pic & bio: northerntrust.com
Peter Mladina is the Executive Director of Portfolio Research for Northern Trust Wealth Management. He is responsible for the application of leading research to the wealth management investment process. This includes research, tools and methods that support asset allocation, portfolio construction, investment selection and best practices in portfolio management.
Previously, Peter was the Director of Research at Waterline, a boutique wealth manager acquired by Northern Trust in 2010. In addition to advising high-net-worth clients, Peter helped Waterline develop an innovative goals-based asset allocation solution and an empirical investment approach rooted in academic research.
Peter is a Professor of Practice at UCLA, where he teaches applied financial economics. His research on asset allocation, portfolio construction and asset pricing has been awarded publication in peer-reviewed journals, and he is a co-author of the CFA Institute’s Level III asset allocation curriculum. He received a B.A. in economics from UCLA and an MBA from Edinburgh Business School (U.K.).
In today’s episode, we unpack how rigorous research translates into actionable strategies for wealth management. Ben and Mark are joined by Peter Mladina, Executive Director of Portfolio Research at Northern Trust Wealth Management and professor at UCLA. With an impressive body of published work and practical innovations like his goals-based asset allocation software, Peter offers a unique perspective on bridging the gap between theory and practice. The conversation delves into foundational topics like asset allocation and factor models, with a special focus on practical applications of research in wealth management. Peter shares insights from his research, including intriguing findings on factor investing and joint tests of market efficiency. From real estate investment trusts to the nuances of the Intertemporal Capital Asset Pricing Model (ICAPM), the discussion covers how these concepts can directly inform financial planning and portfolio construction. Tune in to explore the intersection of academic insight and everyday financial decision-making!
Key Points From This Episode:
(0:00:17) Introducing Peter Mladina and his wealth management research.
(0:04:00) Theoretical and practical shortcomings of Markowitz's Modern Portfolio Theory (MPT).
(0:05:24) How the Capital Asset Pricing Model (CAPM) resolves MPT’s shortcomings, and how the Intertemporal CAPM (ICAPM) resolves the CAPM and MPT’s shortcomings.
(0:10:16) Key distinctions between an optimal ICAPM portfolio and an optimal CAPM portfolio.
(0:15:33) Allocating between liability hedge assets and risky assets, and when it’s sensible for individual investors to try to fully hedge consumption liabilities.
(0:20:14) The role of Monte Carlo simulation and human capital in building ICAPM portfolios.
(0:24:15) Steps for practitioners starting with ICAPM and how to advise their clients.
(0:37:18) Insights from Peter’s papers on factor models: why common risk factors should explain returns across most asset classes.
(0:40:11) The value of looking at asset classes through a factor lens.
(0:41:54) Main factors Peter uses in his research and observations on the zoo of factors.
(0:46:23) Takeaways from Peter’s paper on real estate (and why he doesn’t like it that much).
(0:56:45) Unpacking hedge fund returns and factor models and Yale’s endowment performance.
(01:02:44) Peter’s research on traded portfolios and jointly testing factor models and manager performance.
(01:07:14) How Peter defines success, both professionally and personally.
Read The Transcript:
Ben Felix: This is the Rational Reminder podcast, a weekly reality check on sensible investing and financial decision-making from two Canadians. We're hosted by me, Benjamin Felix, Chief Investment Officer at PWL Capital, and Mark McGrath, Associate Portfolio Manager at PWL capital.
Mark McGrath: Welcome to episode 338.
Ben Felix: Today, we welcome Peter Mladina for a very interesting and practical, but also theoretical conversation about asset allocation and factor models. Peter is the Executive Director of Portfolio Research for Northern Trust Wealth Management. In his job there, he's responsible for the application of research to the wealth management investment process. And that includes research tools and methods that support asset allocation, which is basically what we talked about in our conversation.
Now, his research is published in journals. He's got a ton of papers in the Journal of Portfolio Management, Journal of Wealth Management, Journal of Investing. He's doing research that is getting published, and then he's also applying that in his professional capacity with Northern Trust, which is pretty cool. He also developed a goals-based asset allocation software that he alluded to during the conversation, which sounds pretty cool. It's not available online, unfortunately. But he's able to speak to how it was constructed, and how it works, and why it matters to do things the way that they do.
Peter's also a professor of practice at UCLA where he teaches Applied Financial Economics. He's got a BA in Economics from UCLA and an MBA from Edinburgh Business School in the UK. I, as I mentioned at the beginning of the conversation, became familiar with Peter because he's got a paper on real estate investment trusts where he does a factor decomposition to show that REITs are basically just small-cap value stocks and corporate bonds effectively. I remember reading that paper many years ago. That's a 2018 paper and I probably read it before it was published in the journal. I'm not sure. Or maybe when it was. But either way, it's a long time ago. And so, I've kind of followed his research since then.
More recently, he'd done a bunch of writing on ICAPM and how to apply that practically, which of course is something we've talked about quite a bit in the podcast more on the theoretical side. His attempts to make it more practical, I found pretty interesting. That's basically what we talked about, but I thought it was a pretty cool conversation. What do you think, Mark?
Mark McGrath: It was great. I think the ICAPM stuff, a lot of this is really good financial planning, right? A lot of it is about matching goals and liabilities with cash flows, and that's a large part of what we try to do for clients at the end of the day. We're talking about ICAPM. We're talking about portfolio theory from that perspective. But really, it's portfolio theory combined with individual financial planning. For me, that was fascinating. But the whole conversation was great. He's obviously brilliant. He's done a lot of work across a couple really interesting areas. And it was a great conversation. I learned a ton.
