Lies, Damn Lies, and the Great Depression of 2008

On Monday night, this story from British newspaper The Independent was linked at the top of the Drudge Report.  In a media environment increasingly bent on trashing the U.S. economy (irrespective of whether the data support such trashing) this one wins top honors for junk statistical analysis.

USA 2008: The Great Depression

Food stamps are the symbol of poverty in the US. In the era of the credit crunch, a record 28 million Americans are now relying on them to survive – a sure sign the world's richest country faces economic crisis

The article uses this record number of Americans on food stamps as its primary evidence that the United States is pitching inexorably into economic ruin.

Reports that things are at record levels are terribly exciting.  The article fails to notice, however, that many things are destined to continue hitting new all-time highs.  GDP is at record levels.  Federal spending is at record levels.  Your own age is at a record level.  And of course, the national population is at a record level.

If food stamps eligibility criteria remain unchanged, and holding macroeconomic variables constant, the number of Americans on food stamps will go on to set all-time records year after year after year, so long as the population continues to increase.

The article does concede that there are special, non-economic factors contributing to the increase.

The increase – from 26.5 million in 2007 – is due partly to recent efforts to increase public awareness of the programme and also a switch from paper coupons to electronic debit cards.

But that effect is sumamrily dismissed as the underlying cause.

But above all it is the pressures being exerted on ordinary Americans by an economy that is suddenly beset by troubles. Housing foreclosures, accelerating jobs losses and fast-rising prices all add to the squeeze.

This is a lazy bit of circular reasoning.  Increasing food stamps enrollment is given as evidence of a terrible economy.  And the ostensibly terrible economy is given as evidence that the enrollment increase is economic in nature, rather than a result of the switch from paper to electronic and/or the awareness campaign.

But let's stipulate to the suggestion that the increase is completely organic and that the other factors played no role.  The increase from 26.5 million to 28 million represents an annual growth rate of 5.7%.  Since 1980 (the span over which data is readily available from the Census Bureau and the USDA), nearly one in every three years saw increases at least this large (9 of 29).

As for the record-setting 28 million Americans estimated to use food stamps in 2008, to even begin making relevant comparisons, we need to control for population growth.  To that end, if you consider the percentage of Americans enrolled in food stamps programs, you once again see that we're not in uncharted territory.

Below is a chart illustrating the two metrics described above - the percentage of Americans using food stamps (in red) and the annual enrollment growth rate (in blue) - from 1980 to 2008.  (The gray bands represent economic recessions.)

The Congressional Budget Office's estimate of 28 million implies that 9.2% of Americans will redeem food stamps in the 2008 fiscal year.  This is significantly lower (1.4 percentage points lower) than the peak usage rate seen in 1993.

The 5.7% enrollment growth from 2007 to 2008 is also well below (less than half) peak levels.  Since 1980, this growth rate has reached into double-digits four times.  What's more, these four peak growth years (1991, 1992, 2003, and 2004) did not coincide with (nor did they serve as leading indicators for) recessions.  On the contrary, these years were characterized by economy recovery and solid GDP growth.  It's actually been periods of negative growth in food stamps enrollment that tend to foretell looming recessions.  That may well be coincidental, but it seems clear that swift enrollment growth is not a coincident or leading indicator of economic decline.

Recessions are also not well forecasted by high food stamps usage rates (judging by the economic growth that followed the 1993 peak or the recession that followed the 2000 trough).

I'm not expecting the dogged hand-wringers in the media to present economic data fairly or to draw reasonable conclusions from it.  But is it too much to ask that they avoid manhandling those statistics that actually tend to refute their pre-determined conclusions?