David Wright: 0:02

My guest today is Amy Finkelstein, professor of economics at MIT. And His research focuses on market failures and government intervention and insurance markets. And she has won numerous awards, including a MacArthur Fellowship and the John Bates Clark Medal. Amy is co author with Liron inf and Ray Fishman of the forthcoming book, risky business, why insurance markets fail and what to do about it. Amy, welcome to the show.

Amy Finkelstein: 0:23

Thanks so much for having me. David. I'm delighted to be here.

David Wright: 0:26

First question. When is government compelled insurance a good idea?

Amy Finkelstein: 0:31

Great question. I mean, we all know the cartoon version of economics, which emphasizes the glory of the market, Adam Smith's invisible hand and all that. But I think we also all know that markets have their limits, and sometimes government policy is warranted think monopolies, setting exorbitant prices, bank runs polluting firms, you name it. So our book is about a particular form of market frailty that doesn't get as much airtime in the popular discourse as it deserves. That's the problem of adverse selection, and it can wreak havoc on insurance markets. So just to briefly explain the problem. An insurance seller naturally cares about who their customers are, an accident prone driver, or an accident prone homeowner is going to be more expensive to cover than her more cautious counterpart. And insurance customers often know things about themselves that are relevant to the costs of insuring them, like how frequently they get distracted while driving, or whether they tend to leave pots cooking unattended on the stove. The insurance company may not know about these trades, or may not be able to easily learn about them. And so the result is that at a given price, the insurer offers its product, the high risk customers tend to flock to the market that drives up the price that drives low risk customers out of the market. So now you've got really high risk customers in the market, the insurers have to raise their price again, that drives the lowest risk remaining customers out. And so it goes

David Wright: 2:02

Death Spiral

Amy Finkelstein: 2:03

death spiral, exactly

David Wright: 2:05

I've seen it personally happen to different companies. Oh, yeah.

Amy Finkelstein: 2:09

We talked about some specific examples in the book as well. And you said, Yeah, with a death spiral, the result can be the entire collapse of the market. So that's a long winded way of getting to the beginning of an answer to your question of when is government compelled insurance a good idea? One of the possible solutions to this type of death spiral is mandatory insurance. And we see this in automobile insurance, in homeowners insurance and health insurance. Just to just to name a few examples, you know, so the idea is that insurance can be valuable to everyone, because insurance is about protecting us against the vagaries and vicissitudes of life. And those can happen to even the most healthy person, the most careful driver. So it's not just for the high risk insurance should be for everyone. And when adverse selection makes the market collapse or become severely limited. One potential solution is to mandate insurance coverage. Now, does that mean mandates are always a good idea? Not necessarily. You know, we sort of have all had a front courtside seat at one recent example of mandates with the mandate for health insurance created by the Affordable Care Act in 2010. And in some sense, just to back up a little, the conversations around that were the genesis of our part of the genesis of our book. We were listening to the 2012 Supreme Court, oral arguments about the constitutionality of the Obamacare mandate. And we heard, you know, the conservative justice Antonin Scalia, ask, you know, well, if it's constitutional to mandate health insurance, can we also mandate that people eat their broccoli? And this got a lot of attention? In fact, we, we jokingly sometimes thought we should call our book, you know, why is insurance different from broccoli? And I'm sure one can come up with many reasons. But the key for the mandate is that the insurance company cares very much which customers buy their product, because that directly affects their costs and their viability. Whereas the supermarket does not care who buys its broccoli, its profits depend on how much broccoli it sells, not who it sells it to. And so that's why a mandate, you know, made make sense for health insurance for a selection reason, or for auto insurance or homeowners insurance in a way that it just doesn't for broccoli. So that's the argument for it. Our book, either frustratingly or intriguingly depending on your view deliberately stopped short of You know, pronouncing the the easy and facile solution because as we also talk about everything involves trade offs and mandates are no exception. So what did we learn from the mandate for health insurance, which actually started in my home state in Massachusetts in 2006, you know, preceding the Affordable Care Act by several years, when Republican Governor Mitt Romney mandated health insurance coverage in the so called Romney care of 2006. And then we saw it play out nationally, starting after the Affordable Care Act. I think we learned too well, three things. One, we learned that mandates do work the way the textbooks and economic theory says they would work that they both get more people into the market, and they get the lower risk people. So that's, that's the good news. The the less good news, or the reason that the real world is complicated and hard, is, I think we learned two other things that weren't so obvious from this sort of stylized, at least obvious to me from the stylized economic models. The first is, what does it mean to have a mandate in when I teach it in class? Well, you mandate, everyone has insurance, they have insurance, not exactly how it works in the real world. In fact, we saw this play out in Massachusetts, they introduced the mandate. And some people signed up and they tended to be, you know, somewhat of the healthier people. But it didn't really start to kick in until they put a penalty in place if you didn't actual financial penalty if you didn't have health insurance. And that's when you really saw the mandate affecting getting a lot of people and a lot of healthy people into the market. But even then, not everyone signed up. So Massachusetts, you know, still has some people as do other states who are not insured. So I guess one way to save it, we talked about in the book, there's this rock old what Robin Williams joke about, you know, the police in Britain who don't carry guns, and so he jokes it, though, you know, what do they do when they're arresting a criminal? They shout, Stop, or I'll say stop again, you know, when I think there's an element of that with a mandate that just to have a mandate, per se, is not enough. If it's not enforced. It doesn't get everyone covered. And I think the second problem with the mandate is, it's one thing to say everyone values insurance. And, you know, if adverse selection is making it hard for the market to function, there's a role for mandates. It's another thing to decide what level of mandate how much insurance should we mandate that people have. And here we have a bit of what we call in the book, the Goldilocks problem of insurance, you know, when when one bowl is too small, and other is too big, what is the just right bowl, if you if you mandate only a very small amount of insurance, and we talked about, for example, in auto insurance, you know, Florida has very, very minimal requirements for the amount of auto insurance people have to have compared to other states, like Maine, which is much more expensive, extensive, excuse me, you know, so on the one hand, when you require only a small amount of insurance, and you let people have the option to buy more comprehensive coverage, if you want it, you kind of recreate the whole selection problem, just on the margin of you know, how much insurance coverage do people buy, once again, you get the more expensive customers flocking to the more comprehensive coverage, and pricing it out of reach for the lower risk, people who might still value that additional security. On the other hand, if you mandate that everyone have enormous amounts of coverage, that maybe more coverage than people want, they might prefer to have a little bit of risk and, you know, save some of their money for other things. So mandates have a real role when there's selection problems, but they're not a panacea.

