The myth that conservative welfare reform worked--Part 6

by: Paul Rosenberg

Sun Mar 14, 2010 at 18:30

We now turn to our last topic: the diversity of options and outcomes among US states, even given the constraints that keep us from being like the more successful countries.

In this section, we complete our debunking of the myth that conservative "welfare reform" was a success by examining poverty rates on a state-level basis.  We've already seen in Part 5 that American anti-poverty policy as a whole is a failure compared to other industrialized countries.  But by looking at state-level data we can gain a better sense of what sorts of policies within the American framework are most promising in at least steering us somewhat towards the direction of more successful policies practiced by other nations.

We begin by simply looking at state-level data on a state-by-state level, informally noting how poverty is highest among certain states by various different measures.  We then turn to looking at systematic groupings of states, following a methodology similar to that employed to analyze groups of nations, first pioneered by  Gosta Epsing-Anerson in his 1990 book, The Three Worlds of Welfare Capitalism and then adapted by a variety others to propose other typologies for various different purposes, both in Europe and in East Asia.

We begin with a series of maps based on census data collected by Annie E. Casey Foundation at the Kids Count Datacenter website.  First is the most basic map of child poverty, showing clearly that the greatest concentration of child poverty is found in the South, with somewhat lower rates along the Eastern Seaboard (except for South Carolina), and additional high poverty in the border states of Kentucky and West Virginia, as well as New Mexico:

This pattern alone should make one thing painfully obvious: the political party (GOP) and movement (movement conservatives) whose power is centered in these states is singularly not to be trusted as experts on how to combat poverty.  Quite the opposite.  The policies they've chosen to implement when they have the power to do so are particularly ineffective and not to be copied elsewhere.  Indeed, they should be eliminated and replaced where they are practiced.  Any sort of fact-based approach to poverty reduction would have to begin by recognizing this as an obvious truth.

Next, we focus on the poverty rate for young children-those under 5.  The overall pattern is much the same as that for all children in poverty, but it's worth noting that Montana is added to the list of states in the worst category:

Looking specifically at child poverty in single-parent families, we again find that the basic pattern of where the worst poverty is concentrated follows that of child poverty in general, only except that New Mexico and Texas are no longer among the states with the worst records:

Shifting to child poverty in two-parent families, we get the first and only different pattern from all those that went before--even though there is still considerable overlap.  Added to the states in the group with the highest poverty levels are California, Arizona and Montana (seen before in this group only for under-5 child poverty), while removed from this group are Tennessee, Alabama and South Carolina:

How Different Approaches Fare Across States--A Cluster Analysis

"Cluster analysis" is a statistical approach that allows researchers to aggregate larger amounts of data from a group of different entities (nations, states, voters, etc.) and then discover how they naturally clump together.  PEW uses cluster analysis, for example to come up with its political typologies.

For a more detailed look at state-level results, using a cluster-analysis approach, we turn to "Packaging Support for Low-Income Families: Policy Variation across the United States" by Marcia K. Meyers, Janet C. Gornick and Laura R. Peck [Journal of Policy Analysis and Management 20(3): 457-483.].  Gornick is a director of the Luxembourg Income Study, and I'm most familiar with her work.  In fact, I originally planned to use one of her (co-authored) papers in the previous part of the diary series, but I had too much information, and not enough skill in organizing it, so that part got left out in the end.  Suffice it to say, she is very familiar with the practice of grouping similar countries together the better to study how different policy regimes perform.  Most notably, she has played a central role in advancing the argument that the typology introduced by Epsing-Anderson in Three Worlds of Welfare Capitalism does not properly capture the different ways in which gender-related policy regimes are differentiated.  So her involvement in using a cluster analysis approach to state-level policy differences and poverty rates is quite natural.

As the abstract for this paper explained:

This paper addresses a gap in state-level comparative social policy research by analyzing policies that support low-income families with children. Variation in state policy "packages" is measured by considering three characteristics of 11 social programs. Individual measures of policy are found to be weakly and inconsistently inter-correlated at the state level, but when cluster analysis is used to analyze multiple dimensions simultaneously, five clusters or regime types are identified that have distinctive policy approaches. These range from the most minimal provisions, to conservative approaches emphasizing private responsibility, to integrated approaches that combine generous direct assistance with employment support and policies that enforce family responsibility. A comparison of a subset of programs at two points in time (1994 and 1998) suggests that states made substantial changes in cash assistance and taxation policies after the 1996 federal welfare reforms. The magnitude and direction of these changes remained consistent with the state clusters identified in 1994.