Ben Felix: I like that he's really focused on the practical side of things. And he's working in wealth management. He's working quite literally to help people practically implement these ideas. I find that to be pretty compelling just about his overall story. He's done some more recent papers on what he calls joint tests of factor models and performance. He's basically taking market efficiency as the benchmark. What outcome would we expect if markets were efficient? And then he's testing different factor models on live equity, and fixed income, and multi-asset portfolios.
Instead of it being a more academic exercise, he's looking at actual traded portfolios. I won't give away the result there, but his findings are super interesting. Make you think about just the whole idea of factor investing. And we do talk to him about that in the context of his research results. But anyway, that's probably enough for an introduction. Anything else you would add?
Mark McGrath: No. Otherwise, we'll spoil too much of it. I think that's good.
Ben Felix: All right. Let's go ahead to our episode with Peter Mladina.
***
Ben Felix: Peter Mladina, welcome to the Rational Reminder Podcast.
Peter Mladina: Hi, Ben. Thanks for having me.
Ben Felix: Very excited to be talking to you. I read your paper on REITs many, many years ago, which we're going to talk about later, but I've been familiar with your work since then.
Peter Mladina: It's funny how some of the papers I like the least are the ones that everyone else likes the most.
Ben Felix: Okay, we'll come back to that later. To start off, can you talk about the theoretical and practical shortcomings of Markowitz's modern portfolio theory?
Peter Mladina: That's in the context of what I believe strongly is a much better approach to asset allocation than what most practitioners are utilizing in sort of the workhorse method. The main shortcomings are – just stepping back, there's three at the highest level. The first is Markowitz's MPT is an asset-only framework. How is it that the optimal asset allocation solution doesn't even consider the funding objective? What you were trying to do with those portfolio assets? That's shortcoming number one.
Shortcoming number two is that it's a single period model. It's optimal for one period, whatever that period you're optimizing around is. In most cases in practice, that's a one-year period because we're optimizing around one-year capital market assumptions. Arithmetic means, which are technically one-year returns. And then the third would be there is no sound theory or good theory for constraints.
And in practice, we know MPT optimizations are highly sensitive to inputs. And so, as a practical matter, you're forced to kind of impose some constraints, which is fine. Optimizations require constraints. However, the truth is due to that input sensitivity. What ultimately happens is whatever your constraints are, you're going to ultimately drive your asset allocation, which fundamentally overrides the theory of MPT.
Mark McGrath: How far does the CAPM go in resolving the shortcomings of modern portfolio theory?
Peter Mladina: The CAPM still suffers from some shortcomings of MPT, not all of them, but then adds some additional shortcomings. It's still an asset-only framework, the optimal CAPM portfolio. And we have to sort of differentiate the CAPM as a portfolio theory versus the CAPM as sort of a theory of pricing assets, like a factor model. Here we're talking about portfolio theory. That is one cool thing about the CAPM, both it sort of unifies asset pricing and portfolio theory. Factor models and asset allocation.
But to answer your question, it's still an asset-only framework. Has one shortcoming with the CAPM. It's still a single-period model. That's a shortcoming with CAPM. You don't have the same issues with there are constraints because you're assuming the optimal risky portfolios, the market portfolio. That's an improvement. But you have other problems. At the portfolio level, you have issues of cash aversion, where most investors don't want to have high allocations to cash as sort of their safe asset in a multi-asset portfolio.
And then we also know empirically, there's at least more than one risk factor, more than just a market factor that explains asset returns. The literature is full of them. I'm sure we're going to get to a number of them later today. But at a minimum, we know for sure there's at least a market factor for equity and equity-like assets and a term factor for hydrating bonds that are very different based on any empirical test you can run. Solves one shortcoming. Adds maybe two more.
Ben Felix: How does an ICAPM resolve the shortcomings of the CAPM?
Peter Mladina: In the case of the ICAPM, you have improvements over both MPT and the CAPM in the sense that now we're going to identify a funding objective. Maybe that funding objective, if you're pension plan, is a liability. Or insurance company, it's a liability. Maybe if you're a private investor, it's a goal or a series of goals you're trying to fund like lifetime consumption or gifting. But whatever we're doing, we're going to basically look at the purpose of the assets in the portfolio.
In our view here, I believe anyway, and here in my employer, Northern Trust, we believe essentially assets should have a purpose. It should serve a purpose. And that purpose is to fund something ultimately, whether it be a liability or a series of goals. We can step back and say we have now a funding objective. Let's construct a portfolio around that funding objective. The optimal asset allocation solution is no longer an asset-only return versus risk perspective. It still incorporates elements of return and risk. But now, in a way, we're going to optimize it so that, in the most optimal way possible, we are meeting the funding objective.
The improvement or enhancement of the ICAPM over both MPT and the CAPM is essentially now it is an asset goal or asset-liability framework, a liability relative or goal relative framework, not an asset-only framework. That's enhancement number one. In terms of doing that, now we've turned it from a single period model, optimal for one year or one month or whatever may be the case, and we've turned it into what we call an inter-temporal or multi-year model where it's going to be optimal over the full period. Whether that's the next 10 years, the next lifetime, the next 100 years, whatever might be the case.
And so, doing, we've added a set of liabilities or goals. They have the character of time associated with them. We're going to create an optimal portfolio around those liabilities and goals that exist through time. And so, we've moved from a single-period model to a multi-period model.
Thirdly, we had the problem in MPT of no good theory for constraints. But with the ICAPM like the CAPM, we can rely on market efficiency, at least as a benchmark to help us see what the optimal risky asset solution ought to be in the portfolio. We can either use that and take that portfolio as the optimal risky asset return-seeking portfolio, or we can use it at least as a benchmark to set constraints. So now we have a good theory for constraints in determining what that portfolio should be.
