Not Another Politics Podcast – Episode 22

It’s long been thought that giving people resources through government programs will get them more involved in politics. But this has always been a difficult question to answer in a controlled environment. That is until the 2008 Medicaid expansion in Oregon.

There was a research initiative done on that expansion, and our boss the Dean of the Harris School of Public Policy, Katherine Baicker, was involved. We parse through the results with her to see if we can get a new perspective on this question.

Transcript

William Howell:

I'm Will Howell.

Wioletta Dziuda:

I'm Wioletta Dziuda.

Anthony Fowler:

I am Anthony Fowler. And this is Not Another Politics Podcast.

We are still living through the aftermath of the 2020 presidential election. One notable, potentially promising thing about American democracy is that we had record levels of voter turnout. We had higher turnout than in, essentially, any modern American election in recorded history. That's a notable milestone in and of itself. But even with this record level of turnout, it was still the case that fully one-third of eligible voters didn't show up to vote.

There's a big debate as to why that's the case, and one argument that's out there is that there just aren't enough policies and government programs that reach these people that actually make them realize that government is a part of their lives. Maybe if there were, we could imagine different kinds of government programs that would bring these people into the fold and get them excited to vote that we haven't rolled out yet. That's an interesting thing to talk about. Do we have any evidence that there is something like a government program that could reach these people and get them to vote in ways that they haven't before?

Wioletta Dziuda:

Yes. Moving forward, we all hope that the new administration is actually going to implement some reforms, and perhaps reform even healthcare system. That also raises the question of whether we can expect the high turnout to continue even in a situation in which we don't have these divisive politicians, this divisive rhetoric, but we just simply have good policymaking. Will, you talked to someone who actually asks a question along these lines.

William Howell:

I did. It wasn't just somebody. It was the dean of the Harris School of Public—

Wioletta Dziuda:

Who is this person? Can't think of any name.

William Howell:

That's right, I talked to our boss, Kate Baicker, who is a health economist. She's done a lot of work on the effects of extending healthcare benefits on long-term earnings of people who receive these benefits, the health, obviously, of the recipients of these benefits, and she also looked into the effect of the extension of healthcare benefits on one's willingness to engage in politics, one's willingness to subsequently turn out and vote. Her findings are pretty compelling, and they speak to precisely these things which have to do with when you extend the reach of the government and the benefits of government to individuals, does that bring them into the political fold and encourage them to actually engage in politics going forward?

We're so pleased to have with us Kate Baicker, who is a labor economist?

Kate Baicker:

Health economist.

William Howell:

A health economist. Let's be clear. The dean of the Harris School of Public Policy, and the closest thing to a boss that I have and that Anthony Fowler and Wioletta Dziuda have as the members of this podcast.

Kate Baicker:

How's that working out for you?

William Howell:

We're holding on. We're doing the best we can. But we get to talk to you today, Kate, about a paper that you wrote, and it was published in the Quarterly Journal of Political Science this past fall, called The Impact of Medicaid Expansion on Voter Participation: Evidence From the Oregon Health Insurance Experiment. What happened in Oregon?

Kate Baicker:

What happened in Oregon was an unprecedented opportunity to study the effects of Medicaid on all sorts of outcomes. Before the ACA, Oregon was one of the states that had chosen to cover low-income, non-disabled, non-pregnant adults in its Medicaid program, but they'd run out of money, so they had closed the program to new enrollment as of 2004. In 2008, there had been attrition from the program. They had new resources from tax revenues. They decided that they were going to open the program for 10,000 more low-income adults to enroll.

But they knew that a lot more than 10,000 people might want to enroll in the program, so they opened a waiting list and they drew names from the waiting list by lottery. That is a perfect randomized controlled experiment through which to evaluate Medicaid. People signed up for the list in early 2008, and throughout 2008, some people were selected and got to enroll in Medicaid if they were eligible, some people were not. Our treatment group is those who were drawn in the lottery; the control group, those who were not drawn in the lottery.

Some of the people who were in the treatment group, who got selected, didn't end up on Medicaid. We're always going to use the lottery selection as the way we figure out the effects of Medicaid on outcomes. You can think of us as measuring the effect of two different things. One is potentially being eligible for Medicaid, and the other is being enrolled in Medicaid. The fact that the lottery was randomized lets us examine both sets of effects without worrying about the inherent bias that would be present if you just did an observational study of who was enrolled in Medicaid and who wasn't.

