Study findings emerged from several methods and data sources, including analyses of spending trends, econometric models of state spending on different types of social welfare functions, and the case studies of the six poor states (Arizona, Louisiana, Mississippi, New Mexico, South Carolina, and West Virginia). Five of the six case study states were in the fourth, or poorest, quartile. The sixth state (Arizona) was at the bottom of the third quartile. All had high social needs as measured by poverty and unemployment rates. The states also showed important differences in geography and recent changes in fiscal capacity and need. The case studies relied on discussions with public officials, administrative data on program spending and caseloads, and state budgets, reports, policy information, and other public documents.
The study’s major findings are as follows:
- States of less fiscal capacity spent less per capita on social welfare than states with higher per capita incomes. These differences between rich and poor states resulted largely from differences in states’ spending from their own sources of revenue. The distribution of federal funds neither greatly diminished nor greatly increased state differences in spending.
Poor states spent less per capita than rich states did on social welfare programs. As the chart at the top of Exhibit 3 shows, when averaged over the entire period of 1977 to 2000, per capita spending on social welfare was positively correlated with state fiscal capacity. The wealthiest 13 states (Quartile 1) spent an average of $639 per capita on social welfare programs over this time period, after expenditures were adjusted for inflation. The poorest 12 states (Quartile 4) spent only $408 per capita.
This $231 difference resulted largely from differences in states’ spending of their own tax revenues. Federal grants did not reduce absolute spending differences between rich and poor states. Average federal grants to the wealthiest states were $371, higher than the average grants to the other three quartiles, including the poorest states, which received an average of $339. However, because poor states spent less money overall on social welfare, that $339 constituted a large proportion (83 percent) of their total spending on such programs. By contrast, the $371 per capita from federal sources spent by the richest quartile of states made up a much smaller share (58 percent) of their total social welfare budgets. That is, more federal money went to rich states than to poor states, but poor states relied more heavily on the federal government to support their social programs.
Exhibit 3. Per Capita Spending on Social Welfare and Non-social Welfare Functions,
The chart at the bottom of Exhibit 3 shows that state fiscal capacity was also positively correlated with state spending on non-social welfare functions. The differences across quartiles again were due to how much states spent of their own revenues. However, federal spending played a smaller role in this component of state budgets. Although federal spending averaged over two-thirds (69 percent) of state spending on social welfare functions, federal grants typically made up only about one-eighth (13 percent) of total state spending on non-social welfare functions.
The study’s econometric models generally confirmed the importance of state fiscal capacity for spending on social welfare. State fiscal capacity was related to overall spending on social welfare even after controlling for other variables, such as social need (measured by per capita unemployment and poverty), federal grants per capita, population density, year effects, and long-run “state effects.”1 State fiscal capacity was even more strongly related to spending on non-social welfare functions after controlling for these other factors, perhaps because states relied more on their own-source revenues in supporting non-social welfare functions.
- State fiscal capacity bore a stronger relationship to spending on non-health programs than on health-related programs.
When we broke total social welfare spending down into more specific categories, the relationships between state fiscal capacity and state spending varied greatly. As the chart at the top of Exhibit 4 demonstrates, states of lower fiscal capacity showed consistently lower levels of per capita spending on cash assistance and non-health social services when compared to wealthier states. Spending on health-related functions showed a weaker relationship to fiscal capacity than spending on non-health functions. Levels of spending on Medicaid were highest for the richest quartile of states, and spending levels were lower for Quartiles 2 through 4. Yet Medicaid spending showed no differences among Quartiles 2, 3, and 4, indicating a weaker relationship to fiscal capacity than found for cash assistance and non-health services. In the case of payments to public hospitals, the relationship was the inverse of what was found for social welfare spending. The poorest states spent the most per capita on public hospitals.
Spending Per Capita and Per Poor Person on Different Types of Social Welfare Functions,
Averages for Fiscal Capacity Quartiles, 1977–2000
Because most social programs targeted low-income people, per capita spending levels failed to capture differences in the degree to which states met social needs. To understand spending from that perspective, we compared state spending per poor person. Differences between rich and poor states were more consistent, as shown in the chart at the bottom of Exhibit 4. Disparities between the top and bottom quartiles were even greater for spending on cash assistance and non-health social services, while spending on Medicaid and public hospitals showed positive though weaker relationships to fiscal capacity. For example, the poorest states spent only one-fourth the amount that the richest states spent on cash assistance, while the ratio of spending on Medicaid in poor to rich states was about one-half.
- Between 1977 and 2000, state spending on social services changed in major ways, and the changes differed between rich and poor states.
