Speaker 1: 0:11

Welcome to the Naadam reasonable podcast. I'm your host, David Wright and actuary reinsurance broker. This is a show of interviews with people who have something to teach us about managing our businesses and ourselves. There's a lucky one out there, folks, so let's get to work.

Speaker 2: 0:32

The show is brought to you by beach where I work and have worked my entire career. We were a global reinsurance intermediary and we pride ourselves on creative thinking, deep analysis and client service. Those three qualities are actually intimately related because you can't spend the energy digging deep into a problem unless you really care about the client. We find that when you really do understand the problem, the solution becomes obvious even if it seems a little bit unorthodox to the outside world. At first, the nature of insurance, this it takes solutions we know and trust and try to force them onto the problems of the day. We just don't settle for that. You can see further at beach, GP.com.

Speaker 3: 1:10

Oh,

Speaker 2: 1:11

if you're an actuary, you liked me. Probably dread the professionalism continuing education requirement. I think that the best time to satisfy this isn't podcast time while on your commute, why walking your dog while mowing your lawn while doing the dishes had over to not unprofessional.com were for the price of a cow's Webinar. You can get content dedicated to continuing education for actuaries especially professionalism. See, that's not unprofessional.com. Thanks for listening and thanks for supporting the not unreasonable podcast. My guest today, Steve Milton Hall, is running for a position on the casualty actuarial society board of directors. He made an interesting observation to me actually when I was asking him, why would a casualty actuary want to vote for you for the board? He said, actuaries have been through three revolutions. Three analytics. Revolution is during his career catastrophe modeling, capital modeling, and predictive analytics. Actuaries are the traditional analysts of the insurance industry and across these revolutions and needs assessment, we're 50 slash 50 winners on capital modeling in the sense that actually is run the capital models of insurance companies tie on predictive modeling in the sense that actuaries do run predictive models, but data scientists had been hired in many occasions to run predictive models and insurance companies and complete losers on catastrophe modeling. I think that that's a pretty strong way of putting it, but cat modeling has spawned an entire industry of private organizations that are outside of the field of actuarial science and so steve might have a point there with that. In any case, for the analytics revolutions of the future, Steve believes that the actuarial society should be front and center and in his view and improved focus on education. Implementation and research is the way to do it. Steve got experience in all of these areas, being both a practitioner and an academic during the course of his career, so vote for Steve Milton Hall. Fellows of the casualty actuarial society will receive an email and you can vote anytime during the month of August. Go to the website,[inaudible] dot org. My guest today is Steve Milton Hall. Steve is assistant professor of risk management and insurance and director of insurance data analytics at the school of risk management in a Tobin College of business at St John's university. Steve has 25 years experience in the insurance industry. Prior to joining St John's in 2016, he was the global ceo of analytics for a on plc based in Singapore and head of an bank then filled analytics. Steve received his phd in mathematics from the University of Chicago. Steve, welcome to the show. Thank you for having me. So this is going to be what I like to hope hope is episode one of many of the not unreasonable book club. And hopefully Steve, you can be my cohost and this little endeavor today. We've got a series of books and papers that will be putting links up to as we cover them. We might have more than we'll cover today. We'll see. Uh, we're starting off with a book called radical markets, which was Steve Summer reading project when he was on holiday and the suggested it as a, as a good starting point and um, so I've now started up on it and so we'll jump into that. So the book is divided into a few different sections. We'll touch on that on them individually. Overall, the idea of the book is you can solve problems you might not have thought of before thought to solve before with a radical use of markets. And we'll talk a bit about that. The first market that the, that they cover is one in property rights and they have an interesting way of applying markets to, to, I guess dislodge the People's, uh, uh, inefficient attachment to their own property. Steve, how am I doing with that description? What do you think?

Speaker 4: 4:45

Yeah, I think that's, that's a fair way of putting it. Um, they, the basic idea was that you'd have a wealth tax that would be self assessed. So you would post your photo reservation price on, on assets you own, particularly property and you would be taxed according to that value, but if someone else came along and they wanted to buy it, they could buy it at the value that you'd posted to that. That's your sell price. It kept you, kept you honest in your assessment

Speaker 2: 5:11

and if you put your price too high, your tax bill goes up. If you put your price to blow, somebody wants to buy your property.

Speaker 4: 5:17

Yeah. And, and they were seeing it as a way around sort of eminent domain problems and hold outs with construction. You've bought every other house along the train the right of way or whatever. Uh, and uh, you know, this would be a way of ensuring that those projects that would maybe have a greater social value will be able to, uh, proceed. What do you think about the argument? So, I mean, it's, I like the concept. It sounds great. I'm cute. It is cute. I, you know, as a person who lived in the same house for 25 years that I had an unreasonable attachment to, I sort of didn't like the idea that someone could come along and, and just, you know, move me out of my house. And I did think that they missed a couple of points. One was that to presumably if I could post a low price, if someone named came along and bought it off me, they would then have to post the price and I could just buy it back. Right. So there would be a sort of trading, it would almost be like sniping on Ebay auctions where you'd have your, your listed price, but then in the back you'd have an actual price you bid up to. So I think it would need to be a little bit more of a complicated market. Then they said, but you know, the, the concept would, which would still work. Um, the piece that I struggled with was they identified the use with the highest value as the I'm used to which the person who would be willing to pay the most would put the property. And I don't think that actually is true right. Often the person who is willing to pay the most for property has got their sums wrong. They've made some assumptions that aren't true. Right. There's a sort of buyer beware, a winner's curse. And is that sort of bringing that down to, I'm willing to pay the most. Does that actually mean you've got the highest value use for the property? Interesting. That was one observation. The other problem I handle is, you know, imagine that you're sort of, you know, 75 year old widow who's lived on the coast in Florida since it was basically a swamp in a, in a very modest home, which is now in an area that has become hugely gentrified. Should she be kicked out of her house because through no fault of her own, it's been gentrified around her. That seemed to me. Oh no problem.

Speaker 2: 7:29

I, I, I agree with it. With the sentiment of your response there. Very much so because I think that I also have an unnatural attachment to where I live and I think there were very good reasons for that. I mean, you think of if we're confining this to a conversation about people's homes, which is not enough, so the authors would probably respond to this by saying, well, well let's just start with commercial property, which they do make the point at one point in the book, and we can come back to that in a second, but I think it's worth saying that there. There are good reasons in many communities develop around where you live, right? You have friends. I mean I'm just putting my kid into an elementary school nearby and now we're meeting all the parents and man, do we ever feel better about where we live because we have friends and our kids have friends and role the kind of the same people. I mean they're friends were the same people and so that's worth a lot to me and there's a non monetary benefit to that which is colossal, which is unmeasurable because it's a subtle social fabric of the community which we do all with something we all want and and threatening. That I think is one. Not, it's just a bad idea, but two would never work. They would never be able to pull this off,

Speaker 4: 8:33

so there was that problem. The other problem they had with some of that math got a little flaky. They they talked about an impact of this would be to lower property values, but I think about a third and then they talked about that, you know, your assessed value will be sort of net of any debt you have on the property, but if you lower property values by a third, anyone with the mortgage is water and the whole sort of mortgage fills the entire financial system. I think it was an issue. It was an interesting concept, but maybe more of a piece than a practical though.