William Howell:

Right. By evaluating Medicaid, you and your coauthor Amy Finkelstein, and a number of other people with whom you've been working, looked at the health outcomes and the economic outcomes of the program on participants. Broadly, what did you find?

Kate Baicker:

I'm an economist. I always look at things in terms of costs and benefits. What are the costs of expanding Medicaid? Well, it's the extra healthcare that people use. What happens to healthcare use when you expand Medicaid? You might think, "Well, when you make healthcare more affordable by covering people with insurance, surely they're going to use more." Other people argued that Medicaid would be so effective in getting people out of the emergency room and into the doctor's office that it would actually save money. People would spend less on healthcare. Well, what we found is people spend more on healthcare. When you expand Medicaid, when people get onto the program, they go to the doctor more, they use more prescription drugs, they go to the hospital more, and they go to the emergency room more, not less.

William Howell:

Okay, so it isn't about shifting people away from the emergency room and into doctor's offices. There's an effect all the way through.

Kate Baicker:

There is. This was very surprising to policymakers. I think a lot of advocates of expanding Medicaid had hung their hats on the idea that it would save money as well as saving lives, and there is just no evidence that expanding Medicaid saves money. Maybe part of that disconnect comes from the misperception that emergency rooms treat uninsured people for free. Well, emergency rooms have to treat uninsured people who show up in need of critical care, but they're allowed to present them with a giant bill at the end. That payment or risk of payment was enough to deter a lot of uninsured people from seeking care in the emergency room for situations where there was some discretion.

William Howell:

What were the outcomes for people who participated in the program, the health or economic outcomes?

Kate Baicker:

The costs were the expanded healthcare use, but what are the benefits? What's the point of expanding the insurance? Well, we looked at two different types of benefits.

The first one I think is really underappreciated by people, outside of health economists perhaps, which is the financial security. Remember, insurance isn't just supposed to get you access to care. It's also supposed to keep you from getting evicted from your apartment because you paid your medical bill instead of your rent. People who have insurance ought to be more financially secure than those who don't. Now, we found that having Medicaid substantially improved people's financial security. They were 25% less likely to have bills sent to collection. They were less likely to have to borrow money or skip paying other bills because of medical bills. They were less likely to say they couldn't get care because of prior unpaid bills. They were much better off financially being insured.

William Howell:

Great. And in terms of health outcomes?

Kate Baicker:

There, it's a pretty nuanced story. Their mental health was substantially better. They were 30% likely to be suffering from depression, as measured by our depression screener. This is a huge improvement in unmet health need, especially among this population. The physical health outcomes, though, were much more nuanced. We didn't find any evidence of measurable improvements in blood pressure, cholesterol, diabetic blood sugar control, obesity, all of those chronic physical health conditions that people hoped would be improved by having access to Medicaid. This resulted in much hand-wringing in public policy circles. I think it paints a pretty clear trade-off for voters and for policymakers to think about, which is when you expand Medicaid, it comes at a substantial cost to taxpayers, but it comes with substantial benefit to new enrollees in the form of financial security, mental health, self-reported health, but not discernible improvements in chronic long-term physical health conditions. That's—

William Howell:

That's the trade-off.

Kate Baicker:

That's the trade-off.

William Howell:

Yep. Okay. Well, you said the word. You said the word voter, so we need to talk about voting. You and Amy had the thought that, well, maybe this program had effects that carried over into the political arena.

Kate Baicker:

There are lots of reasons to think that interactions with public programs like Medicaid or food stamps or welfare might affect people's voting behavior, and they go in potentially different directions. It could be that if you have more resources because you're not as worried about your medical bills because you are healthier and working more, and I'm just hypothesizing, you would come out and vote more. You would have more flexibility. You would be more engaged in civil society. That's a story. Another story is you feel bad about the government program that you're on, either because it's not delivering the benefits you had hoped for or because there's some stigma associated with it. Maybe that makes you vote less.

The lottery was in 2008. We looked at voting from 2008 through 2010. We could not look further into the future because of the further expansion of Medicaid eligibility in Oregon that happened after 2010. We found that Medicaid increased voter turnout in 2008 substantially, by something like 7% overall, which I think is an appreciable increase in voting behavior. Now, was this because more people registered or because there was greater turnout among already registered voters? We have a hard time teasing that apart, but the evidence suggests that perhaps it was new registrations, that being selected in the lottery and registering for Medicaid also increased your voter registration.

