Republican and Democratic politicians offer very different agendas and proposals, but do they translate into real differences in outcomes? John Holbein finds that party control of government does not have any near-term impact across dozens of social and economic outcomes. But Jacob Grumbach finds that recent party control is associated with big changes in policy in some issue areas, sometimes producing real differences in directly-related outcomes like health insurance rates. Parties influence policy and some related outcomes, but perhaps not enough to declare one party better at improving well being.

Studies:Noisy Retrospection” and “From Backwaters to Major Policymakers
Interviews: John Holbein, University of Virginia; Jacob Grumbach, University of Washington

Transcript

Matt Grossmann: This week on The Science of Politics, do Democrats and Republicans get different results? For the Niskanen Center, I’m Matt Grossmann. Republican and Democratic politicians offer very different agendas and proposals, but do they translate into real differences and outcomes? Is one party better at delivering economic growth, social progress or improved wellbeing? Research shows they do influence policy and that sometimes matters, but it might not add up to one side’s consistently better performance.

Today I talk with John Holbein of the University of Virginia about his American Political Science Review article with Adam Dines, “Noisy Retrospection.” He finds that party control of state governments does not have any discernible near-term impact across dozens of social or economic outcomes. Red and blue states look different, but it doesn’t reflect real party impact.

I also talked to Jacob Grumbach of the University of Washington about his Perspective on Politics article “From Backwaters to Major Policymakers.” He finds that recent party control is associated with big changes in policy in some issue areas, sometimes producing real differences in directly related outcomes like health insurance rates. They took two different ways toward addressing the problem. Holbein and Dines looked broadly at partisan impacts across many different outcomes, finding little evidence that one side performs better.

John Holbein: So in our paper we explore whether political parties at the state level influence societal wellbeing in the year after a party takes power of a given state. So we’re comparing states controlled by Democrats to states controlled by Republicans and we’re looking at things like the health of the economy, schools, crime, health, social outcomes and the environment. And we use about 50 years of data, ranging from the 1960s to the early to mid 2010s, and various sort of quasi experimental methods to explore this question.

And what we find is that, in at least the short to moderate term after states change power, states controlled by Democrats and states controlled by Republicans are substantively and statistically indistinguishable in terms of how well things are going on in the state. So we think this work has a lot to say about how citizens can or cannot make decisions when they go to the ballot box to make decisions about who’s best equipped to govern states.

Matt Grossmann: Grumbach looked first at policy trends following up on outcomes for those where he saw growing divergence.

Jacob Grumbach: The main findings in this paper are that policy outcomes have really polarized across the states. There’s greater variation in major policy outcomes across states then there were a generation ago. And divergence in policy outcomes is really well predicted now by party control of state government. And overall I sort of push back on traditional view in the mid century US, which was that state governments were not all that important in determining people’s policy outcomes. This comes after a long period of nationalization of policy, major Supreme Court rulings and things that standardize public policies across the states. Sort of flagship examples like civil rights laws, civil rights and voting rights act standardizing, for example, public accommodations law across the states, or something like social security itself.

Matt Grossmann: Grumbach’s paper is part of a broader look at polarization and its impact.

Jacob Grumbach: One is that I’m just very interested in the causes and consequences of the nationalization of state politics, one of which you can see in this policy polarization across states based on party control and the causes. You Matt, are an expert in this area as well and have done a lot of thinking about the causes of these dynamics, but I’m looking forward to additional work uncovering I think three major types of causes of this broad shift since the 1970s, and especially since the 90s and 2000s in the states.

One is mass level sorting and polarization across states, where now, at least at a cross sectional level, the policy outcomes are more highly correlated with opinion now, at a single snapshot in time. When you do a difference to difference, it’s a bit less clear, but there could be some explanation and folks like Dave Hopkins have talked about this and Dan Hopkins as well, both have books about sort of a geography and mass politics in the states that are really important in this area.

A second level of change is in the environment of organizations and interest groups. Some of my work, a recent LSQ paper by myself, Grumbach, has really looked at increased coordination of individual level of campaign contributors in the states that individuals who give the state legislators are now much more likely than the early 2000s even, to also be giving money to and be affiliated with activist groups like an environmental group or the NRA or a anti-abortion group or whatnot across these areas. And that these individuals who are sort of organizationally mobilized are likely to be, these activists, are more likely to have their preferences be heard by politicians and be able to get a little bit of access and lobbying in. And that access and lobbying with an organizational affiliation, like being a member of the CR Club or NRA, is likely to be more affective than some random individual atomized out there in the ether trying to make that same case to a state legislator.