As Exhibit 5 shows, Medicaid spending grew rapidly among all states in the late 1980s and early 1990s, even after controlling for the high rates of inflation in health care services. But the strongest growth occurred among the poorest states, those in Quartile 4, probably a consequence in part of federal Medicaid expansions in eligible populations, services, and special funding for health care providers during those years. In the middle and late 1990s, Medicaid spending growth slowed or even stalled among the wealthier states — not surprisingly, given the strong economy and possible downward pressures on Medicaid caseloads exerted by welfare reforms. But the poorest states continued to show increases in spending even in the late 1990s. By 2000, differences between the richest and poorest states in per capita spending on Medicaid had narrowed substantially.
Changes in Average Per Capita Spending on Medicaid,
by State Fiscal Capacity, 1977–2000
Changes in real per capita spending on cash assistance varied even more across states, as Exhibit 6 illustrates. In the wealthier states of Quartiles 1 and 2, spending on cash assistance either declined slowly or remained static until the late 1980s, when spending began to grow. Spending then declined in the middle and late 1990s as caseloads dropped due to welfare reform and economic growth. Richer states thus reaped substantial fiscal savings in late 1990s from reductions in spending on cash assistance.
Changes in Average Per Capita Spending on Cash Assistance,
by State Fiscal Capacity, 1977–2000
(adjusted with GDP price deflator)
By contrast, spending on cash assistance was already low in poor states, so even large caseload declines in the late 1990s yielded small fiscal savings. State fiscal capacity became less correlated with spending on cash assistance programs at the end of the 1990s when compared to the early 1990s and before. In contrast to the upward convergence of states with respect to spending on Medicaid, states of different fiscal capacities converged downward in their spending on cash assistance after the early 1990s.
Spending on non-health social services programs, by contrast, showed growing divergence among states of different fiscal capacities, as Exhibit 7 shows. All quartiles gradually increased their spending on non-health social services between 1977 and 2000. But growth was faster among the wealthiest states, especially during the late 1990s, while spending growth on services in the poorest states (Quartile 4) lagged behind.
Changes in Average Per Capita Spending on Non-health Social Services Programs,
by State Fiscal Capacity, 1977–2000
(adjusted with GDP price deflator)
These different trends produced enormous changes in the composition of social welfare expenditures — and in the relationship between state fiscal capacity and how states allocated their dollars across social welfare functions. Exhibit 8 shows these shifts by tracking the percentage of total public welfare spending for different program functions in the wealthiest and the poorest quartiles (Quartiles 1 and 4) during the years 1980, 1990, and 2000. Over these two decades, Medicaid came to absorb a much larger share of the budgets of both rich and poor states. Cash assistance spending shrank as percentage of total social welfare spending in all states, though most precipitously in rich states, thereby reducing differences between rich and poor states.
Average Percentage of Total Public Welfare Spending on Different Functions,
by State Fiscal Capacity (Quartiles 1 and 4), 1980, 1990, and 2000
One striking change was the growing disparity between rich and poor states in their spending on non-health social services. In 1980, poor states spent a slightly larger percentage of their social welfare budgets on such programs than did rich states. By 2000, poor states were spending a much smaller percentage of their social welfare budgets on non-health services. Most of the remainder of their social welfare budgets — nearly three out of every four dollars — was devoted to Medicaid.
- Econometric analyses found that different factors influenced different social programs.
Based on estimations of econometric models of state spending on different social programs in all states from 1977 to 2000, we found different types of programs to be influenced by different state characteristics. State fiscal capacity was positively related to Medicaid spending (i.e., rich states spent more than poor states, other things equal). We also found that Medicaid expenditures were positively related to unemployment. Medicaid spending responded strongly to federal grants for social welfare, perhaps reflecting the attractiveness to the states in matching federal Medicaid dollars. The regression equations also indicated that Medicaid spending was higher in states with lower population densities.
Unlike Medicaid, cash assistance spending was negatively related to state fiscal capacity.2 Spending on cash assistance, however, was significantly and positively related to unemployment and federal grants. Thus, like Medicaid, cash assistance spending rose with unemployment, other things equal; and, though less strongly as Medicaid, cash assistance spending was encouraged by federal spending. Unlike Medicaid, however, cash assistance expenditures were higher among states with high population densities than among rural states.