William Howell:

The effects associated with actually enrolling, the estimated effects are four times as large as the effects of simply being offered the opportunity. The 7% increase is the effect of actually enrolling. It's about—

Kate Baicker:

A little less than 2%.

William Howell:

A little less than 2%.

Kate Baicker:

Check for dividing by four. That's mechanically because winning the lottery increased Medicaid coverage by about 25 percentage points. Not everybody who won the lottery got enrolled.

William Howell:

Right. Okay, so that's the overall effect. How does this vary for different subpopulations?

Kate Baicker:

We found two subpopulations in which these effects were even greater. The first was men. The effects on men were more like 18%. We also found greater effects for residents of Democratic-leaning counties. That was about 10%. These effects were not evenly distributed across the population. They were concentrated in those two groups.

William Howell:

This effect, you observe it in 2008. You don't observe it, though, thereafter. It's this short-term burst in the presidential election which disappears.

Kate Baicker:

Yes, that's a great follow-up, that it dissipates by 2010. We don't know whether that's because there was only a boost that shows up in presidential elections and it would've been there again in 2012, or it's an effect that just dissipates over time.

William Howell:

I'm wondering if you're willing to lean back a little bit and to say, "Huh, what does this mean for the likely effects of other kinds of major health or welfare state programs on voting behavior?" Is there something special, do you think, about this being Medicaid per se? Would you expect to see effects in other domains, or do you want to hew closely to the data and say, "This is what we've learned, and this is all we've learned"?

Kate Baicker:

I'm happy to speculate, as long as we're clear that I'm just speculating. This is not something we can learn from our study.

But thinking about how other programs operate and the way we see people interact with food stamps, with cash assistance benefits, I would imagine that that very much affects people's views of how well their government is serving them and how important it is that policymakers' preferences and priorities line up with one's own. I wouldn't be surprised if there were similar results for other programs that affect people's well-being in a similar way.

William Howell:

If we could circle back to these two broad categories of reasons why we might be observing these resource effects versus these interpretive effects, what do you want to put your money on? In the realm of speculation here, do you want to pin this to resources? Do you want to pin it to a direct effect associated with how people are interpreting the benefits that flow from or the costs that flow from state activity?

Kate Baicker:

If I had to bet, and again, speculative—

William Howell:

Here we are.

Kate Baicker:

Can I say that enough times?

William Howell:

Keep saying it.

Kate Baicker:

I'm speculating here because it's not in the paper, so it's speculative. I would guess that this is more about the interpretation, the perception of government effectiveness and government's role in people's lives versus the direct resource effect.

We did do a number of qualitative interviews, hour-long freeform discussions with people about how people getting access to Medicaid affected their lives, how they felt about things, how they interacted with the healthcare system, how they were able to use their benefits. People talked about feeling like second-class citizens when they were uninsured. They were embarrassed to be uninsured. It made it hard for them to talk to their healthcare providers. They felt sheepish about not being able to pay for healthcare. This made them feel really bad about their place in the world.

That suggests to me people feel differently about their citizenship, their government, their community when they are insured versus when they're uninsured. That's putting weight in my mind on the interpretive path, but again, that is speculative. It's just a feeling based on these interviews and the stories people told about their lives, which were quite moving.

William Howell:

Voting is, in part, predicated on people feeling good about entering the civic space and their status within it. What we might say is that people who are really feeling awful and really feeling alienated and disenfranchised and marginalized are less likely to vote, in some ways paradoxically, because you think that's their instrument for effecting influence, but it doesn't work that way, in small part because you want to feel good about your status in the civic space, and when you don't, you're less likely to show up.

But what you have then is an intervention that attends to those concerns, that then brings people out, but curiously, brings people out just for the short term, and that may be happening... Would you sign on to this? What may be happening is that then, having become a part of the program, that just becomes woven into the fabric of their lives and then they move on. It's a short-term stimulus in terms of their thinking about their status as citizens, even though there may be longer-term implications in terms of their health and financial well-being.

Kate Baicker:

That's certainly one very plausible pathway. It's also possible that as people interact with the healthcare system, the reality that all of their health needs are not met sinks in too. We didn't find detectable effects on blood pressure, diabetes—

William Howell:

That's interesting.

Kate Baicker:

... cholesterol, and so it may be that people—

William Howell:

Got excited at the prospect and then—

Kate Baicker:

And then get a little disillusioned that they are still struggling a great deal.