Speaker 2: 9:00

Yeah. One other one other historical quickly tell us if a good friend of mine lived in a town called hoboken, which is known to us as being living in the New York area and that was a. As a. As a community, it's wonder when enormous gentrification, where last 25 years as most of the New York City area has and there's this friend of mine who lived in this, call it a row of townhouses on block and he and the rest of the folks in the little block were given notice from a developer who said, I will pay you three times what the market value of your house or some some huge multiple like you mean it was. It was like, Whoa, okay. Right. And The lady next to him said, no, and you have exactly the problem they're talking about in this book, which is to say that because one person can say no, it threatens everybody's. Everybody in some sense, liberty of being able to sell their house or not for the highest bidder. And so that she's a woman who is, you know, who knows in her eighties or something, let's just sort of picture that way. And my friend went to her and said, why not wearing doing that? And she said, because I have nowhere else to go. I've never lived anywhere else. My kids are all gone. Where am I going to go? And it's like, good point. Yeah, right. What are we gonna do with that person? I mean, it's easy if you're setting up and the technocratic ivory tower, they say, well that's an inefficient outcome. We're going to have to move this lady out of the way and any kind of abstract away to this. Like what are we all here for anyway? Right. I mean there's a point here for, for somebody to have a slightly better, you know, nicer highway, right? I don't know if you've ever heard of Robert Moses or, or read up on that. So he was this a guy that ran the parks, the Newark Park Association, I think it's called in the, in the city, a real power broker in the New York political life in the mid 20th century. And one of the things that he was enamored with his cars and so he built the highways and one of the things you wanted to do was build a highway through the middle of Manhattan and he was famous for trampling on property rights and eminent domain use and that kind of thing. And there's a woman, a famous woman who stood up to him and she lived in the village name was Jane Jacobs, and Jane Jacobs is one of these famous people, right? The Santini, urbanization, a kind of person development in any development. She's much more complicated person than that, but it's the very same issue where she used to knit for communities and he's standing up for development and had this big clash and she won. I did not know that against Robert Moses. Um. Alright. So, uh, and we also say about this section on the property rights thing. That's good. Keep moving. Okay. So the next one is we're probably going to skate through a couple of these, um, uh, voting, uh, so buying votes with a cod quadratic cost. So you want to summarize that one, Steve?

Speaker 4: 11:20

Yeah. Quadratic voting is an idea that you should have, rather than just one person, one vote, you are given a, a allowance of what they called voice credits, which are basically votes. You can think of it as you're given a number of votes, um, and you can issue, you can vote multiple times on a given proposal, but you, the cost of a vote goes up quadratically with the sort of strength of the opinion that you want to issue. So for example, I, if I just want to vote once, he calls me one voice credit, but if I want to vote twice, it costs me for voice crack. Not at all costs. You Zero Provo, not at whole company. Save Them, right? You can save them as well so you can save your votes up for the issues that really matter to you and you can have a slightly larger vote on those issues, but it, it doesn't go up linearly. It cost you. Each marginal vote costs you more and more. And, and, and the, the point, the quadratic function, Katrina has got the property roughly that the margin is, is an right, it goes up because the summit, the first indigenous goes up like n squared. Um, so they, they, you know, I thought the nice illustration that they used on this was that if you do survey data and you ask people, you know, do you really don't care know, don't care a little bit, kind of neutral, Kesem, Carillon, that's sort of one to five scale. Typically the responses you get are you shaped a lot of people really don't care. A lot of people care a lot. Well that's probably not true, right? That doesn't actually ring true. Most people are kind of neutral about most things. So if you give them, you give them a survey and you allow them to vote like this with code, right? They can vote plus one minus one where they can vote plus two, but it cost them four votes. They're forced to think about, okay, I've got this allocation of opinion that I can express and how do I want to play my chips? And you can do surveys with this and you actually then get a distribution of, of opinions across the issues that's much more normally distributed with much more what you would expect to see. And so this seemed like a, you know, seemed like an interesting idea.

Speaker 2: 13:28

Yeah. One of the things that occurred to me about that is what was interesting, and this touches on an actuarial topic, which we should touch on once in a while here, Steve, is there, their standard of evidence for that this is working, is that they get a graph out the end, which looks like a normal distribution.

Speaker 4: 13:46

No, it's like that. It's not a very good statistical test. They had a number. I mean the other issue that annoyed me about this was that they just boldly asserted that most reviews in Amazon are also u shaped. And that's absolutely not true. Right? It's exactly right. Let's move on. Why don't we move on to. One of the things I wanted to mention about that,

Speaker 2: 14:05

which was there's a comment in there they just didn't understand and I'm wondering if you did where they said that for some reason this is supposed to make make outcomes less extreme. And the thought that I had was, I mean the media, just evidence of, of the graph shape, right? Where this going to become a more smooth looking graph, more normal looking distribution as opposed to a more, I don't know, it distribution with a tail events, uh, or sorry, tailor outcomes, one star, five star and I just didn't understand that. Like if you're, if you're, if you're allowing people to opt out or opt in with more than they're kind of one person, one vote. Doesn't that make it more extreme? I just, I just didn't understand the logic. I don't know if maybe you did.

Speaker 4: 14:44

I, I think that the point simply was that to express a very strong opinion costs you way more. Whereas it's free to, you know, you can tell people I'm strongly for or strongly against. It doesn't cost you anything. You're sort of forcing a cost in India expression of the opinion and some people will be a little more tame about how they express their opinion. So that's how I understood it.

Speaker 2: 15:08

Yeah. And maybe the, the binary voting outcomes we can. It's only available to us now is zero one. Maybe there's something there about that being. Mostly we use Zeros.

Speaker 4: 15:19

Yeah, a lot of zeroes. Right. And you also talked about the ability to vote against someone, quite sure if that was like shorting a vote and it created a candidate on the other side. That's pretty cute. Yeah. It was potentially an interesting idea.

Speaker 2: 15:30

Maybe a theme that emerged for me in this book was, and this is something that you mentioned to me when we were talking about initially discussing this book is a, there's a lot of like neat ideas, but man, I'm not sure any of them work.

Speaker 4: 15:42

Right. Well, I'm on this particular one. I was curious and I looked up one of the papers that was referenced and it was had Nash equilibrium in the title and I was quite surprised that it was 30 pages of fairly dense looking math. Right. So I think on this idea that actually, you know, there's at least some solid theory that underlies it, whether it work in practice or not, I don't know.

Speaker 2: 16:01

Yeah. And obviously any, you know, thinking politically, anything that allows people to buy votes. I mean that just sounds

Speaker 4: 16:07

that would be tough. Toxic. Although they. The social dividend from the revenue. Yeah. Yeah.

Speaker 2: 16:14

Uh, well we'll move on. So we're going to go visit. I'm pretty quick too, I think, which is migration and so maybe we can summarize the argument a little bit here where they're trying to encourage migration because the feeling there is that there isn't enough migration, immigration or immigration, um, and that the idea was that you'd be able to see individual sponsor an immigrant to come in and work for you.

Speaker 4: 16:34

Did I get that right? Yeah. So at the moment, uh, corporations can sponsor individuals when they were like, well, why not allow individuals to sponsor immigrants to come in this chapter? Frankly, it had me feeling very queasy. I mean, it was, it sounded very much like creating a sort of underclass in society. And I, I really quite turned off by, by this chapter. Yeah.

Speaker 2: 16:58

So let's move on to a stock market voting maybe another quick one for us, which is to say this is an interesting argument that I'd seen a couple in a couple of other instances where there, the, the premise here is that institutional investors call it passive index fund. Investors are actually call it the majority or plurality of the ownership, a significant one of most companies. And so if you're a bunch of index funds who are passive investors and they own all of the airlines, I mean airlines, the classic example here, then there's actually an incentive for them to encourage collusion and turns fixing and oligopolies tic behavior because they don't care who, who ends because they own all of them. And so you have this functional monopoly or oligopoly, I guess because of the ownership is common. Um, what do think about that argument?

