I haven’t beaten the drum lately for Lane Kenworthy — perhaps the best researcher out there on the economic effects of income and wealth distribution. His years of careful, diligent (and voluminous) statistical and analytic work, tapping the best data sets available, and his cogent, coherent explanations of his findings, should get a lot more attention in the econoblogosphere. Lane Kenworthy rocks.
He’s especially good at trying to suss out causation, which he will be the first to acknowledge is always a difficult business in a discipline that’s inevitably dependent on retrospective data — where you can’t rerun the experiment, much less run it from the start with a randomized control group. (And natural experiments/control groups like the ones that Arindrajit Dube exploited to look at minimum-wage effects — adjacent counties across state lines with different minimum wages — aren’t thick on the ground.)
Nevertheless there are some excellent statistical techniques that can give a good indication of causation. Well-executed, they can really move your Bayesian priors. At the very least, they’re excellent at ruling out causation. Put simply, if there’s a significant negative correlation between presumed-cause A and presumed-effect B (or no correlation at all), you can feel fairly confident that A didn’t cause B. It’s difficult to prove causation with correlation; it’s much easier to disprove causation — to falsify a hypothesis.
But enough with the philosophical throat-clearing. Let’s look at one recent paper (PDF), a multi-country multi-regression analysis comparing rich countries, looking at income inequality and middle-class income growth. He finds that from the late 70s to the mid 2000s (all emphasis mine for easy scanning):
…an increase of 1 percentage point in the top 1 percent’s share of pre-tax income reduced growth of income for the median household by about USD530. In the most extreme case-the United States-the top 1 percent’s pre-tax share increased by 8 percentage points between 1979 and 2004. According to this estimate, that may have reduced median household income growth by a little more than USD4,000. The actual rise in the United States during those years was USD8,000, so the estimated impact of rising income inequality is not trivial
In other words, if the 1%’s share of income had not grown by 8%, median household income would have grown by $12,000 instead of $8,000. This bears out Lane’s rather intuitive, common-sense assertion earlier in the paper:
Household income growth is not a zero-sum game because the pie tends to get larger over time. Disproportionately large gains at the top, however, are likely to come at least partly at the expense of those in the middle.
Always careful, he adds:
At the same time, the data suggest that the income-reducing impact of a rise in top-heavy inequality has been overshadowed by the income-boosting impact of economic growth and of increases in net government transfers.…even after adjusting for these other influences, change in top-heavy inequality is not a very good predictor of growth in middle-class incomes.
So yes: income inequality in and of itself seems to have reduced middle-class income growth significantly. But obviously, of course, that’s not the only economic effect at play. (Only a wild-eyed, ideologically blinded, axe-grinding, bought-and-paid-for Republican would make that kind of foolish claim about some particular economic effect.)
Which brings me to another recent paper (prominently citing the previous one), that questions the Left’s rhetorical emphasis on (in)equality:
I fear the American left’s recent move to put income inequality reduction front and centre might be harmful rather than helpful. It may foster a conviction that the key to addressing America’s social, economic and political problems is to reduce the top 1 per cent’s share or the Gini coefficient. That could distract attention from more direct and effective efforts to address those problems.
Such efforts include fully universal health insurance; improvements in eligibility, duration and benefit level for various social-insurance and social-assistance programmes; wage insurance; early education; enhanced financial support for college; a minimum wage indexed to prices; an expanded earned-income tax credit indexed to average compensation; and monetary policy less tilted towards inflation avoidance. Policy changes like these would go a long way towards improving economic security, enhancing opportunity (and mobility) and ensuring shared prosperity in the US. Inequality of political influence could be lessened via direct reforms, such as reversal of the Citizens United decision, introduction of a strong transparency rule and public funding for congressional election campaigns.
I think Lane’s right. I’ll say it again: if you talk about fairness and equality, Americans change the channel. (They’re only somewhat more open to hearing about “opportunity.”) They want to hear about prosperity — especially widespread prosperity. And the programs Lane points to have a decades-long history of delivering widespread prosperity. Expanding those programs (and funding them with a tax system that actually is progressive) would make us all more prosperous.
And that’s exactly what Lane’s first paper demonstrates. No: just reducing inequality through redistribution doesn’t make everything peachy. No duh. (Though in the current environment of concentrated wealth and income it does improve things a lot in and of itself.) If you really want to increase prosperity, you use methods of redistribution that increase prosperity — like the programs that Lane details above. (Plus publicly funded infrastructure, research, etc.)
So the two things aren’t mutually exclusive. You implement programs that deliver widespread prosperity in and of themselves, and distributive effects also deliver the prosperity benefits of reduced wealth and income concentration. It’s a virtuous cycle, rolling forward on a path to American prosperity. Rinse and repeat.
In brief, widespread prosperity both causes and is greater prosperity.
Cross-posted at Angry Bear.