Now, I want to circle back to which pathway seems most plausible. The resource mechanism for political engagement could be more broadly cast to think about psychic resources, cognitive resources. People reported being really stressed about being uninsured in the qualitative interviews. I was putting that in the interpretive bucket, but there is real cognitive load that comes with that. If you think about resources more broadly cast, there is more psychic space to engage with other things when this enormous burden of stress and worry is reduced. That's another pathway. Maybe it's not such a bright line between the two after all. It's different from monetary resources, but it's resources that are vital to functioning.

William Howell:

Okay. If we think about Medicaid expansion as a get-out-the-vote opportunity, these returns are not great, just in terms of the amount of dollars that are spent, that in the end we get a two percentage point return for rolling out a major initiative, and that knocking on people's doors or sending them reminders that the election is coming up may simply be a more cost-effective way to get these particular returns. Are you encouraged? You're encouraged.

Kate Baicker:

I'm always encouraged. No, that's not true. I don't know how to think about the cost effectiveness. I'm not an expert in get-out-the-vote mechanisms, so I don't know how many doorbells you have to ring to get one more person to vote. Clearly, the dollars spent on Medicaid had manifold effects. People were much less depressed. They were much more financially secure. They got access to more care. They were more likely to say—

William Howell:

Huge returns.

Kate Baicker:

If it has this incremental effect of also increasing voter turnout, is that expensive or is that cheap?

William Howell:

Or is it just a bonus? You put it in the realm of bonus, I'm guessing. It's like, "Great to see that as well."

Kate Baicker:

I think so.

Anthony Fowler:

Okay, so let's get into the paper. Will, do you want to just walk us through the numbers a little bit? We heard in the interview about the design. Healthcare was randomly assigned to some people, some people lost the lottery, and so we're going to compare people who got free healthcare from the government versus didn't. Just walk through the numbers a little bit more so we understand how big are these effects. How much does healthcare increase turnout in 2008?

William Howell:

You bet. Let's do this. There are two kinds of comparisons. One is simply comparing those people who were offered the opportunity to get Medicaid versus those people who were not offered. That's straightforward because of the randomization. What she finds is that there is a 0.7 percentage point difference between those two populations. That is, people who were offered the opportunity to get Medicaid were, on average, 0.7 percentage points more likely to turn out to vote.

There's another kind of comparison that might be interesting, which is what's the effect of actually taking up the opportunity to get Medicaid? Now, she doesn't have randomization there, but what you can do, what she does do in the paper, is use the lottery as an instrument to predict the probability that you took up the treatment. There what she finds, there was a 2.5 percentage point difference on average between those who took up the offer and those who didn't.

Now, if you want to talk about these not in terms of percentage points but percent, then you have to think about the change relative to a baseline. The baseline for this population was about 34% were turning out to vote anyway. When you talk about differences relative to that baseline, then that 0.7 becomes roughly two and a half, 2.1, and the 2.5 becomes about seven and a half, 8% difference between the two populations.

Anthony Fowler:

Okay, so these are not huge effects. If you just compare the people who won versus lost the lottery, they voted more, but not by a lot, by 0.7 percentage points.

William Howell:

So why would we think we would see any effects whatsoever when we think about the rollout of a Medicaid program on voter turnout? This isn't just a case of the authors pulling out a random outcome of interest and just re-running regressions they've been accustomed to running. There's a whole literature that speaks to the—

Anthony Fowler:

It could be, but nevertheless—

William Howell:

It could be. But there is a whole literature that speaks to how policy sets in motion political changes.

Anthony Fowler:

Yes, absolutely. This is a huge literature in political science, with lots of claims as to why we should expect to see something really big. Of course, it makes sense that politics and policy are closely connected to one another, and major government programs should affect people's political behavior, partly because you care about those government programs. Maybe once you become a recipient of a large government program, maybe you find that you like it and it's a good program and you want to vote to keep it going.

Anthony Fowler:

One of the famous examples that's been studied a lot and written a lot about is Social Security and Medicare, which a lot of political scientists claim essentially created senior citizens as this important voting bloc that vote at very high rates, often vote together for the same party, and are willing to vote directly on those couple issues that they care about very directly. If you buy that story, if you buy the story that big government policies can create active voting citizens, this should be a nice opportunity to observe it.