At the end of the day, you have two kind of sub-portfolios, this liability-hedging or goal hedging-portfolio, and this return-seeking risky asset portfolio. You construct the two in a way that's efficient relative to funding your liability or goal profile, and then you will basically select the allocation to each based on funded status and risk tolerance. That does away with the other shortcoming of the CAPM where investors have potentially aversion to high allocations to cash in the portfolio because we've redefined the risk-free asset. The safe asset is no longer cash, but rather some sort of high-grade bond portfolio that's very closely cash flow aligned with the liability profile. In that way, in a number of different ways, we've actually significantly mitigated the issues with MPT and the issues with CAPM by moving to the ICAPM.
Mark McGrath: You already touched on this, but maybe we can elaborate a little bit more. How is the optimal ICAPM portfolio for a given investor different from the optimal CAPM portfolio?
Peter Mladina: Let's start with our definition of risk and what the risk-free asset ought to be. In the CAPM, the definition of risk is some concept of volatility or low volatility. That's why the risk-free asset is the CAPM is cash or treasury bills. That's commonly viewed as a safe asset because that has the least or it has arguably no volatility to it. But in the ICAPM, that's not really the relevant definition of risk. The relevant definition of risk is a liability relative or a goal relative definition of risk because ultimately the risk that matters is the risk around funding that final liability or that final goal profile. The risk that ultimately matters is the dispersion in final funding outcomes. Is there significant dispersion as future surplus or shortfall in funding outcomes?
The first and most important thing is to redefine risk from this asset-only view of volatility to this liability relative risk that really ultimately manifests as dispersion in future funding outcomes. With higher dispersion and future funding outcomes, you're more likely to fail in terms of funding that liability. You can think of it that way, but another way to think about it is what's that definition of risk? It's a different definition than standard deviation. It's actually better characterized as tracking error. We've changed our definition of risk from asset only to liability relative by moving from standard deviation to tracking errors, our definition of risk, and it's tracking error relative to funding that liability or goal profile.
What asset is risk-free in relation to funding that liability profile? Well, the asset that has no default risk and is perfectly cash flow matched with the cash flows of that liability or goal profile is your risk-free asset. Let's step back and think about it. Let's say you had a liability due in 10 years for a million dollars, the asset that's risk-free relative to funding that asset might be a zero-coupon 10-year treasury bond. At least in theory, it has no default risk, and will perfectly mature at a million 10 years from now.
There's no dispersion in future funding outcomes around using that asset. It may have a lot of volatility between now and the next 10 years. 10 years from now, it's going to mature at $1 million and fund your liability without any risk. Cash would be your definition of risk if you have an asset-only view. But if you're trying to find a million dollars ten years from now, that's zero-coupon ten-year treasury bond is your definition of risk.
Therefore, we want to create kind of a liability hedging risk-free asset from that perspective. One that has essentially no default risk and is perfectly cash flow match. Now in practice, there may be very many reasons where something like that is impractical. Most liability profiles are more complex than sort of a single bullet 10 years out. Not all bond maturities may be available to meet your liability or goal profile. Additionally, most bonds pay coupons. There's reinvestment risk. There could be other practical reasons to use different types of bonds than just treasury bonds.
You can get very close to the risk-free asset in practice by using something we'll call liability hedge instead of the true risk-free asset. As long as you're using high-grade bonds that have very minimal default risk and you're aligning those bonds in a way where if you're at least on average duration-matched, you can very, very close to an intertemporal risk-free asset by essentially minimizing the tracking irrelative to liability using high-grade bonds with minimal default risk and duration matching those bonds on average with the duration of a liability or the growth profile.
A couple of things so far just to recap, most importantly with the ICAPM, we have to redefine risk to be liability or goal relative. And then we have to redefine the risk-free asset to be the asset that funds that goal or liability profile with the least amount of risk. And that's not cash. Once we've defined that optimal liability hedging portfolio, now we can look at the rest of the portfolio. There may be some component that I still want to bear risk around. And what is that portfolio? What's the optimal return-seeking portfolio of risky assets? And we can still optimize that portfolio relative to the liability.
Now, if you're a big believer that markets are efficient, you essentially can take the market portfolio of risky assets to be your return-seeking portfolio in the ICAPM. However, if you choose to deviate from that market portfolio of risky assets, you can at least look to it as a benchmark to set constraints in your own optimization procedure. You may choose to deviate because you have a different worldview about expected return and risk than maybe you see in the market portfolio of risky assets. You may deviate because you pay taxes. You may deviate for many reasons. Maybe you have other outside assets to consider human capital, other major concentrated positions. There could be a number of reasons you choose to deviate, but you can still use that market portfolio of risky assets to set constraints and run an optimization procedure that gives you essentially the optimal portfolio of risky assets.
And so, at the end of the day, you have this optimal liability hedging portfolio, and you also have this optimal return-seeking risky asset portfolio. And depending upon both risk preferences and your funded status, frankly, you can choose what proportion of each to align with a liability or a set of goals, whether they're higher priority lower priority.
Ben Felix: Walk me through – practically, I've got a client who wants to retire and whatever, 20 years, how do you allocate between the liability hedge asset and the risky assets?
Peter Mladina: There are different ways to do it. But getting to the core methodology of the ICAPM, what's great about it is it's a process that works both with pension plans, with a single homogeneous liability, and it works just as well with a private investor who has a series of heterogeneous goals that have different priority levels associated with them. Some goals may be super high priority, and therefore the investor wants to take less risk around them. Other goals may be more aspirational, so the investor's willing to take more risk around them.
