The forthcoming article “The Dynamics of State Policy Liberalism, 1936-2014” by Devin Caughey and Christopher Warshaw is summarized by the authors here:
Many political science theories rely explicitly or implicitly on models of policy change. This is true of both of the determinants of government policies, such as shifts in public mood or changes in the eligible electorate, and of the effect of policy feedback on political and social outcomes. Moreover, many of the most ambitious theories focus not on individual policies or policy domains, but on the character of government policy as a whole. In short, most theories of policymaking are both dynamic and holistic: they are concerned with changes in the general orientation of government policy.
However, the literature on U.S. state politics relies almost exclusively on policy indicators that are either measured at a single point in time or else cover only a partial subset of state policy outputs. Static measures are poorly suited to studying causes of policy change over time. And while domain-specific measures may provide useful summaries of some aspects of state policy, such as welfare spending or gay rights, they are imperfect proxies for the overall orientation of state policy.
In this paper, we develop a holistic yearly summary of the ideological orientation of state policies, which we refer to as state policy liberalism. This measure is based on a unique dataset of 148 policies, which covers nearly eight decades (1936–2014) and includes policy domains ranging from social welfare to abortion to civil rights. Based on these data, we estimate policy liberalism in each year using a dynamic Bayesian latent-variable model. Despite the disparate policy domains covered by our dataset, we find that a single latent dimension captures the bulk of the systematic variation in state policies. Indeed, our dynamic measure is highly correlated with existing cross-sectional measures of state policy liberalism as well as with issue-specific scales on gay rights, welfare benefits, anti-discrimination laws, and abortion policies.
We interpret our measure of policy liberalism as capturing a set of ideas and issue positions that, in the context of American politics, “go together”. Relative to conservatism, liberalism involves greater government regulation and welfare provision to promote equality and protect collective goods, and less government effort to uphold traditional morality and order at the expense of personal autonomy. Conversely, conservatism places greater emphasis on the values of economic freedom and cultural traditionalism.
Our dynamic measure of state policy liberalism opens up multiple avenues of research not possible with cross-sectional measures. Most obviously, it facilitates descriptive analyses of the ideological evolution of state policies over long periods of time. The map below shows the geographic distribution of state policy liberalism in 1940, 1975, and 2010. Blue shading indicates liberalism and red shading indicates conservatism. The map shows that the geographic distribution of policy liberalism has remained remarkably stable, despite huge changes in the distribution of mass partisanship, congressional ideology, and other political variables over the past seven decades. Throughout the period, Southern states such as Mississippi have had the most conservative policies. This holds not only on civil rights, but on taxes, welfare, and a host of social issues. By contrast, the most liberal states have consistently been in the Northeast, Pacific, and Great Lakes regions. New York, for example, has long had among the most liberal tax and welfare policies in the nation, and it was also one of the first states to adopt liberal policies on cultural issues such as abortion, gun control, and gay rights.
The overall picture of aggregate stability, however, masks considerable year-to-year fluctuation in policy liberalism as well as major long-term trends in certain states. These details can be discerned more easily in a plot of the yearly time series of four states— Mississippi, Idaho, Vermont, and New York—along with the average policy liberalism across all states. As this figure illustrates, states’ policy liberalism can change substantially between years. For example, until the mid-1960s Vemont’s policies were a bit more conservative than the average state, but since then Vermont’s policies have become steadily more liberal relative to the nation. Whereas it had been a laggard in passing racial anti-discrimination laws in the 1950s and 1960s, in more recent decades Vermont has been at the forefront of adopting gay marriage and other rights for homosexuals. Its welfare benefits and regulatory policies exhibited a similar evolution. The liberalizing trajectory of Vermont and other Northeastern states, such as Delaware and Maryland, have made the region’s policies much more unifomly liberal than they once were. By contrast, several Midwestern, Mountain, and Southern states have followed the opposite trajectory. Idaho, for example, became much more conservative over this period. In the 1930s–1950s, Idaho actually had some of the most generous welfare benefits in the nation, but by the early 2000s they were among the least generous.
Our yearly estimates of policy liberalism are illuminating for their own sake, revealing historical patterns in the development of state policymaking that would be hard to discern otherwise. But they also open up research designs that leverage temporal variation in state policies to explore questions involving the causes and effects of policy outcomes. For example, scholars could examine how the cross-sectional relationship between public opinion and state policy liberalism has evolved over time; estimate the state-level relationship between changes in opinion and changes in policy; or analyze how interest groups or electoral institutions moderate the link between public opinion and state policy. Scholars could also evaluate the policy effects of electoral outcomes or the partisan composition of state government.
The relevance of our paper extends well beyond the field of state politics. In addition to facilitating the study of topics of general significance, our measurement model could be applied to policymaking by local governments as well as in cross-national studies. Even more generally, our dynamic approach to measurement helps to illustrate the value of data-rich, time-varying measures of important political concepts like policy liberalism.
Devin Caughey is an Assistant Professor in the Department of Political Science at the Massachusetts Institute of Technology (caughey (at) mit.edu).
Chris Warshaw is an Assistant Professor in the Department of Political Science at the Massachusetts Institute of Technology (cwarshaw (at) mit.edu).
Speak Your Mind