Policy Seminar With Casey Mulligan

>> Speaker 1: Welcome everybody, to this second meeting of the economic policy working group. So happy to have Casey Mulligan here, all the way from Chicago, a cold country, weather is very nice here, so appreciate that. And we'll be talking about Opioid Pricing and the Dynamics of Public Health. I can't think of something more difficult, more important to do.

But Casey was here, I just looked it up, just about a year ago, January 19th. He spoke about the backward art of slowing the spread. We'll see if there's a relationship with one year that's gone by. But Casey is a visiting fellow at Hoover, he's Professor Chicago, and most welcome.

So you may have a few questions as you go, but get started.

>> Casey Mulligan: Yeah, we should be a dialogue. I'm kinda gonna build a logical structure here, so I wouldn't wanna lose anybody in beginning and in the middle. So we sent around a paper, but I thought it'd be better for you guys and for me too, to bring together a couple of different papers and focus on what they have to say about the dynamics of public health.

I'm gonna start with a macro picture. This shows life expectancy over the past 40 years. This really is a macro picture cuz every single person in the United States contributes to that number. And life expectancy has been increasing for years. You could go back even before this picture, life expectancy is increasing.

Really, you can make the argument that's the major story of economic growth and living standards. But recently that trend, and this is all pre-pandemic, that trend not only stopped but reversed and turned negative. And opioids have a lot to do with that. I've kind of circled it, it's not a bright line in when the opioid issue became serious.

But I've kinda circled the opioid years here, and opioids had been used at the level they were used in, say, the year 2000. You wouldn't have had that drop, in fact, some kind of increase, would have continued. I wanna make the case today that we can understand a lot of this by looking at really increasing returns in opioid consumption at the individual level.

The time we have, I wanna focus on three results of that increasing returns perspective, ones you get something that looks like diffusion curve. If you saw just the quantity data over time, you might call it an epidemic. It's not that there's anything contagious underlying the process, although maybe there is, but it's just a nonlinear dynamics.

Second finding, really it's gonna be a group of findings, is it can look like the opioid demand curve slopes the wrong way. Meaning you increase an opioid price and people respond by consuming more opioids, and we'll be able to talk about when that happens. That's not something that would expect to happen all the time in all situations, but I'll kinda lay out the parameters for when that can happen.

And then some surprising corollaries to that, like why the race gap would change sign over time, why and when and among who. And then the third set of results I think we'll have time for is to understand why fatality rates from opioids would diverge for youth and adults.

So a single price change can reduce mortality among youth while increasing it among adults. So that single framework is gonna say all three of these things. So I'm gonna begin with a consumer budget set from microeconomics and I have the quantity of opioids on the horizontal axis and the vertical axis has the quantity of all other goods.

And in my setup there's gonna be two sources of that q opioids. One source is gonna be prescriptions and the other source will be illicit manufacturers. Especially, you read about it in the news, they will refer to heroin and illicitly manufactured fentanyl. Now what I've drawn here first in red, is that well let's suppose for a moment that we only had one source instead of two sources, just prescriptions.

Then we have a usual budget set, whatever the price of prescription is, you buy more prescriptions, you have less of other goods at more or less a constant rate. But now maybe there's a second option, the illicitly manufactured option of getting opioids. And this option has a fixed cost for the individual consumer, maybe also at the market level, but for the individual consumer it does.

I think of this relating to euphemism, I'll call it self dosing skills, measuring these small quantities using a needle on yourself. This is something you gotta get used to, and once you able to do that then you have these extra opportunities that open up. Now the way I've drawn it here, the marginal price is lower for the illicit we manufacturer.

Now it doesn't have to be that way, in fact, many years many situation is not that way. Then they wouldn't cross like this. And we wanna pay attention, well, in what situations is the marginal price of illicit, less than marginal price of prescriptions? Then we can bring in preferences and think about choices.

I've drawn one type of outcome here, choices on B, person chooses prescriptions as their source and that picture shows the quantity that they choose. But then in this picture we can look at well what would happen if there were an increase in the price of prescriptions. We're gonna rotate the red budget set and look what happens.

And this is the way I drew it here, people buy more opioids because it now justifies switching to that cheaper source and then they respond by buying more. So that's the first kinda preview of the demand curve kind of appearing to slope the wrong way. It's not that I've featured irrational people or somebody with weird preferences, they just have these two options and this really nonlinear budget set.

Now let me go back to the budget, just the budget set. Take the preferences off there for a second. This is gonna give our first glimpse maybe at the increasing returns. There's a critical quantity here, two star where if you're gonna be above that quantity, you wanna source from the illicit manufacturer.

If you're gonna be below the quantity, you wanna source from the prescriptions. So we have a situation here where the high quantity people face a lower price and the low quantity people face a higher price. That's the source of the increase in returns. That's normally not the way we think about supply and demand issues, when you buy more the price is gonna rise on you, but here it's gonna fall.

That's why using the phrase increasing returns. Now, this is important enough but I wanted to show the same idea, same model in a different space. So I'm gonna change the vertical axis from all other goods, a quantity, into a price or a marginal cost.

>> Speaker 3: Casey?

>> Casey Mulligan: Yeah.

 

>> Speaker 3: Would it be useful to have a vertical segment at the beginning, the right hand side of the black line, and then have the lower slope? Cuz that seems to be more what you're describing, there's a fixed cost, and then you have a lower margin.

>> Casey Mulligan: The fixed cost is the result of other goods.

 

>> Speaker 4: Yeah, but if you have q star equals 0, then your budget constraint is up at where the red line is. You should have a black dot up at the intersection with the red line and then a black circle.

>> Casey Mulligan: Okay, right.

>> Speaker 5: I don't think it's gonna make any difference for what you've done.

 

>> Casey Mulligan: No, it does. I don't believe that black line actually touches the axis. But my artwork wasn't anyway intended to show that detail. Right, and the model that I sent around in the paper is done algebraically, and then it says, you have an option of not paying the fixed cost and then not using that source.

Definitely, yeah, thank you. So here's drawing it the other way, where we look same horizontal axis, but the vertical axis is now a price. And now we can see the situation. Well, if I change my quantity from below q star to above, that changes my price in a downward direction.

