Stewart: Welcome to another edition of the Insurance AUM Journal podcast. My name is Stewart Foley. I’ll be your host standing with you at the corner of insurance and asset management, with none other than the queen of risk, Karyn Williams. Welcome Karyn.
Karyn: Stewart, thanks for having me today. We have a lot to talk about.
Stewart: We do. And here’s the thing, Karyn did not give herself the name the queen of risk, I did. And I just think what you’re going to find out today is that Karyn’s firm and Karyn’s approach to risk and risk management is unlike anything that has really been done prior. Right? So let me just set the stage a little bit. We both know, and I think anybody listening to this podcast knows that traditional mean-variance optimization for risk management doesn’t work, right? It doesn’t work when it’s supposed to help you, which is market dislocation. We’ve all learned about, “Oh, diversification, it’s supposed to help you,” blah, blah, blah. And all of that portfolio theory that was built decades ago, the people who built it knew it didn’t work, if the assumptions changed any. Now you have a PhD in finance. Let’s talk about why you think that is all wrong.
Karyn: Thank you, Stewart. Well, that’s a great lead in. There’s, as I said, a lot to unpack here in this conversation. And so maybe the thing to do is to start with what I observed in practice for a very long time. And when I started, I made the jump from the academic world into Wilshire Associates, which we practiced implementing some of the concepts that we’re now talking about. And what I had seen, time and time again, was regret, disappointment, confusion, and generally an inability for a lot of people to talk about risk, period. And so over time, and it took a lot of different perspectives in the industry for me to really understand and absorb this, was a view from governance, a view from the people involved and then a view from the systems that were used.
Karyn: If I reflect, there were challenges in all of those. It’s generally true, from a governance perspective, that people are pretty clear about what they want their investments to accomplish. They want it to pay out, they want to support various goals, they want to pay fees, they want to cover inflation. Those are actually relatively easy. In fact, I remember spending years and years getting in the last basis point of return, very clearly defined. But with respect to how do we get there, the engine, if you will, to get there, the risk, that was very, very unclear for most people. And so we’re starting with why is that? The tools long have been, as you said, being variants. And also, as you said, Harry and others who had brought similar kinds of concepts in Sharpe, think about the arbitrage pricing theory, these really are academic grounding for our understanding of risk, just as an intuition, there is a systematic risk in the marketplace.
Karyn: It’s not a prescription for how to actually build a system that would help you to make a decision, huge insight about how the world works. And so I think that having that translated exactly into a tool was compelling. So if we just take a step a little, why would people do that? It’s actually really simple. It’s simple stuff, in a way. So it’s analytically, very tractable. It allows for risk to be described by various attributes, think portfolio, think marginal risk or contribution to risk. There are some metrics you can derive that are actually really straightforward.
Karyn: You also can explain, if you will, the results of your portfolio in risk terms. So you can generate information ratios and Sharpe ratios. And so it actually simplifies the analytical world, but it’s not really very good for decision-making because people don’t use, okay, so who are people? Who is at the investor table, if you will, with this choice, risk choice, portfolio choice? For institutions, as you might know, and you think insurance institutions, you’ve got an investment committee, which may include very bright people who are investment professionals and sophisticated, and they understand. Beyond that, you could have a board where you have non-investment professionals, very bright business people, but who don’t speak the language of investments, who don’t speak the language of standard deviation and risk. And so it’s not a very familiar kind of discussion or choice to make.
Stewart: Yeah. And I think from which you speak right? You were the Chief Investment Officer of Farmers Insurance, you’ve sat on boards. The governance angle is a good one. And I know that you’ve been in the room when this has happened and I’ve been in the room when this has happened. So you get through your presentation and you’ve gone through and blah, blah, blah questions, and then there’s no questions for a minute. And then some brave soul raises their hand and asks a question that is incredibly obvious that you were talking dramatically above their head the entire time.
Stewart: Right? To which somebody on the other end of the room goes, I’m glad you asked that because I was going to ask that too, or I didn’t understand that either.
Karyn: That’s right.