David Wright: 8:53

You wrote a paper many years ago, talking about maybe maybe you just sort of move past this kind of understanding, but how the, there's a classic, original paper, I think, by Germany and Wilson, the 70s talking about how the public option or the public mandate, government issued mandate can create, you know, sort of the the kind of theoretical sort of framework here was you have two different equilibria in the possible market, you have a call of separation or pooling or Libya, right. So the separation is what we're talking about here is this desfile, where you have the risky people go to high risk policy, and then low risk people who are lower in this sort of, they separate into camps. Right. And I have a lot more to say about that later on. But there's the pooling mechanism is is superior in a sense that it allows more people have a walkaway price. Right? And if you're the high risk person, it might just be just an absolute basis, just too expensive. And you're not and I see this as well, when volunteer coverages This happens all the time. And it's something that people will say they want more coverage, but then you show them what it's gonna cost them and they're like, well, not Wow. I like that, you know, I would like more coverage at the price that you know, the cheap price person you are, for whatever reason, right, and my experience is more impedance even health. So a little less kind of, you know, morally loaded to exclude people. But it happens all the same same dynamic. Now, what you've mentioned in your paper there was that there's a way for the a partial insurance from the private market to kind of facilitate a pooling outcome. I honestly didn't quite totally understand the proof by graphic, you know, inference there, in the paper itself, but maybe you could sort of talk a bit about that, whether you think that's kind of real, or, you know, how your understanding might have evolved? If not?

Amy Finkelstein: 10:38

Yeah, that's, that's a really good question. I think, you know, that was a particular academic paper working through a particular theoretical theoretical model. I think the the broader point is actually much simpler than, you know, I made it in that in that paper, unfortunately, or, fortunately, it's simpler. Unfortunately, I didn't make it simpler, which is, it's exactly the point that I was talking about, that you can get everyone to some minimal level of insurance through a mandate, but you will then you know, recreate the problem of selection on the question of the amount of coverage, right, sometimes people think, oh, adverse selection can destroy a market, and the market can disappear. One thing we talk a lot about in the book is, you know, sometimes selection doesn't destroy a market. But that which does not destroy the market doesn't always make it stronger. A lot of what happens is, because, of course, many insurance markets exist, many of them exist without mandates. And yet, what you see is all kinds of limits on the nature of the policy, which can seem incredibly frustrating to a would be customer or a potential business opportunity to a would be entrepreneur, but are actually there because of selection. So one example we talked about in the, in the paper is dental insurance, which is, you know, for lack of a more technical term extremely crummy, right? It's not what insurance is supposed to be. Insurance is supposed to cover, you know, the unexpected and potentially financially really ruinous events. And instead, what everyone's dental insurance, if they have it covers is routine cleanings and cavities. And, as my co author Ray Fisman found out when he needed dental implants, you know, the policy doesn't cover it, which is exactly the kind of unexpected and expensive thing that insurance is supposed to cover. He also quickly figured out why because when he needed dental implants, he thought, Oh, I've never had dental insurance, you know, I'll wait and sign up for a coverage policy that covers my implants, and then I won't have to pay as much for them. And that's precisely why this market doesn't exist, it's because it's something that people can anticipate or learn that they need. It's not, you don't have to handle it immediately. You can wait till the open enrollment period and get it and that's why these types of policies don't exist. It's also why we have open enrollment periods. If you've ever wondered why you can only change your health insurance plan in in October, it's precisely to try to help reduce this a wall, just wait till I need it. To get insurance, it's once you know you need it. It's really tempting to buy insurance. But at that point, it's no longer insurance because the event has already occurred.

David Wright: 13:27

I have personal experience selling dental insurance to people over the phone when I was working at an insurer tech startup. And it is it is I mean, it's more than exactly just exactly what you're saying there. Now, when I was thinking when I was experiencing this, and seeing everybody showing up trying to sign up for a policy just before they're about to go in and get their full mouth replacement, or get braces for their kids or whatever it was right. The thing that occurred to me there was that it wasn't really, it wasn't really an insurance product, right? They're just looking for a freebie. And the numbers, though, large, they're not really big in insurance kind of world. And we're talking about some something worth 1000s of dollars, which is, you know, that's a lot of money if you're not seeing it come in for most people. But it's not like compared to the other insurance policies you buy, right? I think in property and casualty more. So I mean, that were the benefit payouts are hundreds of $1,000 is kind of light, right? So it's hundreds, millions, right? And these are where we really do need to buy a product and so the way it got my head around the dental insurance problem was that this is a very modest income smoothing kind of product when it's properly deployed. And for the stuff that gets really bad. You go to the hospital, I mean, you now you're in for general surgery, I got a friend of mine who's a who's a dentist in town. And you know if if he when he was you I don't know what they call them a resident or something. You know, he worked in hospitals and and in hospitals. That's what kind of the real work happens and then you go and if you know if the problem is severe enough, it just sort of merges in with general health care, or health insurance, right, that's seems to be the insurance part of dental is just health insurance.

Amy Finkelstein: 15:06

I mean, I think you're hitting on a really important point that we try and touch on in the book, which is, you know, one of the reasons sometimes these selection pressures destroy a market and other times they the market can hobble along, precisely has to do with, you know, how big the risk is. And if it's really, really big risk, you can get low risk people being willing to pay a little extra even above, you know, their actuarial costs just for that security. So when you talk about something like, you know, really catastrophic risks, like you know, your home burning down. On the other hand, we also talk about markets in the book that have collapsed, that we think due to selection, which I think you would agree, are, are not small potatoes. So for example, a market for layoffs and insurance if you if you are going to lose your job, right, so that, that has potentially catastrophic implications for households finance, and as we talked about, in the book, there have been examples of entrepreneurs trying to offer this type of insurance, you know, to pay out above and beyond the, what the government will pay, if you lose your job, they were aware of the selection issues, they put in waiting periods, you know, you know, you can't, you can't, you know, claim the insurance within six months of buying it. And, you know, the CEOs are saying, You tell me, if anyone can see six months in advance that they're gonna lose their job, you know, I'd like to meet them. And in fact, it turned out, they met a lot of those people. And those markets collapsed.

David Wright: 16:36

Every every venture capitalists and insurance should have a copy of your book, if only for the catalogue of dumb ideas that turn up every few years and somebody funds collapses for exactly the same reason, every single time. I mean,

Amy Finkelstein: 16:49

yeah, we saw it. We saw an example actually, after we'd written the book. You may have seen it, where Zillow was trying to get into the market of selling its own of selling houses. And, you know, once again, that there was a adverse selection death spiral, like who tries to sell their house on Zillow, you know?

David Wright: 17:06

Yeah, I mean, people as predatory, you know, market participants. Another concept here that I want to bring up, which is something that the interesting, I don't know if it's an irony about your book, is that the one big difference in health insurance versus property and casualty is kind of comes to this concept of is healthcare really health insurance United States, kind of a weird place for I grew up in Canada, where a lot of these questions are completely off the table, right?

Amy Finkelstein: 17:36

Our co-author Ray Fisman is also from Canada.