Now, there's other ranges where you change your quantity and have no effect on your price. Maybe in the red region, your marginal cost is constant, maybe in the black region, your marginal cost, again, is constant. And that's where we see how you might get nonlinear dynamics, is people are moving through the constant marginal cost, reason you get one sort of time series.

And then as they move into the other region, you get an accelerated quantities, and then they slow down again. It's kind of diffusion type occur. So I'm just gonna show that here, we can do that with cost shifts, but I'm gonna show with the demand shift. So I'll put a demand curve in here, we start shifting it over time.

And at first, there's this early phase where, well, how much does the quantity increase? Well, it increases by the same amount the demand is shifting, because the marginal cost is a constant. And we may get into a middle phase where the same demand shift results in a much bigger quantity change, because people, as they consume more, they get a lower marginal cost.

And that makes them consume more again, a multiplier type of effect. Then you may have a later stage where people on the constant marginal cost case again, and then the rate slows down.

>> Speaker 6: The price here of the illicitly manufactured good is a subtle concept. There's the funerary cost, there's the monetary cost, there's the psychic cost of doing something illegal.

There's the stigma which is going to be affected, presumably by how many other people in your social network are using it. All of that can be changing at the same time. Just part of what's happening during this period, that feedback on the level of your black curve.

>> Casey Mulligan: Well, some of those things you mentioned are marginal costs, others are also fixed or quasi fixed.

And I mentioned maybe increasing returns at a market level, that would be related to the friends and things like that.

>> Speaker 6: Then when you operationalize this, there's hard work to be done in thinking about how to measure that price. Right, so the illicit manufacturers, are you just gonna take the street price?

 

>> Casey Mulligan: The main thing I'm gonna lean on is that there was an era when the illicit was expensive, and then we're now in an era where it's cheap. And you're asking me how would I know if it's 5x cheaper than prescriptions or 2.5x cheaper? I'd have to get into those issues.

But I think to say what I'm gonna say today, I'm not gonna have to have that level of precision.

>> Speaker 7: But the illegal stuff is often of low quality, mortality risk is substantially higher even now than legal stuff is, right?

>> Casey Mulligan: Some people say that now, when you're having fentanyl, you're getting a lot bigger quantity.

So here the question would be the risk per unit quantity.

>> Speaker 7: Well, my problem is, do you know how much you're getting?

>> Casey Mulligan: That argument has been made, maybe true, I'm not gonna try to dispute it, I'm not sure we should embrace it either.

>> Speaker 7: I've had fentanyl, it was fine.

But it was after a medical procedure.

>> Speaker 7: Most of you have probably had it.

>> I don't know.

>> Casey Mulligan: I've had it too.

>> Speaker 8: So just as a clarifying question, see if I understand the model. If you have a needle exchange program or some kind of subsidized approach to allowing people to more safely get these illicit drugs, then the nonlinearity tends to go away or is reduced in your model, is that right?

 

>> Casey Mulligan: And that fixed cost would be reduced and then the illicit would just dominate the prescription and you would get more traditional demand.

>> Speaker 9: Going back to Steve's question, I think some of the mechanisms that he was describing and that I'm sure you described as well, do you have the feel of turning something that's an individual decision into a pandemic, right?

Cuz I'm fueling the local drug dealer, then there are more local drug dealers, and it's easier for me to get stuff. And if it's costs are lower and maybe there's less stigma and all of those things make it more deciding individually.

>> Casey Mulligan: I agree with that, I'm showing an individual model today.

A lot of what's done in literature is not individual level data. So if it's happening at a neighborhood level or county level, predictions are pretty much the same. But if you had individual data, you could start to focus on some of these distinctions.

>> Speaker 10: It seems like the dynamics would be different, in the sense that you might be able to get more kind of charting type behavior more quickly into this kind of situation.

Might be useful. Network effect it's more like macro aggregate.

>> Casey Mulligan: When you have the complementarities across people, the aggregates are more elastic to whatever stimulus, I totally agree with that. What I'm making here is, even without those things, we can have the epidemic or diffusion type of dynamics just because the individual faces the nonlinear.

Okay, so let's look a little bit at some prices, this is some more macro pictures. And I wanted to start out with more familiar price series, this is the CPI for more or less textiles, boys apparel, over the past 40 years or so. Notice the axis here I'm using is log scale.

We're gonna look at some massive price changes, so I had to do things in logs. And there's been a very big change in the price of textiles, they're a lot cheaper, you can buy now, 150% more boys apparel with the same real income than you could 40 years ago.

 

>> Speaker 11: And that's imports or technology?

>> Casey Mulligan: Both, the synthetic revolution, I think is important in textiles. In the same question, I'll come up with fentanyl and heroin, there's both technology change there and imports. Here's the most extreme one we know of is television sets changed by a factor of 100 during this period, hence the big, beautiful television set we have here.

 

>> Speaker 11: That seems too much.

>> Casey Mulligan: You have to take that up with the Bureau of Labor Statistics.

>> Casey Mulligan: And the series I put together for illicit opioids, heroin and fentanyl is somewhere in between these two, but still we need the log scale being some big changes. Today, I'm gonna focus on this last change, which is rather sudden and quite significant in itself, factor of three or core change, really, fentanyl becomes available and in many areas totally displaces heroin and the prices twice.

 

>> Speaker 11: So this is not BLS. The other lines work.

>> Speaker 12: Can you say about the data a little bit now?

>> Casey Mulligan: The illicit opioids is a DEA, they go out there and sample, they've been doing that for years, purity and price of heroin especially. And then that heroin increasingly has fentanyl in it, and therefore more morphine gram equivalents of opioids per gram of whatever the dealer's handed.

And that's really the source of the big drop toward the end, that you go to the heroin dealer and you're getting at first mixed, and then now in many places, there's no such thing as heroin.

>> Speaker 13: So it's a different product.

>> Casey Mulligan: Well, you have to ask the people who use it, they do notice some differences, but the morphine symptoms is especially what they're after that, pretty good substitutes in that sense.

Again, you see these regions where heroin is just totally displaced by fentanyl, that's why I sometimes have trouble believing these nose droids to say, people didn't know they got fentanyl. Wait a second, in your area there's no such thing as heroin, you think you got heroin, it's not anywhere around, it's been totally beaten out, but it makes a good story, I guess.