And the nice thing about the ICAPM is that you can have full hedges, you can have partial hedges, and you can have no hedges. You can fund an aspirational goal, for example, without any allocation to the goal-hedging portfolio. You can really impose your level of risk preference. What you would do is you would go through a discovery process with that investor and you'd identify and value all of their goals. And you would understand that – you gave the example of their lifetime consumption and retirement. It might be broken down into different phases. Phase one, while they're still very active, traveling, doing many things. Phase two, while they're much older or more sedentary. There may be health care issues to also consider. You can think about gifting goals, gifts to generation two, generation three. You can think about gifts to your alma mater. We're going to want to go through that process of understanding and quantifying those goals, valuing them. And that's a process of also understanding the priority level for each goal.
Higher priority goals will get higher goal hedge allocations. Lower priority, more aspirational goals will get lower goal hedge allocations or no goal hedge allocations. Perhaps they're funded completely by the return-seeking risky asset portfolio. Goal-by-goal, we can go through and understand risk preferences. And then at the same time, as you understand these risk preferences, we have an additional good piece of information that helps with portfolio selection because it can give you essentially your funded status. Both at the goal-by-goal level and in the aggregate portfolio level, we can have depending on the assets you have and the risk you're willing to take around the goals that you're trying to fund. You still at the end of the day have budget constraints. And so, you can think about how much risk I prefer to take or maybe how much I may need to take to fund my entire profile of goals.
There's very much a back-and-forth and a set of tradeoffs. And as you move through the course of your life, this process is also very adaptive, and that's very important. As expected returns become realized returns, funded status may change, and you can make adaptive tradeoffs and revisit the degree to which you hedge each goal. Obviously, if you completely hedge each goal, you can go away and sleep at night. You never have to make adaptive trade-offs. But the degree to which some or all goals are funded, at least in part, by the return-seeking risky asset portfolio, there's still risk around funding those goals. So adaptive trade-offs over the course of your life are still important.
At the end of the day, practically speaking, goal discovery, with each goal, we're going to identify how much of a hedge you want to impose. Anywhere from 100% to fully secure those goals and never have to worry about them again, to 0% for an aspirational goal. Evaluate funded status, and revisit as we move through the course of your life. Because you still are incurring risk to some degree in most cases to find those goals unless you've completely hedged away everything.
Mark McGrath: And do you think that's sensible for individual investors to try to fully hedge those consumption liabilities?
Peter Mladina: Where you see investors fully hedging away their consumption liability is high net-worth, ultra-high net-worth investors who have so many surplus assets that oftentimes they may choose to hedge away all of their goals. And then surplus assets are the ones that are just held in the risky asset portfolio. That's where we see it more commonly. But even in those cases, typically there may be some element of a partial hedge even though it's largely hedged away.
For many other investors, the mass affluent and normal retirement savers, that's where I get back to you may be forced to take risk to fund your retirement goals. Just be aware of the risks you're taking. How much of the goal you're hedging away and securing essentially? And the degree to which you may have to make adaptive tradeoffs over the course of your life as expected return becomes realized returns. Maybe you'll get lucky and get better returns than you expected on the risky asset side. Maybe you'll be unlucky and you have to revisit your goals and modify them. There's value and adaptive approach. But again, the degree to which you can hedge away your goals, you don't have to worry about them again as long as you properly construct that goal hedge, have minimal default risk, and to be duration matched on average.
Ben Felix: What role does Monte Carlo simulation play in building ICAPM portfolios?
Peter Mladina: We can use Monte Carlo simulation to eject risk. There really is no risk or minimal risk on the liability hedging side. All of the risk is on the return-seeking risky asset portfolio. We're able to build Monte Carlo simulations that capture that dynamic. They're different than the traditional MPT Monte Carlo in the sense that these ICAPM Monte Carlo's capture the de minimis risk of the liability hedging portfolio and then the excess of goal risk of the risky asset portfolio. And we bring it all into the present.
That's two main differentiators, is that the risk here is risk in relation to funding the liability or goal profile. And the Monte Carlo is built that way. Your risk is tracking error, not standard deviation in this Monte Carlo. It's not a Monte Carlo into the future, but rather a Monte Carlo that brings you to the present to augment or supplement your deterministic view of funded status. We're able to evaluate funding status stochastically via the Monte Carlo. So you can see the risk around your current funded status. The degree to which, in today's dollars, you have a shortfall or a surplus across the distribution of outcomes. We think a present value Monte Carlo is meaningful because most investors intuitively understand dollars they have today in terms of surplus or shortfall and in terms of decision-making. It's much more intuitive than thinking about dollars I might have 30 or 40 years from now.
Mark McGrath: How does human capital fit into ICAPM portfolio construction?
Peter Mladina: Human capital, particularly as we apply the ICAPM to goals-based investing, is really important. Most private investors have a lot of different potential assets that can fund their goals, their lifetime goals or consumption goals, retirement, gifting goals, etc. They're portfolio assets. They're financial assets. But they may have many other assets they can use, from personal property they might have, personal assets they may have. But also, most importantly with most investors, human capital.
And at the end of the day, human capital is an asset and we can even capitalize it. I mean, there are other complexities around it, but we can take the value of that future of labour income, bring it into the present, present value it, and we can look at it as an asset just like we look at portfolio assets to help fund goals.
And so, in an ICAPM framework, we can do it a number of different ways. Fundamentally, at the end of the day, what we're doing in a goals-based application of an ICAPM is incorporating that asset as sort of this outside asset, so that really the portfolio of financial assets sits right next to that portfolio of human capital assets and both are there and enable you to fund goals.