And it’s associated with a rise in these interest group activists and campaign finance is associated with state legislative polarization. So that’s a sort of mezzo level explanation. I’d add to that sort of work by folks like Alex Hertel-Fernandez on highly elite groups in the states, like The American Legislative Exchange Council or Americans for Prosperity and the Koch brothers network, that have also set the agenda in the states. And then finally they’re changing relief incentives too. So if you’re in a national partisan context, if you want to move up from state legislature to national politics, now your incentives are to play more ball with the national party agenda probably than before. We’ve seen similar changes in media that can potentially create a nationalized policy context decline of state and local media. Many huge causes to this transformational change and I hope to be more sort of quantitative in this area moving forward.

Matt Grossmann: Holbein’s project started from questions he got when teaching students about retrospective voting and economics.

John Holbein: So this was actually generated from a teaching experience that both Adam and I had had independently from one another. So both Adam and I teach an intro to an American politics class to undergraduates and some masters students as well. And we teach this course, we spend a day talking about this big question in political science, of whether or not citizens are sophisticated enough to hold government accountable. So during that talk we talk about this idea of retrospective voting. So for your listeners who are unfamiliar, retrospective voting is this idea that if we have a well-functioning democracy, when citizens decide who they’re going to go vote for, they look in the rear view mirror at past politician performance and they sort of punish policymakers for poor performance. So during the classes when we would talk about retrospective voting, we’d say, “Well it’s been pretty well established in political science research that voters punish elected officials if the economy’s not doing so well.”

And one of the common questions we get from our students from our undergraduate and masters students was, “Well, how much control does the government have in between elections in this short period of time on things like the economy?” And that would usually be followed up with a question like, “How long should we hold the previous government responsible for the current performance of things like the economy?” So we would give them an answer, but we’re never really satisfied with that answer. So that’s what sort of spurred this project. And what we found was pretty surprising to us.

Matt Grossmann: Holbein says there’s now a consensus that parties make differences for policies, but mixed evidence on their effect on outcomes, even though the public assumes that and it needs quick impact.

John Holbein: Then pretty clearly I think that, at least over the time period of study, from the 1960s to the 2010s, that Democratic and Republican governments are implementing distinct policy agendas. That they, when you have a Democratic government in power controlling the state legislature or the governor’s office, you see them implementing different types of policies. So there’s a great paper by Chris Warshaw at George Washington and Devin Caughey at MIT, that shows that when a Democrat is in, or Democrats are in control rather, they implement a distinctly more liberal policy agenda. That being said, it’s not clear whether or not that actually trickles down or transfers to societal wellbeing. So the scholarly consensus on this question is a little bit mixed, but you do have some scholars like Larry Bartels, political scientist at Vanderbilt, who say that political parties have a great deal of influence on societal wellbeing, things like the economy.

So you’ll see Doctor Bartels argue that parties have such a strong effect on things like the economy that we have to worry about political parties doing things like manipulating the economy leading into elections. But if we get away sort of from academic research and look to popular wisdom, the story is pretty much the same, right? The popular narrative is that the party in power really matters, right? We look to states like Virginia where I live and when we see a chamber of the legislature coming down to one seat and one vote, we watch really closely, because we have this sense that the party of power shapes things like the performance of the economy, how well kids are doing in schools and things like that. So for those reasons a lot of people take it sort of as given that these changes in policy direction will have distinct effects on societal wellbeing.

John Holbein: But as we started into this research project, we found that that literature was somewhat limited methodologically and so our results sort of challenge and poke at this assumption, that we’ve sort of made as political scientists and as public more generally, that political parties can influence these types of societal wellbeing metrics and the time pressure we place on elected officials by holding our elections every two to four years.

Matt Grossmann: He was expecting to find big effects on the economy, schools and crime based on how different states look.

