The Uneven Patterning of Welfare Benefits at the Twilight of AFDC: Assessing the Influence of Institutions, Race, and Citizen Preferences

Abstract

Scholars have been slow to test welfare state theories on the extensive subnational variation in the United States during the recent period of retrenchment. We assess institutional politics theories, literature on race and social policy, and public opinion arguments relative to levels of support in states' Aid to Families Dependent Children programs from 1982 until its elimination in 1996. Pooled time-series results demonstrate that the determinants of spending during retrenchment are mostly similar to those driving development and expansion. Pro-spending actors and professionalized state institutions limit benefit curtailment, while jurisdictions with larger African- American populations have lower benefits. Additionally, liberal citizens positively impact support and strengthen the effects of state institutions, but this effect is attenuated in states with larger African-American populations.

NOTES

Notes

¹ This study uses variables from 1981 to 1995 to predict outcomes in the period from 1982 to 1996.

² Because Stata uses the Cochran–Orcutt method of correcting for first-order autocorrelation, the first instance of each case is dropped from the model; thus, the analysis only includes 700 state-years.

³ Data on public employee union membership by state are not available during the early years of our sample.

⁴ These ideology scores are based on the rankings interest groups assign to legislators.

⁵ While it would be useful to consider the racial composition of the population in poverty as well as more detailed characteristics of AFDC recipients, we are unaware of such state-specific measures that cover the entirety of our sample.

⁶ Because conservative Southern Democrats were key players in shaping the original ADC program (Quadagno 2000) and these state politicians were often slow in their changeover to the Republican Party during the latter years of the twentieth century, we ran supplemental analyses including an interaction term between Democratic Control and Southern State in models 1 and 5 of Table 2. The interaction term was not significant, suggesting that lingering Dixiecrat politics was not one of the driving factors of benefit levels during our period.

⁷ Two additional models were run. In the first, the sample was limited to only those states during years in which they had not submitted for a waiver. In this model, the effect of Experimenting Neighbors was negative and significant. In contrast, when the model was limited to only those states during years in which they submitted for waivers, there was no significant impact of Experimenting Neighbors on AFDC.