Poverty and Inequality, in Charts

Jared Bernstein is a senior fellow at the Center on Budget and Policy Priorities in Washington and a former chief economist to Vice President Joseph R. Biden Jr.

Many War on Poverty analyses, including my own, invoked increased inequality as a factor keeping poverty rates higher today than would otherwise be the case.  Many commenters, however, seemed suspicious of this link between poverty and inequality, and as I look back at much at what’s been written, my own piece included, I see it was left largely unexplained (I wrote that “by steering any given level of economic growth away from the low-income families,” higher inequality “leads to higher poverty” â€" which sounds a bit cryptic, I grant you).

So let me remedy that omission.

This chart below by the poverty scholar Sheldon Danziger has a few moving parts to it, but I think it helps get at the underlying relationship between poverty and inequality.  Professor Danziger does something very simple here: he predicts poverty rates based solely on real growth in gross domestic product (per-person G.D.P., to be precise).  And he finds that overall economic growth was pretty much all you needed to predict the poverty rate from the late 1950s to the mid-’60s.

After that, however, if you based your poverty prediction on G.D.P. growth, you would have concluded that poverty ended at some point in the mid-1980s, which … um … didn’t happen.  Of course, G.D.P. growth slowed down some in the 1970s, and some analysts back then blamed the slowdown for this uncoupling from poverty trends.  But Professor Danziger shows that’s not the problem, as the line with the circles in it simulates the poverty/G.D.P. relationship, holding G.D.P. growth constant at its pre-slowdown rate.  That simply shows poverty ending a few years before the other prediction based on actual G.D.P. growth.

So no, it wasn’t just slowing G.D.P. growth that led to the disconnect.  It was also â€" among many other factors, of course â€" growing inequality.

Official poverty rates and predicted rates with pre-1973 G.D.P./poverty relationship. Source: Danziger, 2007.

The mechanics are simple.  If less of the economy’s market-generated growth â€" i.e., before taxes and transfers kick in â€" ends up in the lower reaches of the income scale, either there will be more poverty for any given level of G.D.P. growth, or there will have to be a lot more transfers to offset inequality’s poverty-inducing impact. (Such analysis riffs off the distinction between primary and secondary distributions that I discussed in an earlier post.)

Inequality serves as a wedge or a funnel in this model, redirecting growth from a broad swath of households across the income scale to a narrow slice at the top. The chart below paints a pretty clear picture of that dynamic, showing low, middle and high family incomes growing together and then growing apart (levels are indexed to 100 in 1973 so as to track the relative trends).  It’s not hard to see how that maps onto the poverty result that I and others have been describing.

The next chart plots what I think are interesting differences in this space between the economy facing the original poverty warriors and today’s economy.  The figure compares real G.D.P. growth and inequality (the share of income going to the top 1 percent before taxes and transfers) averaged over 1964-66 and the most recent three years for which we have these data.

Source: Bureau of Economic Analysis and Piketty, Saez.

Not only are we now faced with slower growth, but that lesser growth rate is much more narrowly distributed (for broader historical reference: the real annualized G.D.P. growth rate over the full 1960s cycle was 4.5 percent versus 2.6 percent over the 2000s cycle).  We’re baking a smaller economic pie and cutting less equal slices.  As underscored by the simple Danziger exercise, at any level of G.D.P. growth there would be less poverty reduction because of the inequality wedge.  But we’ve also got less growth to boot.

One last chart.  In my War on Poverty post, I referred to the chart below by the Economic Policy Institute, which quantifies the contributions of many of the factors driving poverty trends. Actually, it’s the institute’s adaptation of an earlier methodology introduced by Professor Danziger and the economist Peter Gottschalk.  Of the factors they look at, they find that rising inequality is quantitatively the largest, including overall growth, family structure, education and race.

They back out this result by basically asking what would have happened to poverty rates over this period â€" 1979 to 2007 â€" if we held the distribution of income (that is, inequality) constant but let all those other factors change. Then they compare the poverty rates that result from that simulation to the actual rates that prevailed in 2007.

Economic Policy Institute Note: The bars show by how much the poverty rate increased or decreased due to rising income inequality, income growth across the income distribution, and changes in the education levels, family structures, and racial composition of the U.S. population.
Source: Authors’ analysis of Current Population Survey Annual Social and Economic Suppleent microdata. Analysis based on Danziger and Gottschalk (1995).

It is, of course, not realistic to try to isolate one factor at a time, since they interact with one another. But it’s revealing that when you do try to isolate these poverty determinants, inequality is the most important.

I hope all of this at least analytically clarifies the important relationship between inequality and poverty.  If so, then you may well agree with those of us arguing that we’ll be hard pressed to make much more progress against poverty without tackling the sources of inequality.  Quite interestingly, similar points were recently made by two conservative politicians, Senator Marco Rubio and Representative Paul Ryan, though their solutions are, not unexpectedly, different from mine.  More on that in future posts.