David Wright: 17:39

There are there problems in Canada, I will say about the health care system there. But figuring out how much you pay for it, as an individual is not one of them and whether you're covered or not. Right, so but the the interesting problem you introduce here is insurance, usually, like the kind of social mission of insurance here is subsidy. So it's all about subsidy. It's all about trying to get the people who are most expensive to healthcare, they need, you know, Medicare is kind of the classic sort of solution from mid century, mid 20th century, to the worst of that problem, which is the senior folks. But in the rest of insurance, subsidy is not discussed in P&C, because it's there. But implicit. So the real thing as an actuary, and I was an actuary for a bunch of years, the real thing you're trying to do is you're trying to get critical mass of a portfolio to stabilize the variance of the claims payout. Right, you're trying to get an accurate estimate for the population cost. And so subsidy exists, but to the degree that it's I mean, different people have different kind of takes on this, but it's not a first of all, it's just not an article of discussion very much, which is who's subsidizing whom inside of a pool, but you select the pool to minimize the variance of the cost of that pool, right. And so you tend to wind up just selecting for, you know, your your segmentation. And let's say a claims cost estimate exercise where people who just are kind of like now as an insurance carrier, you might not have enough people that look alike. And so you have to sort of lump more heterogeneous folks together implicitly as a subsidy. But the goal of it is not subsidization. The goal is to try and get an accurate estimate for the cost. Because because you have 100 people, and one of them have a clean every year and it's like, well, what the heck doesn't know the 99 cost? Well, we don't know it's the same as the one divided by 100. Right. And that's kind of the way it works. And the subsidy, though implicit is not a not an objective.

Amy Finkelstein: 19:23

It's interesting, it's interesting when you're talking about the the challenges of sort of estimating the, you know, the cost of your pool, because, you know, we found you know, going back, you know, 1000s of years, some of the, you know, very, you know, earliest actuarial attempts which were done often for life insurance, or then for maritime shipping. Even there they you know, you had you had some of the, you know, great mathematical minds, you know, working on it, and yet they weren't great social scientists and so they would come up with these, you know, actuarial table All but not consider selection problems. And so one example we talked about in our book, you know, from the 17th century, which was using life tables, created by Edmund Halley of Halley's Comet fame. So he's done other great stuff in his life and the life tables were accurate for their population. But then when the, you know, government of this is Louie, the 14th, and France and other governments are trying to finance you know, basically issue bonds in the form of annuities. So, you know, products that will pay out until the person dies. Well, they were bought up by on behalf of, you know, young girls in in France, who had already survived smallpox, and guess what they had, like, you know, much greater than average life expectancy, and it kind of bankrupted the French government, and arguably, According to some historians, you know, ended up being sort of the impetus behind the French Revolution. So, you know, these, these issues of are not just accurately estimating the risk on a given population is a important and hard actuarial problem. But you got to then always think about, well, who is my population? And why might they be different from what I expected? Or you'll, you know, go the way of Mary Antoinette?

David Wright: 21:17

Yeah, that's up so of that. So one other kind of really important concept, that interesting get your take on is the reason why I, here's my opinion, the reason why we are so concerned about the subsidy portion, and, you know, if it's a, let's put it this way, if it's an ultra high net worth, California resident who's got a$60 million house in Beverly Hills, and they can't get all the wildfire insurance that they want, you're like, whatever, you know, like, I don't care, I don't know, you care, I'm sure the politicians don't give a flying bleep about it, right. But if it's a senior, or it's totally disadvantaged person in some way, who has, I don't know, cancer in some version of the story, maybe a heart disease or something solvable or fixable, they break their leg something. Right? They show up at the hospital, we take care of them anyway. Right. And so this is, I think, a pretty rich topic, like, we do use a term that use which I haven't used in this concept, but he says that we bail them out. Me. So it seems to me that like the reason why we want to get him in the healthcare system, it's because we're gonna bail him out anyway, later on after the fact. And that's a and that seems to me to be kind of like a social objective that we carry out in certain circumstances. Have you thought about this? And like, when and why do we choose to bail people out?

Amy Finkelstein: 22:40

Yeah, this is this is a great question. And in some sense, takes us a bit beyond the the question of selection, right. So this book focused on the problem of selection, which which pervades all kinds of insurance markets, property, casualty, life, and health, etc. There are other problems that can be more specific to specific insurance markets, such as, you know, the the need to or the, as you said, the sort of social obligation to bail people out. And actually, my co author, Lauren, I nev, and I have another book we're working on, on exactly this topic. I think the two key examples you've kind of hit on them are healthcare and natural disasters. And in both cases, there's a sense that, you know, something has happened to people through no fault of their own, and we have a social obligation to come in and help them and we see this play out all the time during hurricane season when the politicians fly down in their hard hats and tour the area and declare it a FEMA disaster area. And, and that's this, that's a reason that people may not buy natural disaster insurance, because they're going to be bailed out ex posts, and a lot of what they're buying would be paying for what they would get anyway. And you see the same thing in some health insurance settings as well, as you're alluding to, this is something that economists call the Samaritans dilemma. I don't know if you've, you know, heard of have not heard that. So the parable of the Good Samaritan in the New Testament is one of you know, a traveler is is beaten left for dead by the side of a road. You know, several people pass by and do not help them and then the, the eponymous Good Samaritan comes by, picks him up 10s His wounds takes, you know, make sure he's healthy and sends him on his way. This is, you know, a parable that's told about you know, altruism the, you know, charity towards others morality, etc. So, okay, where's the dilemma? Leave it to economists to find one. The dilemma is precisely that. Well, if you're, if you know someone's gonna come help you out. You may be willing to take on more risk, you know, travel on riskier roads or, you know, not purchase the health insurance or the natural disaster insurance you need. And as I mentioned, my co author, Lauren and I are working on it on a different project, because we do think this is, you know, a fundamental issue preventing a lot of the healthcare policy debate in the United States. But it's in some sense very different from from the selection problem, which will occur even in markets where we don't feel some social or moral obligation to help people like health insurance, which we, you know, talk about in the book or auto insurance, where in all of these, you have the issue that people know more about their risk than the than the insurer, and that can wreak havoc in these markets.

David Wright: 25:47

Do you, Do you have a kind of like a, I don't know, any sort of I, dia for? How do we choose this to pay attention to some of these or not? Because I think it really impacts it kind of precedes economics. So to be fair, it's not something that you know, it's your job to study. But I become increasingly kind of normalized in a certain sense, right? Because in the insurance industry, what we do is we handle problems that the government, I mean, P&C, and in some other related industries, I mean, most of what you buy is compelled the vast majority of what you buy, you're forced to buy by somebody who's not you even life insurance. I mean, I bought life insurance, not necessarily because there was something I worried about my wife, maybe, but she's the one who's gonna, you know, who cares? I mean, not that I don't care about, obviously, what things are, but you know, it's not quite as present in my mind. But she's imagining a world where I get hit by a bus, and she's like, I'm screwed. And so like, there's a social decision making process, which forces the insurance upon us, and that's what we handle in the insurance industry. It's kind of famously inside the business, they say, it's, it's, it's bought not sold insurance. And what that means is that, you know, the customer shows up saying, I want to buy insurance from you, or from somebody, and I'm gonna pick you, and you don't have to actually talk them into the risk management benefits of insurance. You know, I like to joke that I would never hope to never again need to convince somebody the risk management, but if it's an insurance, because me and my experience, they don't give a damn, because they don't, they don't understand, they don't care. They're worried about other things. This is in the future. You know, to me, the availability bias and override of cognitive biases, which are well document all over the place. These prevent us really from, and I don't mean this as a criticism necessarily have the same problem myself, like, it's just a fundamentally hard thing, like, how many risks do you worry about, and you're in your life? And that's a really difficult thing. And so we start forcing people to buy things, because we're going to bail them out after the fact. Anyway, that's like a really critical decision. And I think like a primary motivating decision for defining the sort of primary decision for defining the insurance industry as we see it. I'm just wondering whether you thought about the model there.