 

>> Speaker 14: Have you got the same kind of price level for prescription?

>> Casey Mulligan: I'm gonna show you a prescription next, I don't have such a long time series, but again, that is something I show on a log scale. There was a big earlier than that drop I'm focusing on for heroin and fentanyl, you had to drop in prescription price from the consumer point of view, and Medicare Part D is a big part of it.

Medicare Part D was created, ends up purchasing the majority of prescriptions that are sold in the United States are purchased either through Medicare or Medicaid. Medicare is much bigger than Medicaid, but Medicaid's a factor in there too, and that's driving the price change. Now, that's a money price, there's a lot of regulation going on here, they're changing kind of the inconvenience, hassle costs, like Steve was talking about on the illicit side, but it's also true on the prescription side.

They're trying to put barriers to using, or maybe they would say abusing prescriptions, but this is just a money price. Now, these are indices, so you can't compare one another and say, opioids are cheaper than TVs, they're more expensive than TVs, they're indices, they're buffet changes. What I want us to lean on today is that, if you go back 20 years, 15 to 20 years ago, prescriptions were the cheap one, and heroin was there, but quite expensive.

And then today, or 2019, 2018, we're in a situation where prescriptions are expensive per morphine gram equivalent, the illicit market's getting you a lot more per dollar.

>> Speaker 15: Can I just ask a clarifying question, because I don't know the answer? On the supply side, is it that heroin is less elastically supplied because it's manufactured in a different way, it's kind of a natural product, whereas fentanyl is a synthetic?

Presumably it's manufactured quite different, I'm just trying to understand what the supply curves are for these different forms.

>> Casey Mulligan: These pictures, I actually took from another paper where I focus more on that side, fentanyl is perfectly elastically supplied cuz it's industrial chemicals. Of course there's an opportunity cost of making fentanyl, but it's like people have a little bit less gas for their cars or something like that.

Whereas with heroin, there's a manufactured part of the process, but they're also growing poppies in the field and you need more acreage and that's a lesson. So, the incidence of some of these policies could be pretty different in the fentanyl world than the heroin world. I'm not gonna talk about that today, but if it's interesting, I think it's important.

 

>> Casey Mulligan: Especially in the pandemic, the ability of the market to respond with a large quantity in a short period of time made a big difference in the pandemic, we saw that.

>> Speaker 16: What do we know about quantities?

>> Casey Mulligan: Most of my quantity information comes from death certificates, sorry to be morbid.

 

>> Speaker 16: In terms of drug seizures, the sense, how many seizures does the aggregate, it's kind of like the distribution of the size of the seizures.

>> Casey Mulligan: One thing that I have looked at is the seizures, they show the breakdown, how much meth, how much heroin, how much fentanyl, how much marijuana, how much cocaine.

And those I've kind of looked at, and the fentanyl quantities there expand massively, it's a share of what ends up in the laboratory at the police station. I'm oversimplifying a little bit, but my NBR working paper, I show that I forget the acronym for it, but it's the drug reports from the crime labs, it's all proportional.

Of course, most drugs don't get seized and never make it in the prime, that's why I focus kind of on the proportional analysis.

>> Speaker 17: I think it would be good to see that picture, because you showed us this huge drop in life expectancy, in the sense that if you show us this massive expansion supply of these ideas.

 

>> Casey Mulligan: Well, it follows the death certificates, what I show in the Nvarro working paper, it follows the death certificate. So at the same time, peoples death certificates are recording, this guy died from fentanyl, we're also seeing the crime labs are finding a lot more fentanyl per unit cocaine per unit marijuana, it lines up very nicely.

 

>> Casey Mulligan: It's a statistical matter.

>> Speaker 17: Yeah, I'm just saying, putting the picture would not.

>> Casey Mulligan: I have that in the paper, I didn't bring it today. The only quantities I'm gonna show you today are from death certificates, people deceased. Okay, so now we see a little bit what's happened to the prices now let's get some more of these predictions from the same theory.

I'm not changing the theory on you.

>> Casey Mulligan: We talked about there are two prices to think about they're moving differently. So let me draw a picture here where we hold the illicit price constant. We just ask, well, what happens to quantity as you vary the prescription price? Well, back in the day when the illicit price was quite high, kinda didn't matter what the prescription prices that was people source you get a kinda standard demand curve and that's what I've drawn here.

But we also know that the illicit price fell after 2010 a lot and that's gonna change this relationship. So now I'm gonna draw a second blue curve, moving along that blue curve, the illicit price is constant, but at a much lower level. And here illicit becomes an option for people, especially for the people facing the higher prescription prices, so you get this kind of a Laffer curve.

In fact, analytically, it's almost exactly a Laffer curve type of relationship, where over some range, higher prescription prices will mean less quantity, but over another range it'll be more. So what? Yeah.

>> Speaker 18: Just question out of my ignorance so if it's not important, say so. But there's another source of dynamics here, which is anybody who, for legitimate reasons, takes fentanyl or one of these other opiates, there's some small chance they become addicted.

And I don't know whether that's true or how important that is, but if it is, then the big decline in the opioid price prescription markets that you showed also is entering here. And there isn't really an object like the one you've depicted here because it's shifting over time as a function of past prices.

 

>> Casey Mulligan: Yeah.

>> Speaker 18: Is that an issue or?

>> Casey Mulligan: Past prices are gonna matter.

>> Speaker 18: So we're just abstracting from that.

>> Casey Mulligan: No, we're gonna use that a little bit. Let's come back to it and I'll make my comments then. The race comparisons are gonna be very interesting from the addiction perspective.

So let's come back to that when we get the race numbers. So what happened over time is you got an increase in quantity. In theory, not surprising because we had one of the prices fell. Okay, you're not surprised, but look what happened. We're going to a point in the demand curve where the prescription price no longer chokes off quantity.

There have been a number of studies not done by me finding it. In fact, in about the year 2012, 2013, 20 14, as they ramped up prescription regulations and made it harder for people to get prescriptions. They switched, almost in the aggregate, one for one, maybe even more than one for one over two at the time, it was heroin.

So they were finding that demand curve is pretty flat. Maybe it's even sloping up. I'm not sure the studies are fine enough to tell that difference. And they also said it has something to do with the illicit market as being different. Yes.

>> Speaker 19: Just a modeling decision about this.