Stewart: And it really speaks to your point, which is we all know intuitively what risk is in some manner. But how do you get it? How do you quantify that? How do you put risk management into an entity, like an insurance company, that’s complicated, there’s a lot of asset classes and oh, by the way, there’s these things called liabilities that is the core of their business? And so those things fit together, but you’ve moved out of, and I don’t want to take you off that line of thought where you were, and I like how you throw out the name Harry, is Harry Markowitz, the granddaddy of them, right?
Karyn: Yeah. Yeah.
Stewart: So how do you come at it as opposed to the mean-variance framework that is the most common approach?
Karyn: Well, there are a number of ways early days that I started to come at this, I guess I would say kind of bottom up. And I realized in coming at this from the bottom up was insufficient. So what is bottom up? Bottom up is first teaching the committee or committees, because I, like other insurance CIOs reported to not just one investment committee, but potentially several committees and boards. And so one of the first journeys we took with them was to understand that risk was shared among asset classes, that fundamental drivers of performance are shared. That’s powerful they had never heard that before. Their facility with asset class is actually pretty good, somewhat limited. And if you get into maybe some newer descriptions of an asset class today. But generally speaking understood that these should be working together to diversify. But once you started to talk about the fact that a growth risk factor lives in multiple asset classes, that oh, all of a sudden, we have a lot more at risk than we believed.
Karyn: And so I know that you’ve had other conversations about this, for example, the Northern Trust conversation about risk factors. And so that was the way I started to come at this. And that was when I was at Farmers, and it was fantastic. We helped to transform the investment program from one that was really roughly conservative because they couldn’t see the kinds of risks that we were talking about, like let’s put a lot in cash, let’s put a lot in reserve, but then getting clear on how much we could take and then putting it to work in factor space was really powerful for the organization. And we got everyone behind. It took some time.
Stewart: Well, I think it took some time, but at the end of the day, you were really early, too.
Karyn: We were. Yeah.
Stewart: You were super early in that factor, in that risk factor approach. I have yet to meet an insurance CEO that didn’t say we have a conservative investment philosophy.
Karyn: Yes. Yeah.
Stewart: What the hell that means, heaven only knows. But at the end of the day, what’s always been viewed as, and I’m doing the air quotes, the conservative approach, is different today with rates with the 10 year note at 90 basis points or whatever it is today. And those big allocations to high grade bonds is a very risky strategy in this kind of environment. So it’s interesting that while you had a big learning curve at Farmers, you guys were early on, I mean early, early days.
Karyn: Yes. And I have to get the attribution, right. So Farmers is an affiliation with Zurich Insurance and a lot of the original groundbreaking work was within the investment management function and team really at Zurich. And we were in a really nice position to have the freedom to do the right thing, which was not being conservative, but also not being crazy with risks that we would want to take to improve the investment program. But actually, first of all, let’s understand, how much can we take? And that’s a very different approach to investments than say, well, we want to produce X percent income, Y percent growth, and we want to have alpha.
Karyn: And so yes, you can certainly set those out as objectives, but getting clear on what is available to put to work is the real innovation here. That’s completely distinct from, and only by chance related to standard deviation. It would just be the rarest of circumstances that the amount that you could put to work actually met that standard deviation number that you would produce in a mean-variance optimization. So anyway, the bottom up work was essential, but that wasn’t enough. There’s a lot more work to do. And let me just tell you a little bit about the journey for Hightree.
Stewart: Hightree Advisors is the name of your firm, right? I just want to make sure that when you say Hightree, everybody knows what you’re talking about.
Stewart: This is the firm that you founded. You’re the founder, predominant partner. There are others, but it’s Hightree Advisors, and that’s you. So how did that come about?
Karyn: Yes. How did that come about? I’m still wondering.
Stewart: I hear you.
Karyn: You’ve always been an entrepreneur, but now I’m officially an entrepreneur. But so I knew that I wanted to bring that bottom up kind of intelligence as an independent portfolio, if you will designer, for organizations. Very much in the spirit of what I had the opportunity to build with colleagues at Farmers. And yet, I knew that I was still going to be dealing with, even if you use a factor-based approach, you could still very much end up in the mean-variance world. And I didn’t want to end up there because all of my experience had taught me that it was in the deepest drawdowns, which happen all the time, where the potential lies in the innovation, thinking about the tail, thinking about the sequence. So it’s not just that these events happen more often than a normal distribution would suggest. It’s also that there are sequences of drawdown.