Amy Finkelstein: 28:00

Yeah, no, as I said, I think this is this is particularly salient in in both natural disaster and health insurance. And so we thought about it in that context, as you said, the interesting question you're asking is why we feel, you know, compelled to act as a society and bail people out of certain situations and not others is, is is, you know, definitely beyond the scope of economics. It's really, I think, a question for social psychology or moral philosophy. But what they have my understanding of what they have to say about it, is, you know, there's sort of two things that tend to, you know, be a common denominator. And when society feels obligated to act, one is this notion of a moral emergency, or something that's up close and personal. So one of the examples, the legal scholar, Michelle, Michelle Dalbir, get gives in her work on the history of US policy and natural disasters, as she talks about, you know, in the moment of Hurricane Katrina, with all the horrible pictures, there was an overwhelming flood of, you know, public policy response and public opinion, to help the mostly very low income, disproportionately African American, poor individuals in New Orleans who were living in the exposed areas and subject to some of the worst of the disasters. And yet, you know, the soak, in some sense, their chronic condition rather than their acute condition of the sort of systematic discrimination and poverty that had left them vulnerable. There was no, there felt there was no sense that we needed to address that issue. We addressed the acute selling, you know, picture on TV, but not the underlying chronic condition and you know, Oh, I think so that's, that's one thing. And there's this, you know, tradition in social psychology of noting that, you know, if you ask people, you know, if you give them the vignette of, you're driving along in your fancy new car, and some guy is bleeding by the side of the road, and if you don't take him to the hospital, he's going to lose his leg. But if you do take him to the hospital, he's going to bleed all over your nice new car. Most people say, No, you're a jerk, obviously, you should take him to the hospital. Yes, if you say, you know, you could spend 50 bucks and save 100 kids who are dying in Africa, most people, although obviously they feel that's a good and right thing to do, they don't think you're a jerk for not doing it. And again, that the notion is given of the up close and personal versus, versus the sort of more distant. So that's one thing that people talk about, again, I'm just sort of summarizing a literature outside of economics. And then I think the other thing, particularly in US culture, is this is this notion of, you know, when people are suffering through, quote, unquote, no fault of their own. And so again, the idea of, you know, in the moment of Hurricane Katrina, nobody is blaming people for, you know, having their house flooded, that's seen as a, you know, an act beyond their control. And I think similarly in when someone is, you know, very, very ill in an emergency, but then somehow the more general chronic problems of poverty, joblessness, homelessness, etc. There's, you know, there's a tradition going back to the 19th century, and the British poor laws of viewing that as well. You know, those people, you know, we need to separate the deserving and the undeserving. I'm not condoning that at all. And I'm just trying to explain, I think that the notions of why sometimes we feel compelled to act. And other times we don't,

David Wright: 31:49

It's definitely the case that we do and do not in certain circumstances. As I say, it's something that I think about an awful lot. You know, I had this, you're mentioning there about the car thing, at this experience. I mean, a long time ago now, but 20 years ago, and I was dating my now wife, and we're walking along a beach. And there's a, you know, vagrant kind of looking guy, um, he didn't look great, right? So homeless or something. And he was, there's like a rock, kind of, like, I don't know, barrier towards the water. And he started having a seizure, like, you hit the ground, and it's kind of like, you know, thrashing around and smashing his head off a rock, and he starts bleeding. And I just, you know, I didn't I don't even remember doing this, but I guess, I'm no hero here, maybe, but this to happen, I took my shirt off, and I put it under his head. And, you know, just sort of cushion him? Well, I didn't know what to do. I was like, this guy, I know, what's happening here is he's gonna have a problem with his, with his brain, and worse than he already has. And my wife looked at me like I was on my mind. And she said, Do you have any idea that your blood from this person on you like, what do you think's in that? And I was like, you know, what, I didn't think too much about the consequences, maybe of what might happen to me, I just sort of acted. And now I have this bloody t shirt of mine, because other people came around. And you know, I was like, Alright, I'm done here. It was just sort of this thing. I just sort of did it. And it might not have been a good idea, actually, for me to do that. And it was mine. But anyway, so if somebody else had different perspective, thinking about risk a different way, you know, the, the story you told there about your story, I want to talk about paper, actually, that you wrote, where I think it was maybe an article actually, in New York Times, where you're talking about Medicare, or sort of Medicaid, and how subsidizing medicated versus giving somebody actually given them money. Poor people give you pretty different outcomes. And the concept of giving them cash instead of more insurance is more beneficial to them, if I understand the result, right? Socially, because they're already getting the insurance or the health care, rather, through these good samaritan type, you know, bailouts, we can call them that a little bit of a negative connotation. So maybe there's a better word you can give me for that. Because, you know, not really opining on the social mission of that, but they're already getting health care. So they don't need more health care necessarily what they need, is this help with their life? Isn't that strange that we've created a system whereby, you know, we've sort of introduce all these constraints, and we're trying to help them and it turns out to not be as good even though we probably would argue that's actually a good thing. Right?

Amy Finkelstein: 34:27

Yeah, it's not that they don't need more health care. In fact, you know, I've done work, you know, from a randomized experiment of what happens what happened in Oregon when some people were randomly given Medicaid coverage and other people weren't. And there are clear benefits to the low income uninsured, of getting health insurance, you see that they have, for example, much better protection financially, they don't have as much out of pocket expenditures as much medical debt. They report themselves in better health. It's more that The benefits aren't as great as they would be absent that the safety net that exists. And it exists. Because, you know, as we talked about, we feel, I think, for good reason, a real social obligation to come, you know, to the aid of people when they're sick, and don't have the resources to pay for it. But it does mean that, you know, then when you're facing the choices of the government of should we expand Medicaid or give people cash, a lot of what you're doing when you pay to expand Medicaid is paying for care that people would have already received for free, that would have been paid for by some third party, you know, which in the first instance, is probably the hospital in the state. But then there are complicated and difficult questions of how much that feeds back into private insurers, you know, private insurance premiums, and, and the like, all of this is very, very interesting, and important aspects of insurance that, you know, of which there are many that aren't really about selection, I mean, this would exist, even absent the the selection problem, the selection problem just makes it very difficult to try to get private markets to function, whether it's for layoff, insurance, divorce insurance, dental insurance, pet insurance, you know, auto insurance, all of those.