So in the prescription market, the way that they limited demand is by making it more difficult to get prescriptions from doctors. Lots of my friends, they tell me about drug seeking behavior for patients, the doctors. If you give a prescription now, you're on some DEA list, and they're just double checking, they're linking and seeing how many.

So it's just much, much more difficult. Is the right way to model that a price increase in the prescription, or is it the right word, model at a fixed cost? Just the same way you're modeling a fixed cost effect of fentanyl in the market?

>> Casey Mulligan: I think of now, that's a little bit later than we're looking at here.

And certainly the studies I mentioned, we're looking more at where the pharma company changed the formulation of its prescription so they were more difficult to abuse.

>> Speaker 19: I think it would have. You'd probably monitor that red line in your first graph going down that gap.

>> Casey Mulligan: I mean, there's some of each.

You go to a doctor, he gives you some pills, and then you run out, and then you gotta go to another doctor and talk him into it, right? And then, so there's a marginal cost and a fixed cost. You get a skill at this, you start to figure out which doctors are easy.

So I think there's both types of costs, but that would be those kind of regulations were coming in. Especially, I think the paper I sent around has a whole data set on the federal regulations. They were coming in around 2018, 2017, 2019. Which is too bad, in a sense, because that's when they change from a brake pedal to a gas puddle.

If your intention was to reduce the number of death certificates, it's too bad. This is a big deal for policy. When there was an era where pulling this policy lever would reduce mortality. And now the same lever pulled in the same direction increases mortality, that's something a policymaker would want to know.

 

>> Speaker 19: Why is that? I don't think I agree with that. Why is that true? How have we identified the fact that, first of all, I don't know what the policy lever is, but second, how did we identify that the policy lever caused the increase in deaths? I don't understand.

 

>> Casey Mulligan: There's a couple different issues we're talking about here. One is what is the effect of changing the prescription price? And that's gotten a lot of study. And it also gives us some light on kind of what part of the model we're in here. Are we on that before 2010 type of curve or are we on the more Lafford curve?

And yes.

>> Speaker 20: Prescriptions went down in the past decade, that's true.

>> Casey Mulligan: Well, more like around the Medicare part D in that era.

>> Speaker 20: Well, when part D was adopted, that was a price decline. Prescriptions went up. That's true.

>> Casey Mulligan: I thought you meant the price, okay.

>> Speaker 20: Right, price went down, the prescriptions went up.

Then in the 2010 to 2019 period, prescriptions went down.

>> Casey Mulligan: Yes.

>> Speaker 20: Okay.

>> Casey Mulligan: So we're switching to something else. So that this horizontal axis is the total quantity, your quantity, whether regardless of where you sourced it from. So yes, prescriptions go down and these studies are focused on heroin.

It's before fentanyl heroin goes up.

>> Speaker 20: Right, but what was the policy lever? I don't understand.

>> Speaker 21: I think what Casey was saying is that they started making it more difficult to get the Rx. Therefore, people switched to the illicit and consumed more of the substances and that increased consumption resulted in.

 

>> Speaker 20: So this is a broader big picture question maybe, which is that there are sort of three facts here, right? That prescription use went down, illicit use went up, and illicit use per use got way more dangerous. But those are the three facts, and this model is consistent with those facts, I believe that, but there's lots of other models that also consist.

 

>> Casey Mulligan: I agree.

>> Speaker 20: Okay, so, but then why would we use this model to predict out into the future when there's lots of models that are consistent with the facts? I mean, you know, I'm saying,

>> Casey Mulligan: I got 30 minutes, I'll talk about one model, and it's gonna say some surprising things that people have ignored.

For example, the Stanford Commission that was just a few months ago, didn't mention this at all, that the policy lever that used to increase, reduce deaths might be increasing deaths. And there also didn't talk about the race things that I'm about to show you.

>> Speaker 20: I'll let you go, I'll let you go.

I'm sorry.

>> Speaker 3: I think what Tyler was trying to say in Casey's model is the idea that once you pay the fixed cost to switch to the illicits, then even if the fentanyl were not stronger, you would still have paid the fixed cost, so then you're in this different marginal cost environment.

 

>> Speaker 19: Yep, but what I'm saying is that you don't need that feature to explain any of these empirical regularities. Those empirical regularities can be explained by all kinds of things so, since we haven't rejected those, I don't know.

>> Speaker 21: Do you have a preferred alternative explanation?

>> Speaker 19: Well, look, Casey, I'll tell you my preferred.

 

>> Casey Mulligan: Okay, so now let's, let's look at some of the race data so, this is a chart of drug deaths over time. Separate series for blacks and for whites, and for many years, the mortality rate was higher for white, case and Deaton, in their book. They started writing their book when they had about this much data that I'm showing you, they put a whole chapter in there.

What the heck is going on, why is the mortality rate lower for the blacks than for the whites? And they have a kind of preference or sociological explanation, maybe with kind of a puzzle they're trying to solve. They don't really look at prices, prices may not only help explain this, but it also tell you what's going to come next.

It does tell you what's going to come next, whether it's the only story for what comes next. We'll see, so I'm gonna go back to the picture now. I'm gonna put blacks and whites in there, I put blacks in at a higher prescription price. A lot of people have argued, due to access to insurance and things like that, that blacks face a somewhat higher prescription price than whites do, and then this picture would say, well, they would consume less.

Now, you didn't need me to come today and say, hey, if there's a group that pays a higher price, they have a lower quantity, you don't need me for that. But what happens after 2010 when you have a different prescription price relationship because of what's going on in the illicit market, putting the same curve there, we're going to jump to that other curve mold.

The races will jump to the other curve, and with this theory, makes a quite a bold prediction. It says, okay, yeah, they're gonna be increased quantity for both races, but it's going to increase more for blacks. Even bolder than that is the blacks are going to pass up the whites, not just saying that they're gonna catch up.

During a period when the white mortality rate is gonna be going up a lot, the black one's gonna go up even more to pass them out, that's a bold prediction. Maybe you can get it from other places, but that's a bold prediction, and let's look at whether it's correct.

 

>> Speaker 11: In the previous slide, it looks like in 2015, blacks had a lower death rate.

>> Casey Mulligan: Yeah.

>> Speaker 11: Okay.