Speaker 4: 17:47

So I think it's good argument. I mean, I, you know, and I liked the way they phrased it, which was these, you know, the guy from state street or blackrock or vanguard or fidelity could, you know, has access to the ceos. They own six plus percent of the stock of a lot of these companies and they can just say the same thing to the ceos and it's, you know, it's nothing illegal. Um, but they can have the guide thinking, well maybe I shouldn't go after market share and sort of zero sum game if I'm not growing the pie, we can all cut costs. And the net result is all our shareholders are worse off. So obviously it's a balance between the individual as the ultimate shareholder owner and the individual as the consumer. And how has that kind of squared away, and I, I think one of the points on this that was important was that the, because ultimately you know, you know, you can say well we can solve this problem simply by forcing the institutional investors to kind of pass their votes through to the ultimate owners. So the owners express an opinion and then the institutional owner is forced to vote the opinion of their shareholders rather than sort of having the proxy conduct a poll. And they did, you know, they make the good point there that encourage to encourage people to vote. They have to feel like their vote matters on the margin. Yeah. And it's very hard if you're one of$100 million, whatever, x, Y, z Corp shareholders to do that. So the, the institutional guys have that incentive because they own enough that they can see they're having an impact, but the individuals may maybe wouldn't and that their solution was to allow the institutional investors to own a substantial share of a company. But if they own a substantial share of the company, it could own. They could only pick one company in each industry group, which is w as heavy handed. But I might read this, was that this was a sort of typical, let's introduce a little bit of a regulation here and then you need regulation to make sure the regulation works in soon. You've got this big unworkable bureaucracy. That bad idea also felt very much along those lines. Yeah.

Speaker 2: 19:52

One of the things I was reading, uh, because I've, I've some reason will come across different articles and commentaries on this idea over time. And one of the interesting takes I heard on this was that, and I felt this urge to the first time I heard it, where my response was, yeah, but does voting matter or how much does it matter? And what's interesting at that point is you're kind of appealing to agency cost as a solution to this problem. So you're actually saying, oh yeah, but the markets don't work that well and the shareholders aren't actually in control. And he kind of stopped yourself. Wait a minute, is that a good argument or is that a bad argument?

Speaker 4: 20:22

Well, it tends. I think that the sort of corporate control, um, there's the incremental change and there's the revolutionary change and it's the revolutionary change, which is the takeover. Somebody comes in and Cleans House with the senior management and that's, I think what actually sharpens, opens people's minds and that,

Speaker 2: 20:40

or you have an activist investor. I mean, think in our world, amtrust with Carl Icahn turns up, uh, you know, shakes everybody, uh, out of a stupor, I guess I don't know what it is. And I guess the price raise that way. I mean, that was not an institutional investor action and was able to persuade people even with a, with a minority of the votes. I mean he didn't have that much, but he was able to kind of shake people up a little bit and get something done. And I think that the activist investor, I'm the phenomenon of the activist investor is a pretty strong argument against this kind of thing because they can sort of wake people up a little bit more than they might otherwise.

Speaker 4: 21:13

Yeah. I mean I think the two chapter sort of was in within the context of monopolies and anti competitive behavior regulation and I think they made the good point that maybe we've taken our eye off that ball a little bit. You know, it was a huge issue in the late 18 hundreds sort of through Sherman Act and what have you into this sort of 19 fifties. And then one of the crawlers of the sort of Reagan, Sacha deregulation was that we stopped worrying about that question. Why should we put our trust in markets? And I think they do raise an issue that, you know, you look at a lot of industries are very, very concentrated. I mean, how many car manufacturers are there in the world now? That's a handful, right? And you've always got this problem of balancing economies of scale. It's cheaper to make things in bulk versus that, uh, you know, monopolian in market power. Um, but I thought it was a valid point that it, it should probably be getting more attention. I wasn't convinced that the solution they had was probably be the practical way to go.

Speaker 2: 22:11

I agree with you and we'll touch on some monopoly topics a bit later on. I'm sure. Uh, but let's skip over to the last section of the book, I think on data. So this was interesting and thought provoking section. The observation here being that quite a lot of the value that sits within a lot of the large tech companies call it the Google and facebook, is that they are selling the data that they are taking for nothing from individually, you know, quote unquote nothing. Which I actually have a problem with that because they're offering a service for free so they're not, you know, it's not as though the individuals who are giving up their personal data to Google and facebook, which that Google and facebook are using to, to, to, to sell advertising services or whatever else it is. There the are giving something back to the consumer in the form of a service which consumers paying nothing for. Um, but in any case there, the point I made there is that Google and facebook, there's, there's a lot of value being aggregated in these organizations. The value of the data. And that's not something that's recognized in any monetary way.

Speaker 4: 23:07

Yeah. I mean, I, I thought this was the most interesting chapter and the little phrase that they use to sort of coin it was questioning his should data be regarded as capital or data as labor and they're not completely clear on what they mean by that. But I think the idea is that if data is capital, it's owned and consolidated by these big corporations, by the googles, facebooks, and it's sort of centrally controlled versus data as Labor is, is more of a individual. I want to, of our own data and potentially selling our own data in the same way that we sell our labor on a daily, hourly kind of kind of a rate. Um, I, I thought there were a number of interesting things about this. I mean, they, to your point about we do get something in return. They had a very nice discussion of how the business model evolved that, um, you know, that I think their argument would be basically the ship's sailed on paying people for provision of data. Sure. We're all used to not paying for the services they provide and if they were to pay us for the data that they give us, we'd have to pay them for the services were provided and nobody wants to sort of open either of those two cans of worlds. Right. And, and, and I, yeah. Why? I personally am very happy with the bargain I get from Google. I'm not convinced that the data I give them is particularly valuable, but what I get out of Google is fantastic. I know it's an incredibly useful thing. I think google has single handedly increased the productivity. I would argue of of programmers and of academics everywhere because I can now do a complete literature search, pretty much download every paper I want in an afternoon using google scholar. It's just brilliant. I didn't know how many citations it's got. I can link through, follow all the references, I can get these just amazing. And then programming, you know, stack overflow. How many billions of hours of programming time has that saved. You could find your, your solution department. So yeah, I think that they've. They've actually provided sort of good good service on, on the, on the other end. One of the interesting things though that they, they did talk about with the way the data is provided for free is that it has an implication for the quality of data is companies that can can collect. So they talked about a lot of the data is collected sort of as a byproduct of entertainment and they try and encourage people to know, tag you photographs isn't hugely important, right. Having a library of photographs that are tagged and you know what they are all is just unbelievably important thing, but it's hard to. People don't have to tag their friends in the photographs that they're going to share with their friends because everyone knows who everyone is. Right. And, and because they're not explicitly paying for that, the, there's a limit to the quality of the data they can get and they need to sort of the, you know, the providers need to work around that set of limitations were. So it was an interesting observation.

Speaker 2: 25:49

[inaudible] and data as exhaust from a commercial process is a, is a common one. And I, my, my gut reaction to that and I see the argument probably a lot, which I don't totally, I don't really like it because or I don't find it that compelling because I think that a lot of times data isn't actually all that valuable. And I think that there's a, there's a lot of excitement and enthusiasm around data and having data and using it for something, but this is something that really resonates with the SSD or really resonates with me as an actuary where you can get a lot of data but it can be dirty and ways that you don't understand and that can really contaminate what your conclusions you are. You are drawing

Speaker 4: 26:32

selection bias in particular. But I saw they. Did you know the, their thesis to some degree was that I'm. So they drew