John Holbein: I was thinking that there would be these big effects on the economy. Citizens act like there are big differences between state parties. Politicians take credit, like there’s a big effect, like they have a big effect. And states have a lot of levers at their disposal to try and affect change on things like the economy. I also was thinking that there would be effects on things like school performance given that school control is mostly a state and local policy area and then crime as well. So we thought in all three of those domain, the economy, schools and crime, that there was potential for an effect. And when you look sort of just without any sort of statistical adjustments, without any quasi experimental comparisons, there are big differences between states controlled by Democrats and states controlled by Republicans.

So over the period we study, states controlled by Democrats have higher, significantly higher unemployment rates, lower incomes, higher rates of crime and worst school outcomes. But of course that could be attributable to other factors. It could be not the political party in power, it could be things like the culture of the state or the broader nationwide trend in these outcomes. If we have a recession, right? Or if schools are just improving over time naturally for other reasons. The composition of students are getting sharper or crime is hitting a high level for unrelated reasons to the party in power. So we really thought that we would see something with the economy goes in with crime. It turns out that those big differences when you just sort of naively compare Democrats and Republicans, they don’t last when you start to control for cultural and broader societal trends. But we were a little bit surprised by that.

Matt Grossmann: But Grumbach says we shouldn’t expect big differences until very recently, when policies started to diverge.

Jacob Grumbach: The Caughey and Warshaw Measure is great too, where it actually shows in the earlier time periods, up until, to some extent the 90s, sometimes see Republican state governments have, on average, more liberal policy outcomes on Republican controlled states. Really, some of us wouldn’t have expected much divergence in socioeconomic outcomes in the early period where, for example, George Wallace segregation now segregation forever. George Wallace was governor of Alabama into the 1980s. You had states like Arkansas, Mississippi, Louisiana are blue in state legislatures and also and often in terms of governorships. Until the 2010 wave election to present. And then in terms of policy change, major policies that we assumed were big deals or associated with states are not. So I was fascinated to learn that New York only passed a no fault divorce in 2010. A major socially liberal state. Time really matters here in generating our theories of policy difference and socioeconomic difference. Because you had Nelson Rockefeller, liberal Republican governor, you had George Wallace, very conservative Democratic governor. Up until recently, the state policy changes or party control changes are very slow and sticky in this way.

And then also, like my work suggests, we should be very thoughtful about which policy areas we should look for corresponding divergence and socioeconomic outcomes. Where you’re unlikely to find it on something like education or criminal justice my paper would suggest.

Matt Grossmann: Grumbach was addressing research that lumps many policy areas together. Looking at issue specific trends instead to see where polarization is strongest.

Jacob Grumbach: One important contribution I make I think is to separate these policy outcomes by policy area. So typically in a lot of American politics scholarship to gain some statistical precision and power will collapse policy outcomes to one or two policy dimensions, like a left right liberalism dimension. But as we’ve known throughout American history, especially when it comes to the states, there’s been often many different dimensions of political and policy conflict across race, issues like abortion and reproductive rights, economic policy in the welfare state. So separating that was important. And I do find, consistent with some APD scholarship and historical scholarship, that the parties have diverged across the states and policy outcomes in most of these policy areas. But in two of my 16 key issue areas, I did not observe increased polarization by party. And those are criminal justice policy outcomes and education policy outcomes.

And there’s pretty good qualitative evidence that the parties at the elite level, especially in the states, haven’t been as polarized on those issues in their policy agendas through even the nineties and 2000s. So there’s a lot of great work. I focus on criminal justice in the states, especially the tough on crime laws of the 90s and the fact that states of all forms of party control, pass those tough on crime laws and that follows an earlier convergence of the parties on criminal justice, highlighted by folks like Vesla Weaver and others. And then on education, similarly, there’s been some good work on the fact that the interest group coalitions around things like charter schools really work with both parties. So you don’t see a huge difference in things like charter school laws, or even education spending across Democratic and Republican states. That is changing your party control does not seem to change education or criminal justice outcomes that much.

Matt Grossmann: He finds big differences on social policies where conservatives have innovated.

Jacob Grumbach: There are systematic differences in social and economic policies. One of which you highlight in your new work, which is that to the extent that conservative or Republican State governments have made affirmative policy change and sort of added new [inaudible 00:16:16] policy to the books in the States, it’s largely on the social issues like abortion and to the extent they’ve been more effective on economic policy. It’s in sort of blocking changes that would be considered expansions of the welfare state and things like that. To some extent you’ve seen major conservative gains in labor, but really it’s much more about abortion policy. Conservative States have really innovated new ways to sort of increase the cost of obtaining an abortion in their States.

