Poverty Data Release Preamble

September 14th, 2013 at 11:47 am

This Tuesday the Census Bureau will release poverty, household income, and health insurance coverage results for last year.  My CBPP colleagues and myself will be developing and releasing analysis throughout the day, but here’s some context within which to consider the results.

First, while there’s useful information in the official poverty rate, it’s seriously limited.  Thankfully, the good folks at Census provide enough supplemental information that we can help offset, to some extent, that limitation.  Let me explain.

The key point is this: the official poverty rate does not count the impact of the anti-poverty programs that have become increasingly effective over the past few decades.  Conversely, it includes others that have become increasingly ineffective.  So, over time, the official rate has become biased away from actually showing the impact of policy measures designed to reduce poverty.

My colleague Arloc Sherman clearly explains and analyzes these issues in a new paper:

Since the 1960s, official poverty status has been calculated by comparing a family’s pre-tax cash income with a poverty line that has remained essentially unchanged in inflation-adjusted terms since it was first established. In the 1960s, this basic comparison of cash income to the poverty threshold was a reasonable proxy for whether a family could meet basic needs such as food, clothing, and shelter. Today, analysts…agree that this measure has become outdated, making comparisons over 50 years very misleading.

One of the most obvious flaws of the official poverty measure is that it only considers cash income and doesn’t count non-cash benefits, such as SNAP and rent subsidies, both of which help poor families far more today than they did in the 1960s. Nor does it count tax-based benefits for low income working families, like the Earned Income Tax Credit (EITC), which did not exist until 1975, or the low-income component of the Child Tax Credit…

Congress and presidents of both parties have since expanded these tax credits.  At the same time, one of the main sources of means-tested income that the official poverty measure does include — cash welfare assistance for families with children —has fallen substantially since the late 1960s.

Thus, the official measure includes a key benefit with a diminishing antipoverty role while failing to include tens of billions of dollars a year in non-cash benefits that were added over the same period.  If we adjust the poverty and income data to count the newer, non-cash benefits (SNAP, rent subsidies, the EITC, and the refundable Child Tax Credit)…we get a fuller picture of how the lowest-income Americans’ economic circumstances have changed since the 1960s.

Here’s what the picture looks like, compared to the official rate.


As you can see, under the more inclusive measure, the poverty rate falls faster than under the official poverty rate, a clear example of the growing bias alluded to above and its impact on our ability to accurately recognize the progress we’ve made.

Arloc makes this point, for example: “By the expanded measure, even the poverty rate for 2011 — a year of very high unemployment —was below the estimated figure of 12.0 percent in 1969, which was the lowest rate in the 1960s. The unemployment rate in 2011 was nearly three times as high as in 1969.”  I’ve made a similar point regarding the under-appreciated effectiveness of the safety net, partially amped up by the Recovery Act, in the recent recession.

These observations surely reveal the emptiness of the old Reaganism, “we fought a war on poverty and poverty won.”  But, with a little more digging, they also reveal how much poverty reduction has become a function of the secondary, as opposed to the primary distribution.  I describe those terms here as follows:

The primary distribution is the income, wage and wealth status of households generated by market outcomes, before taxes and all kinds of other policies kick in to help those on the short end of those outcomes (i.e., transfer policies, like food stamps or unemployment insurance).  The secondary distribution is the income distribution after taxes and transfers.

In the poverty context, these concepts invoke two related and central developments: the rise of inequality and the wage stagnation in middle and low-wage jobs.  With greater inequality channeling more of the economy’s growth up the income scale, any given level of growth is less likely to lead to poverty reduction.  And the mechanism through which that occurs is the diminished availability of jobs paying wages that would lift economically vulnerable folks above the poverty thresholds without help.

The figure above adds some empirical support.  Before inequality took off in the late 1970s, poverty rates fell with growth.  It’s been shown, for example, that if you used GDP growth to predict where poverty was going, you got a very strong fit over the early years in the figure (see figure 2 here).  But since then, poverty has wiggled about in a cyclical manner, while today’s rates remain (under the inclusive measure) and about where they were in the mid-1970s.  Since GDP growth can no longer can be counted upon to flow to all parts of the (primary) income distribution, it has become an inadequate predictor of changes in the poverty rate.

Take Tuesday’s data, for example.  Real GDP rose about 3% last year, but because so much of that growth accumulated at the top of the scale, I only expect the poverty rate to come down a few tenths of a point.

So here’s how I view the larger lay of the land: we’ve implemented a broad set of anti-poverty policies that have been a lot more effective than is generally recognized—or officially measured—at helping many of the poor meet some of their basic needs.  These programs must be defended against current political attacks that would gut them.

But unless we attack the underlying disconnection between growth, jobs, wages, and poverty reduction in the primary distribution, we will be asking too much of the safety net and the political system that supports it.  To be very clear, that doesn’t mean, as in much of today’s conservative rhetoric, we can simply assume the market will provide the jobs and wages that must complement safety net supports.  We know that assumption is wrong.  It means that progressives must target the primary distribution with policies that will make work pay for all comers.  The safety net is working better than you thought, but it cannot be our sole anti-poverty strategy.

Print Friendly, PDF & Email

2 comments in reply to "Poverty Data Release Preamble"

  1. Stephen Weinberg says:

    The other main lines of criticism of the poverty rate are the failure to correct for regional variation in cost-of-living and various problems with how the poverty line itself is set. (It’s important to remember that the poverty rate gives us at best a benchmark for comparison, but NOT a definition of “poor” or “acceptable living standard” or, certainly, “comfortable living standard.” The statistical meaning of “poverty” is just not the same thing as the English language meaning of poverty.)

    I can understand your decision to focus this post on the programs targeted to help the poor, since the rhetorical point is that these programs are very useful. I accept that point in general, and in practice I see the importance of defending these programs very hard just now. But I cannot for the life of me understand why Medicaid is not on the list of in-kind poverty reducing programs. Even when Medicaid just provides health care, it is providing a huge benefit that wouod otherwise be tremendously expensive to buy; this is why health care and health insurance inequality in the US is non-monotonic in income, and one important reason that the official poverty line is a very poor measure of “acceptable standard of living.” going beyond health, Medicaid is also a major payer for housing for the elderly, in the form of long-term care facilities.

    In NYS, Medicaid Director Jason Helgerson has been working on having Medicaid assume larger roles in housing and employment assistance, on the principle that it’s very hard to keep people healthy when they’re homeless and unemployed. That sort of Medicaid role is still nascent, but will make Medicaid an even more important tool for helping the poor.

    And the Republicans appear to be targeting Medicaid for destruction at least as hard as any other part of the safety net, with the possible exception of SNAP.

    • Jared Bernstein says:

      Excellent point…it’s harder to valuate as in the CBPP paper but I’ll have more on it later in the week.