Karyn: So it’s the pain and it’s the duration. And I thought, well, gosh, we can do a lot more. So what did I do? I said, all right, I’m going to approach one of the smartest people that I know. And I literally went out to Caltech and he said, “Yeah, I’d be happy to help you think through this.” I was just looking for a thinking partner, in Jaksa Cvitanic who’s out there. And I actually wrote down all of the things that I wanted a new approach to have or not have. And I wanted to be able to capture this idea that there could be multiple objectives and multiple limitations so that there would be potentially soft limits on a drawdown and hard limits on drawdown, that you might have the ability to specify a number of different targets, for example, on the plus side. So multiple, if you will, objectives with respect to the value of a portfolio at some future date. So that was kind of number one, is that why just do?
Stewart: I think that when it’s something that sounds very simple. But when you start expanding the variables in that way, the math gets orders of magnitude heavier, right?
Stewart: Potentially or no? No?
Karyn: No, ours did not. The way that we actually simplified things quite a bit. The other thing that I wanted to avoid was assuming that I knew Stewart Foley’s utility function, like to specify your utility function, there are, as you probably know, tens if not more types of utility functions that exist on the basis of which we would derive a function to optimize. So the problem I had with that is that specifying utility function means also to get from you somehow a risk aversion measure or parameter. Well, I don’t know what that is. And there are surveys, especially in wealth management, where like, okay, we’re going to do a survey today. We’re going to ask you all these different questions and we’re going to say, “Oh, well, you fall into this very conservative class of investors or very risky.” At that point in time, I don’t know that that’s any, and then you’ve got to translate that.
Stewart: Absolutely. And oh, by the way, when you tell the board of directors the equity market can be down 30% and they all nod and say, “Yes,” and then you’re down 28%, that’s a whole different conversation, right?
Karyn: That’s right.
Stewart: Depending upon the situation at hand and the way that markets are and the mood in the market and whatever, that’s not a constant, right?
Karyn: That’s right.
Stewart: I’m putting words in your mouth, I don’t mean to, I’m asking. That’s not a constant, is it?
Karyn: No, these are not constant. But at the same time, it’s not like we can, I’m not necessarily interested in forecasting and I don’t want to put people in a box.
Stewart: Right. Exactly, yeah.
Karyn: So the question is, well, where do you go from there? So let me add a few more things. I also wanted to make sure that we didn’t have to fall into this trap of thinking that the market was normal, normally distributed, that returns were normally distributed because we know they’re not.
Stewart: Yeah, clearly not. Yeah.
Karyn: And so, I said whatever we create here as a tool, as a decision support tool, has to be flexible enough to, if we made it real, that we could have whatever distribution we want. We could say that it corresponds to this return generating process, or we could say that we’re using history, but it would have to be realistic. And there were a few other things along the way. But basically, what we ended up working out was this measure we call portfolio Pi, thinking about the whole, the whole portfolio and what we want that portfolio to accomplish. And so portfolio Pi is actually kind of simple. Maybe the other thing to mention, sorry to go back one step again, is it was critical for me, critical, that whoever was to learn the measure, that they could understand it, even if they were a non-investment professional.
Stewart: And I think that’s critical.
Stewart: When board members are being asked to provide a governance function and vote and approve decisions that they may not understand fully, I think that this is a particularly important point.
Karyn: Yes. And Stewart, you probably saw this too. If they don’t have something like this measure that they understand and can track, what did they do? They focus on their peers, which the peer set may have a completely different set of objectives and ability to take risks. They focus on, and I thought this is maybe what you were going for before, oh, that’s all great, but what happened to Apple?
Stewart: Yeah. That was always my joke is like you manage a billion dollars of fixed income and 7 million worth of equity, and the equity portfolio dominates the discussion. It used to drive me wild.
Karyn: Yeah. Well that happens everywhere.
Stewart: But it’s true. Yeah, it’s really true. And I do think that one of the things that the insurance industry does a lot of and probably should do less of is this looking at peers, right?
Stewart: Because what’s good for your peer and what’s good for you. You know what I think it is at the end of the day? People don’t want to stick out, right?
