*****WARNING, THIS STORY HAS BEEN RETRACTED. IT APPEARS THAT THE REGRESSION GENERATOR ON THE GSS WEBSITE HAS NO RELATIONSHIP WITH REALITY WHATSOEVER. AGAIN, NONE OF THE STATISTICS QUOTED IN THIS ARTICLE ARE EVEN IN THE BALLPARK OF ACCURATE***

I regressed age, gender, income, race and ‘have ever been unemployed in the last ten years’ on support for reducing income differences between the rich and the poor using data from the US General social survey(1).

In non-technical terms everything had a statistically significant effect except age, but the effect of unemployment in the last decade was very, very much larger than the others. Having been unemployed, even controlling for all these other variables, increases your support for redistribution by 1.7 points on a seven point scale- an effect large enough to be both practically and statistically significant. In technical terms, check out the regression equation details in the appendix if you like.

What has this come to teach us? There is a prima facie case for thinking that past unemployment status plays a huge role in the politics of redistribution even controlling for age, gender, income, and race though of course the nature of the causal relations remains opaque in correlational analysis, even with controls.

This makes a lot of sense to me. The desire for redistribution- especially the kind of redistribution achieved through political rather than union action- isn’t primarily about the desire for more stuff. Rather it is about the desire for security. Those with first hand experience of insecurity want to be protected against it in the future.

Given the disastrous effects of unemployment on well-being, this is perhaps unsurprising. On evidence based the Holmes and Rahe stress scale, being dismissed from work is rated as the eighth most stressful thing that can happen to you. Libraries of evidence attest to the severity of the negative effects of being unemployed on psychological well-being.

Unemployment and previous unemployment is often invisible except as an aggregate statistic. Few people like to be publicly identified as unemployed or previously unemployed given the potentially negative. Thus we find ourselves in a world where unemployment plays a crucial role in politics, but has few open voices associated with it.

If unemployment really is this important the left needs to think about -as a priority- how to connect with, organise and activate the unemployed, and connect concerns about economic security with demands for greater income equality.

Footnotes

(1) To balance between the need for a large sample and to ensure data is as recent as possible I used the 2012, 2014 and 2018 surveys.

Appendix, regression equation

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 1.0766112509 0.1274153901 8.450 7.34e-15 ***
age -0.0010378398 0.0013817247 -0.751 0.453505
race -0.1773152563 0.0369196629 -4.803 3.15e-06 ***
sex -0.1834569178 0.0508871525 -3.605 0.000398 ***
realinc 0.0000016759 0.0000006995 2.396 0.017543 *
unemp 1.7246522016 0.0462150965 37.318 < 2e-16 ***