>> Casey Mulligan: And furthermore, this, this approach not only says this difference between blacks and whites, but to say which types of people, older people versus younger people. We were talking earlier about how well people have higher demand levels are gonna be earlier to do their switching business.

But we know that the older people, I don't mean like great grandmothers, but people over 30 or 40 years old, they consuming more because they've built up a tolerance. They're addicted, like Steve said earlier, where the younger people, your teenagers, are consuming lower quantities, they have in a sense, to get the same feeling.

They need lesser quantities, so the switching is gonna make less sense for them, so let's look at that. This is an important point, I wanna explain the same thing in a different way, more algebraic way.

>> Casey Mulligan: There's two prices that determine the quantity, something changed in the illicit market that changed the sign of the effect of the prescription price.

Well, that's true, and a key difference between blacks and whites is that prescription price, then it's gonna change the sign of the race gap. That's the same argument made in an algebraic.

>> Speaker 9: So here's before we go on both of these slides. So if we look at your previous slide, the 2019 aggregate demand curve is a backwards spending aggregate demand curve.

But that's just how you drew it, right it doesn't have to backward bend at that point, it could have kept going some other direction, right? So I know that you could get this effect whereby blacks eventually supersede whites in terms of the drug induced deaths. But that's just a schematic diagram, it's not like we know that.

 

>> Casey Mulligan: I haven't, I actually skipped that part I should have mentioned. And in the paper I have a theorem or proposition five that says that if the only difference between blacks and whites is the prescription price, blacks must pass up the whites when the illicit price gets low enough, and that's a theorem,

 

>> Speaker 19: But it's not the only difference between whites.

>> Casey Mulligan: I understand, but we shouldn't be surprised by this I didn't hear the people saying, who were puzzling about, why are the black whites lower than the white rights, saying, by the way, the blacks are just about to pass up the whites.

 

>> Speaker 18: I think you're making a useful point, there's a different explanation for the phenomenon that you put on the table. I guess what some people are asking for is, and I don't know here, I don't know the answer to this, is there enough spatial variation, say, across states in the Rx price, that you could replicate your black white comparison across the 50 states or across the 50 states by race?

Is there enough variation in the data to make that an informative exercise and see whether we see this pattern repeatedly where you would expect it?

>> Casey Mulligan: That could be investigated, the literature on the difference between blacks and whites in terms of the prescription access is not a state level analysis, maybe it could be taken to the state level.

 

>> Speaker 18: My understanding to get it may be wrong is that prescribing practices differ a lot across areas, so that the implicit cost of obtaining the legal drug might differ across areas, and the timing of the change in these prescribing practices might differ across states in ways that would allow you to extend your.

 

>> Casey Mulligan: But would it be different for blacks and whites?

>> Speaker 18: No no, but it would be different across states, so you could just. I wanna take the same analytical framework you've given us and try to use it to explain spatial variation.

>> Casey Mulligan: Well, that's what the studies I mentioned were doing that, okay.

They weren't focused on race. That's how they tried to make the claim that you're on this flat part of in the aggregate. We're on the flat part because they're looking at these different states and some of them had more oxycontin than others and when the oxycontin product changed to make it harder to abuse.

What mortality changes do you get differentially by stage? That's the whole literature. I haven't contributed to that, but I'm using it heavily.

>> Speaker 18: Okay, but is the variation that that literature uses helpful in disentangling this explanation?

>> Casey Mulligan: I don't know what the other explanations are.

>> Speaker 18: It seems to have some of mine.

So I presume there are some in literature.

>> Speaker 19: But yeah, I mean, you can explain all this by just practice patterns changing and physical, right? Those two things explain everything you don't need that generates and the black white differentials are essentially all because of contamination of non opioid illicit drugs.

That's essentially what's happening.

>> Casey Mulligan: I mean, you have a two part story to explain two data points. That's fine?

>> Speaker 19: Right, as opposed.

>> Casey Mulligan: You wanna use a different theory for the race gap than for these other things.

>> Speaker 19: No, the two things that happened was that practice patterns changed and fentanyl got cheaper.

Those happened. That explains all of this. That's all I'm saying.

>> Casey Mulligan: You, so through contamination, I was listening you. Contamination is an additional bell and whistle on the explanation machine, which is fine, doesn't make it wrong.

>> Speaker 19: So is it the case that in states where the prescription price fell by more the deaths increased by more?

Do other studies show that? Because that would be some evidence.

>> Casey Mulligan: No, they didn't. I mean, that's-

>> Speaker 19: The other way around, sorry. And you could add to that to see if the relative mortality of blacks versus whites changed in response to those state level changes.

>> Speaker 3: The other nice thing about the state test is that there are 50 states.

Well, okay. 48 continental states or whatever, whereas there are only two races. So it's given the fact that you might have bigger differences among the states and their Rx prices, that would give you more variation as a test. So I think I Steve's proposed robustness test seems like a good one.

 

>> Casey Mulligan: Yeah, that hasn't been done. They've been using all the races pulled together.

>> Speaker 19: When they do all the races pulled together, does it go in the direction that you're saying by state?

>> Casey Mulligan: What I'm saying, I'm just repeating what they found at circa 2012, 2013, 2014, you raise the prescription price and you don't reduce deaths.

You have a big switch from prescription deaths to the other type of deaths.

>> Speaker 19: You don't increase deaths either.

>> Casey Mulligan: I'm not sure they're fine enough to. Is it a one?

>> Speaker 19: No one's done other consensus, so I agree. Somebody should do what Steve's saying. It shouldn't be hard.

 

>> Casey Mulligan: Yes, they're confidential. You may have to sign with the agency and everything, but that's what people are using State level certificates.

>> Speaker 1: Lindsey show us the next chart.

>> Casey Mulligan: So then the right one case and deed and are right in their book about why the white rate is above the black rate.

That no longer ends up being true passed in 2019. Now this, my approach not only says that would eventually happen if the illicit price got low enough, it also says what groups it would happen first among. So here in this picture, I've changed it to a gap. So I'm not showing a separate black series and a white series.

I've taken the difference between the two. This is black-minus white. So a negative means the black grade is lower than the white. What I've also done here is control for some demographics, which kind of important here is controlling for the census division. There's nine of them in the country.

