Throughout the world, including in the United States, we are witnessing an increase in “democratic backsliding” – the state-led erosion and weakening of democratic institutions and norms. How can this trend be opposed and undermined? In our new article, we examine a potential economic consequence of backsliding: whether backsliding harms people’s evaluations of job opportunities. Were this this case, backsliding may make it more difficult for backsliding states to attract high-quality workers.
The most obvious approaches to combating democratic backsliding are political. But, it can be hard to notice backsliding until it is too late and the most vocal opponents are often those targeted with restrictions on political voice. Witness how several states have recently made it harder to vote. An external actor, like an international organization or the federal government, could take steps to limit backsliding but they must have the capacity and the will to do so. That will is often lacking.
Economic checks on backsliding are another a possibility. Both nations and the U.S. states compete for mobile, talented individuals to grow their economies. Individuals in affluent countries profess a clear preference for democracy over other forms of government. Though economic factors are typically of primary importance in migration decisions, people – especially the young and highly educated – also consider the presence of natural and cultural amenities.
Might democracy be an amenity that individuals consider when deciding where to pursue their livelihood? Some scholars have argued that federalism preserves freedom by allowing individuals to migrate to areas with better governments. And it does seem that violence drove many Blacks away from the South during the Great Migration. On the other hand, the violations of democracy we see now are minor compared to what Blacks experienced in the Jim Crow South. Contemporary migrants may not care much about democracy and sort into different states based on other amenities, or economic considerations may simply dwarf any concerns about democracy.
To investigate this, we performed two conjoint experiments using a sample of 368 university students and 750 individuals from the Amazon MTurk platform. Each respondent rated 10 pairs of hypothetical employment opportunities. We varied whether the job was in a state experiencing backsliding. Here, we used actual examples from states in recent years and months: a voter identification requirement, a law restricting teachers’ collective bargaining rights, a law increasing penalties for a protest that obstructed traffic or access to airports, a law limiting the ability of the Governor to make appointments, and the passage of a redistricting plan that gave the majority party two extra seats despite its declining vote share. To contrast with the backsliding treatments, in some profiles respondents read about a corruption scandal in the state or that the state had passed a benign piece of legislation (for more bike trails).
Alongside the information about backsliding, we randomized the partisanship of the area (Trump or Clinton vote), salary and other attributes of the job and location. We then asked individuals to rate the attractiveness of each offer and to indicate which job they would prefer.
Perhaps unsurprisingly, the job-related factor with the largest effect on respondents’ ratings was salary. We also observe, however, that for both samples all types of backsliding led to a significantly lower evaluation of the job, with voter id having the smallest negative effect and attacks on union rights the largest. The results of this experiment are presented in the figure below, which plots a quantity called the average marginal component effect for our key backsliding variables alongside the associated 95% confidence intervals. The AMCE is essentially the effect of each attribute on the outcome. We compare these effects to the corruption versus bike trail effect size in the top of each panel.
Figure 1. The Effect of Backsliding on Job Rating and Job
In the right-hand panel we see the results for which job would be preferred and again we see generally negative effects, though these are typically not statistically significant for university students seeking their first post-college jobs.
The estimated magnitude of these effects is substantively important. Our experiment revealed that a shift in salary from $75,000 to $90,000 is associated with an AMCE of approximately 0.13 to 0.17 (depending on sample) for the job selection outcome. The anti-collective bargaining law has a AMCE of approximately -0.05 to -0.07. Thus, people’s aversion to living in a state that strips collective bargaining rights from teachers is worth about $7,000.
Virtually all backsliding is being pursued by Republicans, and thus we may expect to see that Republicans are less sensitive to backsliding. In the figure below with the MTurk sample we see that this is indeed the case. Democrats consistently have a more negative reaction to backsliding than Republicans, though these differences are not always statistically significant. Backsliding politicians may actually find the fact that these actions repel Democrats (and “d”emocrats) to be a positive side effect.
However, it should be noted that several critical professions, like doctors, are staffed increasingly by Democrats.
Figure 2. The Effect of Backsliding on Job Rating and Job Choice, by Party
Clearly, individuals do not always act on experimentally-induced preferences, but conjoint experiments do often approximate real-world outcomes. Further, our design guards against purely expressive answers with respect to backsliding because we did not give individuals any reason to believe that we were primarily interested in this factor, and it was just one of many.
Overall, our results show that backsliding may make it harder to attract certain types of migrants and that businesses would need to pay a backsliding premium to hire them. While this would be an admittedly limited check on backsliding – since it would take place over the long-term–our results nevertheless suggest that backsliding regimes may pay an economic price.
This blog piece is based on the article “The Economic Costs of Democratic Backsliding? Backsliding and State Location Preferences of U.S. Job-seekers” by Michael J. Nelson and Christopher Witko, forthcoming in the Journal of Politics, April 2022.
The empirical analysis of this article has been succesfully replicated by the Journal of Politics and replication materials are available at the The Journal of Politics Dataverse.
About the authors
Michael J. Nelson- The Pennsylvania State University
Michael J. Nelson is Jeffrey L. Hyde and Sharon D. Hyde and Political Science Board of Visitors Early Career Professor in Political Science, Associate Professor of Political Science at Penn State, and Affiliate Faculty at Penn State Law. He studies the power of courts in democracies, especially the United States. I am the author of two recent books: The Politics of Federal Prosecution and Judging Inequality.You can find further information regarding his research on his website and follow him on Twitter @mjnelson7.
Christopher Witko- The Pennsylvania State University
Christopher Witko is Associate Director of the School of Public Policy and Professor of Public Policy and Political Science at Penn State. He studies inequalities in the policy process and how political and policy actors both respond to and shape economic outcomes, like inequality and unemployment. His most recent coauthored book, Hijacking the Agenda: Economic Power and Political Influence, examines how business and the wealthy influence the policy agenda in the U.S. Congress. You can find further information regarding his research on his website and follow him on Twitter @Cmwitko.