Devolution Paradox and the US South

Abstract

There is an inherent trade-off in federalism regarding representation and equity. With the devolution of responsibilities to regional governments offering increased representation, there comes the inequity of services provided by some governments and not others. However, citizens in these federations may want the best of both worlds—desiring regional control but without the regional policy variation. This tension, dubbed ‘the devolution paradox’ in a study of federal countries in Europe, was less apparent in one key group of regions—those with historic ties through language, culture, or experience. This article examines US respondents to identify whether the devolution paradox is evident in this country and looks specifically at the South, a region with long-standing distinctiveness. We find that there is evidence of the devolution paradox among the American public and that the South is indeed different.

Key Words:

• Federalism
• South
• devolution
• public opinion

Disclosure Statement

No potential conflict of interest was reported by the author.

Notes

1 Kincaid and Cole (2005) and Cole and Kincaid (2006) examine responses by region to questions concerning trust in government, worst tax, governmental efficiency, respect, and power.

2 Moreno (2008) and Henderson et al. (2013), which both look at the effect of regional attachment within European states, are exceptions.

3 While the work is ongoing, the research findings are strongest at the aggregate, rather than the individual level. Thus this analysis of individual level data is especially important.

4 Because the focal independent variable in this study only measures attachment to the state in which a person currently resides, it would not capture whatever effect there may be from attachment to another former Confederate state. This measure would also not capture an effect from high state attachment to a home state that was part of the former Confederacy for people who now live outside of this region. Thus, the estimated effect for the study's focal independent variable on the devolution paradox, and support for non-uniformity for specific policies, is likely lower than it would be if it operationalized multiple state attachments. Future research that assesses attachment to multiple states could help to clarify this matter.

5 We think it is plausible to consider party identification a post-treatment variable for at least some people if high state attachment arises at a young age. Certainly, a past component of Southern identity was identification with the Democratic Party. Over the last 40 or so years that identity has become supplanted to a substantial degree by identification with the Republican Party. Nevertheless, the importance of this variable in the literature and its possible concurrent development with state attachment necessitates its inclusion. It also serves as a robustness check for our findings. If our focal independent variable has a statistically significant estimated effect independent of party identification on support for non-uniformity for certain policies then it strengthens our causal inference about the effect of high state attachment moderated by residence in a former Confederate state.

6 Due to the moderate correlation between this variable and that of trust in state government (a correlation of .439), we do not include the latter variable in any of the model results in this article. However, when we include this variable in each model, we find virtually no difference in the size of or statistical significance of the coefficient for this study's focal independent variable.

7 These percentages are based only on the portion of respondents who did not select ‘don't know' or who did not fail to answer one or both of the assignment or uniformity questions. Some 169 did not meet this criterion for the fighting unemployment/unemployment benefits comparison, while 140 did not for the crime/punishment of youthful offenders comparison. Only 122 answered ‘don't know' or failed to answer the questions used to compare food safety opinion.

8 We interacted attachment to a state with residence in another region and the interaction never even approached statistical significance.

9 The first estimated increase is the predicted probability that we calculate for this dependent variable, going from the minimum value of the interaction when state attachment is coded = 1 and residence in a former Confederate state is coded = 0, to the interaction's maximum value in which each component = 1. The second estimated increase is the predicted probability that we calculate for this dependent variable, going from the minimum value of the interaction when state attachment is coded = 0 and residence in a former Confederate state is coded = 1, to the interaction's maximum value in which each component = 1.

10 One reviewer suggested a three-way interaction among race, state attachment, and residence in the former Confederate South. We did analyse these but found mixed results (statistical significance for the interaction looking at unemployment benefits and punishment of youthful offenders, but none for nutritional standards and gay marriage) and given the difficulty in interpreting triple interactions, we are not reporting the results.

11 The first estimated increase for each policy area occurs when going from the minimum value of the interaction when state attachment is coded = 1 and residence in a former Confederate state is coded = 0, while the second estimated increase for each policy occurs when going from the minimum value of the interaction when state attachment is coded = 0 and residence in a former Confederate state is coded = 1, to the interaction's maximum value in which each component = 1.