Wioletta Dziuda:

You can tell the story that when I don't have healthcare, when I don't even have access to healthcare, I'm motivated to go and vote because I want to change my situation for the better. Once I get this good policy, for example, I get healthcare, if this was my number one issue, I might be less inclined to actually spend my time on politics and trying to figure out whom I should vote for because I got the most important thing in my life from politics, so the stakes are lower. It's not obvious to me whether receiving some benefits from the government makes me thankful and more inclined to vote, or that it makes me thankful and I say thank you very much and I move on.

William Howell:

She says that.

Wioletta Dziuda:

Now the stakes are not there.

William Howell:

I think these fit under her category of interpretive effects, which is that, "Look, you won this benefit, you got it, and you might want to participate in political activity to ensure that those benefits continue to flow." That's one story. Alternatively, you might get the benefit and realize that, "Oh my God, our healthcare system is a mess, and this isn't what I thought it would be. Now I have to wait for long periods of time when I thought I was finally going to solve my health problems. In fact, I'm not." That might lead to disillusionment and anger, and she recognizes that. In the category of what she calls interpretive effects, they could go either way.

Wioletta Dziuda:

In their experiment, it seems that most of the effects were positive, but even if I assumed that we know that they are happy, that they actually like the program, I still don't know which way I should expect the turnout to go, "I'm happy, so I want to reward the politicians" or "I want to hold on to what I have, and I want to make sure that I have some good politicians because now I see the government works" or "I'm happy. No one is going to take it away from me because I think it's something that creates entrenchment and it's hard to take away, so now the stakes are lower." It's interesting that they were able to look at this question, but I want to point out that if you asked me before I read this paper, I would say anything goes.

William Howell:

The other category that she points to are these resource effects, which is that it has nothing to do with your understanding of government or the stakes involved in an upcoming election. What matters is that, by virtue of having participated in this program, my finances have eased, she shows that they have, and therefore I am in a better position to expend the cost associated with voting, either because I can pay for the gas to get to the station or because I have a little bit more free time because my anxiety is a little bit lower, these kinds of things. The immediate effect is improving my economic life, and that, in turn, improves the odds that I'm willing to actually vote.

Wioletta Dziuda:

To what extent do we believe that this is the main explanation for this 2.5 percentage point increase in voting in 2008, given that we know that this effect went away very quickly? It didn't seem to occur in local elections that happened in 2008. What's your thinking? I know that this is speculation, as Kate was very clear about in the interview, but what's your gut feeling? Is this the explanation for what we see?

Anthony Fowler:

It's possible. I'm certainly willing to believe that that is a viable candidate for explaining this result. The fact that you don't see an effect in 2010 is interesting. Maybe you say, "Okay, maybe it's not a resource thing, because they're still better off in 2010. Why aren't they voting more?"

Well, you can still tell a story. That's a midterm election, not a presidential election. It could be that we are mostly talking about very low SCS people who, maybe it's very hard to even induce them to get interested in the local election, but they can get interested in a presidential election. It's possible that those things all interact, and maybe it is partly a resource story, that I'm less economically distressed and now I can pay attention to the presidential election and vote in it.

William Howell:

Hey, if you're getting a lot out of the research that we discuss on this show, there's another University of Chicago Podcast Network show that you should check out. It's called Big Brains. Big Brains brings you the engaging stories behind the pioneering research and pivotal breakthroughs reshaping our world. Change how you see the world through research and keep up with the latest academic thinking with Big Brains, part of the University of Chicago Podcast Network.

So the effects are greater for men and for people who live in Democratic counties. If what you want to do is say that this is a resource effect, then the obligation is to say, "Why would we see it in 2008, not in 2010? Why would the effects we observe for men but not for women, and for people who live in Democratic counties and not for people who live in Republican counties?"

Anthony Fowler:

It is worth saying, again, these are small effects. We're not talking about huge numbers of people that were mobilized to vote to begin with. We don't really have a lot to go off when we're trying to make a guess about why are we getting this effect, I think. We don't really have any direct evidence that speaks to it. We would like to know if we had another experiment where we gave people private health insurance that wasn't connected to the government in any way, would that have had the same effect, in which case that would seem like a resource story, or is it really something about knowing that it came from the government that drives this effect? I don't think we know. I don't think we have a whole lot to go on here.