In the paper itself, the authors further clarify their research questions:

Data from 1994, the eve of the federal welfare reform, are used to consider several descriptive and methodological questions:

1. How do states vary in the adequacy, inclusiveness, range, and quality of programs that have short-term effects on families' economic resources and poverty risk?
2. To what extent are state-level policy choices correlated, both across policy characteristics and across programs?
3. Considering multiple programs and policy characteristics simultaneously, are there coherent groupings or clusters of states that provide similar policy packages?
4. How well do these groupings predict state policy choices two years after the passage of the federal legislation?

The goal of this paper is to develop a detailed description of state variation in a subset of social policies.

After surveying the literature, they decided on a set of 11 programs to analyze, with a total of 31 factors to track.  There are various different types of cluster analysis that can be used, and they chose one specifically designed to produce a small group of relatively similar size.

The clusters that the authors came up with are as follows (verbal descriptions of the cluster follow the chart):

The first cluster of states--denoted as "minimal"--was particularly stingy:

Cluster 1 states (Alabama, Arkansas, Kentucky, Louisiana, Mississippi, South Carolina, Tennessee, Texas, West Virginia) can be described as providing the most meager or minimal support in nearly all dimensions. These states provided AFDC and child care benefits that were well below the national average, even after adjusting for cost-ofliving differentials. Their average rates of program inclusion were among the lowest in the five clusters. And the tax burden on poor families was the highest. These states were also below the all-state average on seven of nine measures of policy commitment

The next cluster--denoted as "limited"--was stingy in some respects, but near the national average in others:

States in cluster 2 (Arizona, Delaware, Florida, Georgia, Missouri, North Carolina, New Mexico, Nevada, Oklahoma, Virginia) provided limited support. On average, these states provided slightly more income support than did minimal cluster states, although average inclusion and adequacy were at or below the all-state average. On Unemployment Insurance, and inclusion in programs that enforce private responsibility, states in the limited cluster were notably below the all-state average. These states provided support at levels near the national average in the areas of child care, preschool, and tax relief.

The third cluster--denoted as "conservative"--was actually more like the conservative welfare states of Europe than the movement conservative politics that would deny any sort of government support.  They were low to middling on some measures, but actively targeted the promotion of individual responsibility:

States in Cluster 3 (Idaho, Indiana, Kansas, Montana, North Dakota, Nebraska, South Dakota, Utah, Wyoming) are characterized as having a conservative policy approach. These states had average to low performance on policies relating to income support. They were notably low in the provision of employment support through child care and preschool and were average in tax policies benefiting the working poor. The states in the conservative cluster were high, however, in performance on policies that enforce private responsibility, with the second highest average rates of child support collection and the highest average rates of enrollment of AFDC recipients into mandatory JOBS activities.

The fourth cluster--identified as "generous"--had above average performance in all areas, except for a subset that was near-average:

States in cluster 4 (California, Colorado, Connecticut, Iowa, Illinois, Massachusetts, Maine, Michigan, New York, Oregon, Pennsylvania, Rhode Island, Washington) provided a generous package of assistance. These states scored higher than the national average in the generosity and inclusiveness of cash assistance programs. They were also well above the all-state average in per-child spending for child care and preschool. In other program areas-including tax policy, child support, and JOBS-their performance was near the national average.

The final cluster--dubbed "integrated"--scored well above average across the board. These states used every available tool in the state-level anti-poverty toolbox:

Finally, states in cluster 5 (Maryland, Minnesota, New Hampshire, New Jersey, Ohio, Vermont, Wisconsin) had an integrated approach to policy, with scores at or well above the national average on all dimensions. States in this cluster had policy commitment score above the national average in six of nine program areas. Rather than making tradeoffs among policy approaches, these states generally combined generous and inclusive benefits in cash assistance and in-kind programs with progressive tax policies, commitment to employment support through child care and JTPA, and enforcement of private responsibility through the highest level of child support collections and a strong commitment to the JOBS program.

