State political cultures and institutions might also affect state spending on social welfare programs. But prior research often showed unstable results, and it failed to cover the wide range of program areas dealt with in this study. In political science, the investigation into the effects of political and institutional factors on redistributive policies and expenditures began with V.O. Keys study of southern politics (1949), where Key argued that one party dominance in the South limited political competition for voters and thus incentives to mobilize low-income families. Also, intra-party divisions made enacting major policy initiatives more difficult for parties once in power. The result, Key hypothesized, was a lack of real responsiveness to the interests of the have nots. This argument inspired a series of studies, beginning in the late 1950s, which attempted to isolate the roles of party competition, party control, and other institutional or political variables on redistributive policies or expenditures (often Aid to Families with Dependent Children [AFDC] spending) while controlling for the effects of state wealth or fiscal resources, state need, and federal grants (Dye, 1966; Dye, 1979).
In the earlier studies, the estimated effects of political and institutional variables proved small or nonexistent. Mogulls (1989) review of the literature found that neither party competition, hypothesized to create incentives to mobilize lower income strata, nor government control by the Democratic Party, viewed as more supportive of spending on social welfare programs, corresponded consistently to state welfare spending in multivariate analyses.
More recent studies amended these conclusions by using more refined models and measures that attempted to isolate the conditions under which political and institutional variables were likely to exert impacts. Brown (1995) claimed that inconsistent and weak effects of party on welfare and other redistributive policies might be due to differences across states in party coalitions. Where parties were divided by class, party control over government was more likely to influence state and local AFDC spending. Brown found that the effects of Democratic Party control of state government were greater where parties were class based. He also found, like other studies, that welfare effort was reduced by the percentage of the states population that was black.
Plotnick and Winters (1985; 1990) argued that the effects of political variables were underestimated because total welfare spending, the dependent variable typically used in the early studies, was not controlled by governors and legislators. State officials determined such policies as benefit and eligibility levels. Once these factors were established by law, demographic and economic changes interacted with policies to produce total expenditures. Using a measure of income guarantee for the dependent variable-the cash value of AFDC, Food Stamp, and Medicaid benefits for a nonworking family of four-they found that the size of the guarantee was significantly and directly related to interest group strength, measured by per capita memberships of liberal interest groups in the state, and inter-party competition but that Democratic Party control over state governments and local cost-sharing arrangements exerted no impact. They also found that the population density of the poor was positively related to the size of the guarantee, while smaller guarantees were associated with large numbers of illegitimate births in the state and high proportions of non-white families on welfare.
Other studies followed Plotnick and Winters lead and focused on choices under state control. Gais and Weaver (2002) examined state policy choices under welfare reform and also found that the racial composition of the welfare caseloads and a measure of state conservatism were associated with stronger sanction policies, shorter time limits, and immediate work activity requirements for welfare recipients. Kousser (2002) found that Democratic control of the legislature had strong and positive effects on changes in discretionary spending under Medicaid-but not on mandatory spending. Like other studies, he also found that the size of a states minority population was weakly though significantly related to lower levels of spending. Kousser found that the party of the governor had no impact on spending, nor did a measure of state ideology.
The literature as a whole suggests that political and institutional factors might influence state policies and expenditures on social programs. However, the estimated effects have been unstable across the studies. Some of the instability seems to be due to differences in the measurement of dependent variables (e.g., results are more consistent when dependent variables measure actual state choices, such as eligibility criteria and benefit levels, rather than total spending). But some of the instability still seems inexplicable, perhaps because of little real analysis in the studies of how states make decisions. Another weakness in the literature is that, with few exceptions (Kousser, 2002; Barrilleaux & Miller, 1988), most of the empirical analyses have focused on traditional cash welfare benefits, while the theories have usually treated social or redistributive policies as an undifferentiated whole. Thus, little theoretical or empirical work has been done on whether and how different kinds of social welfare programs-whether health or non-health, cash or services-might be affected by different institutional or political processes or conditions. Finally, no research has been conducted on how state fiscal capacity might interact with political and institutional variables in affecting social welfare spending. Yet such interactions would seem likely, because low fiscal capacity, almost by definition, would appear to limit the range of state political choices.