Non-health social services — mostly services such as child care, child welfare, and energy assistance, programs serving the homeless, and a wide variety of others — responded to a different set of factors. State fiscal capacity and federal grants for non-social welfare programs strongly affected spending on these diverse services; also, spending was higher in low population density states. Spending on non-health social services was not sensitive to unemployment, like other program areas. In fact, a negative, albeit insignificant, relationship occurred between unemployment and service spending, other things equal. A negative, though significant, relationship also appeared between such spending and the percentage of people in poverty. These relationships suggest that a state’s total economic resources were critical in understanding spending for non-health social services.
In sum, the multivariate econometric analyses suggested the following:
- Unemployment pushed up spending on cash assistance and Medicaid but not on non-health social services.
- Growth in state per capita income enhanced spending on Medicaid and non-health social services but not necessarily on cash assistance.
- Rural states spent less of their money on cash assistance programs and more on health and non-health social services.
- States were especially responsive to federal grants for Medicaid.
- States of different fiscal capacity responded to different factors — or responded to the same factors in different ways. Wealthy states, for example, were more affected than poor states by economic cycles, fiscal capacity, and federal spending on social welfare programs.
By estimating regression equations for states of different fiscal capacities (i.e., for each of the four quartiles), we found that our general model of spending fit wealthy states more closely than poor states. In other words, poor states showed weaker relationships between the independent variables in the equations and all types of social welfare spending. For example, fiscal capacity, federal grants, unemployment, and population density were all significantly related to spending on cash assistance for the rich states in Quartile 1. Yet, only population density showed a significant relationship to cash assistance spending for the poor states in Quartile 4. Also, unemployment was positively and significantly related to Medicaid spending among Quartile 1 states, but the same coefficient was insignificant when the regression was estimated for Quartile 4 states. Similar differences in the degree of fit were found in equations for non-health social services, payments to public hospitals, and non-social welfare services.
But we found important exceptions. For Medicaid, federal spending for social welfare was at least as important for poor states in affecting state spending as it was for wealthier states. Population density was also important in many of the equations for poor states. Finally, our analyses of state effects revealed that poor states showed long-run differences in the ways they allocated their social welfare budgets, especially between cash assistance and health care programs, including Medicaid and payments to public hospitals. For example, poor western states showed higher levels of spending on cash assistance programs than did poor southern states, and they also showed lower levels of per capita spending on Medicaid. Poor states thus seemed to undergo smaller short-run changes in their spending patterns but did vary from one another based on relatively stable attributes, such as population density, region, and demographic composition of the state’s population.
- Case studies of six states of low fiscal capacity and high social needs indicated that the basic trends in spending found among poor states before 2000 continued after that year.
When we examined what happened in the six poor states after 2000 — and through 2003, when the case studies were completed — we found that the trends identified between 1977 and 2000 generally continued. Per capita Medicaid spending increased in most states between federal FYs 2000 and 2003, and the growth was greater among the six poor states in our sample than for all states. Using administrative data from the federal agency responsible for Medicaid, we found, as the graph at the top of Exhibit 9 shows, that the average nominal growth in spending among the six poor states between 2000 and 2003 was slightly higher than the average growth in Medicaid spending for all states.
Per Capita Spending on Medicaid and TANF Basic Assistance,
Averages for the Six Poor States and for All States
Per capita spending on Medicaid, 1997–2003.
Per capita spending on TANF basic assistance, 1998–2002
Some of this growth was due to the fact that some of the poor states (notably New Mexico and Arizona) that had spent relatively little on Medicaid in the late 1990s greatly expanded their programs in the early 2000s. The most striking change occurred in Arizona. State voters overrode years of legislative opposition to Medicaid expansion in 2000 by enacting a citizen’s initiative, Proposition 204, that greatly expanded Medicaid coverage to include households up to 100 percent of the federal poverty level. Enrollment grew from 575,000 in January 2001 to 902,500 in January 2003. Louisiana also expanded eligibility for children and aged and disabled people in FY 2003, and it increased eligibility among pregnant women to include all those under 200 percent of the federal poverty level. Other major factors driving up Medicaid costs — and noted by our respondents — included expansions in enrollments due to the weakened economy and SCHIP outreach efforts and the substantial increases in prescription drug costs.
In recent years, TANF basic assistance constituted an important part of the cash assistance component of the Census data. Spending on this assistance continued to drop among all states between federal FYs 2000 and 2002, with an average decline of 13 percent, as the graph at the bottom of Exhibit 9 shows. However, as was also true before 2000, no average decline occurred among poor states in cash assistance spending. Mean cash assistance expenditures remained comparatively static among these states, as before 2000.