Matt Grossmann: Let’s dig in, Holbein and Diane’s look for short term effects that voters could respond to. It’s harder to assess any long-term effect.

John Holbein: So we sort of have the sense that we want policy to have an effect that quickly and when there’s an economic crisis or schools are doing badly. Or there’s a spike in crime where you sort of expect our elected officials to act swiftly to pass policy, implement that policy and see changes right away.

So that’s why we looked at the short term. There’s also a methodological reason that makes it kind of tricky to see long-term effects. Anytime you’re trying to trace out the changes in party control over time and trying to map that to things 10 and 20 years down the road. That’s just really methodologically tricky to do from an empirical perspective. It’s not so much that we think that this is impossible that that affects won’t come in the long-term. It’s possible that they do. It’s just really, really hard to try and trace the sort of who creates those effects and when do they arise.

It’s certainly not in the timeline of elections as we show in our paper, but it might be long thereafter and that sort of makes it, I think even more tricky for voters and for other vested parties to try and figure out who’s creating high performance in the economy, who is responsible for low performance in the economy.

It’s just this really tricky question to answer and when voters go to the ballot box, all they have at their disposal is what happened between now and the last election. So we think it’s really important to look in the interim. It turns out to be really tricky to try and look in the longer term. Given the fact that parties do change power in it over time, that’s part of what we’re using in our analysis is the fact that party control does change at the state level. It’s actually really hard to map those changes on the things that are happening long thereafter.

As best we can tell, if we can sort of say a party changes power in a given year and let’s just pretend like, oh, they’re not going to change for the next six or seven years. We still don’t see any effects, but we don’t know if that’s because the parties are switching back and forth going from Republicans to Democrats in that period. That’s the tricky methodological issue.

Matt Grossmann: They use two methods to work around stable state differences and look for impact of parties. Both show the same no results.

John Holbein: States are not the same. Right. Massachusetts is not the same as Alabama, right? The different states have different cultures. They have different compositions of citizens. They had different policy environments to name just a few. So if we just go out and make these naive comparisons that we were just talking about, we might be picking up on things that are unrelated to the party in power, but other sort of better instead about other contextual forces.

So to account for that we’re doing to approach that you mentioned that. So the first is a difference in difference. So this is an approach that leverages panel data or data that spans over time and it looks for changes in our policy outcomes when a political party in power switches. So when it goes from Democrat to Republican or Republican to Democrat, it looks to see if on average the policy outcomes, we look at change over that time period relative to States where the political party in power stayed the same.

So there’s sort of this two stage comparison, you’re looking at the change over time in the States where the party in power changes and you’re comparing and benchmarking that to States where the party in power doesn’t change.

So that’s the intuition behind the difference in difference. In the second approach, the regression discontinuity design, we’re looking at States that are really close to being sort of on the margin. They’re like right on the margin of being controlled between Democrat and Republican. So we sort of say like the assumption behind a regression discontinuity approach is that things could go either way based on forces outside of the types of things that would influence our outcomes of interest. So it’s effectively like the party in power came down to a coin flip. So we think of Virginia as a great example of this comes, down to one seat in the house of delegates and one vote actually.

Not all States have things like that, but there are a lot of States as it turns out that are really close to this threshold where it’s sort of almost random. At least it’s outside of the control of the key actors at play here who actually gets control of the legislature and the governor’s house. And so it turns out these two methods allow us to control for different factors like the culture of the state, broader economic trends in the nation and other social differences that are unrelated to who’s in power. There’s a trade-off to using the regression discontinuity and difference in difference methods here as your astute listeners will probably pick up on. We’re making comparisons between States that are close to switching or that actually do so over time. So there are many States in this category in the United States. Most States fell into this category where it could feasibly sometime in the future switch from red to blue or back blue to red.

But this does keep us from making comparisons between States that have been staunchly blue for a long time, like Massachusetts and States that have been pretty staunchly red like Kansas.

Matt Grossmann: Basic Texas versus California comparisons fool us because there are lots of non-party differences.