Wioletta Dziuda:

If we truly believe that there is something more than the resource effect, that somehow just learning that the government is functioning motivates me to vote, then I would say that perhaps the design is not the best to pick up this effect because I would envision that if I'm one of these 90,000 people who applied to be put on the wait list, I'm already informed that actually this expansion of Medicaid is happening. I'm informed that, look, 10,000 people are going to get Medicaid, and you're as likely to get it as someone else.

Yes, I get unlucky that I didn't get selected, so I might be a little bit less happy than the people who are selected, but when it comes to my information about what the government is doing for people, I have similar information to the people who actually got selected by the lottery. I could see how [inaudible 00:28:12] that if you had ability to compare the control group to people who actually never even heard about this lottery because they are above the poverty threshold, I wouldn't be surprised if you also see some sort of change in turnout.

Anthony Fowler:

Do we know what the people in the effective control group learned? I assume the state did not send them a letter that said, "You were in this lottery, but shoot, you lost out. Sorry, no healthcare for you."

Wioletta Dziuda:

At least in 2009, in the second wave of this expansion, they were explicitly told that, "You could be part of the lottery. Would you like to be part of the lottery?" That's a good question. My understanding when I was reading, and maybe I just made an assumption based on this 2009 lottery, everyone whose name was on the wait list, they knew that this lottery was happening, then I would say that they might be underestimating the effect of government program expansion and the effects of, "Okay, now I know government working," because every one of those 90,000 people should know that there was an expansion.

Anthony Fowler:

You raise a really interesting point, which is maybe... Usually, we think it's great when we get to randomize some treatment because it's great for causal identification and ensuring the treatment group and control group are comparable. But the fact that it was randomized, and the fact that it was known or it could be known to the subjects that it was randomized, changes things, as you say.

If you know that you got healthcare, but it was only because of the random lottery, I don't know, are you pleased with the government, or would you just think you were lucky? Similarly, if you know that you didn't get healthcare because you lost the lottery, are you mad at the government for even doing this lottery? Actually, maybe are the people in the control group, as you say, are they more likely to vote because they realize, "Oh, hey, there's this government program that I could be eligible for, and I should make sure I vote to expand the program"?

You're right, I think it really complicates things a lot. It makes it hard to generalize the estimates we're getting from here to the broader of question of, do large government programs that aren't randomized in this way, do those increase participation? It also potentially makes you worry about those underlying assumptions you need in the IV analysis. Is the exclusion restriction violated here? Is there something just about winning the lottery that affects your behavior over and above whether or not you actually get healthcare as a result of it? There's all kinds of interesting things to think about.

Wioletta Dziuda:

Yeah. In my mind, that might be an explanation for why the number, when we just compare people who won the lottery and lost the lottery, is actually relatively small. As Will said, it was like 0.6 percentage points, because people might recognize that I learned a lot about this program, but the final outcome was just a matter of luck. Then this additional two percentage points that they are finding when they just look at people who indeed were part of Medicaid, but that might be explained by additional resource effects and health effects and so on.

William Howell:

It's worth underscoring why this point is relevant for the two key comparisons that they're making. When it comes to the intent to treat, where you're just comparing those who were offered versus those were not, what you're suggesting is that the treatment isn't just the offer; the treatment is also the denial of an opportunity, that some people who were on the waiting list and may have been hoping for it were in some ways maybe affected by the fact that they considered they had a shot at something and then it was taken away from them. That may have—

Wioletta Dziuda:

That's also a question.

William Howell:

... changed their behavior. That's relevant for how we interpret the intent to treat.

Then there also is, for the treatment on the treated, the assumption is that these differences that we observe are driven by just those who took up the offer and actually received the Medicaid, which is one in four of the people who actually were offered it. If, in fact, those people who were offered who didn't take it up were, nonetheless, in some way affected by the offer, maybe they were happy, like, "Oh, you see, life is good. I'm really fortunate," or maybe that opened up other kinds of opportunities or, more specifically, led them to seek other kinds of opportunities that they wouldn't have otherwise received. Then we've got problems in using the intent-to-treat effect as the baseline to ratchet up from in order to assess what the treatment=on-the-treated effect is.

Yeah, these questions about what the nature of the treatment is and who's being treated are really important, both for matters of interpretation and for matters of estimation.

Anthony Fowler:

I'm going to take a strong position here. You guys fight with me if you want to, but I'm going to take a strong position and say that this is a striking study, mostly for the fact that we get a very small estimate. We have a really nice opportunity to credibly estimate the effect of a very large government program that should be very salient to the recipients of it, that should meaningfully affect people's lives, people say they care about it tremendously, and yet this huge government program that all the recipients should know that it's coming from the government, nevertheless, it doesn't affect their political behavior very much.