David Wright: 36:22

And people, you know, in some ways, I think it motivates the the inquiry into selection, because, you know, the default of, let's call it, the default, if the government doesn't come in to mandate insurance in advance, you get post event coverage, in a certain way, right. I wonder like, one thing I haven't really seen, I've looked a little bit for this occasionally. But how much do we know about the maybe you can define the word efficiency? Because it seems like an interesting kind of way of thinking about efficiency for insurance. But let's just think of it generally just as aggregate costs for the moment now, and I'm interested in getting into the concept of efficiency of insurances, as economists define it. But for just general economic efficiency, like isn't like the cost of something? What is the difference between pre event financing, paying for an insurance policy, putting the money in a pool that pays you when an event happens? versus post event? We don't necessarily find anything, and we just sort of show up with their handout and get bailed out? Surely the second one is more expensive, unless kind of efficacious or efficient for society. You know, post event, coverage must be bad, right? I mean, do we know how much we know about that?

Amy Finkelstein: 37:25

Well, it's an interesting question. If you if you think of, you know, what you could say that one way to say if we, if we set it formally and upfront that everyone, you know, if their houses destroyed by a hurricane is going to get, you know, a certain level of rebuilding of their house done, whatever you call that, whether you call that, you know, an ex post safety net, it's also it's also in some sense, you know, compulsory insurance that everyone has it, and it's just being financed, you know, by the taxpayer, right, rather than by paying, you know, a premium upfront, you're just paying your taxes. So I think the the issue with that is, is exactly are the are the challenges, I should say, without are exactly the challenges that you open this discussion with, about, you know, what are the pros and cons of government compelled insurance, the Pro is it can potentially solve this, you know, selection problems that if you simply said, Please, you know, people buy homeowners insurance if you need it. Well, the people who know that their house is really rickety, that they haven't been, you know, putting in the adaptations needed to withstand a hurricane are going to buy it and drive up the price and drive other people out. So, so that's the Pro, I think the con is, you know, well, not everyone wants the same level of insurance. So you'll get something from the government, but perhaps you would have wanted to buy a much, much more comprehensive policy. And that is now you know, not available or priced, so that only the people who are really, really high risk of needing it, you know, it makes sense to buy them whereas you might want it just as a low risk, but low risk, but but high risk aversion, homeowners so one of the things that's kind of neat, is it a lot of these policies, you know, even though they sound very different can be effectively the same thing. And one way to think about what FEMA is doing is it it's effectively providing everyone with a, you know, government compelled government financed out of out of, you know, our tax dollars fairly minimal policy that will pay out in the event of a, you know, federally designated disaster now, you know, you're on your own if your house is hit by a tornado, and it's not a federally designated disaster area. So that's, you know, that's the type of problem that limited policy.

David Wright: 39:51

One thing I thought you're gonna say there are invoke a concept of moral hazard in those circumstances. If you have an undefined blanket coverage If you have no specification for whether somebody, you know, in different kinds of word of world of calling it desert weather, they deserve to get the coverage, right? Because they could just ratchet up the risk and say, To hell with it, I'm going to, you know, I got all the coverage I want. So I'm going to build my house. You know, what the big basement and the flood plains of the Mississippi River?

Amy Finkelstein: 40:22

Yeah, yeah, this is what what this issue of, we sort of obliquely call it, you know, wait for our second book entitled, instead of risky business, risky behavior, but we're not we're not actually going to write the book. But yes, this issue that, which is an issue for, I think, all insurance, and it's an issue, regardless of whether the government is or is not involved that, you know, once once you have protection against risk that can, you know, reduce your incentives to engage in risk mitigating behavior, or, you know, encourage you to take more risk. In the example you give of, you know, your willingness to locate in a high flood area, or even given where you're located, how much you're going to, you know, sort of build up and elevate your house versus not is a very real problem. I actually have two PhD students, Abby, oh, striker, and Anna Rousseau, who have written an excellent paper on that. Just just this, you know, just this past semester, so you could have them on your podcast to talk about these about these issues. Yeah.

David Wright: 41:29

So the, the problem of, of actually, like controlling moral hazard, as I am conceiving of it here is one where insurance is not just a It's not just a payout, right? It comes with constraints. So it says, in an insurance, you tend to have things like at least in P&C, a more so than health, maybe you have reading variables, right? So you say, well, we're going to, we're going to insure you for this much. But if you double the size of your dry cleaner, your insurance will also double, because you have twice as much risk as you used to have at the very least, the concept of exposure driver is is a is kind of a constraint, at least on the price, or, you know, just the product to suit the greater risk. And you kind of exclusions, and you have controls, and you have all sorts of things that you aren't allowed to do this, you know, if you're going to get whatever motorcycle insurance, you can't do, stunk work. Right. And so you kind of bundle in this controls against moral hazard, which is I think this is really related to the sex selection problem in a certain way. I mean, how do you see these sort of moral option of interacting here?

Amy Finkelstein: 42:31

That's a great question. I mean, one way to look at it is a lot of the, the sorts of ways that insurance is distorted because of selection can also occur because of what you call moral hazard. So to give an example, waiting periods, right, that, you know, we give the example in the book of, you know, of my husband, who, you know, tried to upgrade his his roadside assistance policy to you know, cover, you know, long distance towing when his car broke down, and he wanted to get towed to his, you know, childhood mechanic, you know, so that that's, he can't be found out, you know, that, in fact, yes, he could upgrade his towing insurance, but it wouldn't start immediately. It has a two week waiting period. And that's precisely to prevent, you know, bad customers. He's a good husband, but he's bad insurance customer, righty are high risk and your

David Wright: 43:28

your co-authors and spouse, really, you're hanging out with great insureds.

Amy Finkelstein: 43:31

Exactly, exactly. So but But another reason to have to have you know, waiting periods are or deductibles elimination periods or exclusions is so to give you some pause before you you know, just go joyriding your car, you know, off road and might and might blow a tire. There are other aspects of insurance products that seem particularly designed to deal with selection rather than potentially also this moral hazard or behavioral component. One of my favorites that we talked about in the book is the phenomenon that often health insurers will bundle with their health insurance, product, a discounted membership at a gym. And you might think, or they might want you to think that that's because oh, well, they want you to be healthy and, and you know, working out at the gym will make you healthy. In fact, it turns out it's simply for a selection reason that and Cooper and Trivedi have a very nice article in the New England Journal of Medicine showing that when when insurers offer these bundle gym memberships, guess what kind of customers they attract the ones who are already healthier, because they're the ones who at least can delude themselves into thinking really, they're going to use that gym membership, whether or not they do and then there's other excellent work showing that in fact, if you randomly offer these gym membership policies that they don't actually You know, this isn't an employment in a workplace wellness context, they don't actually do much for worker health or productivity, at least in you know, the first few years, but what they do is attract, you know, the healthiest employees to sign up. So it really is much more about the selection of getting people to reveal who are the customers that yeah, that the insurer wants? Or who are the customers they don't want as opposed to the, you know, actually changing behavior in a moral hazard sense.