John Holbein: When we look to other States, we’re often picking up on the composition of citizens there. I also do education policy research, so in a school environment people are always looking to sort of the other school and that often has to do with like the types of students that are in that school. I think if you sort of aggregate it up to the level when we’re sort of having state envy, for lack of a better word, we’re often thinking about who is there, the composition of the state rather than who’s governing it.

Matt Grossmann: But Grumbach does show a recent party effect on health outcomes matching the effects of known policy changes.

Jacob Grumbach: So I find some divergence, it’s probably a little before Medicaid expansion and I was actually a little surprised. It does seem like SCHIP started in 1998. Does play a role and there’s some divergence by party control bit before that in the mid two thousands before Medicaid expansion. But still you’re right, I think it’s almost all about Medicaid expansion.

And this is the flagship component of the Affordable Care Act, which itself is probably the landmark piece of domestic legislation in the US of the past generation potentially maybe, I don’t know, Clean Air Act, but probably that more recently some health economists have come with a number of studies on mortality rates as a result of Medicaid expansion or non-expansion. And you do see this, my finding, translating to real physical health outcomes, mortality is in public health research, which I’ve gavel in mortality findings are the hardest ones to find.

It’s so sticky and so hard to change with policy. But you really do now see the estimates are around 16,000 preventable deaths that occurred in the non-Medicaid expansion States. And again, Medicaid expansion is essentially perfectly predicted by party control. So that’s an area where we should expect divergent socioeconomic outcomes versus other ones, not so much. But the poverty rate in the great Holbein and Diane study, they do show pretty robust findings on party controls effect on the poverty rate where democratic state governments reduce the poverty rate relative to Republican state governments. But that’s in the appendix.

Another issue here in that is how we think of like what is a large policy effect or party control effect on socioeconomic outcomes. And it’s really hard to change socioeconomic effects with policy. They are huge constraints. State governments, this is one reason why wanting to frame my paper around from backwaters to major policymakers is because States are highly constrained by things like fiscal federalism, low legislative capacity to pass major bills and things like that.

You do see, so the poverty rate finding is substantively this is in terms of standard deviations across States, right? So outcomes are standardized by standard deviations across States and the difference across States, it’s so wide. The US is so decentralized. The difference between Mississippi poverty and Massachusetts, other Northeastern poverty, it’s huge. These are almost different countries and to move in the poverty rate like 0.2 standard deviations, like by changing to democratic control, I find that a huge effect. If you’re thinking about such, that standard deviation is so large. Is like a 5% of the population being in poverty is like a standard deviation. That’s a lot of people in poverty. I consider that a big effect. And that’s also in associated with policies in the welfare state and redistribution. Any area that we should be looking for divergence in socioeconomic outcomes.

Matt Grossmann: Holbein says they do find in a follow up that affects on the poverty rate, maybe marginally significant, but one of 40 outcomes is less than you’d expect by chance.

John Holbein: So we do find some evidence that democratic States can sort of reduce the poverty rate. But I would caution folks on this it’s possible that this effect is just by chance by looking at a lot of different outcomes. So we look at 47 in total in this paper, there’s going to be one or two that are significant by chance. There are a couple of methods that we can do to try and rule this out and once we apply those methods, it looks like the poverty rate effect isn’t really that robust. It isn’t that striking, or it isn’t significant or substantive.

But I think that gets to a broader discussion of like why we would want to look at a lot of different outcomes together. It turns out that it’s really hard to think about which outcomes to look at because we did a lot of weighing back and forth to think about like, okay, where is a realistic effect?

Where would we expect things to show up, and in some of the types of things that we expected to see affects things like where income goes to the top of the income distribution. No difference there. We might expect that Republicans would sort of favor the rich on this dimension, but we don’t sort of see that in the short term at least. Poverty rates, again, not really a robust effect. This suggesting that their differences aren’t as big as we might expect, might expect Democrats to target that better.

So I would just caution folks in trying to say which political party is having better effects or bigger effects or the effects that we want to have. We need to be comprehensive in what we look at because otherwise we might be getting a skewed perspective of what affects political parties have.

Matt Grossmann: He says parties do affect policies, but the outcomes are in danger of being cherry picked.

John Holbein: There’s a broader literature on whether parties can affect policy outcomes and as you would expect, like the findings are somewhat mixed, you’ll sometimes find people finding significant effects like Jake does on sometimes I’m finding no significant effects like we do. So there’s a great review piece by Niklas Potrafke and the journal public choice that came out last year.