Speaker 5: 26:44

a parallel between the statistical value of data and this sort of machine learning on the statistical side, you know, typically what you're trying to do is you're trying to come up with an estimate of some parameter and as your volume volume of data increases, the accuracy of your estimates will have decreases with the square root, square root rule for the standard deviation. And you pretty quickly get to a, a estimate that serviceable and good enough for what you're going to use it for. And so the, the sort of first quantum of data is very, very useful. But the marginal utility of the data falls off quite quickly and they drew the parallel. When you go to the sort of machine learning world, um, a couple of different things. There's been a sort of revolution in how machine learning works. So if you go back maybe 10 years and you think back to sort of the deep blue, a chess program and made prior to that machine learning works by having a few highly qualified, highly trained individuals sort of put a lot of thought into the like basically the scoring function. So for chess, how do I score out the quality of this position of the pieces? Right? And they would tinker with that and that, that would be the approach they try to do. So it was relatively few people providing very complex input. The contrast to today where we've gone to, and this is the sort of Alphago Alpha zero model, is we essentially get a massive amount, a very low quality data we get, we get all of these billions of photographs that are tagged with the face or the famous, is that a Muffin or a show? Our image. And we let the machine learn from that so we don't try and tell the machine what to do. We let the machine learn by itself and I think everybody's been surprised over the last sort of five to 10 years that we've somehow just reached a critical point for the volume of data and that now all of a sudden these algorithms are really starting to work way better than anyone would have expected. I think the guy who invented, they had the quote right from the guy who invented a neural networks and you said 10 years from now, I don't think anyone will be using your networks and whatnot. We're all using that works all the time. They weren't. They weren't brilliantly well, but you need this massive, massive amount of data to make them make them harm. And then they talked about, so what is the sort of marginal utility of data in that world? And that if they had a couple of interesting examples. One was if you're doing voice recognition software, even image recognition, but basically voice recognition, I'm 99 percent accuracy is pretty much useless, right? I'm not going to use a a dictation machine. There's only 99 percent accurate. I got to correct how many words? 90 nine point nine is getting pretty good, but that's a thousand, but that's one area. A page. I got to read it very carefully. You need super, super high accuracy, but if you can get super, super high accuracy, the tool becomes super, super valuable.

Speaker 2: 29:42

Yeah, so now you're not in a situation of decreasing marginal utility of information. It along as kind of mostly mostly useless and then all of a sudden you hit a billion images and the thing explodes. It's unbelievably useful. He can identify everything. I thought that was a fascinating kind of way of looking at this development. It makes me think if so for this, for this podcast project of mine, I use Google's Api to transcribe them and I can them a Wav file and I have to chunk it up and I had to write this little script to do it and it costs me about a dollar, a podcast transcribed and the transcription is probably a lot. I don't know, 95 percent accurate, so it's not something to publish, right. I can't put it on the website and say, here's the transcript of the podcast. I have to go through it carefully. I use it to get quotes out from my little essays and stuff, but it's, uh, it's hard. It's hard to use. It's not a commercial grade application yet. I use voice dictation software for a while. I was very excited when I realized that I could actually talk very, very quickly and it could actually keep up with me because all computer. Yeah. But it's just not quite accurate. And there were words the insurance wants it. It never could learn. Seeding caused it all sorts of confusions. Yeah. Yeah. And I think that the, the maturation of that is. I think you're right. I like that idea, that of the, of extraordinary disproportion improvement and its utility for a relatively small increase in kind of the last little than the last mile problem, right? The 80 slash 20 rule the other way around the 2080 rule. Right. That's the part that's really valuable. And then, you know, one of the points they made in the book in terms of sort of commercializing the data relationship was I need a massive pool of data, but each individual's contribution to that data is worthless. Facebook does not care if, if, if I opt out of providing them data because they've got one and a half billion other people to work with and the chances we could get one and a half billion people all together to, you know, data union that's not gonna happen. So again, sort of speaks against the commercialization of the data provision and I think that it, it touches on, maybe we could transition to another one of the books we wanted to talk about. Um, I don't know if we're following the plan that we laid out earlier, but it's, it really is drawing in my mind this picture of you need to have a really big company to have a really big data. And, and the amount of new information that is required to train these algorithms properly is colossal. And so you're, you're almost requiring google and facebook be huge. Right? So the returns to scale of an organization that can control as much information or much larger than they were say 25 years ago. So. Okay. I a couple of things there. One is I think Google and so you're kind of talking. I think that there's a little bit of hindsight bias going on that I don't. I think Google and facebook are today in a position that is way way better than they would ever have dreamt possible 10 years ago. They may have dreamt that they would have the volume of data that they have, but I'm not sure they would have been bold

Speaker 4: 32:40

enough to dream that would be as potentially useful as it seems like it's, it's being. So I think they had the model of what is it, revenue first data users first revenue later, right? The Url, right. You're right, your uses revenue later. Yeah, I think they've been in a way they've just been lucky that they, you know, we, we couldn't have necessarily predicted it was going to work out. And then on a sort of. You're talking about large companies necessarily maybe for insurance and whatever. I always like to think about, okay, well, so how does this apply like this? Get back to actuaries? Yeah, let's do that. How does this apply to the insurance industry? And uh, I always, I believe that the, a lot of these big data things that we see working incredibly well, they worked very well in a, in a sort of domain, reasonably limited domain of problems. You know, we talked about voice recognition, image recognition, you know, these types of problems. There's a clear answer, there's a definitive answer and it's an answer that's generally selected from a finite set of possibilities. It's a deterministic, more or less system. Um, I'm always going to get the same answer. The answer doesn't evolve over time and the answer doesn't depend on sort of my understanding of the context. When you think about insurance, insurance just doesn't work like that. We insurance operates, I like to call it in the twilight zone, right? We, if something is clear cut, you generally don't need insurance for it. You manage it, you manage the risk away or you just accepted. Insurance is always going to be dealing with uncertainty. Ambiguity. Um, it, it's never going to be clear cut the, the muses, me, the oldest.com. People ever like, oh, we want to make the claim process could super objective. Yeah. Okay. So Ethan risk can do that for travel insurance. That's easy, but pretty much anything else, any claim you have, it's not going to be clear cut. And then it's not a map. It's not floor, it's just reality. My praying my car. Well, what was the status? You know, what, how, what was the shape of my car before the accident? How many of these things were preexist you've got to, you know, when you, when you have an insurance claim, there's elements out there that will try and take advantage of that and you have to manage. You can try and screw you.

Speaker 2: 35:03

I think that. So just to reference the book that I'm referring to in my mind here at least says capitalism with a capital a book came out about a year ago or so I think. And that's about the increase in intangible assets and the observation that I think motivates a lot of it, a couple of them. One that firms are getting larger and another one that the measured amount of what they call intangible capital and they measure that in specific ways, which hopefully we'll get into in a minute is increasing. So the amount of of assets which are not physical assets if the proportion is growing, but coming back to insurance companies and I think the thing that people misunderstand the most about insurance companies is is that insurance is such a highly local thing, right? So there anytime you look at. I mean in the United States sometimes the United States is an interesting example here because on the one hand is the most extreme version of this because you have state regulation, right? On the other hand, I feel like the idea of state regulation almost masks a true underlying variability with the political, with the political map because the state regulation to me is appropriate. Not, not actually because that have been political units, but because the people who live in those states are different and they behave differently and they use insurance differently and they consider different kinds of organizations and people. Outsiders in my organization was recently a reinsurance broker. My organization was recently acquired by an insurance broker and I've been spending some time speaking to insurance broking people and I'm a reinsurance actually from other space basically coming down and talking to insurance brokers. And the thing that amazed me is how personal and regional their relationships are that, that, that actually deliver the insurance product. So that isn't, is I think a pretty powerful argument against standardization and data and, and actually, so travelers is very large, but travelers is probably actually a collection of pretty small operating units. Not Pretty Small, small for travelers maybe, but that, that behavior regionally and have regional relationships,

Speaker 5: 36:57

right in insurance is a. When people talk about insurances, his personal business, um, there's a reason for that, right? And that is with most products, I sell you the product. I don't care what happens in like, you know, as long as the cash checked, check cashed. I don't care with insurance, I sell you the product, I absolutely can. How you're going to act now. And we all know moral has it, you're probably going to act differently as a result of having the insurance. So your, your character and what I know about that math is. And, and that's why insurance is a, is a, a personal business. It's all about knowing, knowing the customer,