This is relevant because the distribution of people of blacks among the regions of the United States is not the same as the distribution among whites in the United States. And the fentanyl rolls out somewhat differently in these places. So it makes somewhat of a difference to control for that I've done that here.

But still, you see the gap almost change sign by 2019. This is for all the ages. Now let's break it down by age, because remember this theory said that it's gonna happen first among the older people because they have the higher quantity levels, and so that's what I've done here.

I've added red is the older people, 45 and older, and the blue is the younger people, 44 and younger. And you can see the passing up happens earlier among the older. So this is a race gap among older people and then blue is a race gap among younger people hasn't changed sign yet.

The white rate is still above the black rate among younger people. Maybe a little overpass.

>> Speaker 1: But the dispersion has changed big time.

>> Casey Mulligan: Yes, I mean again, part of the increasing returns tend to generate dispersion. It makes the high consumption people consume more and the low consumption people consume less.

 

>> Speaker 22: So Medicare would be a natural. What's happening with. Yeah, so like access to these kind of these painkiller drugs in hospice for instance is much easier as a Medicare policy. Interesting to redo this with 65 and up.

>> Casey Mulligan: Yeah, the 45 plus is largely driven by 45 to 64.

And the elderly population is an interesting case. In fact, Dan mentioned the prescription is going down a lot and it has generally happened, but the elderly has been at different dynamics on the prescription.

>> Speaker 22: The kind of fraud I'd heard about in the elderly is you have relatively easy access to prescription or painkillers, then you resell it on the black market.

 

>> Casey Mulligan: Yeah.

>> Speaker 22: So they can have some special.

>> Casey Mulligan: I've seen that happen in Walgreens in Chicago with my own eyes. There weren't many grandmothers doing that, but enough grandmothers doing it at scale.

>> Speaker 22: You just need some entrepreneur to walk around the neighborhood and say, can you give me your extra pills?

 

>> Casey Mulligan: There was a study that found that when Medicare Part d rolled out in places with more elderly people living around there, that the death rate of the non elderly people went up. Okay, so that the last. Somebody have a question?

>> Speaker 22: It's not okay.

>> Casey Mulligan: Now, the last result to show is the young and the old, youth, adults and minors, the same picture that we did before.

The adults are gonna have a higher demand level and therefore they're gonna respond differently to a prescription change. So I wanna go back to that regulation or change in the formulation around 2010 that increases. More than non-pecuniary part of the Rx price. So that goes up. Doesn't affect the illicit price, as this was about a regular pharmaceutical product.

And for the younger people, any group with a low demand level, it's gonna reduce their consumption. And you get the kinda naive law of demand prediction. But for people with higher level consumptions. They're gonna potentially could respond by consuming more. And so you get this divergence. So here's the death certificate data.

The younger youth, minors and minor us up through 2010. Here you see some standard error bars. Cuz when you're dealing with youth, there's small enough samples that the sampling error is actually interesting. Everything else I've shown you, the samples are so huge that they even bother to show you these microscopic standard errors.

They're kinda tracking each other. They're at different levels, of course, for a variety of reasons, but they're tracking each other. And then after 2010, the youth deaths fell, at least until the pandemic. And the adult just went up. And we see that as well. I think the prescription regulations were intended to reduce mortality from opioids.

Who knows the intentions? But I think that was the intention kinda worked among younger people. It's harder to see an effect among older people and overall opioid deaths. If you look at deaths from prescriptions, that would be different. Because of the switching that's happening among the adults. And then you come back to the addiction issue that Steve raised.

There'll be a dividend from this in later years. You'll have cohorts or warranties to addict because the younger people didn't have prescriptions. That might be something that could happen. Although I'll go back to the race picture. I know this story that well. One of the favorite things we did in the White House is to criticize the companies.

Both Democrats and Republicans like to do that.

>> Casey Mulligan: The companies have been blamed. Well, they pushed out OxyContin. People got addicted and they couldn't give it up. And they switched to fentanyl. The blacks weren't getting the OxyContin. And somehow they passed up the whites who were getting the OxyContin.

There must be something more of the story than just blaming the companies.

>> Speaker 23: On the black white comparison in the literature, not in what you're doing. The fact about the difference in access for blacks and whites is purely a question of insurance. Is that how this is?

>> Casey Mulligan: The Stanford commission, which actually.

I agree with a little bit on this issue. They point to lower insurance rates. And that's been documented in other studies.

>> Speaker 1: Yeah, yeah.

>> Casey Mulligan: Insurance, but then they're also say, well, maybe the part of the game here was you go to the doctor and you say, I'm hurting.

And he's supposed to, or she is supposed to listen to you and say, you're hurting. I'll give you a pill and not question that. And it's been claimed that, well, the black patient gets more questions. And the white patients is kinda allowed to be the best.

>> Speaker 23: Yeah, I know they support that argument about equity.

But I'm just wondering, how do people know that?

>> Casey Mulligan: I'm not sure how they know.

>> Speaker 23: I mean, insurance rates I can understand cuz we can document that.

>> Casey Mulligan: Yeah.

>> Speaker 23: But I've always wondered about this question about equity and prescription describing behavior, is there?

>> Casey Mulligan: There's some studies that'll look at patients coming out of the same hospital.

And going home with different amounts of prescriptions to take home.

>> Speaker 15: So, Casey, isn't a nice, testable implication of your model here going back to the states? Or maybe you have to be divisions or something like that if the states are too small. That if I'm a black person in an area where maybe I face high prescription prices.

But there are a whole bunch of white and maybe white elderly people around me, they're the ones who are really switching into the market. They pay the fixed cost to make that market exist. And then once the market exists, I have access to a better market. Whereas if I live in some place where there are no white people and I'm a black person, okay, there are no places in the United States where only black people live.

But anyway, there are too few white people to support the market. Even though I might want to switch from following your model, right? If there's a higher Rx price. So I would like to move over more and increase opioids more. No one has paid the fixed cost to make the market exist.

So I don't actually do the switch. Because the switch should really happen with the whites kind of leading the blacks into the illicit opioids, right? There are big differences among different areas of the United States in the black to white ratio.

>> Casey Mulligan: I use the phrase switch to make it sound like a longitudinal thing.