David Wright: 45:28

So, you've just invoked a brilliant conceptual structure there, which is another massive difference between health and other other lines of business, particularly Property Casualty and life even, actually, which is the concept of intermediation and, and they call it affinity, right. So you have a broker, this I mean, they kind of are insurance sense of the word intermediary, I suppose, the economic sense. So what what often insurance carriers do, and my most of my experience here is on P&C, but they will choose their distribution sources very carefully. And the district district, they do so to mitigate the selection problem. And so they'll have a, you know, if you are, let's say, in the kind of classic sense, you're the insurance agent in your town, and independent agent, you have a bunch of carriers, you're kind of working with, you know, the people, and they know you let's there's two insurance agents, there's the good guy and the bad guy, girl, good girl, bad guy. And, you know, you go to the, you go to the good gal, because you know that her customers are good people, we can trust them. They're not, they're, you know, they tend to have you know, of lower claims rates, they tend to be a little quicker to respond. They pay on time, all the time. And it's sort of always at least measures of quality are correlated, right? And you're the bad guy and you like his customers, a bunch of deadbeats, couple of them into jail driving records, you know, blah, blah, blah, on it goes he just sort of, you know, it's like, it's like when you're a parent, right? I don't know if you have kids, but I do. And it's like who my kids friends, your social circle, there's a lot of information in that, right. And so insurance carriers pick and choose their distribution sources, especially when personified by a person to filter out the customer base. And then, you know, one of the things that always frustrated me when I was a reinsurance actuary was how little insurance companies kind of really recorded data about this. So they would pin all this energy, which to a reinsurance actually looks wasteful, frankly, but I knew it was wrong at the time, on picking the regions and figuring out like talking to them and all this and like, isn't it just in the data? And it's not? It's not? For the reasons you point out in your book. And it's not something that exists in health? I think, because it's just so I think, pervasive, I mean, I'm not sure exactly kind of why I'm thought about that too much. But it is everywhere in P&C and life.

Amy Finkelstein: 47:36

Yeah, it's interesting. And you can see that what you're describing is, you know, the selection problem now playing out at the level of the intermediary, like right, again, some you know, even what you're calling the quote, unquote, bad intermediary who has high risk, you know, customers, right? They still would benefit from insurance, you just like to be able to offer

David Wright: 47:59

Everybody's got to buy liability insurance then too bad, they're going somewhere.

Amy Finkelstein: 48:02

But this notion that a lot of the one of the things we talk about in the book a lot is, you know, what is the private information? How is it that individuals, you know, can really know more than these sophisticated insurance companies with their, you know, mountains of data, their armies of Actuaries, they're, they're sophisticated algorithms, etc. And I think your example is in that sense a very interesting one, because it speaks to some of the sort of soft or, you know, the soft impression or information that's very hard to collect, even in this age of big data. So we give, we give an example in our book as well with auto insurance, of private information that I had, and that my co author Liran had cut in opposite directions. So, you know, my private or soft information was when I when I first applied for auto insurance, in graduate school, I had a they asked me, how long I had my license answer. 10 years, they asked for my driver's record, answer spotless. What they didn't know is it was spotless, because I hadn't driven in the past 10 years, that I had gotten my license when I was 17 years old, in Manhattan growing up in New York City, because my mother told me, I absolutely had to get a driver's license, it was a quote, invaluable form of ID. So I literally I learned enough about how to parallel park and you know, use turn signals to get my license. And then I hadn't driven in 10 years, hence my spotless record now, not just myself that if I had, you know, as your example of the intermediary, anyone who knew me also knew I was a terrible driver, but my insurance company didn't right so that's, that's the that's one type of private information. My my co author, Liran, you know, had the opposite problem. He has two teenage sons and, you know, the older one went off to college, and the younger one was supposed to get the car but his insurance premiums went up to now reflect that he had two teenage boys, you know, on the car. And you know, one teenage boy is bad enough two teenage boys is really bad. And we tried to explain that the older one was off in college and the family agreement had been that the younger one got the car once the older one went to college, but he was in college within, you know, a few miles of his parents house. And you know, the way the insurer you know, or the actuary looks at that is they know, he's at risk of driving the car. So the fact that, you know, Lauren could swear up and down that there'd been a multi year family agreement that one kid got the car, Julie went to college and then wasn't allowed to touch it, you know, he still looked high risk from the purpose of from the perspective of the Actuary. So those are examples where I think you're, you know, small town intermediary, who knows, you know, who's kind of in the community would know that, like one kid wasn't allowed to drive the car would know that I was a terrible driver. And my thought thought big a secret, right? And what might be able to actually try and you know, get rid of some of that private information.

David Wright: 51:02

Well, I guarantee you that insurance company learned that lesson the hard way. They probably saw huge amounts of increases in claims cost. And they figured out that answer, and this is one of the reasons why insurance incumbent insurance carriers last so long, it takes a long time to learn all these hard lessons for a company truly, truly does it. I do however, think that private information is probably more powerful in health than you have some pretty interesting evidence about how people I forget who it was that did the study, whether it was you, or, or when it may be some one of your colleagues looking at some data from studies of people's health look at the Health Survey, that kind of, you know, within the the study was conclusively determined, as I looked at it, that people understand how long they're gonna live. You know, they can they can really, you know, right, they can take

Amy Finkelstein: 51:56

Yeah, yeah, this was this was no, this was not a set of mine, there was Was this the life insurance one, there was also a health insurance, a long term care one. This was this was actually a former student of mine, Nathan 100, who's now about to be my colleague at MIT. Where Yeah, he basically looked in survey data, and where they asked people, you know, what's your, what is the chance you're going to die in the next 10 years? What is the chance you're gonna go into a nursing home, etc, and then follows them over time and looks at what happens. And he also, you know, can see everything that the insurance company asks is a very detailed, you know, survey data. And yeah, people are able to, to, to predict, sometimes, you know, often even better than the insurance company, what what is going to happen to them, one of the most interesting aspects of his study, is he's dealing with the fact that which is kind of at least two economists a puzzle that very high risk individuals are often rejected from insurance coverage. So you can't get life insurance, if you've had a stroke or long term care insurance if you've had a stroke. And, you know, we asked, you know, why don't they just charge a higher price? This isn't private information you can this is something they've asked about on the forum and then said it when you say you've had a stroke, instead of saying, Okay, your premium is higher, they say, No, you can't have insurance. And so he tries to understand this puzzle of rejections. And he actually finds it's also one of private information, when I remember when he first tried to explain this to me when when I was his thesis supervisor, and I sort of condescendingly said, Nathan, it can't be private information, it's there on the form, it's something that you add the insured. No, but as he, you know, explained first to me, and then much compellingly in his work. People use there are many different ways of recovering from a stroke. And the concern is that the amount of private information people who have had strokes have is much greater than, than you know what you can know if you're healthy. As he puts it, it's a very nice line, he says, look, there, there's only one way to be healthy, there are many ways to be sick. And so we don't mind ensuring the Healthy People because we're less worried about their private information. But people who have had major past health events, there's a lot of private information into how well their recovery is going whether they're likely to have another event. And so those people tend to get totally shut out of the market. In fact, he was completely right for this, you know, selection problem that I had poo pooed as not being possible in this situation.