The econometric analyses indicated that non-health social services were most likely to show a cyclical effect (i.e., be vulnerable to cuts when government revenues declined). Our case studies confirmed that the greatest pressures on spending in the poor states were imposed on non-health social services and expenses. Administrative expenditures were often severely curtailed during the recent state fiscal crises. South Carolina’s department of social services saw a 26 percent reduction in staff between FYs 2001 and 2003. Arizona reduced its human service workforce despite substantial increases in Medicaid and cash assistance caseloads. Administrative cutbacks were greater among poor states than among others. For example, TANF administrative costs per capita declined by 22 percent among the six states between FYs 2000 and 2002, while the average decline for all states was only 5 percent.
Legislatures often cut state spending on child welfare programs, including staffing and payments for protective services, foster care, and adoption. However, the effects of some of these cuts were sometimes mitigated by the greater use of federal TANF dollars to support such programs. A number of programs deemed less critical to basic agency goals were cut or eliminated, including parenting and fatherhood programs, employment services, job development positions, and pregnancy prevention programs. Some discretionary cuts in child care were imposed. However, because most of these states put small amounts of state money into their child care programs, only the two states that had put larger sums of TANF or state money into their subsidy programs in the 1990s (Arizona and Louisiana) made major reductions in their child care spending.
- Case studies found that, in poor states, different processes influenced spending on different social programs.
The case studies of the six poor states revealed important differences in the processes affecting spending on different social welfare functions. Cash assistance spending in low fiscal capacity states was largely determined by policies established through state legislation. These policies were changed infrequently in the six case study states. A flurry of policymaking around AFDC waivers and TANF cash assistance occurred in these and many other states in the middle and late 1990s. By 2003, however, TANF and its policies drew little attention from legislators in these six states and changes in earned income disregards, maximum benefits, time limits, and sanction policies had changed little since the late 1990s. For example, maximum benefit levels for AFDC and TANF assistance had changed only once or twice in most of the six states during the last 20 years. Debates over these policies also seemed more ideologically and racially charged than most other social policy questions.
Our site visits indicated that Medicaid spending processes were vastly different. Federal mandates, considerations of federal match rates, and active participation of organized provider groups that served as advocates for various parts of the program often dominated Medicaid spending. Perhaps because of the involvement of such groups, Medicaid was also a more politically salient program area. Unlike “welfare,” even the post-reform versions, health programs regularly engaged the attention of elected officials, including legislators, governors, and top administrators.
Among the six poor states in our sample, organized providers were especially strong in the four rural and southern states where Medicaid played a major role in the health care industry and local economy. In these states, hospitals and other health institutions played a large economic role; nursing home residents were more likely to rely on Medicaid; and a greater proportion of births, often more than half, were paid for by Medicaid. Also, most health institutions in these states relied on Medicaid, making it more of a universalistic program, in contrast to the situation in wealthy states, where some institutions depended heavily on Medicaid, while many did not.
Attractive federal match rates were often noted in debates over Medicaid budgets in poor states, especially during periods of budgetary stress. By themselves, however, match rates appeared to constitute insufficient conditions for understanding levels of support for Medicaid in poor states because equally strong match rates had been available under AFDC and were still available under child welfare programs, though with much less expansive effects. However, the combination of attractive match rates, federal minimum requirements, and strong constituencies seemed critical and meant that, even in poor and conservative states, medical assistance programs could enjoy substantial support.
Other factors might have bolstered support for Medicaid. In most of the states, the agencies that administered Medicaid were separated from those that administered welfare and non-health social services, and that separation insulated health programs from the often negative views about social service agencies held among legislators.
Our case studies indicated that, for the most part, non-health service programs elicited less political interest. These programs did not evoke the ideological divisions affecting cash assistance programs. Nor did they elicit participation of major provider groups in these low-fiscal capacity states. As a result, decisions in poor states about what programs to support in this diverse category of non-health social services were often delegated to and controlled by state administrators rather than legislators and governors. Choices regarding non-health services were more likely to reflect a sort of technocratic politics. Decisions were often made on the basis of state administrators’ assessments about the relative merits of different programs and their centrality in achieving major goals.
However, though administrators had some control over which programs grew and which were cut, they had little control over general fiscal constraints, and these changing constraints often dominated recent budgetary processes for this group of programs. Since the last recession, legislatures and governors typically instructed social service agencies in the six case study states how much they must cut from their total budgets, not precisely where to reduce services. Administrators often made cuts based on judgments about what services were critical to major agency goals, what programs were successful in terms of performance standards, what programs were needed to create or maintain a coherent “package” of benefits or services, what programs involved high stakes (e.g., “life or death” issues), and how and whether programs can be administered with fewer staff.