We've got a huge program, and we've got a huge literature that says there should be big effects. Nevertheless, there's not a whole lot going on here. One way to interpret this is this is actually a really compelling study, that maybe that old literature is somewhat overblown. Maybe these big government programs don't affect politics as much as people thought.

William Howell:

I want to argue with you.

Wioletta Dziuda:

Me too. Here we go. Gloves are off.

William Howell:

Okay. I'll point to two things, and then, Wioletta, you knock them out metaphorically. I think the literature, when it talks about how policy changes politics, when they talk about voters, what they're mostly talking about has less to do with turnout and more about how you vote, that you become aligned with a party you weren't previously aligned with. We don't have any data on that here, and there may well have been huge effects there that this—

Anthony Fowler:

They do talk about both, to be fair. They talk about both. Turnout is a big part of the senior citizen story. We're supposed to learn from that literature that the reason that old people vote at such high rates is because of Social Security and Medicare.

William Howell:

Okay, but there's this other piece. There's this other piece which is relevant for the turnout part, which is that the policy changing politics, what often people point to is the development of new organizational interests and interest groups, that come to represent senior citizens in this case, that's born of the policy. It isn't automatically that people say, "There's the policy. I now am going to be more likely to vote." It's because then a set of organized interests build up around that policy to protect it and then rally the beneficiaries of it to turn out to vote.

We're not going to see those kinds of effects in the short term. We didn't see them when we rolled out Social Security in the short term. This played out over long periods of time. Now, it's very hard over long periods of time to have the research design that they have. We're able to estimate the short-term direct effects doing a randomized field trial, whereas to say, "What does this mean for 20 years out?" we're not there yet. It's harder to do. Wioletta, what do you think?

Wioletta Dziuda:

I think something along the lines that you suggested, and I completely buy Anthony's—

William Howell:

Who's you? Who are you with here?

Wioletta Dziuda:

You, Will.

William Howell:

Okay, good. Good, I've got company.

Wioletta Dziuda:

But I'm still sympathetic to what Anthony said. I believe that this story has some merit, that by implementing a policy that benefits, or perhaps harms, a certain group of the society, you are making this group more cohesive, further, and their common interests become salient. It's easier to organize those people, and then you solve the collective action problem in this way by making them realize that they can vote as a bloc and then they show up and vote. I completely buy that this story might exist, and you are telling me there's literature that suggests that this story occurred.

But I think here what we have, we have a group of people who had exactly the same interests, and we are just comparing those who didn't get Medicaid to those who got Medicaid. If I were to examine this story, I would say they all should realize that they should vote as a bloc. "Yes, I'm unhappy because you got lucky and I didn't, but we still have exactly the same interests." For the people who already got it, to maintain the support for Medicaid expansion. And people who didn't get it, to continue supporting it so that we expand it even more. Maybe that's why we don't see this kind of coordination effect there, because we just have people with exactly the same interests in the dataset.

Anthony Fowler:

This is, again, where it's very relevant how this information was provided to the subjects in the experiment, because if when you win the lottery, if you get a letter that says, "Hey, you're one of the very lucky few that won this lottery," then, yes, you won't sit around and think, "Oh yeah, we should all get together to form a coalition." If you just get the letter and it says all of a sudden you have this big government program that you didn't have before, you might say, "Oh, this is nice. I should find people like me, and we should..."

Wioletta Dziuda:

But it's not obvious to me that that's relevant. Basically, no matter what I learn about this lottery and whether I won or not, no matter what happens, you are moving all these people in the same direction. It's comparing people who didn't get it to people who got it. It isn't really going to tell us whether this coordination somehow occurred by the virtue of this program bringing up salience of this group of people. All of those people are in the same boat and they have exactly the same interests, so—

Anthony Fowler:

Except the literature really is about the effects of becoming a recipient of some government program. It's not just that you could become a recipient or that you have shared interests with other people, because everyone could and is a recipient of some major government program probably. But the literature is really about the fact that I as a politician can change the voting population and maybe even benefit my own reelection chances by bringing out some big program that introduces a whole new set of people into the voting population and gets them all voting-

Wioletta Dziuda:

No, but it gets to the story—

Anthony Fowler:

It's really about receiving the thing. That's what the literature is about. Even before Social Security and Medicare were passed, senior citizens could've organized with the idea that maybe there should be something like Social Security and Medicare, but they didn't. We're going back to this whole question of, yes, is there even a compelling logic to this literature to begin with?