David Wright: 54:32

Well, what we see I mean, what I see in P&C, and we're doing a lot more experience there than in this segment, is you get specialist carriers that handle these these kinds of segments. So we're it's compulsory anyway, in this case, struggling to volunteer product, so there's probably just not enough of that population. You have two effects going on. So one of them is once the price gets to an absolute number that people don't like they just walk and they don't buy it because it's going to be expensive. In compulsory markets. You just sort of have specialist carriers that compete over that business and it will be as cheap as it can be, but still a lot more money. And let's say you're a, you're running a dump truck company, and you got you got a DUI, you know, like, it's either you pay up and maybe that is that number is really high or you go to business, because you can't get auto liability insurance because they already go to a specialist carrier who says, we're going to rehabilitate you. And here's all these things. And they did crazy things me like they'll, they'll that they will make sure you go to AAA. And they're like, they have all these crazy requirements in the policy, just because they understand what the problem is, the problem is you got a drinking problem. And so we will allow you to continue operating your business, but you know, under these controls, to make sure that you don't have another DUI, because that's going to have a claim for sure. For us, right. And so, you know, like, you know, if you want a deal, here's what it looks like. And so we're gonna identify the real risk, and we're going to actually underwrite that. But in voluntary markets, you're just out there, like, forget it. I'm not, it's just gonna be too much money for me. And insurance carriers. I've seen two generals, insurance carriers, they're scared to quote, a high price for the reputational effects, because you're gonna say, they're screwing me and you're gonna tell your friends and they're gonna look bad. And like, we don't want that we'd rather just not even play in this marketplace. And so you go to, like I say, a specialist marketplace, and they can, you know, they'll do what, what they'll do. I do want to add one other thing about this private information thing in P&C anyway, for, let's say, auto might be a classic example is my take on that is, for the most part, the information people get about the riskiness they get from insurance companies, because, you know, your case is a is a, you know, a pretty stark one and a very good one, of course, you haven't driven hadn't driven in 10 years, but for most people, like you all think you're better than average. Right? It's kind of a little Lake Wobegon thing. Yeah. And so how do you get real information? Because you're, I have another claim, my claim, and so all of us have another claim together, some of us are better risk than others. But we have characteristics, which are, you know, like, we have a slightly, I might have a 10% probability, and you have a 5% probability, but over the course of like, you know, 10 years that might not, we might not tell the difference between us, right? Relation of people and insurance, nobody has an information. And so it's only by going to get quotes that we learned about our riskiness. There's a feedback loop.

Amy Finkelstein: 57:12

Yeah, that's really interesting. There is that feedback loop for sure. And yet, you know, it's still seems that, you know, if it was only that you were getting the information from the insurer, then you shouldn't have as a potential customer, any informational advantage. And for sure, the examples I gave myself and my co author are, you know, idiosyncratic. There's always a tension between, you know, the anecdote, which is, hopefully somewhat, you know, memorable and compelling. And the, you know, more careful, rigorous academic study. And so what we try to do in the book is, is include both. So the example, with auto insurance, the actual evidence from this is work that Alma Cohen has done on automobile insurance in Israel is actually quite consistent with your intuition in the sense that for new drivers, beginning drivers, it doesn't look like there's much private information. But for experienced drivers, you do see evidence that, you know, people who, who who have private information that they're worse drivers are more likely to, you know, say upgrade their, to a to a more comprehensive automobile insurance policy. And again, these may, you know, to your point, actual auto accidents or claims are relatively rare. So it may not be something that company observes. But, you know, well, if you've ever driven with me a few times, you'll get a sense, even if I don't have an auto insurance claim that, you know, I'm at high risk for one. And so I think it is that information that can come with with with experience that can wreak havoc in the market.

David Wright: 58:47

What is efficiency in insurance is an efficient insurance market.

Amy Finkelstein: 58:52

That is that is a great question. I guess what I would say there are many different ways to define efficiency, I think, for our purposes and for thinking about the problem of selection. I would say that, you know, going back to something I said earlier, even safe drivers and healthy people benefit from insurance, they benefit you because that just because you're a safe driver doesn't mean an accident can't happen. It just means it's less likely to happen and we would all benefit from or almost everyone would want to have some security against the sort of potential for you know, having to pay a large claim, do it for an automobile accident, a home that's destroyed, expensive dental implants, what have you. And most people are therefore even willing to pay a little bit more than the amount they expect to get back from the insurance company. So if you have a one in 100 chance of having $100 claim Have, you might be willing, you know, to pay more than the, you know, expected, you know, payout from the insurance company to to make sure you don't ever have to pay that. So what's inefficient is when people can't do that when people aren't who would value insurance and being and would be willing to pay more than it would cost the insurance company to cover them. So the insurance company would be willing to sell them the policy when they can't buy it, because selection has driven up the price, under the assumption that everyone who's buying it must be really high risk. So the inefficient outcome and the real tragedy if I can use if it can be a little melodramatic work well, problem with selection is that it can destroy the market and have people who would benefit from insurance not and be willing to pay more than it costs the company to provide that insurance unable to buy the insurance. That's the, you know, inefficiency caused by selection, or, you know, the tragedy in some sense of you know, when the market doesn't work, well,

David Wright: 1:01:02

Let me throw a kind of theory at you. This is is I pretty fast, this has to be wrong in some way. But this is how I kind of live my career in insurance. And it is as the following. Nobody wants to buy insurance. Nobody ever wants to buy insurance, nobody would buy insurance, it there is no there is no rational process by which somebody arrives at the conclusion that they will pay more money for insurance premium than they will get back. That's, that's like an opinion I have. But what because what I see is I see always a super irrational process in the sense of a social process, which forces people to buy insurance, you don't tend to make the decision for yourself. And whenever I have pitched insurance on voluntary on compulsory basis, it's easy. You don't have a choice like well, I don't care what you want to buy, you're buying it because you have to and you can walk into my shop and you'll buy insurance from me. But when you buy voluntary coverages, inevitably, the if you don't pass a cost benefit analysis, and we're assigning zero benefit to future payouts. So saying next year, am I gonna make money on this policy? If the answer is no, there's nothing for us to talk about, right. And I've lived through these conversations, many of them to individuals over the phone or the internet, to big companies to medium sized companies. They all they all approach the concept of a voluntary purchase the same way, and they never buy it virtually, with some exceptions. And those exceptions tend to be socially driven. For example, if all of your competitors are buying this insurance, you don't necessarily know if the value proposition is negative to you. But they're all every else is doing it. Right? Then that's pretty powerful evidence, or if somebody somewhere else decides they want to buy this, and they're making you do it. Right. And that doesn't necessarily have to be legal compulsion, there's contractual compulsion. And there's other sort of forms of doing this, but nobody ever wants to buy insurance. What do you think about that?