Speaker 2: 37:31

if there's a. So there's a famous court ruling called Paul versus Virginia. Do you remember this from when you were studying for the[inaudible]

Speaker 5: 37:38

exactly. Remember the name? Don't question me. Okay. So this, this is,

Speaker 2: 37:41

this is in the original reading, my original reading out of it where you have to memorize a whole bunch of facts about this thing. It's kind of the original sin for state regulation where, uh, I don't know that the, that the authors intended to be negative, but maybe they didn't. Maybe it isn't negative. Maybe that was just my reading of it where they. Because this Guy Paul, I forgot his first. Samuel. Paul, I think was living in New York, but he went, I think he was selling insurance policies in Virginia and, and was trying there some dispute and the question of whether it be regulated in Virginia and in resolution resolve in Virginia or New York. And the court ruling went, the Supreme Court, Supreme Court ruling was insurance is a contract as to that that is delivered locally. So there's this pretty powerful, um, assessment by the legal authority there that insurance has regional. What amazes me about that, because at first I thought, I think that maybe anybody who studies something imagines they know more than they do and you're thinking, well, should be federal regulation because insurance is, you know, it's just what we call it an insurance everywhere. It's the same product everywhere and it shouldn't be regulated federally. And I've come around to that. I, I don't, I don't agree with that anymore. I think that there are local regulation of insurance is actually pretty appropriate and I think that might be a contrarian view, but I think that the supreme court got that one right and they tried to unwind it. Maybe 150 years later in the weekend, the mccarran Ferguson Act or something like that in the twenties. Um, or was it, you know, it was. So there's, so there's an underwriting, I mean it's been a few years since I took the exam, so I don't remember these two carefully but to clearly, but they unwound and rewound it again and reinstituted state regulation. So we've gone back and forth on this United States. But I think that unbalanced, it's being regulated locally appears to be an efficient because you're duplicating services, right? Travelers again, for example, is regulated, but every single state, doesn't that seem ridiculous? But

Speaker 5: 39:22

maybe they moved. That's right. Maybe it should be. What do you think? I think that's a very American view of things. Sure. Um, it always amuses me that. So America is, you know, the free market, right, where you run everything. Yeah. Yeah. Competition is both. It seems to work everywhere except in insurance. And you know, we allow competition in markets where there's what, three auto manufacturers and five gas companies and for airlines or what have you, there are 3000 property casualty insurance companies. Yeah, that actually is a competitive market, right? You can go out and get probably 10 plus realistic quotes on every single policy you want to write. It's absolutely a competitive market and yet it's regulated to death. Let's take a broader view than just, you know, a state versus the let's look internationally, which no one ever does enough look internationally in my opinion, if you can learn a lot from seeing how things work in other countries. Some in the UK that has essentially no regulation for auto rates, it is legal in the UK to run a one 800 number and offer different quilts to every other color. You can gauge price elasticity. Unbelievable does the market for and what is the result of this. The result of this is that the insurance companies make very little money, so we should wonder, you know what people complain about regulation, there's a market that genuinely is competitive. I think the last time it ran under 100 combined was, it was a long, long time ago and I mean there's, there's a slight downside to it, which is that they tend, they make their money. Then on selling maybe products, you know, inappropriate, there's been a number of inappropriate selling settlements in the UK. They sold a credit card to credit insurance knowingly to people who wouldn't be able to claim from it because they were self insured for example. So self employed. Um, so it, you know, you need some sort of market conduct to regulation. But I would question why, why do we have any rate regulation?

Speaker 2: 41:13

Oh yeah. I mean, I don't, I don't necessarily dispute that. Honestly. I think that I agree with you that less rate like your regulation is benefiting consumer. I think that one thing about the UK example that is I think not quite appropriate is that the UK is a pretty small country geographically and culturally it'll certainly be distinct, but it's more like one or two American States, right? The United States is big and there aren't really any other countries that are like that. Maybe you have Canada, that's kind of that way. You have Russia, you have China. Um, I don't know that there are too many other examples of developed economies that are, that diverse internally. You have actually that much difference between different regions and they're far apart. And so I think that finding right analog is internationally is actually a bit harder than, than maybe it looks. Do you think?

Speaker 4: 41:57

Uh, I don't, I'm not sure I would agree with that because. Okay. So even if you, I argued that the UK is just one state. It still doesn't have any regulation.

Speaker 2: 42:04

No. Well pricing. I mean it has some relation. There will be consumer protection and the reason solvency regulation. Yeah, I mean auto rate regulation, again, I'll definitely grant the point that that's ridiculous for the most part, but that's not what all regulation is. I think that the interesting thing about insurance regulation to me is that it is solvency regulation and you know, we were working a little bit with a client of ours who is filing very high rates for auto liability coverage and the regular, he said the regular, he was worried that they would let him file high enough rates and you said you can fight whatever rates you want, you can do is you can't increase them too fast once you already had them in, but you can start with however you want to start because they're worried about how backlash from consumers about changes. Um, once you've already established something and people being disrupted in their own financial planning for their own lives, whatever reason. Um, so I think that it's a, it's a, the thing that makes insurance regulation very different from other regulation is that it's not actually entirely in the consumer's court.

Speaker 4: 43:03

Well, it, it's, it's also, um, the rate regulation. There's full regulation, there's underwriting regulation, it's a mandatory product. And then the solvency regulation. Yeah. So the, the, even the regulator doesn't have, you know, the, the regulator wants low rates, insolvent companies. Yeah. They have to make a decision that you don't have, you can't have both. I do think that, um, you know, regulation around allowable rating variables is very important and that we should, you know, what the question that, uh, about price discrimination and using variables that are not directly related to loss but are related to other behavioral characteristics. That is something that in my mind should be regulated quite carefully.

Speaker 2: 43:48

[inaudible]. Uh, so maybe we can draw this back into the, this, this book at this point which is capitalism with capital. And one of the ideas that I just love thinking about is the idea that intangible assets have increased as a share of total assets and insurance companies. I think maybe that's true. I mean, what do you think about, let's say an insurance company is growing and scaling across a geographic across the United States that has to file lots of lots more filings. And you know, is that, is that insurance companies maybe are not a great example because they're entirely made of intangible capital right there. They're entirely made of, of uh, well, I would argue they're actually intelligent. Yeah. They might have

Speaker 4: 44:26

tangible capital, right? The cash, the cash is the only capital. So I think that this is an interesting

Speaker 5: 44:32

discussion. Yeah. And it comes down to sort of, you know, when you set up an accounting standard, which has kind of really what we're talking about here, what is it, the objective of the accounting standard. And therefore, what should I count as an asset, a rethink about an insurance company's statutory accounting is around ensuring that claims are going to be paid. And so it can only count assets that can turn into cash essentially. Any other asset, if that asset is not going to turn to cash, it's useless. It shouldn't be on statutory balance sheet. Hey, y'all have perhaps that maybe there's this very real. To me, I worked for a kemper insurance, which is a large$4,000,000,000 growth company in 2000. One, uh, that went into runoff know, literally ran out of run out of money and so I had spent a lot of time building a model for them of their balance sheet and it was really brought home to me. Every single item on the asset side or the liability side at some point has to turn to cash in order to be useful. And if it's not turned into cash from a sovereignty perspective, worthless. Now there are other metrics that we might want to use a, some sort of quote unquote value as a going concern perspective. Absolutely. Their intangible assets are going to be part of what creates the value. And, and you sort of actually get an amusing paradox, right? That a company can have a massive intangible value which isn't on its balance sheet, but it gets bought. And then the company that acquires it has the book, the good world, which basically is that, you know, you sort of make it real at that point. And there's a. It seems like the accountants need to sort of work through that, that that fact that if I've, if I've purchased it, it's on the balance sheet and if I haven't, it's not on the balance sheet that, that that's a little problematic. And then I think, you know, stepping back to it even broader perspective, we get to the question of measuring GDP and measuring, you know, how the country is doing, where, you know, go back to look at Google, think of all of the industries that Google has basically taken out. The yellow pages will be one of my favorite examples, right? How many people did the yellow pages employ between selling the ad space, printing it, distribution, and let's face it, the, a terrible solution to the problem of actually looking anything up. Right? I want to get my smartphone out, I want to google it or I get reviews. Is it open now is a busy now you know, how do I get there is a better solution. But from a GDP perspective, it looks like we've gone backwards because there's a whole lot of people who are not employed. There's a whole lot of physical stuff that's going on, it's been replaced by electronic stuff that doesn't really make it into the GDP statistics, but who would want to go back to the world? You know, I would not want to give up at this. We are clearly better off. We just not measuring it. Right? Yeah. No, that's right. I agree with that. But Let, let, let, let me go back in and try and defend my, my view again in insurance or intangible assets. Right. So that where I was getting out there, it was