So he’s very thorough looking at the previous studies that have looked at the effect of party control on policy outcomes. So there’s actually a table in that paper that runs 41 pages long. It’s quite comprehensive. So what you see, if you look at that synthesis of the broader literature is three things. So the first is that scholars, when they do this sort of analysis of the effects of the party in power, they tend to look mostly at the effect of the governors themselves because it’s a little bit easier to model that and get data for historic data on governors and they tend to ignore the party of state legislatures.

So the second thing is that many scholars in this area is they tend to make comparisons that don’t allow us to control for factors that might bias our estimates. And then the third thing that you sort of see when you look at this broader literature as a whole is that people tend to look at like one or two outcomes at a time. So there are like really strong theoretical methodological reasons why we wouldn’t want to do that to just look at one or two outcomes that it would skew our conclusions of the effects that state governments have. So the literature as a whole is given us kind of an incomplete picture of how parties or States do or do not influence societal wellbeing. That’s why we wanted to do as long of a time series as we possibly could, look at as many policy outcomes as we could.

Just so that we weren’t get painting a picture that was skewed by the types of the government entities that we were looking at. The types of outcomes we were looking at or the methods we were using to look for effects. So I liked Jake’s paper a lot. I think the broader context of the literature suggests that we’re sort of going about this process of looking for party effects in a slightly incomplete way, that we should be more comprehensive, look at more outcomes, look over our broader time span, and think about like looking more than just at governors, even though it’s easier to get data on them.

Matt Grossmann: It’s hard to pick the right outcomes. So I think it’s better to look across many at once.

John Holbein: We overestimate our ability as social scientists to sort of identify the right outcomes to look at, that that decision is actually turns out to be informed by the results that we see.

So sort of identifying these outcomes that are proximate versus not. In the example you gave about States choosing to implement Medicaid expansion versus not, right. It could be tempting to look at some health types of health outcomes, mortality rates, death rates, things like that.

And then we might sort of say, oh, well if there’s a significant effect on those, those are approximate outcomes, right? We might be sort of working backwards based on the results, but then look at other outcomes in the health domain and say, oh well, they’re not as approximate because there’s not a significant effect there.

So I guess what I’m trying to say is we sometimes overestimate our ability to say what outcomes are approximate and which are not, which is one of the reasons why we tried to be thorough. The second is like a broader response. I think we actually want to know the answer to the broader question not so much of like do Democrats and Republicans have an effect on this one outcome. But our broader question we want to know is what is the effect on societal wellbeing. And it’s not easy to measure societal wellbeing. There’s no one single scale that measures all components of how well communities and individuals are doing. So you’ve got to be kind of thorough. You’ve got to look at multiple different metrics to see the complete picture of how things are going. You could think about outcomes where there are trade-offs, growth in the economy, versus the health of the environment being one example.

So I think it’s a little bit dangerous from a theoretical perspective. And also just from a bigger picture perspective. And again there’s sort of also this methodological risk that we’re just picking the outcomes that are statistically significant because those tend to be the ones that are easier to get published.

Matt Grossmann: Grumbach likes the large scale analysis and the short term focus. But he also thinks we should look at policy specific theories for likely specific outcomes.

John Holbein: The Holbein Diane’s paper is great where really carefully done methodologically and very transparent about where the data is available and not. It’s really great and it’s great to see attention to know results and sort of to what size can you have effectively rule out with this data and things like that. It’s really great.

So I also think the fact, their point about sort of electoral accountability is their main substantive theoretical focus. And that’s really crucial. And I actually, I can’t wait for them to engage with there are sets of formal theorists that engage with empirical behavioral work. So there was a great dialogue between, for example, Aiken and Martell’, and Ashworth & Bueno De Mesquita on interpreting sort of how theoretically an informed or uninformed electorate would operate in terms of generating responsiveness or quality democratic outcomes. And here too, this is not my area of expertise, but convergence on outcomes related to public policy is kind of Downsianism, right? It’s a second step of Downsianism, where if you have similar policies because you converge on the median voter across the parties then you may find similar socioeconomic outcomes. And similarly, if you believe in retrospective voting and you punish all the people who do worse than some objective criteria on these performance metrics, then you might see convergence in party performance on these metrics. That’s a theoretical debate that I hope to see emerge going forward. And that’s, I think, the main focus there.