Now, it adds some complexity because there are things you have to think about just like we align an ICAPM portfolio with the cash flows of the goals that an investor may have. We need to do the same thing with human capital. We need to align it so that human capital funds when investors' goals. And this adds complexity when we have a bunch of different types of assets' portfolio assets, financial assets, and human capital assets. They need to work together. They need to look at the sequencing and mix of assets and how they're funding goals. How are they mixed together and how are they sequenced in terms of finding goals?
If you think about it for a second, oftentimes investors have future human capital. They save it. It goes into an investment portfolio and then it eventually funds goals. Alternatively, you may have a mix of portfolio assets and human capital assets that contemporaneously fund a goal flow together. It's just another asset to consider that sits next to your portfolio of financial assets. Ideally, you'll capitalize it like you do. It brings it to the present, so you can think about it right next to your portfolio financial assets in terms of how it funds goals in the future.
The challenge is what is the risk of human capital? What's the appropriate discount rate and the complexities of sequence relative to all the various goals, the profile of cash flows that compose the liability? It adds some complexity, but it's really important to actually solving the optimal solution for private investors.
Ben Felix: I'll be honest. I've read your papers on this. I've read John Cochrane's most recent paper on this, and we're not using ICAPM in practice with clients at the moment at our firm. And I don't think many practitioners are. What steps do you think practitioners should be taking to start using ICAPM and the advice they give to their clients?
Peter Mladina: Kudos to you for reading the papers and understanding the concepts. You mentioned John Cochrane. I know he's been a guest on your podcast. I'm a practitioner first and foremost. I also teach, but it's more of a side job. A lot of this research has relied on some of the original research of John Cochrane. And I need to give him credit first and foremost. I think a lot of my contribution is bringing it to life in the practitioner world, working out this theory. Is it valid empirically? And then how does it work in practice? You asked a great question.
Firstly, read the papers, an ICAPM framework for asset allocation in the Journal of Portfolio Management, an ICAPM for goals-based investing in the Journal of Wealth Management. Empirically, forthcoming I have compensated portfolio risk or the compensated risk of multi-asset portfolios forthcoming in the Journal of Portfolio Management. I think if you understand those frameworks, you'll have a really sound anchoring in the theory and frankly the empirical evidence as to why this is a superior asset allocation.
Now, how do you bring it to fruition? The first thing I think is to start thinking about risk from a goal relative or a liability relative perspective and start to move away from standard deviation as your definition of risk. Start working with clients and explaining risk from that perspective. And what might be a set of assets that really is their safe asset? It's not cash. It's going to be something that is safe relative to funding their goal profile. In practice, something that is composed of high-grade bonds that's duration-matched on average.
In theory, something with no default risk, perfectly cash flow matched with the cash flows of your liability or goals. From there, all other assets are available to create that return-seeking risky asset portfolio. It could be other types of low-grade bonds. High-yield bonds can be part of that return-seeking risky asset portfolio. But more commonly, equity and equity-like assets, private investments, etc., can all be part of that return-seeking portfolio.
Thinking in terms of these two sub-portfolios, that safe liability-hedging or goal-hedging and that return-seeking risky asset portfolio. How do we build that return-seeking risky asset portfolio? Well, if you're a big believer in market efficiency, it's very easy. Just take a look at all risky asset classes out there, which is pretty much every asset class minus liability-hedging assets. What's the relative cap-weighted market value of all of those? Well, that's the market portfolio of risky assets. If you're a big believer in market efficiency, stop, you're done. That can be your risky asset portfolio.
There's an optimization process that we document in the paper where maybe your optimal return seeking portfolio is a little different than the next person because of your liability profile. That's an important consideration. But there could be other reasons. You're an active investor and you have a strong point of view of expected return and risk. So you want to deviate from that market portfolio. You can optimize around that. Or you may have other real-world constraints like you pay taxes or you have outside assets that are a significant portion that you can't really do much about it. You want to work around that.
I do think that creating software around this and having an interactive approach with a client is really optimal. That's something we do here at Northern Trust. From what I understand out there today, there really isn't something. There are goals-based approaches, but not one that's really well anchored to the ICAPM like we do here at my firm.
A software-based approach also helps a lot with that goal discovery process and working through the implications of identifying goal hedges per goal. Do I want to hedge away 50%, 75%, 100 % of each of those goals? What are the implications in real-time? And then how do I make adaptive trade-offs?
Ben Felix: How important is it for the liability hedge asset to be a real return asset, like tips in the US? And I ask the question, because in Canada we don't really have a real return bond market. The government does phase that program or is in the process of phasing that program out and it's just not that popular of an asset class for Canadians because of the structure of the market.
Peter Mladina: The first thing I would say is, is your liability or goal nominal or real? Is it inflation-sensitive or not? At least in the US, if you're in the world of pension plans, particularly corporate pension plans, those are actually typically not a liability. You would want to use nominal bonds because you have a nominal liability you're trying to fund. You move in the world of private investors, first of all, we should differentiate between sort of a legal liability like a pension plan has and a goal which tends to be more discretionary.
As a goal, you have much more flexibility in how you want to fund it. But most goals among private investors are in fact real goals. They are inflation-sensitive. An optimal solution, in the US anyway, would use treasury inflation protected bonds primarily. Now it's interesting, because in private wealth, investors are allergic to paying taxes. In the U.S. in particular, there's a very strong preference for municipal bonds. But municipal bonds are subject to a significant amount of inflation risk. That's a problem.
And behaviorally, there is the challenge in high-net-worth investors, at least to some degree, their preference for avoiding taxes through municipal bonds by utilizing treasury inflation protected securities. The reality is, is that if you are looking for a true goal hedge that mitigates all goals, not just mismatches of duration risk, or not just undue term risk, or default risk but also inflation risk, you'd want to use inflation protected bonds.
