Poverty Has One Dimension: Income

Over at Brookings, Richard Reeves, Edward Rodrigue, and Elizabeth Kneebone are trying to come up with a new framing around poverty, perhaps to replace Brookings' old comical "Success Sequence" framing. In this new framing, Brookings insists that poverty is a multi-dimensional thing and that the focus on income poverty is "facile." This is exactly wrong: poverty is a one-dimensional thing concerning income and efforts to describe it as multi-dimensional traffic in trendy faux-profundity and faux-nuance that needlessly complicates what is fundamentally a very simple topic.

According to Reeves and crew, multi-dimensional poverty has five components:

  1. Low income (aka the only dimension of poverty).
  2. Limited education.
  3. No health insurance.
  4. Living in an area with lots of poor people.
  5. Unemployment.

Why we need to redefine poverty to include a bunch of things that aren't poverty is not very well explained by the authors. Initially they had this to say:

Our hope is that a richer, multidimensional formulation of the problems of poverty and disadvantage, and in particular the way disadvantages cluster together for certain people or groups, or in particular places, can help to inform policy.

The main thrust of policy will be—and should be—to try and reduce the number of people who are disadvantaged on each of these and other dimensions. Our point is simply that it is important to consider ways to de-cluster as well as to reduce disadvantage. These goals are perfectly compatible. Policy ought to aim at lowering the proportion of people who face disadvantage X and the proportion of people who face disadvantage Y. But it should also aim at lowering the correlation between X and Y.

But then later on we get what I believe to be the real explanation of this needlessly complex treatment:

Polices aimed at tackling poverty often focus solely on raising income. But an equally important goal of anti-poverty policies is to de-cluster disadvantage, and reduce the consequences of having a low income on other aspects of life. In other words, make income poverty matter less.

Ah there it is: a new Brookings' framework for why we can "fight poverty" without adopting institutions to distribute the national income more evenly. I can already see the David Brooks op-ed gushing about this policy innovation.

Education and Healthcare

When it comes to education and healthcare, it's not difficult to see where Brookings is coming from as a general matter. In the normal welfare state, these things are delivered by the state as services and are thus not income. So in that sense, the state can make movements on these things without working the income channel (except of course to levy taxes to fund each).

But it's hard to see how they figure into the "declustering" argument. There is no other developed country in the world where lack of health insurance is correlated with anything. This is because in normal countries, everyone has health insurance.

Brookings' declustering argument says that we could make strides in multi-dimensional poverty reduction by keeping health uninsurance in the US at its current level and just making sure that health uninsurance is more evenly spread throughout the population instead of being so disproportionately shouldered by those with low incomes. But how would you go about uninsuring rich people in order to more evenly spead the uninsurance burden? The nature of being rich is that you have money and so you can purchase health insurance. Would we hold a lottery to ban some rich people from getting it while also extending Medicaid a little further, thereby reducing the correlation between health uninsurance and low incomes without reducing the overall level of health uninsurance?

Education presents similar issues. How would we decorrelate adult incomes with education while also keeping the same educational composition (and thereby win a reduction in multi-dimensional poverty)? Unlike healthcare, we can't really strip rich people of the education they've already attained. And like with healthcare, you probably wouldn't want to.

So for both education and healthcare, the Brookings' conceptual innovation ends up being entirely meaningless. Nobody wants to merely "decluster" those things from low income. They just want to extend their absolute reach.


The unemployment point is even stranger than the education and healthcare point. This is because unemployment is correlated with low incomes because unemployment causes low (labor) income. How on earth would you decluster this correlation except to increase transfer incomes to unemployed people (i.e. use the "facile" poverty reduction strategy of simply boosting the incomes)? I mean seriously, how would you do it?

If you were to employ unemployed people, then they would presumably stop being low income, or certainly be higher income than they were. And so this type of policy would not "decluster" low incomes and unemployment as it would move both variables simultaneously.

The only way to "decluster" the two variables is to make it so that there are more people with high incomes who do not work (e.g. the idle rich living off of capital income) or make it so that more people who do work have low incomes (e.g. by somehow making them work for free or otherwise confiscating their labor income). Presumably Brookings doesn't want either of these, does it? If not, then how else does it intend to decorrelate low income from unemployment?

These may seem like pedantic points, but they really aren't. The fact is that Brookings has not thought this through. Of course unemployment should be reduced. The policy mechanisms for doing so are not particularly mysterious: monetary and fiscal stimulus and active labor market policies. But unemployment is not poverty. It is not a dimension of poverty. And the hand-waving about declustering it from low incomes either generates the facile conclusion that we need more transfer income for unemployed people or the absurd conclusion that we need more idle rich and lower compensation for the working.

Stick to One-Dimensional Poverty

As it is, I don't see any value being added by Brookings' framework. But even if it did add value, it also takes away a lot from poverty policy. There is a tendency in the think tank world to massively complicate things by adding in nuances and rubs that make issues more mystical than they actually are. This might seem like a more sophisticated way of doing things, but it actually just makes the problem seem way harder than it is and gives people plausible deniability to avoid solving it.

As Philip N. Cohen documents, this is especially true when it comes to poverty. David Brooks remarks on this are sadly typical:

Poverty is a cloud problem and not a clock problem. This is a Karl Popper distinction. He said some problems are clock problems – you can take them apart into individual pieces and fix them. Some problems are cloud problems. You can’t take a cloud apart. It’s a dynamic system that is always interspersed. And Popper said we have a tendency to try to take cloud problems and turn them into clock problems, because it’s just easier for us to think about. But poverty is a cloud problem. … A problem like poverty is too complicated to be contained by any one political philosophy. … So we have to be humble, because it’s so gloomy and so complicated and so cloud-like.

Poverty is so complicated. It has so many dimensions. We have to be humble and admit we don't know how to solve it. Maybe it's just a matter of "declustering." Maybe income isn't even what poverty is about. And on and on and on. This doesn't move the ball forward on poverty, it just pushes it deeper into the quagmire.

Sticking to the facile one-dimensional poverty that defines poverty as low income doesn't run into this problem nearly as much. Under the facile approach, poverty is a very simple engineering problem (or "clock problem" in Brooks' parlance). In a developed country with a very high per-capita national income, the way you beat poverty is you adopt economic institutions that distribute the national income fairly evenly. This can be done in a number of ways, but as a general matter, the most successful and easiest way is to increase the tax level and increase the welfare income level. It's that easy and nothing is gained by pretending otherwise.