Speaker 2: 47:29

that insurers are intermediaries, financial intermediaries, right. In the sense that they don't do anything other than take the money from consumers and give it right back to them. We know premium comes in and claims goes back out. Right? So there's, there's, there's obviously a net asset component is that surplus, um, of course. But most of the money is certainly a liability insurer that sitting in there on their balance sheet. Most of their, most of what they are made of from a financial standpoint is, is a obligation is to other people and money. They're holding for a little while the end, when I talk about most of what insurance companies are, I suppose what I mean there is the, the, the enterprise value, the value over the cash in the bank account because everybody can have cash in a bank account. What does it sit in an insurance company is the reason for that is that insurance companies do something with it and they're not. What they're doing is merely shuffling money around and organizing things and talking to people. And those are all intangible skills and, and, and the value that they're bringing is one of certainty. That's kind of mushy concept, right? Uh, as opposed to, you know, the, the counterfactual that I think that the, um, the example of what a, the really hardcore county counting argument against an insurance company would be if we just sort of abstracted a way to the national balance sheet, right? The United States. And here we can maybe touch on another paper that we're talking about. This arrow paper, Ken Arrow paper, which will me to ask you to touch on is when you kind of zoom way out, you're looking at, you know, you have risks and you have an, you have capital and you kind of want to match when a risky situation occurs, you know, the capital will flow to it and we'll pay for pay, pay off the losses that happened there, that, that just ignores insurance companies. And so what our insurance company is doing, they're doing something there because they exist. And I say that that value is entirely intangible. What do you think

Speaker 5: 49:13

the whole webcast of discussions that we can have got another 20 minutes. So for this, let's talk about firstly, uh, so insurance clearly do have a intangible value because you see it every time they transact, right? They turn on their trading at, so the one and a half to two times book and then when they transact they transactive control premium of that. So they've got an intangible value, which is generally, you know, let's call it somewhere between one and one and a half times their sort of stated booked value. Yeah, I think so. I completely agree with the insurers are intermediaries. I think you downplayed a little bit about, you know, what do they do? Sure. So the people in finance have to talk about, there's no free lunch. There actually is a free lunch, free lunches, diversification. Diversification is the ultimate free lunch. It's an absolute miracle. And Insurance, what insurance do is they give people access to the miracle of diversification. Yeah, but that's quite tricky, right, because what I I need to as the management at the insurance company insure essentially equity between my policyholders that somebody who's risk I allow onto the balance sheet needs to be paying for the risk that, that brings to the balance sheet so that my other insurance are not disadvantage relative to this new person who comes along. So I think of insurance, that sort of principle functions that they serve are. So going back to regulation, right? There's a, there's a whole lot of things where they have to provide a service that checks various regulatory boxes and people always say, oh, you know, regulates terribly. She just do away with. I know you shouldn't have rate regulation, but I do one inch, I do want drivers to have to have insurance because if they called me a taught, I want someone I can go and recover from. Right. And, and so insurers are enabling that, a surface which, which is, which is very important and then they're managing sort of the fair access between insureds to, to, to this pool. Um, so, so I think that's a, that that's very important. The, the Arrow paper that you mentioned. So it's, I think the theory of risk bearing small and great risks is the name of it. Um, every is I think only a 10 page paper. And um, so for those of you who maybe haven't heard of him, Ken Arrow was a Nobel prize winning economist, extremely famous hymn. And a Dobro essentially proved that the, a competitive market equilibrium is a parade, optimal allocation of resources. So the whole, whenever anyone talks about to a competitive market is the way to go. You're, you're sort of going back to Arrow and arrows. Paper has a number of implications for how risk should be shared. Okay. And what his paper says is, risk will end up being shared through a large number of bilateral contracts. And the net effect of it is that all risk will be thrown into a pool and then everyone will quote, share the pool. And if you do that, you obviously get rid of all the diversifiable risk has gone away because it's all gone into the pool and you're just left with the systematic risk. What's the size of the pool is the only risk variable that's laugh you only thing. And then informed pricing at that point. And then, you know, people's sort of the, the share that I'd kept back is inversely proportional to my risk aversion. So if I'm more risk averse, I'm prepared to accept a smaller share back in exchange for having gotten rid of all, all of the risks rather than the more risk or less risk averse. People sort of take greater volatility. So he, he, uh, and, and this was arrows theory that, you know, it's implemented through these things called Arrow Deborah securities, which you may have heard of, which are a security that pays$1 in one particular state of the world. Okay. And that's the sort of fundamental insurance contract, if you will, it prices out all the states of the world. And then from that I can press any security because any security is just a combination of these fundamental Arrow Deborah securities. So this is a wonderful theory and it works kind of nicely accepted as it looks nice in theory I should say. Right? And so, so Ken Arrow's sort of observation around this. Well, the implications of this is that there should be a lot of risk sharing going on. Um, and we also noticed that there's no place for an insurance firm within this Arrow debreaux world, and yet we see them everywhere and yet we see them everywhere. So I think it's a fascinating question of sort of why are there insurance companies at all? In theory, nevermind, you know, no mining practice. The theory gives us a nice, you know, causes theory of, of um, contracting carsten specialization and what have you for most industries, we sort of understand why it makes sense for it to be a firm, but fundamentally why would I exchange in the Arab world? I, I'm allowed to contract and my contracts never default because they're always expressed in relative terms. They're always expressive. I will give you 10 percent of what I have, not that I will give you$10 and so I can't, I mean there's, there's a um, enforcement issue potentially, right? Actually getting me to give you the 10 percent I've promised, but, but I always would have the capability to pay what my contract obligates me to pay. Why do we move from that world to a world where I contracts through an insurance company that then can default. It doesn't, it doesn't make any sense. There's reasons for it, but it's a. that's an interesting question to sort of ponder, which is really where he gets in his paper.

Speaker 2: 54:47

Yeah. My, my pet theory for what insurance companies do is the word we use is underwriting. I think is most of what they do and my, the way I like to put that his insurance company stopped customers from screwing insurance companies, which means customers from screwing each other in the pool. Yeah. It's an evaluation and moral hazard and saying, is this guy a bad guy or good guy kind of thing. And if they're, if they're good then great, we'll get you some kind of price and if they're bad then there's no price for, for a bad risk. And so I think that there's a capacity for humans to deceive each other and try and try and make a, make a play for the pool. You're saying, well I have a claim on this pool in the event of an event of some outcome happening to me or somebody else and maybe if I just kinda like lied a little bit about what was happening and then I might be able to get access to that pool for myself. And so the moral hazard problem is, is everything.

Speaker 5: 55:42

Right? And so you're a smart guy, Kansas guy. You both came to the same one. I did read a paper first. It's moral hazard, it's adverse selection and then it's also this whole complexity and ambiguity and contracting right there. The writing the contract is, is difficult and so you give up a little bit in the quality of the risk transfer you get in order to have the practical implementation through an insurance company.