Now, you could use derivatives like swaps and things like that if you don't have a market for government inflation protected bonds. You could use high grade bonds with swaps. You have to do a little bit of portfolio management engineering to get there. That's available, I believe to you guys, in Canada. But again, it requires a little bit more construction liability hedge to get there.
Now in practice, again, getting back to the U.S. where, again, a lot of high-net-worth investors have a strong preference for municipal bonds, it turns out that you don't necessarily have to have 100% tips allocation. Because the reality is, is that even with tips, because of taxes, you don't have a perfect inflation hedge. You only have a partial inflation hedge to the degree of one minus the tax rate. Even with tips, you don't have a perfect inflation hedge. That therefore leaves room for municipal bonds in your goal hedge. It just turns out you still need a pretty material allocation to tips to kind of craft that optimal goal-hedging portfolio. It tends to be, in the U.S. anyway, a mix of high-grade municipal bonds and tips.
Mark McGrath: A few minutes ago we were discussing human capital and the way it plays into ICAPM. And Ben, you said not a lot of practitioners are using ICAPM in that way. With our clients, we combine risk tolerance and then risk capacity when helping to design portfolios. And risk capacity for us is more quantitative. When we think about human capital, we're thinking of the stability of that human capital over time. Things like whether you have disability insurance in place. How risky is that income stream for the individual in the future. And obviously, those who have more stable human capital might have the capacity to take on more portfolio risk. Is that kind of a light form of implementing ICAPM?
Peter Mladina: Yeah, that's a light form. Anything you can do to incorporate human capital for investors that have large human capital components, I think that's good. That notion of risk tolerance and risk capacity, we also use it here at Northern Trust and in our goals-based methodology. We think about risk tolerance as something that's more psychological. Different investors can have different risk tolerances, and that's something a little bit more psychological, but psychology is important. To finding a portfolio that's optimal to an investor and one that they can stick through in good times and bad.
But risk capacity is a little bit different, more quantitative. Do you have the capacity to bear risk? And a lot of what times that has to do with how much surplus assets you have. You could have an ultra-high-net-worth investor, they can easily fund all of their lifetime goals regardless of what happens to the market, for example. They have a ton of risk capacity, but behaviorally or psychologically, they may have low risk tolerance despite that capacity. And then other investors may have very little capacity to bear risk but have high risk tolerance. The two aren't necessarily highly correlated. And in finding a portfolio that is optimal for an investor, you have to consider those things.
And so the ICAPM does that in identifying your degree to which you want to hedge your portfolio is going to partly be based on risk preference or risk tolerance. But then as you look at your funded status, the implications of funded status, both deterministically and stochastically through Monte Carlo, that gets back to risk capacity. And you may have to modify what you had picked in terms of hedging your various goals. You might have picked them with the concept of risk tolerance in mind, but now you have to go back and modify them based on risk capacity.
And all of that, getting back to your question about human capital, is important to incorporate the best you can. But really, the best way ultimately to do this is to treat human capital as if it were any other asset. And any asset is really just the present value of all of its future cash flows. We can bring it back to the present. We can capitalize it. And then we can look at the sequencing of those cash flows. And we can look at our financial assets, our human capital assets, and we can construct an optimal portfolio of financial assets in consideration of the risk around your human capital.
Different human capital has different levels of risk. But still, we have to get back to, well, what's the right discount rate? And I don't think that has really been well established. A lot of the earlier literature has sort of argued that human capital has bond-like risk and it's safer than your financial asset portfolio. I think that's true if you had a diversified set of human capital, like a diversified set of equities, which you can't have in your portfolio. But you don't. You can't diversify away your human capital. It's very unique to you as an individual. Therefore, it has a lot of idiosyncratic risk associated with it that to me makes it, if not equity-like for most people, something more of a hybrid between a safe asset and a risky asset.
Nonetheless, we've figured out ways to incorporate human capital where the discount rate doesn't necessarily matter. I think at best you can say the discount rate is abstract and nebulous. But there are other more technical ways that we can incorporate human capital so the cash flows match up regardless of what the actual discount rate is.
Ben Felix: Mark's question made me think of another question, and Cochrane touches on this in his paper on this. If we have an investor who is very averse to dispersion around their goal, which would then suggest a fully liability hedge portfolio with some kind of probably long -term real return bond, but they're also very averse to volatility, how do you square those two things?
Peter Mladina: Like a lot of things in our business, I think the best approach is to educate clients or investors on a number of things. It's a constant process of education. Really the first thing here is to educate the investor on the definition of risk that ultimately matters. The definition of risk that matters is not standard deviation or volatility. It is a goal relative definition of risk. And then you have to walk through what the implications are. And that's why I like to use that example of a 10-year treasury, a zero coupon 10-year treasury, that's there to fund a million dollars 10 years from now. Because I guarantee you that 10-year treasury is gonna have some volatility far higher than cash over the next 10 years. But only that 10-years zero-coupon bond is going to mature in a million dollars 10 years from now and fund that liability cash flow without any risk.
Education, understanding that perspective is key and understanding that there will be some volatility between now and funding your goals. And so, what we try to do is provide those illustrations so we can show that, because cash has such huge reinvestment risk, you can show that example. There's a lot of risk around funding that million-dollar liability due 10 years from now if you tried to use cash. You don't know what the future yields on cash are going to be. Therefore, there's a huge dispersion in that outcome.