Then I follow a Social Security number, and then they change their behavior. And a lot of that happens. But also, I'm just thinking of new cohorts coming in a different situation. And there's some cohorts that are kind of on the margin between these. I said the young people are prescription intensive.

And they are, but they're a lot less prescription intensive than they used to be. So you have what I've called the switches, can be more of a comparative static. But still, one could look at that. The studies I mentioned don't look at race. They have it, I mean, they're using death certificates and race is a variable there.

But they haven't. And there wasn't a lot of interest. Another reason to show you the part that Deaton was looking at. There was a lot of interest in opioids among blacks because it seemed like a non-issue, right? This data hasn't been available more than a few months. There wasn't really an awareness that the blacks had passed the whites, let alone participating in this type of health issue.

 

>> Speaker 1: Casey, can you distinguish between the proposition that it is government policy that's making prescription more expensive versus lax government policy that's allowing for illicit prescriptions price to fall? But both of them is a relative price that.

>> Casey Mulligan: We do know that the level of the illicit price has fallen a lot, right?

Now, why is that? I hope it's clear the illicit price matters a whole lot here. So we like to understand why is it falling? And certainly, the technology, I think, has to be part of the story. Is law enforcement part of the story? I mean, you're not allowed to ask that question, but it'd be good to ask.

When we had these criminal justice reforms, there was not a discussion. Might we be reducing prices of illicit drugs? Might more people die from that? There was not a discussion. It would be nice to have one.

>> Casey Mulligan: The criminal justice reform at the federal level was in 2013.

Pure coincidence that fentanyl comes in our country on a permanent basis on the end of 2013. Pure coincidence, of course, that when the Chinese fentanyl came in the United States, 2014, around January 1, it also came into Sweden, same time. Pure coincidence that the Swedish police, drug enforcement people fought back hard and they don't have fentanyl there, and we still have it.

These things ought to be investigated. I'm not allowed to talk about causality, but these things have happened at similar times in similar places.

>> Speaker 19: What happened with the age difference in fatality rates since 2010? You read that everyone reads that there's a huge spike in 2020 and 2021 among adolescents.

But do we know? I think that graph stopped in 2010. We must have some more data.

>> Casey Mulligan: Your printout, you don't have the dynamic.

>> Speaker 19: I was looking away when I missed. Sorry, thanks. I missed that.

>> Speaker 24: I wanted to follow up on the question that condita asked.

So in terms of whether there is difference, in terms of evidence that physicians vary by respond to the pain of individuals. Marcella Alsano, like Olivia Harvard, she has a paper where she created a clinical and randomized whether you saw a same race doctor or a different race doctor.

What she found is that in terms of uptake of preventative services, you actually saw that folks who were coming in with the same baseline level of kind of problems. They were more likely to same race doctor to elect for more preventative career. And so that literature is burgeoning.

The fair side. I think that's a useful thing to look at. I think the other piece, going back to the point about the spatial variation, I completely agree with that, and I think that that would be very compelling. In fact, you can perhaps do this within race. And so if you were to look at figures of whiteness across different regions or different states, were there certain states where we saw a divergence on uptake?

And then is that divergence within the same racial category also correlated with differences across space and price? Because in some sense, your paper precisely is that this has nothing to do with race. This really has to do with the prices. And so why not control for that and just look within race, but then across space, where there's heterogeneity in the prices, and then if you can show that.

I think that persuasive both for whites and then also for blacks.

>> Casey Mulligan: And the illicit prices are tricky to measure. There's some attempt at that and work could be done. DEA, its survey has a regional component to it.

>> Speaker 24: There's some crowdsourcing data to. Unless it prices too, like streetrx.com or something like that.

I know they did some work in 2012. And even if you had something that was cross sectional today. And you could say that there's some permanent level of differences across space, that could be something at least that would be suggested.

>> Casey Mulligan: There's a Princeton dissertation recently that looked at, I think it was Street Rx and they were looking at illicit prescription now that they didn't have fentanyl and heroin in that.

And there was definitely city level, definitely city differences. So these are feasible projects that you're talking about, but they haven't been done yet.

>> Condia: So, Casey, I have kind of a policy question out of this. So essentially this is a story about, let's take your story about unintended consequences.

So you thought you were gonna get one thing, you got another. The interesting policy question is you don't want to say one never makes an intervention because of unintended consequences. Was it at all foreseeable that this might have been the consequence?

>> Casey Mulligan: I think from an economic point of view, the change in law enforcement would affect the price of what those criminals were selling.

I think was fairly clear. I did some estimates around the amount of time in prison. If that's passed on to the customer in terms of prices, what would be the effect on the price? It's not so different than what actually happened. It's billions of dollars. That was part of the reason for the criminal justice reforms.

Well, this is a big cost to the criminal, but what if the customer's paying that cost? We kind of understand the kind of magnitudes at stake. So I think it was something you should have expected. I think some of us did. We were working on this in the White House in 2018 and 2019, and these were kind of issues we were getting into before we had the data I've shown you today and that we have these kind of concerns.

 

>> Speaker 26: Condia, you're talking about the restrictions on prescribing illicit or talking about the sensible reduction in punishment for illicit.

>> Condia: Well, those are two different policy levers, right?

>> Speaker 26: Yeah.

>> Condia: One had to do with criminal justice reform. Another has to do with the way that you control prescriptions.

And one interesting thing is you probably need to tease those out and they're kind of conflated. And what I'm asked, because as a policymaker, you do try to understand about this. I guess I'm asking in a more systematic way if this had been given to you as advisors.

And you were able to, to set up an actual research design that would say, what if I vary this what response am I going to get here? Would you have been able to do that? Is this what you would have set up? And would you have been able to see that the predictive capabilities of predictive outcome is x, y or z?

So there is a way to do this right if you're a policymaker? We don't always do it, but as academics, we do an actual research design and we say, here are the possible triggers of this outcome. It was possible to see that coming.

>> Casey Mulligan: What I'm trying to add is in the kind of a middle step there where how can we generalize from the research designs that already been executed.

These studies that were on the flat part of the Laffer curve, they weren't about race. But I have economic theory can help me generalize that and say, does findings have anything to say about race, even though race is not mentioned in the paper? And I think this framework is a very clear answer to that.