My style of looking at socioeconomic outcomes is more akin to … I mean, there’s just a ton of work of policy analysis and things like health econ and public policy work that generates a sort of policy specific theory on how that’s supposed to affect outcomes. And that’s different than summarizing the party difference across many, many socioeconomic outcomes. So yeah, huge trade off. It’s great to see this work. And I think it’s important though to … Holbein and Dynes’ paper is from 1960 to 2010 or 2014 and I’m not sure how many people’s prior … My prior was that the party effect would be pretty zero or null, because like I was saying, you’re averaging the policy of and sort of socioeconomic effect of Nelson Rockefeller and Sam Brownback who are very different ideologically and in their policy, both Republican governors, against the average of George Wallace and Jerry Brown in California, two Democratic governors with highly different policy agendas.

So I do think going forward there’s some value to contextualizing your theory of party effects on any outcome within space and time. So it would be great to think more about Southern Democrats, for example, in this case, to do a 1960 to the present paper. And there’s not really a discussion of the big change was the South was the solid South and it’s not the Democratic party we’re used to in the 2010s. So I think there’s value in both and I hope this generates really wonderful new work, which I expect already is.

Matt Grossmann: But Holbein says their findings don’t seem to be a product of the long period they study. The null results do not seem to be changing over time, even with parties’ ideology shifting.

John Holbein: This was a big concern that we had as well, how much do we want to include these comparisons between eras when what it meant to be a Democrat and what it meant to be a Republican was very different. So it turns out, we did a sort of a robustness check in our analysis, and showed that even if you look in more recent periods when these policy differences are the largest and have started to come out, that we still don’t see an effect on policy outcomes. This short time period just doesn’t seem to be enough time, even when there are meaningful differences in policies implemented.

Matt Grossmann: But they could be changing in the future. Grumbach expects a lot of action to come in the states.

Jacob Grumbach: We are living in quite the political times in the US and the states are where a lot of the action is. Despite the unified Republican control with the Trump administration, they were not especially effective at passing major national policy aside from the tax cuts. The health care plan failed and things like that. So a lot of the action is still in the states from really all areas and types of coalitions. You mentioned the teacher walkouts, which are sort of a liberal coalition, teachers’ unions, very much in the Democratic Party, sort of interest group extended network, but really occurred mostly in more conservative states. So I actually don’t know in theory how that may … if they are end up influential over state education policy, how that will look in terms of policy polarization by party because they’re sort of liberal actors in often red states. The financial crisis around 2008 really decimated public school budgets and staffing levels and state universities and things like that. Maybe now, having the financial crisis a decade over with, we may start seeing some divergence, state funding for public education.

You do see a tiny bit of emerging divergence, I can’t remember actually, either on K through 12 or higher education spending, but they may widen. Socioeconomic outcomes in health, I think will increasingly widen. Actually, a bigger thing though, is regional economic divergence has become so extreme and there’s been new great studies by … Larry Summers put out a great assessment of economic dynamics across regions. And you’re seeing really a divergence between the sort of coastal professional economies that are in high-tech industry and things like that and what are considered the “left behind” regions with real high levels of unemployment and low labor force participation and real crisis with the opioid crisis and things like that. And that’s all correlated with party control of government. Those high performing states and regions tend to be more Democratic, but it may be something beyond party control.

Matt Grossmann: Holbein gives the example of gun policy where differences might actually eventually show real impacts.

John Holbein: I mean, there’s certainly this dynamic in gun policy, right? There’s been a number of articles that have come out in the past few years that have things to say about the right to carry laws, the background checks, the waiting periods. So there’ve been a number of articles that suggest that those have quite stark effects on rates of crime. So I think that would be interesting to see as well. As we evolve to a world where gun policy means something very different in Democrat states, or at least the types of gun policy we see in Democratic states are very different from the gun policies that we see in Republican states.

It’ll be fascinating to see if these non-existent differences that we have documented over the past 50 years, if those continue to stay in place or if there are big enough effects that are immediate enough. I mean, we were talking about gun violence and gun crime, those effects, there are reasons to suspect that you might see when a policy changes that effects kick in right away, given that the outcomes that we’re looking at are happening in real time. Once a gun policy comes into effect that restricts access or carrying rights, that that might have an immediate effect and that might happen before the next election. So that would be interesting to see an interesting next step for the next wave of scholars in this area to sort of say, “Okay, there’s this area where the policy effect could be immediate. We have changes between how Democratic and Republican states are implementing it. I think moving forward, that’d be really fascinating to take a look at.”