Really, I think the answer is to educate the investor on this issue. Show illustrations empirically with real-world data that helps them come around. Education I think, Ben, is the real answer. Because the truth of the matter is there's no good way to give them both. You can't have a low volatility solution that's also the lowest risk asset relative to finding your goal profile, unless that goal is a year away. In which case, by the way, there is a role for cash if, in the next 12 months, cash is your optimal goal-funding asset.
Ben Felix: I agree with all that. I think that education around volatility not always being a risk is such a big thing for investors. I don't think most investors understand it.
Peter Mladina: It's just, stepping back, asset-only risk or goal relative risk. Standard deviation or tracking error relative to the goals. Why is tracking error important? It's because tracking error is the definition of risk that manifests their time into dispersion of future funding outcomes, which is the risk that really matters. How certain is it that you are going to successfully fund that goal?
Ben Felix: That's a great explanation. All right. I want to move on to your papers on factor models, which, as I mentioned earlier, is how I initially discovered you. To kick off that topic, can you talk about the economic logic behind why common risk factors should explain returns across most asset classes?
Peter Mladina: I think at the highest level, the idea is compensated risk. What source of systematic risk or risk that is common across assets and portfolio of assets are compensated with return premia that explains the return and risk of these assets and portfolios of assets? At the highest level, what are the systematic risks that explain portfolios of assets? I think that's the highest-level notion here.
Now, some of these have good theoretical and empirical validity to them. So you can think of like the market factor for equities, traditionally defined as the return of equities minus the return of treasury bills. Is there a return premium associated with the return of equities minus the return of treasury bills? Well, it turns out, empirically, the answer is yes. There's a return premium. Empirically for many assets and portfolio of assets, sensitivity to this factor seems to explain asset returns.
The market factor kind of hits both of those criteria. In a capitalist system, investors are taking on risk to provide capital to finance companies and they expect compensation for bearing that risk. There's a good theory behind it. And empirically it seems to come with a return premium and explain some of the common variation in assets and portfolios of assets.
Other factors that have been sort of documented in the literature, things like the size factor and the value factor for stocks, for example, they seem to be there empirically. They don't necessarily have a good theory behind them. Some of them have good theory and empirical evidence. Others are maybe more empirical risk factors that don't necessarily have a good theory behind them. The theories have been proposed, but there really isn't a good consensus around a lot of them.
A good factor should have three or four criteria to define them. First, it should come with a return premium. It should be a compensated form of risk. Historically, there should be some statistically significant return premium. A robust, reliable return premium statistically that you can have some confidence that will happen going forward. It explains the common return variation or the systematic risk of assets and portfolios of assets. A factor should be independent of other potential factors. So it's not a redundant source of return and risk.
If you have those three together, you have a common risk factor. But I'd add that fourth criterion where, in theory, we want a good sound theoretical foundation of risk. And why is that important? Because I think with a good theory and strong empirical evidence, it's more likely to persist in the future. If you just have a factor based on empirical evidence without a sound theory, you [inaudible 0:40:01] maybe being the result of data mining rather than something that you can have high confidence is going to persist in the future.
Mark McGrath: And why is it important to look at asset classes through a factor lens?
Peter Mladina: There's still work to be done here that I'm doing and others are doing. But the holy grail in financial economics is to unite portfolio theory with asset pricing. CAPM was the first attempt of that. The CAPM is cool because, at least in theory, it's a portfolio theory. How do you build asset allocation? Portfolios of asset classes? And it's also a theory of asset pricing. How is a stock priced?
Stepping back and trying to unite the two, asset classes are sort of in between portfolios and securities. But in practice, most investors commonly take asset classes as the building blocks to construct portfolios. But we commonly define asset classes in an arbitrary way. There are categorizations of assets that are similar to each other. There are dozens of different asset classes potentially, depending on how you define them, that you can build a portfolio around. But it turns out that there are only a handful of common risk factors that explain the return and risk of all of these asset classes, and thus of portfolios of asset classes.
And so it turns out that asset classes underneath them have overlapping exposures to common risk factors. And it's the common risk factors that ultimately matter in constructing portfolios because they are the drivers of return and risk of assets and securities that ultimately drive and explain your overall asset allocation. At the end of the day, they're the ultimate driver of asset class returns and, thus, portfolio returns. Now, we can get into it a little bit later, but it turns out, at least at the portfolio level, maybe you don't need so many common risk factors to explain multi-asset class portfolio returns.
Ben Felix: There's the zoo of factors, though. What are the main factors that you're looking at in your research?
Peter Mladina: My job at Northern kind of captures at the highest level two things. I teach at UCLA also. But the other area where I work deals with, well, how do we select investments and think out the risks and funds, hedge funds, private investments, mutual funds, separately-managed accounts? Managers – the risk managers take to kind of generate return in the risks that they're bearing and so doing. A lot of the research we've done is trying to understand as we evaluate the performance of risk of various investment products to understand, after solving for the risk investors take, is there any leftover alpha risk-adjusted excess return? Is it random? Is it real? What's the prevalence of those alphas across asset classes? What's the most important risk factors in various asset classes?
There's a very practical reason. We kind of, on an ongoing basis, evaluate really the entire universe of active managers available to us across asset classes. So we have a very good understanding of, for various asset classes, what common risk factors are the genuine common risk factors in traded portfolios?
As you kind of hinted at, there's an entire literature on mining factors from the historical data. And commonly, you'll see for equities, for example, market factors, size and value. Fama-French back in the early '90s published that factor model. Recently, they added profitability investment factors. Back in the '90s, there was also evidence of momentum factors. There's a zoo of factors in equity. In fixed income, there's term factor, default factor, or a credit factor, pre-payment risk factors. There seems to be this plethora or zoo of factors. But do we really need so many factors in real-world portfolios?
