Political Polarization has changed the way states are governed
New research has found that since the 2000s, U.S. states have been more likely to adopt laws enacted by other politically aligned states. In the past, policies tended to spread to states that were close geographically or demographically similar instead.
All but 5 of the first 27 states that expanded Medicaid coverage in 2014 under the Affordable Care Act voted Obama in the 2012 election. By 2022, there are 12 states that have not yet enacted Medicaid Expansion. Ten of them voted for Trump in 2020.
Medicaid Expansion has been a well -known political policy in recent times. The program itself, however, was less polarizing during its launch in the 1960s — back then, there was no correlation between the timing of the state’s initial adoptions and their voting patterns. Is this Medicaid case study just one example, or does it represent a general trend of polarization in policy decisions at the state level?
In our paper, we use a dataset of over 700 state laws to analyze how policies spread from state to state. We use two main data sources. The first is the State Policy Innovation and Diffusion (SPID) dataset, which represents typical state laws but may not necessarily cover policies that economists are likely to study. So for our second resource, we reviewed more than 11,000 working papers of the National Bureau of Economic Research (NBER) to extract the dates of adoption of state-level policies.
From the 1950s-90s, we found that policies were likely to spread between those states in geography near or by demographics similar. This may be due to neighboring states learning from or competing with each other (e.g., states that increase spending when their neighbors do), or policy needs that are spatially or demographically related.
But since the 2000s, the strongest predictor of a state adopting one policy has been whether another in politics it has already been adopted by the aligned states. This pattern persists on both sides of the political spectrum, as well as in policy areas and especially in the NBER dataset of policies.
But these recent trends in policy diffusion may emerge for reasons other than polarization. For example, states with the same ideology may now share information and policy needs or have stronger competition between states.
As two tests for these competing models, we investigated whether policy outcomes such as opioid deaths and poverty rates (reflecting policy needs) became more politically relevant, and whether the transition to inter-state (proxy for information flows and competition) can explain the results. Case no either.
Returning to the theory of polarization, we examine the role of party discipline: what happens when the governor and most of both chambers of state government are under the same party?
We found that when state governments united under one party, they were more likely to pass ideologically -aligned policies to their party — but only since the 1990s. Both party control switches had no effect then.
Based on the evidence, we conclude that recent trends in policy dissemination are likely due to the rising polarization of state politicians. As another notable example, we saw similar signs of polarization in COVID-19 policies at the state level (e.g., shelter-in-place), but there was no role of partisanship in vaccine mandates that existed since the 1980s (e.g., chickenpox mandates in schools.).
This paper builds on a rich literature on political science but also holds important implications for economists, who have long examined the impact of state laws such as minimum wage and voter ID requirements with little regard. the diffusion process. From a methodological perspective, these results may inform the empirical approach behind such policy analyzes, such as the commonly used variation design. In a broader way, the evidence casts doubt on the quality of the match between states and the policies they have adopted. Do states now choose policies according to citizen need, or partisan demand?
Learn more about our disclosure policy here.