Matt Grossmann: Grumbach is anxious to look at criminal justice policy changes, which now could be converging.

Jacob Grumbach: I would love to collect some new data on the post-2014 period. My data time series ends in 2014. In the couple of years leading up to 2014 you see a bit of divergence on criminal justice policy with Democratic states becoming slightly more liberal, so to speak on criminal justice, but that may just be a really highly localized finding, where since 2014 we’ve heard certainly more reports of more libertarian-minded, conservative actors in politics pushing for reductions in prison population and things like that. So it may be more bipartisan indeed post-2014. But at the same time, while there are huge important pushes from conservative actors, even the Koch brothers have held conventions on how to reduce the prison population and things like that, conservative actors want to not spend so much public money on criminal justice and things like that.

There’s been new concern about civil asset forfeiture and policing. Despite all of this, the California judicial ruling about that it essentially has unconstitutional health procedures and overcrowding in California prisons. That really led to one of the most … Even though it doesn’t get as much press because it’s not as openly politically debated, it’s actually led to a huge prison drawdown and major changes in California. It may be the case that still it’s the blue states like California are moving. There’ve been major conservative pushes and I would love to see what it’s looking like in the states.

Matt Grossmann: Both want to look next at federalism. Holbein is seeing interesting state, federal and local interaction.

John Holbein: We’re sort of looking to state governments to do the job of the federal government these days. Given gridlock in Washington, a lot of where we’re looking to for meaningful policy change is happening at the states. But I think this is a really interesting question and this is actually where Adam and I are heading with the next wave of projects that we’re looking at. We want to see how the various levels of government interact. So it might be the case that local governments actually have more control over some of these local issues, like the performance of students in schools for example. But all of these governments don’t exist in a vacuum. So it’d be really interesting to see if there are these interactions between the party control of local governments, the party control of state governments, the party control of the federal government together. There could be interesting, fascinating interactions between those things. So we’re actually collecting data to try and answer that question now. That’s our next big project in this area.

Matt Grossmann: Grumbach finds that nationalization may be making the states less responsive to policy success and more to party networks.

Part of my interest in American federalism is that they’re all the … Traditionally, American federalism has been seen as a super effective institutional structure of government where you have these state governments and through the authority of state governments you have these local governments that can … Part of it is that it offers this safety valve where actors out of power at the national level can do state and local changes, the fail globally act locally type of dynamic. But probably the other main benefit of federalism theorized is Louis Brandeis’ idea of laboratories of democracy. And that idea is that states, given how numerous they are, they offer the potential to experiment in new policies and sort of isolate any policy failures within the state and provide examples to emulate for other states.

Well, what I find is to the extent that was true in earlier time periods, I sort of, again, this 2000s or just maybe the ’90s, you see less and less emulation across states that are not controlled by the same party. Or now states tend to just emulate the policies of other states controlled by the same political party. And one way this could be a problem is that laboratories at democracy, some of the idea was that okay, success stories will be emulated. So if a policy increases economic growth or reduces unemployment, it should be more likely to be emulated because it shows sort of evidence of being a good performing policy. But that dynamic, again, doesn’t really occur if it … there’s not great evidence since sort of the late ’70s either, but there’s especially not great evidence now. Rather, what happens is states look to the most successful states also controlled by their party. So there is some success. Sort of economic success of a policy may matter but only if it’s a signal from a co-partisan government.

And that I think is explained by people like Kate Grimmel have looked at the divergence of interest group networks and my work similarly looks a little bit at that. And that could be state governments don’t have a ton of resources or time so they really outsource a lot of the policy research, like which policies are going to be successful or not, to groups increasingly. And these groups are increasingly aligned with just one or the other parties and may provide additional incentives to not look to out-partisan governments for examples of policies to emulate.

Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center. I’m your host, Matt Grossman. Thanks to John Holbein and Jacob Grumbach for joining me. Please check out “Noisy Retrospection” and “From Backwaters to Major Policymakers,” and then listen in next time.