Yes, you should be concerned that the race gap is going to change sign. And there are other examples like that. When you've seen one price change, how do you recognize other price changes? Like I think Steve mentioned the non pecuniary cost and pecuniary costs. Well, maybe those are in a kind of the same metric, and the responses to the pecuniary costs are maybe telling about.

Responses to the non pecuniary costs are of.

>> Speaker 26: But that would be more fundamental implication of your framework is we need to think about alternative sources of supply. Chalk off the legal whether it's economic sanctions, prescription drugs. And so-

>> Condia: That's true of almost any commodity that you would describe.

 

>> Speaker 26: Yeah, you gotta think about both the existing alternative sources and the potential for creating new ones.

>> Casey Mulligan: Yes.

>> Speaker 26: And maybe that wasn't done enough in this case.

>> Casey Mulligan: This argument I had with the FDA, they refuse to recognize the consequences of their rules for illicit markets.

They just don't do it on principle. Lawyers will tell me that's not our jurisdiction, but yes, a cost benefit analysis has to look at the outcome with your regulation versus the outcome without your regulation.

>> Speaker 19: So what would you do to fix this?

>> Casey Mulligan: I didn't even get any question whether something's broken.

I mean, I had the macro pictures, there's technological progress.

>> Speaker 19: Sure, that's the answer.

>> Casey Mulligan: This is technological progress.

>> Speaker 19: What do you do about it? Does it matter? It was good?

>> Casey Mulligan: What's your stance on the war on drugs is kind of what you're asking.

>> Speaker 19: Yes.

 

>> Casey Mulligan: I mean, I see both sides I used to only see the Milton Friedman side, now I kind of see both sides of it. I mean, a lot of medical problems we've had in the past, whether it be cholera or AIDS, other things, we didn't solve them by kind of becoming like monks and living in a monastery.

We solve them or alleviate them with medical innovation. So I would like to think that this issue will evolve with medical innovation.

>> Speaker 26: You mentioned law enforcement, sorry, what about that?

>> Casey Mulligan: I mean, that's the war on drugs question. I see both sides of it now, like I didn't used to, but I think medical innovation, is gonna be new ways to treat addiction.

 

>> Speaker 26: But is that even allowed to have research on that?

>> Casey Mulligan: There's a lot of policy barriers to medical innovation, right?

>> Speaker 26: I've been studying it.

>> Casey Mulligan: Yes, psychedelics is one area where they're trying to do research. Can psychedelics help people get off opioid addiction? And they got big legal problems in investigating them.

 

>> Speaker 19: They're all drugs like methadone and all that I'm sure you knew that. I mean, I'm just struck by the 2020 effect for kids talking like hundreds of kids basically dying in the pandemic with this huge increase in demand for opioids happening for them. And so this is one of these things where, there is some psychological aspect of this.

There's like a depression skyrocketing in youth, one in four young adults seriously considering suicide around these times, it's gonna be a complicated story.

>> Casey Mulligan: Can I just follow on that angle? Look, I take that your discussion, but also a broader discussion. There's a tremendous variety of sources of supply, of drugs that make people feel better lead them to self-medicate.

And that suggests to me the anti-addiction approach you describe is one basic message. But another one is we need to shift the level of the demand curve through nonprice, through other mechanisms. Like leaving the underlying sources of depression that are driving people to self-medicate. Or that we've lost, perhaps to some extent, the stigma that was once associated with taking certain drugs.

And maybe there was a big cost for losing that stigma because it's basically your framework that translates into an outward shift in the demand. And that's especially dangerous for addictive products when the supply sources are so plentiful and cheap, which is a change from the past. So we've lost the stigma at a time when-, it's not the right word, the ability to get easily addicted is greater than it was in the past.

 

>> Condia: Yeah, and I think just again from a policy point of view, cuz national war on drugs, too, which is. So one reason I think it's important to tease out which of these variables is actually most important in producing the outcome. So what you're talking about here with judicial or law enforcement reform or criminal reform, it's basically sort of mild decriminalization.

And there's a whole now movement toward mild decriminalization. First you decriminalize marijuana, now we're about because prohibition never worked and so forth and so on. That actually, I think might be a more important explanatory of what we're seeing than what the price was. And what I'd really like to know is how much is explained by price, explained by differences in prescription regulation?

How much is explained by race and the response to different race and how much is decriminalization, because that's the only way as a policymaker, people are going to do something. They're not gonna just sit by and let this happen. So the way that I like to think about it almost econometrically is which one's gonna give me the most powerful explanation.

And that's why I think it would be really interesting to try to tease this out more so that there's not a several. It might have been what I'm calling mild decriminalization, it might have been changes in price. It might have been which was driven by changes in prescribing behavior, etc., or prescribing regulation, etc., etc.

That's kind of the interesting policy process.

>> Casey Mulligan: Yes.

>> Condia: Can I point out just a basic point that I think we all ought to be very cautious about interpreting data from the years 2020 and 2021? Because your story, if it's correct, right, should go on post pandemic It really shouldn't be.

Maybe young people, it's spiking up in 2020 because they're out of school and they're adolescents and they're not in high school and so forth and so on. And if it goes back to trend line, then we would say, okay, maybe that was the pandemic, and the pandemic may have affected young people differently than it affected older people.

I'm just saying cautious about 2020 and 2021, and we should get out a little bit more to 2023 or something. It's another explanatory variable? Yes, well, the pandemic is such a, unfortunately, since we only had just the one.

>> Condia: It's not all that useful as an explanatory.

>> Casey Mulligan: Although some of the states differ, I have a paper where I looked at $600 and $300 bonuses.

Drug and alcohol deaths, and those stopped at different times in the red and blue states. And it looks like now they did not go back to trend in the red states, but they did go back, started back toward their trend sooner than they did in the blue states.

 

>> Speaker 1: Can you update this chart, it's only 2019?

>> Casey Mulligan: Yeah, I may have it dives way down in COVID both from COVID and from drugs and from a bunch of other things. It goes down about two years, and the drugs and alcohol by themselves contribute about a third of the year.

Things like diabetes and circulatory diseases contribute about a half a year. COVID itself contributes about a year or a little more.

>> Speaker 1: I mean, that's the most striking chart in a sense. So we're just about out of time, any more questions? What if I bought from the Zoom world?

Anything from the Zoom world? Fascinating Cochrane, he's always the first guy