In answering their fourth question listed above, the authors reported:

Policy Change over Time

If these clusters capture the underlying orientation of state policy officials, they would also be expected to be predictive of policy choices over time. Under the 1996 federal welfare reform, states were given new opportunities to innovate in many areas of social policy. To examine the over-time stability of the cluster solution (derived from 1994 data), cluster performance was compared in two program areas for the postreform year of 1998. Policies in the TANF program, which replaced AFDC, provide an example of traditional public assistance programs. Tax policies illustrate policies that have particular relevance for working poor families.

The researchers found strong indications of continued consistency within the clusters.  The following charts visually present their findings:

From this, they drew two major conclusions:

These results suggest  two conclusions. First, the cluster solution based on 1994 data appears to predict state performance in 1998 in at least two policy areas. As states have responded to devolution, they are continuing trajectories that reflect their original clusters' policy orientations. The ordering of clusters is stable over time. Adequacy and inclusion in the AFDC/TANF program converged slightly between the highest and lowest state clusters. In tax policy, however, the range between the clusters with the least- and most-generous tax thresholds increased by over $1000, and the range between the clusters imposing the highest and lowest tax burden increased by over $100.

Second, evidence is mixed that states are engaging in the often-predicted race to the bottom following welfare reform. The adequacy and inclusion of cash assistance declined in all clusters. But the declines were not of equal magnitude across the clusters: The reduction in welfare adequacy was small in the states in the integrated cluster, and the decline in inclusion was large in the minimal, limited, and conservative clusters while it was more modest in the generous and integrated states. During the same period, tax policy changes benefiting the working poor improved in all five clusters. States appear to be moving at different paces in their reforms to social policy and, across all the states, to be moving in directions that contract traditional, welfarebased assistance, and expand support for the working poor.

In short, it would seem that one of the main reasons we didn't see a "race to the bottom" was not because of the nature of the changes imposed, but because states were inclined to continue their pre-existing policy regimes.  However, it should be pointed out that the did so against a background previously described in this series, which included a continually improving economy, dramatically increased support through the earned income tax credit, and significant additions to flexibility and availability of funds that went into the writing of federal regulations.

To expand upon the results the authors reported, I decided to take their clusters and look at how they performed over time in terms of poverty rates.  Obviously, we have no assurance that the policies followed remained correlated over time, so this should only be regarded as an informal reflection on how different groups of states performed that were shown to belong together at a given point in time.

Still, it is striking that despite some relative movement among the middle three groups, the group providing minimal assistance--that most closely reflecting movement conservative "free market" ideology--had the consistently highest rates of poverty by a wide margin.  At the same time, the group of states using an integrated approach--all the policies available to states under US custom and law--consistently had the lowest raters by a significant margin, roughly half the rate of the minimal cluster:

Across the entire range of years, the poverty rate for Cluster 5 varied between just under 70% to just over 80% of the national average. While this is still higher than other economic powerhouses, it does bring us significantly closer to parity with them.  It shows that even with the cultural, legal and institutional constraints common to the US as a whole, we could do a significantly better job of dealing with poverty... if only we followed the best practices of our states, if only we had the political wisdom... and will.

Conclusion: It Doesn't Have To Be Like This

In short, America's continued failure to deal effectively with poverty does not have to be.  The myth that conservative "welfare reform" was a success is just one more way that we continue deluding ourselves, continue putting up artificial barriers to prevent us from being the sort of inclusive, big-hearted nation that our best values and highest inclinations say we should be.

This is the conclusion specifically drawn from comparing different groups of states to one another, as done in this final part, but it is also the conclusion supported by every other part of this series.  There is simply no excuse for the continued blight of poverty in the richest nation ever seen on the face of the Earth.  And the conservative failure to do anything meaningful about it is not a sign that nothing can be done, and that their failure should provide the new definition of "success."  Instead, it is a sign that we should stop letting ourselves be dragged down and defined by those who've definitively shown themselves to be losers on the national and international stage.