to rise to 47% and 36% respectively. By 2050, 60% of males and 50% of females could be obese.

To state the obvious, it is now 2015. Are 36 per cent of British men obese? Well, we can't say for absolute certain because there is a peculiarly large delay in getting these figures out, but the most recent data go up to 2012 and I doubt that I am leaving a hostage to fortune by confidently saying that they are not. Not even close.

The obesity rate for men in 2012 was 24.4 per cent, less than one percentage point higher than it was in 2007. This is the Foresight prediction plotted on a graph alongside the actual data.

You can see what the Foresight authors did here. Rather like Oxfam, they found a trend that suited them (ie. the 1990s), ignored the most recent data (they had figures going up to 2004 when they wrote the report) and extrapolated a straight line going ever-upwards.

It is a mystery why the Foresight authors thought that male obesity would be so much higher than female obesity by 2015. Female obesity was higher when they wrote their report and it remains so today (although, of course, lower than they predicted: 25.1 per cent—and falling—in 2012).

As with wealth inequality, there is no reason to think that rates of obesity will rise forever, let alone that they will rise in a straight line. But, as with Oxfam's report, the real aim of the Foresight report was not to produce a credible prediction, but to create alarming headlines such as 'Half of adults "will be obese by 2050"'.

The Foresight report's conspicuous failure to make a vaguely accurate prediction of what would happen in five years time should cast serious doubts on their predictions of what would happen in 35 years time. Nevertheless, the claim that half of adults will be obese by 2050 continues to be made by campaigners for various taxes and bans. In 2011, the BBC stated as fact that 40 per cent of adults will be obese by 2030, based on an equally spurious extrapolation in The Lancet.

It is also worth bearing in mind the fact that the Foresight predictions are the basis of many claims about the financial cost of obesity in the future. These cost estimates have already been shown to be highly inflated, but since the Foresight predictions of obesity prevalence have so far borne no resemblance to actual rates of obesity, they should be considered worthless.

Finally, if you are thinking to yourself that obesity rates would have risen to the heights predicted in 2007 had it not been for the valiant effort of public health campaigners, remember that these very campaigners have accused successive governments of failing to tackle the problem. None of the policies that these people have campaigned for have been enacted and, according to anti-obesity industry, the government's Responsibility Deal has been "utterly inadequate" and "will not work".

The inexorable rise of obesity predicted in 2007 has not been averted by 'public health' legislation. It was simply never going to happen.


My thanks to Rob who contacted by e-mail to tell me that the 2013 figures came out last month (see here). They show that male obesity rose to 26 per cent and female obesity fell to 23.8 per cent.

Taking both sexes together, the UK adult obesity rate was 24.9 per cent in 2013.

In 2007, it was exactly 24 per cent.

According to the Foresight report, it will be 32 per cent in 2015. If anyone would like to place a wager on this prediction coming true, I would be delighted to bet against you.

Here's the male obesity graph updated with the 2013 figure...


Christopher Snowdon said...

They also define "wealth" in such a way as to skew the discussion.

For example the richest 38 people have the same wealth as the bottom 50% of the world because the cash/assets of the average poor person is a measly $320. That's because the ultra rich don't spend their earnings, and the poor do.

I probably have more money than the billion poorest people in the world by that twisted logic. Almost all of them will be children, paupers or old people in the Third World, and therefore worth exactly zero when their net worth is summed.

If Oxfam tried the same calculation but with earnings of the top 38 earners compared to the bottom half of workers it wouldn't look half so bad.

And how much you earn is what matters, not how much cash you have.

Christopher Snowdon said...

Oops, I think you may have written 1913 instead of 2013 in the update section.

Christopher Snowdon said...

I caught that too.

Christopher Snowdon said...

Quite right. Thanks. I've corrected it.

Christopher Snowdon said...

There really *ought* to be some sort of significant and severe penalties for researchers who manipulate data like this for political/corporate/grant-seeking motives. People look back with scorn on "tobacco company researchers" of the 1950s and 60s and the tobacco companies got hit with a $250 Billion bill for it. What if the same standards were applied to cases like these or the all-too-rampant cases we see all the time in antismoking research?