Stewart: They don’t want to stick out.
Karyn: That’s right.
Stewart: But your situation may be completely different than the peer group. That’s why I like what you’re doing in the Pi system, I like this holistic approach that has things in simple terms.
Stewart: And oh, by the way, you don’t have an axe to grind, right? You’re 100% independent. You don’t have an axe to grind by asset class or by anything else.
Karyn: That’s right.
Stewart: It’s just, here’s an approach, I can define variables that are understandable and put them in a holistic context.
Karyn: Yes. And this is informed a little bit by working with boards my whole career, I’m also an independent director for a publicly traded company. And so the focus that I particularly have in Hightree on independence and my colleagues, we are 100% behind this idea that whatever information we produce, whether it’s our advice or it’s financial technology, has to be clean.
Karyn: It has to be fiduciary ready. And it has to be aware of the role and job that a governing board has. So we actually, we’ve designed for that concept. And so if you go down the path of saying, “All right, well, what are our objectives, and can we measure our ability to get there?” And I’ll come back to the definition in a minute, then you have a completely different kind of conversation at the board, are we on track or are we not on track, et cetera. And in a way you can start to see the cost of being the same, the cost of simplicity, the cost of making a choice that everyone feels comfortable with, but is actually working against them, oddly enough.
Karyn: So let me go back to the definition. So Pi, very simple over an investment horizon, let’s just say it’s a three year or five-year period, is once the objectives have been laid out is the average chance you’ll hit them. So take objective one. What’s the probability we’ll hit that? Okay, well that’s 90%. Objective number two, that’s 70%. Objective, number three, et cetera. And so the average of those probabilities as Pi. So if you go to any investor, even a retail investor, and you say, “Well, you have a 50% chance of hitting your objectives.” And they may say, “Whoa, I didn’t know that.”
Karyn: And they would probably say, “I didn’t know that. Well, can I increase that?” Yeah, you probably can.
Karyn: And how do you do that? Well, how do I do that? Well, you have to understand what your objectives are and maybe you can get what you want. Maybe fixed income isn’t going to give you the luxury that we’ve had over the last 20 years. And so, you have to think about either making better trade-offs and you want to have a tool that is designed for you to make that trade-off, you as an organization, you as an insurance investor, you have specific capabilities. And so, let me go a little bit further, so capabilities. So Pi, one of the measures you can set of course, is a loss limit. And the way that our team has built the technology, it’s not just considering loss at a point in time, but actually through time.
Karyn: But let me simplify and talk about at a point in time. So that limit that you set is for the investment assets, which as you rightly said, sometimes an insurance covers ten. We had 50 balance sheets, almost 50 balance sheets at Farmers. And so there’s a lot of different moving parts and getting a very clear handle on what the investment size is in relationship to the equity or the surplus is critical. And knowing how much of that surplus do you get to spend? Do you get half? Do you get a third? Do you get a quarter? That’s probably more art frankly now than science. In the future, when I’m old, that’s going to be science. Right now, it’s probably more art.
Stewart: But it does help to define your opportunity set.
Karyn: Yes, absolutely. Yeah. If you can put to work, for example, a billion dollars of risk on a $20 billion portfolio, that tells you a lot about how much equity risk you can take and why it’s super important not just to think about the trade-offs, hitting your return, your income target and your growth target, but also thinking about, “Oh gosh, are we going to blow through it at the same time that the insurance business has a problem?” You’re sharing a balance sheet.
Stewart: And you know what? That is so true and so missed, I think. You are sharing a balance sheet. You are running money inside the belly of the beast, right?
Stewart: The wind blows and the equity market falls and you’ve got double trouble, right? They’re not independent.
Karyn: That’s right.
Karyn: That’s right. And so understanding how those evolve through time, as time moves forward, what’s happening with our commitments, what’s happening with the assets, what’s happening with other parts, key parts of the business that actually people in the company know. They have a good handle on this business. They know what their line of business is likely to look like and how it’s likely to evolve.