Wioletta Dziuda:

Yeah, so what's the story that they're putting forward? Is it the story of resource constraint, that you get benefits as elderly people and that's why now we have more time to engage in politics? Or is the story more psychological, we are raising salience of a certain group, and that group realizes now the stakes? They have common interests, and actually those common interests could be satisfied, and it's easier to organize them as a voting bloc.

I think that's important, because if it's the former, then I agree, we should pick it up in this experiment. One group got resources, the other didn't. But if it's all about you suddenly identify a group with common interests and they suddenly learn that they can actually affect politics because the circumstances are such that it's possible to actually implement policies that are affecting them, then I don't think this experiment would pick it up because both the control and the treatment group have exactly the same interest.

William Howell:

When I think about this broader literature, I think what it productively suggests is that the politics about possible policies, hypothetical policies, things that one might do, are very different from the policies that have been created, and that that's in no small part because the interest group ecosystem that's built around existing policies looks very different from the interest group ecosystem that's built around ideas of that which one might do, just as our understanding of the merits of policy changes as a function of whether or not they're in play or not. The standard story that says, well, there are these prior interests and preferences and expectations that an electorate has that lead to the adoption of policies, and that what this literature refers to as feedback loops is an anemic characterization of our politics. The literature is good. It's right to say that, in fact, that which people receive matters for the nature of their political engagement where we think about that broadly.

Now, having said that, Anthony puts it right, which is, "Okay, if you really believe that story, here is, arguably, the best design that's been put out there to actually test it, and the results that we get back are pretty small." There's something there, but they're pretty small and they're tenuous. They disappear two years later. I still think that there's reason to believe that these other effects play out, and maybe they play out over longer periods of time. I don't have anything like the quality or evidence that our boss presents to actually back up those claims, but I don't want to abandon them either.

Wioletta Dziuda:

The discussion convinced me that there are a lot of questions that I don't know answers to, and I think the paper opens up more questions for me than it answers. I appreciate the clean design, but there's a lot to learn still from, perhaps, other studies. I'm guessing there should be a lot of studies that we could look at about what's the impact of expansion of benefits on people's participation in political life.

William Howell:

I share that view. We're accustomed to thinking about randomized field trials as the gold standard of social scientific research, and in that way we would naively think that it settles matters, it's like, "Okay, we've got something definitive here to hold on to," but that what we have with this study, which is indicative of the conversation that we just had, is that it opens up all these questions. We're not exactly sure what the nature of the treatment is. We're not exactly sure why women versus men show differential effects, or why we see something in 2008 that then goes away in 2010. This is not an argument against randomized field trials, but the notion that they settle matters of longstanding dispute and study by big literature once and for all is a misnomer. I don't think that Kate would suggest that this was meant to settle anything once and for all, but it was a unique opportunity to investigate something of longstanding interest.

Anthony Fowler:

All right. We've already been debating my bottom line, and I'm going to stick with it. I'm going to say I buy the paper more than you guys do, in the sense that there's a big literature, there's a lot grand claims being made in that literature about the fact that government recipients should be induced to vote more, that recipients of major government programs should be induced to vote more. The evidence supporting those claims is not very good, is not very compelling. Here is the cleanest, most compelling experimental evidence we could imagine for some huge program, and look, the effects are just not very big and they dissipate pretty quickly.

So I'm willing to say I think this is a really interesting study that does pose a big challenge to that whole literature, that now the default going forward should be, "Okay, what is the more compelling evidence that supports this story?" If it's just that old people vote more than young people, that can't be the compelling evidence. It's got to be something more compelling that recipients of government programs actually are induced to vote more. Here's the best study we have on that, and the effects were pretty small. I'm willing to say a little bit more than you guys are and say—

William Howell:

That literature is now on its heels.

Anthony Fowler:

That literature has a challenge to try to come up with more compelling evidence, or even tell us what are the testable hypotheses that we should see, because a pretty clear testable prediction of their story doesn't hold up in what I think is a pretty compelling case.

William Howell:

Thanks for listening to Not Another Politics Podcast.

Wioletta Dziuda:

Our show is a podcast from the Harris School of Public Policy and is produced by Matt Hodapp. Thanks for listening.