Speaker 2: 56:05

So you mentioned in passing there a second ago, which we can maybe touch on now, the, uh, Ronald coase is paper, which is another one on our little reading list here today and put the link up to all these things and that was an interesting paper and that one in the Nobel prize I think in a comics which was amazing really for another fairly short and easily readable paper written in 1937 still resonates today. And his question is why do we have firms at all in any industry? Because if the market mechanism allocates things sufficiently to your point there, I think, I don't know if Arrow's theorem had been a, it's afterwards, but we'll just take it for granted that the economic, uh, the economic allocation of resources as efficient because of market transactions, because you have to price system that governs the value and your own coast says, well, why the heck did we have companies then? Because companies are not markets, they are command and control organizations where you have a ceo telling somebody else what to do and they do it and there's no price transaction between them. Why Steve? Why don't give us the answer?

Speaker 5: 57:05

So the, the argument against command and control is sort of amusing. Everyone looks at the government and Oh, the government is necessarily going to be inefficient and stupid at doing things. And it sort of amusing to me, or ironic maybe is a better word that the sort of theory that, that sealed that came about just after the Second World War. That was one entirely on a command and control basis. Yeah. There was the military through the original corporation, the, the, um, government. I think in a way gets a bad rap that regulators, what's the difference between someone working in government and someone working in a, from someone working in a firm can do what makes sense within. They have to do something that's illegal obviously, but they've got a lot of latitude, a lot of discretion about how they act. Someone who works for the government is typically implementing a lot and it can be the dumbest Laura in the land, but they don't have any discretion about those instruction. And so they, they're there. I feel sorry for them because they're there. They're so far behind, you know, that puts you so far behind doing the quote unquote right thing. If you can't get the law to change your hand tight and all of a sudden you're stupid for implementing it. Well No, it's a stupid law. And the legislature stupid because they haven't changed it. You're just doing your job. So I think, yeah, I just thought it doesn't answer your question, but. So the answer your question is, is um, talks about um, contracting costs, specialization, economy of scale. If we went to a world where we didn't have ums and every service I needed, I tried to essentially outsource would outsource everything. We would just be spending our entire lives. We've all worked on statements of work and contract work fulfilled and it's a nice we don't want to be doing that. It doesn't allow us to focus on what we do well and so naturally you begin to kind of bring things within the 10 and you say, well, okay, this is a core function. This is a core functioning. And really his question is how big, how far out does that set of core functions expand? Do I do, for example, what do I do my own HR administration? Or do I outsource that? Well know I'm a software for them or whatever it is. I have no expertise in that. That probably is something that I am so. But other things that are just central, I certainly want to keep within the 10. So I think you know the. There's a range of arguments. I think also you could look upon the firm as a sort of collection of secret recipes for making things that if you, if you, if you read about an insurance, is a great line, if you will. We understand how insurance works, but there's various specialty lines of we're not so sure about. Sure. You do a little bit of googling. You generally run into a bit of a brick wall, you know, maybe there'll be one actuarial paper on this is roughly speaking how I price this, but no one's really going to tell you how they do it. You need to talk to someone who actually does do it. Who is within the sort of film that hasn't. I think it's that aggregation and sort of profiting from private information is a, is a core characteristic of water from is it goes back to what we were talking about, the intangible assets and know what have you from, from the beginning. So I think that's also an important part of what firms provide,

Speaker 2: 1:00:31

you know, coast makes a few comments in the paper. Uh, implications about that, what kinds of firms will be larger and what kinds of firms will be smaller. And he really does. It really does kind of foresee a lot of the argument is that wind up coming in out of capital and widow capital where affirms that, that uh, you know, have, have lower transaction to find a way to have lower transaction costs. Will, will be, or industries that have lower transaction costs will wind up having more firms or sorry. Yeah. More. Yeah. More firms, right? Because you have um, you have more decentralization of the services and firms with high industries, high transaction costs, you know, think like, you know, I think that industries where we were the biggest firms, right. You have the energy firms are the biggest ones. Insurance companies are pretty big actually, which is interesting. You know, maybe I'll pause there and ask you what you think about this. Why are there very large insurance companies? It, it feels to me like a lot of the forces are pushing them in the opposite direction. Well,

Speaker 5: 1:01:26

okay. So I'm not sure I completely agree that in the US there's moderately large insurance companies make. It just goes back to your point about European shouldn't think of European countries is equivalent to the US, but you go to Europe, a viva axar generally allianz, there and Zurich I guess right there are huge in Japan. Another one full of big ones. Well they will merge together in Japan. They started off with more companies and they. But in the US we've got a reasonable number of sort of the top 20 companies. I guess you're right, it's a little top heavy. I think you get um, personal lines is, has greater economies of scale than commercial lines and commercial lines. You have specialized underwriting which doesn't really scale, but you need a lot of capital. But the, you can bring in in different ways. You can reinsurance, you know, you can, you can off balance sheet the capital. On the personal line side you sort of need to build your, you've got a massive number of homogeneous risk units that you can process and there's going to be efficiencies to doing that together. Efficiencies of pricing, just efficiencies in processing all the intangible assets, the the brand that the companies make. So I think personal lines sort of naturally gets to be bigger. Commercial Lines need a big balance sheet, but it doesn't necessarily need to be a big company because because of, because of going back to insurance, every, every risk is unique and I've got to understand each individual risk. A given underwriter can only understand so many risks. I think that actually is an interesting observation about why certain large covers don't exist in the marketplace. I can go out today a a Japanese company or an Australian company can go out and buy property today.$10,000,000,000 of Canada. It used to be seven and a half billion dollars are the biggest programs I suspect is 10 billion today. But if I'm Pfizer or some manufacturing company and I want$10,000,000,000 of liability cover, I can't get that. Well obviously it's not the size of the risk because you know it's provided for care and let's face it for care, you're going to write the check out in two weeks after the event. I mean, the money flowing into Japan after the hooker was vacant. That was real. I think that the reason that there's sort of two reasons why those don't covers don't exist. One is that the return to underwriting skill, which is, which is our scarce commodity, right? The capital, I can get this capital everywhere, but the underwriting skill to go through the books and to, you know, Poor Pars, all the public information about a big manufacturer and decide, okay, this is the price I'm willing to put on them. The problem with doing that is I've now done precisely one manufacturer and it doesn't, it's not going to be the same for the next year. They're all going to be kind of different and then I get to a price, what am I going to think? I'm going to think, well, if they buy at that price, they obviously know something that I've missed and they'd been complicated or you know, they know more about their risk than I do. So I think that's, you know, I know it was always an issue. We always trying to figure out how, how can we upsize liability in particular? Why isn't more liability purchase? I think it goes back to the information asymmetries, but it goes back to the. It's not a good return on scarce underwriter resource and skill to do that because you're fundamentally doing it. One insured at a time.