Karyn: Even under different conditions. And so, it’s not the intelligence doesn’t exist. It does, but there hasn’t really been a good framework for it. And one that, as I said, a board, a group of people in charge, could actually understand. So when we first worked on portfolio Pi together, I kind of harassed Jaksa a little bit. I said, “Well, it’s not enough.” He said, “What do you mean it’s not enough?” I said, “Yes, people understand probabilities vastly better than they do standard deviation.” There’s actually work that shows this, empirical work that shows this through betting. But how important is 60% chance versus a 65% chance? What is that? How is that important? Oh, so what, it’s 5%. Big deal. But maybe that’s really important. Maybe that’s a lot of money. And so we came up with an approach where basically we could translate the improvement in Pi into dollars.
Stewart: Which is something that everybody understands, including boards of directors and everybody else.
Karyn: Yeah. And the CFO, especially.
Karyn: So if you’re really clear on what the objectives are and you know how much risk you have to put to work, and you say, well, if we could improve that, chance of hitting the objectives, then if we shift the portfolio around, maybe we conserve on risk, maybe we avoid some of the tail risks that we have in the portfolio that we’re finally now seeing, what does that look like in terms of dollars? And so we came up with a second term, which is Eta. It’s a Greek letter H. And this is where we get to geek out a little bit.
Stewart: This is my favorite part.
Karyn: The question is, what Greek letter hasn’t been used in the field of finance? Eta would be among the set.
Stewart: Eta’s there.
Karyn: Eta’s there. It’s also with H, Hightree. But, further than that, I wanted to find some meaning in it. And so Eta is a term that is used in thermodynamics, and it’s a measure of efficiency.
Stewart: Oh, there you go.
Karyn: That’s great. So Eta is this other term and it works kind of like a certainty equivalent. How much money would you have to add to your current portfolio, that has a 50% chance of hitting your objectives, each year, such that you have the 60% chance? And so it’s a measure of worth or value of moving the portfolio from where you are to a place that is better serving your outcomes. And then once you have that language, the world opens up. It really does open up. And this is what I find super exciting, and I don’t know if we get to talk about it today, but super exciting for asset managers.
Stewart: Yeah. Well, we’re at the 30 minute mark and I feel like we just started talking. I’m beyond interested in what you do and I love it. Deep in my heart, I’m a finance geek. And at the end of the day, the idea that you can take complicated topics and put them into terms that everyone understands, you are a professor, I’m a professor, the best teachers in the world, that’s what they do.
Karyn: That’s right.
Stewart: They’re able to take complicated things and make it such that you can understand it, or that anyone can understand it. And that’s the power I think of the Pi system and the way that you’ve gone about it, is you’ve created a language of risk that’s not technical really. It’s straight forward. You almost think, “Gee, that can’t be that simple, can it?” Teaching a different vernacular is part of the learning curve.
Karyn: It is part of the learning, and it is, I’m going to push it further. I’m going to say it’s the responsibility of an academic to bring these kinds of concepts to a world that every single person on this planet can benefit from, financial knowledge. And to a certain extent today, it’s inaccessible. And when I take a big step back, and we’re not just talking about insurance investment, we’re talking about investors more generally, and maybe even some of the clients and customers of insurance companies that have investment products, this opens the door for them to better understand better relationships. There’s just much more that you can do. And so I felt very much, and I said this when we first started working on the concepts, I said, it’s really critical. We have a job to do here. And it’s not perfect. There are a lot of practical things that we will be learning in this journey. But I feel like it’s an important step forward, that giving people the ability to talk about things that are the most important drivers of outcomes, that’s where you got to start.
Stewart: Absolutely. We are having you back on and we are actually going to create a series with Karyn on risk and risk management and governance and a whole host of other things, I think. So you can count on seeing more of Karyn. But thanks for being on today.
Karyn: Stewart, it’s super exciting for me to be able to talk about this and geek out with you on investment decisions.
Stewart: I love it. We’re both such geeks. It’s so great. Okay. That’s good. So if you like what we’re doing, follow us, tell your friends. You can find us on all the major platforms. If you have ideas for podcasts, please email us at firstname.lastname@example.org. We’ve also just announced a major joint venture with CAMRADATA that’s going to bring managers search tools and evaluation tools to insurers, free of charge. So for more on that, check out our website. Thanks for listening. I’m Stewart Foley, and this is the Insurance AUM Journal podcast.