Amy Finkelstein: 1:02:49

You know, I have not had your experience trying to sell insurance. And as I said, most of most of my experience, only experience comes more from looking at or buying Life and Health Insurance. Maybe Maybe, it's just, you know, crazy academics. But I'd say for myself and my family, like, we do want to buy insurance, we're willing, we understand that, just because you buy an insurance product, and it never pays out doesn't mean that was a mistake, or quote unquote, a bad deal. You're buying it against a potential, you know, risk that hopefully won't materialize. But, you know, obviously, everyone will think about it differently. And

David Wright: 1:03:39

So let me ask you this question, why don't you buy more insurance?

Amy Finkelstein: 1:03:42

Well, that's a good question. You know, that's something that this gets to the pricing. And at some point, you know, we're not infinitely risk averse, right? We don't want to give up all of our current money to have money in the future, and something bad happens. And so you sort of think about, I don't know, how much, you know, what's the price of the insurance? What's the benefit, and you buy until you think, you know, you've you you thought you've paid enough for what you're getting?

David Wright: 1:04:07

So, so this gets to, which I don't want to be like, critical. That's kind of nuts. Right? I mean, can you how do you know? For sure, person, I'm sure. People, but you know, like, you know, person. And this is why, you know, when I presented it there is kind of presented in kind of like a, you know, I don't know, kind of quasi critical way, I don't actually mean that way, I think this is actually the right thing for society. Because no person, not even any small group, people really have anything close to the amount of information you need to integrate to understand the risks you face. Right? And we're not even talking about the stuff we haven't seen yet. You know, climate change. Let's just take the one that, you know, nobody's actually seen the ice caps melt, right. So we're all anticipating that being a problem amongst many, many other things. And there's all sorts of poll availability entrepreneurs to use a term that I think common came up with me was actually was Cass Sunstein, I think, you know, walking around there trying to say, hey, you should be painting should have this or that risk. And you know, we need to have a the only process that I'm aware of the society has to integrate all that information is through politics and compulsory insurance, right? So no family really is responsible for that.

Amy Finkelstein: 1:05:14

It's amazing because you say that, and it's totally intuitive. It's, you know, it's hard to just add a rhetorical level disagree with you. But then you some of the studies that we talk about, in our book actually, just, you know, contradict that. I mean, so for example, diving has a study on life insurance, she looks at people who buy life insurance, she absorbed, observes an enormous amount about them, their health, their family history, and in some sense, almost more than an insurer could observe, because they're not buying they're not, you know, being asked these questions by an insurer where they might have incentives to shade their information or, you know, change their behavior if it would get them a better price. And then she follows them over the next 10 years. And the and she asked them your don't cheat the survey asked them initially, what do you think your chance of dying in the next 10 years are? And she finds that not only, you know, are people able to predict over and above all this rich information she has, but that those who who are those who buy long term care insurance, in fact, have a higher mortality rate over the next 10 years. So So at one level, I deeply agree with you, it boggles our mind. And we in the case of life insurance, we're sort of left we really don't know what it is that these people know that the insurance company doesn't. And yet the evidence says they know something. In other cases, like we talked about Emily Oster study of long term care insurance. There, there's a very clear example, she's people who tested positive for the Huntington's chorea gene, are more likely to buy Long Term Care Insurance and those who tested negative and people who are at risk of having Huntington's, because one of their parents had them are more likely to buy long term care insurance. So in some cases, we can pinpoint the exact source of the private information. Sometimes we can do it with anecdotes like the one I gave you, you know, when they gave you an auto insurance, in other cases, we can't actually, we don't know what it is. And I agree with you. My sort of prior would also be why don't people know people know nothing, right. And yet, the data says otherwise?

David Wright: 1:07:25

Well, I think that there's a there are we have it, we have cognitive biases, which we use to make these decisions, we do a availability bias is a real thing. I mean, 100%, I was talking to an executive from Aflac, once, and I asked what happens to your personal accident sales, when a celebrity has a skiing accident, as they go up. We sell more sure insurance when a celebrity breaks their leg on the ski slopes. And that's because everybody is now suddenly aware of it. But you know what they do? I mean, they cancel it the next year. Because it's gone. It's sort of in and out. So to me, like, sustainable, you know, the risk doesn't change for you probably like, as long as you don't have private information. So they're, I would say, we're kind of into the cost benefit analysis, they're probably thinking they're getting a good deal. And so they're taking that good deal. But for like real risk management, the risks don't change very often. So we have a decision making process that is persistent, to the same kind of order of the risk, right, so has to be good to stick with it long enough to actually have the risk be properly handled. And we are capricious and flighty creatures, and we change our minds and change our priorities all the time. And we need something that's more permanent, I think, to make insurance decisions.

Amy Finkelstein: 1:08:34

Very interesting.

David Wright: 1:08:35

Let me close on this kind of last thing here, which is I want to talk about your process. Because you, you I you know, you have this phenomenally fascinating Medicaid study you did about 1015 years ago, I think. And you know, to me, like the innovation in that was actually, you know, the strategy conceiving of and getting the data. And so I, you know, I we just weren't interested, you talking a bit about kind of, I don't know, call it the entrepreneurial kind of scrappy side of how do you work with organizations to get interesting datasets you're doing empirical work? Seems to be that's a lot of the value can disagree at the premises if you'd like, but I think that's really interesting.

Amy Finkelstein: 1:09:09

No, I 100% agree. I mean, you know, sometimes I feel that, you know, you know, my real like skill as an economist is not the, you know, the insights, but but the data and getting, and I think increasingly, you see people finding all sorts of creative ways to, to get data or to use data that was already existing, but not, perhaps people hadn't realized, you know, this particular use of it. So if you think for example, the Huntington's work I mentioned that Emily Oster did on long term care insurance. It's not that she and her co authors actually went out and, you know, created a sample of people whose parents had had Huntington's and surveyed them about their long term care insurance, you know, purchase decisions and their genetic testing decisions and that data existed, I believe, you know, for a group of physicians had been following it for years collecting it and following it for years to understand other aspects of the disease and its impact on families. And it was just that entrepreneurial insight that wow, that could be useful for for something else, you know, for studying selection. So, yes, I think I think some of the best economist and best economic papers come out of either creating new data or realizing the potential and data that already existed that that hadn't been exploited. It's funny, because, you know, I got a lot of time on my hands. So that's what we do. I mean, you know, in some sense, you're, you're you're selecting on the outcome. You haven't seen all the crazy ideas, I had to get data that that failed, I couldn't get the data or I got the data. And it wasn't it wasn't useful. So so, you know, I don't I don't have you know, it's not like I have 100% batting average.

David Wright: 1:10:57

But I mean, you're you're, you're exposing yourself the right tail volatility.

Amy Finkelstein: 1:11:01

Exactly, exactly. It's all about it's all about the right tail. That's what I tell my students as well, you got to aim for the fences.

David Wright: 1:11:07

Yeah, we're studying insurance, which is not that. Let's design our careers around something. With more upside to close there. My guest today is Amy Finkelstein. Amy, thank you so much.

Amy Finkelstein: 1:11:18

Thanks so much for having me.