Speaker 2: 1:04:51

Most of are emergent aspect to this. I think where I know some folks that run insurance companies who think very strongly that the, the. There's a. This is a game between the, the call it, the legal judiciary system and the insurance industry where if you wrote a smaller limit, you just have a smaller loss where the whole system's organized too. You're a calibrated to the limit that's available and so the reason why you don't offer higher limits this because they would just take them and and and the one that they want to achieve their economic outcome from a particular trial or a particular event. Liability events. So let's define that. Squishy, squishy, and that outcome is insurance matters, but if they want to screw the company, they're going to go at the top of the insurance. No matter how hard the insurance is, right, if there's not any kind of basis and, and, and our own long actual retailers damages and stuff I guess, but it's only goes in and it's again a big mess. And so that explanation, which is kind of a cynical one for why there aren't very good limits is because they don't need bigger limits. You actually could use smaller limits and you wouldn't change anything. It was just, you'd just sort of less of a transfer of wealth during an event. What do you think that. Okay,

Speaker 5: 1:05:57

I, I think I could see that for larger losses, I think there's, I think we've stuck at a million dollar limit for, for way too long. How can it still be the yellow? Okay. So maybe we should sort of back up to, in my mind, I divided insurance into this property. Then there's what I call it sort of broken and injured and killed lines of business, which they felt sort a little more real to me. And then there was this sort of slighted and insulted lines. We know your Dno, Eli, that has never quite so comfortable with those lines, quite frankly the, but that, you know, when people have been killed and injured and maimed, you know, a million dollars is not what it used to be, that those people should be buying$10,000,000. You, you, you, you know, if you, I'm sure you've seen trucking large losses, right? I mean that horrific probably trucking firms shit, but a$25 million so of limit and it's not. I don't think that falls into your cynical view of things. I would agree probably on some of those other, you know, on, on the slide didn't insulted lines know I could see. But here's an argument

Speaker 2: 1:07:00

is that in some ways where the region in which you are insuring affects how much insurance limit you have to buy. Right? And, and even with the United States, so let's just, we're, we're kind of holding constant the medical system for the most part, and let's say you're in south Texas or South Florida or you're in Wisconsin, right? Place where you spend some time, very different legal environments, right? Probably buy it. They probably need to buy different amounts of insurance limits. To me that's an argument for local culture, really, really attacking the insurance limit, maybe call it illegal culture, the lawyers on the billboards which you see when you drive around the south and, and those guys are our predators of insurance limits and I think that if you shrunk their limits, you probably would shrink the total costs of the. But you know, we have this tradition where it's hard to break, I think this competitive equilibrium whereby you're used to having it so you have to have it and if you don't buy it, you're kind of scared. But what if it goes up the top and insurance company would probably say, well, don't worry about it. It won't go at the top. And they're like, well, yeah, right.

Speaker 5: 1:07:57

I, I, yeah, I think I agree with that. Um, I think the, you know, the poster child example is Michigan. You have, where they have the unlimited pip. Yeah. And I'm at least a basket case a couple of years ago. The NCCA, which is the Michigan Catastrophic Claims Association, they reinsure individual pip claims. It used to be above$250,000 per person. Now it's taken 500,000 per person until just a couple of years ago. The MTCA had higher undiscounted lost reserves than Aig. Mike Guide just for reinsuring pip in one state for personal auto access to, to talking about broke. They had$65,000,000,000 I think of, of a undiscounted loss reserves. Wow. Amazing.

Speaker 2: 1:08:40

Uh, so we're running out of time. Steve. Um, I wonder if you can close maybe if you could talk a bit about. I mean you're, so you're working on, in the world of ideas now permanently, which is pretty cool. I wonder if you maybe talk a little bit about sort of things you're thinking about working on right now.

Speaker 5: 1:08:52

So I'm working on a risk measures and understanding risk measures. I actually worked with the, uh, John Major done mango and Jessie Nicholson four. We did three sessions at the spring meeting that was just held in Boston on a distortion risk measures, which sounds okay. What is that? And what is distortion? Risk measure is essentially saying to measure risk. What I should do is I should think of all my outcomes and I should just make the bad ones more likely. Okay. Okay. And so that increases my main and then the increase in mean that gives me a sort of risk margin. Um, there was a lot of interesting stuff came out of that is maybe it would be more for round two, but one of the key ones was, was thinking about the definition of an event. What is an event? And this sort of really goes back to what we were talking about with, with Arrow and we in finance where people think of events is as tangible things. So it's a, you know, an insurance speed cap models give us tangible events where I didn't say cat five hurricane makes landfall in Miami, Blah Blah Blah. And I can think about the losses that come out of that. And if you look at finance, you go back to your Arrow debreaux securities. They cover tangible events in that. In that way, I'm an insurance, most of the time we don't know what our event is to that detail as we, you know, we've already discussed and we've identified the event with the loss outcome and we insurance outcome some natural oils so that the ground up what maybe from a camp model, you're called the ground up view, right? And then you're going to layer, this is going to be a retention, a little bit heavier, but we've identified it with that. Um, and that closes the, your problems. Um, it, it partly, it gets us to this question that I know actually love about, oh, should we think of value at risk or tail diverse? And it reconciles that problem because the value at risk view is I'm thinking of covering one event is the event that causes a loss of a thousand dollars or whatever. But we never write insurance policies like that. We always writing insurance policies that cover a loss of a thousand or more. Right? So even if you do your little thin layer, you know, people love to think about little thin layer things up. So if I write a policy that's a dollar x, a thousand, it will respond to a loss of a thousand or more. Yep. So it's a tail V, a t bar, kind of a view of things. Whereas a, if we go to the sort of strict finance view, finance would have a policy that would only cover the event of the last being a thousand, so it would cover just that one event and then if the loss was a thousand and one, it wouldn't cover anything at all. Now that has sort of very profound implications for things that in finance you would think would work with in one way, but then insurance, they work in a, in a different way. And we need to make sure as we're translating from sort of pure finance into insurance, that we're sensitive to that difference in the definition of an event and that we're always thinking our event are essentially that always tail events where it came up to close out with where it came up in this, this, uh, counseling was okay. So if I make my bad outcomes more likely, I have to therefore make my last, my good outcomes, less likely. So now if I'm sorry, you imagine, you know, you've got a table and loss of a 100, 200, 10,000 or whatever. They've all got probabilities. Let's say they start off as equally likely and then I adjust them, right? So I increased the probabilities of my bad outcomes and decreased the probability. So my good outcomes, the reason for this is to just represent risk aversion, risk aversion, and you're pricing things with, uh, with respect to these adjusted probability. So John Major sends out this email, I just don't remember like Thursday evening. And he said, well, it seems like we've got a bit of a problem here because what if we take a policy that is just covering the good outcomes? The probabilities are lower than the objective probabilities because probably is add the one, right? If I make something higher, I have to make something low. Um, that seems like I'm going to end up with an insurance policy that I'm going to send them all under cost and we don't like, you know, obviously that doesn't seem like it makes sense. Um, so then he called this the big honking problem, how do we solve this big hunk? And, and I sort of basically didn't. I found this fascinating, right? If you're completely consumed me for about the next sort of 48 hours. And then when I finally realized, well, it, it is a problem if you ever sold a product that only covered the small losses, but you never do, you always sell a product that covers the spouses and above and because it covers small and above, the increase in likelihood of the bad event outweighs the sort of discount you offer on the small events. And any realistic insurance product that you would sell will always be priced above objective cost. But you could have an example that comes up. The one example of a product we sell that I'm only covers sort of certain outcomes is term life insurance, right? If you think of term life insurance, it saying I cover you if you die within this band, so it's sort of like the, the last equivalent for that would be I'll cover your loss if it's between$10,000,$20,000, but if it's more than$20,000, I'm not covering it. We don't sell products like that, but we've turned life and the reason we don't obviously it will be all sorts of hazards, right? You got lots of 21,000 miles and I don't know. No, you know, persuade the guy to make it only 19 and then you got to cover it. Well, with term life you, you're pretty attached to being alive so you're not gonna you know, you're not going to cheat on that one. So it's okay to have that product as a, as a, an insurance product. But that's really the only example I could come up with. Just franchise deductible. But did the franchise deductible part of a franchise deductible also has this right at it only applies to small losses and then it drops to zero. So the sort of solution to the big honking problem was it is. It's a problem in theory, it's not a problem in practice because we don't sell the products for which it would be a problem. Yeah. Neat. Well, we'll stop it. They were at a time. Steve, my guest today is Steve Miller Hall. Steve, thank you very much for joining me. Thanks for having me.