Because like I said, a lot of these tests have been done on portfolios using historical data that weren't necessarily live real-world traded portfolios. In the world I live in, where we try to evaluate real-world investment managers to understand the risks they're bearing and whether or not there's real alpha there, of this zoo of factors in the literature, which ones are genuine common risk factors in real-world portfolios?
Because what we found in our manager evaluation was not necessarily what you see coming out of the academic space nowadays anyway. And what's the prevalence of alphas? What we found was a couple of interesting insights. On the equity side, first of all, you can improve all factor models by moving to a different definition of the market factor and the CAPM. Its return of equities minus return of Treasury bills. What we did instead was we modified it to be an ICAPM definition. The return of equities minus the return of Treasury bonds is a pretty good proxy for an ICAPM definition of the market factor.
It turns out that that definition improved all factor models that we tested. It also turns out that we didn't really need to add any more factors to that factor model. The ICAPM by itself explained the way all alphas across the distribution of funds we tested since the early 1970s. That was an amazing outcome that basically said you didn't need to add size factors, value factors, profitability, momentum to explain away more alphas. That was interesting.
It did turn out that having size and value helped explain more systematic risk and more return valuation than just the ICAPM market factor. But they didn't actually explain away any more manager alpha when we evaluated real-world portfolios. So the best we can say is, in the market, there's really only one market factor that met all four of those criteria that I mentioned earlier. Size and value factors also helped explain some systematic risk, but did not explain away alphas. Didn't necessarily come with a reliable return premium. And then we really didn't see any evidence of either things like momentum, profitability, investment, either explaining away alphas. Not coming with verifiable return premia in live-traded portfolios or they weren't explaining any systematic risk.
For equities anyway, one genuine common risk factor in an ICAPM market factor, I would say size and value of secondary risk factors. They failed in terms of explaining away alphas, but they helped explain some systematic risk. The other factors, I'll just call them anomalies. They seem to be there in the historical data. We're just not necessarily seeing them manifest as verifiable return premium in real-world portfolios.
Ben Felix: I want to pause for a second and come back to the real estate paper that you mentioned earlier. Before I start with my questions on that, why is that one of the papers you don't like that much?
Peter Mladina: Partly because I think subsequent research that isn't cited as much, because it's a little less popular, was far better. What was cool about real estate betas and the implications for asset allocation, I think that what you're getting at is how do you explain re-returns from a factor perspective? And what that paper found was you can explain re-returns by unpacking them into or decomposing them into underlying factors. And it turns out with REITs, looking at the long history they have which goes back to the 1970s, re-returns are fully explained by exposure, market size, and value in stocks, and term and default in bonds.
What you can say is it's sort of a balance portfolio of small-value stocks and high-yield bonds. That in and of itself I think caught a lot of people's attention and they like it as sort of a simple explanation of what real estate is. And then what naturally comes out of that, well, does that also apply private real estate? And I think it's a great follow-on question, because it's a far harder one to answer. And a few years later, and what came out on some of my original thoughts on the paper you're asking about, real estate betas and implications of asset allocation was taking it to the next level, and a methodology I created in that first paper you mentioned, real estate betas, I actually evolved to the next level with a novel econometric technique to actually attribute factor risk across private asset returns. Not just private real estate, but buyout, venture capital, etc.
And with that factor, optimize lag, beta methodology, we're actually able to attribute alphas and betas across private assets, deal with all of the issues private assets have in the reporting. And what we actually found was the same thing, which was pretty cool. In private real estate, you can still see that there's material exposure to market size and value and has a high yield bond orientation.
Using a different, more sophisticated methodology, we essentially found publicly traded REITs and private real estate largely have the same underlying exposures. The beta is maybe a little bit different between REITs, core real estate, and opportunistic real estate, but that mix is the same. That set of datas are the same across real estate, which I think is kind of an interesting result.
Ben Felix: Interesting and pretty intuitive. Why would they be different?
Mark McGrath: I agrees. Super fascinating. What are the asset allocation implications of these fundings?
Peter Mladina: You're using factor models to evaluate real-world portfolios, so actual mutual funds, SMAs, etc. You start to realize, like I said, that the common risk factors at play in real-world portfolios, first, an ICAPM market factor is a slight improvement over a CAPM market factor to empirically understand return and risk of real-world portfolios. That's cool, and that loops back to ICAPM portfolio theory and constructing portfolios based on ICAPM portfolio theory.
Size and value factors, we will use them to evaluate managers for performance and risk evaluation because, like I said, they have a second-order effect in terms of explaining some systematic risk. Even though you give it enough history, they aren't necessarily explaining away alphas. And so, if you're looking at a manager's return over three or five-year periods, a lot of the return variation there may be because of size and value exposure. And so you definitely, over those horizons, want to include factor models that have size and value in them.
But over a long enough period of time, like necessarily seeing you're getting compensated in a verifiable way, above and beyond your ICAPM beta adjusted return. And then getting out to other potential factors, momentum, things like that. If you're evaluating a manager, that's a momentum strategy or something that has a quality bias or low beta anomaly, things like that. We may still use those models to understand what the managers trying to do. But at the end of the day, when you're constructing portfolios from an asset allocation perspective, and if these are portfolios for long-term investors, which hopefully they are, then you have to step back and recognize on the equity side. The best you can say is that the only factor that has a verifiable long-term return premium associated with it in a real-world portfolio, that also explains systematic risk, that is also not redundant, you have a single ICAPM market factor that takes you back to the ICAPM portfolio theory. Different applications for portfolio theory than what you might use for evaluating a manager's track record over a three or five-year period.
On the fixed income side, empirically, we found in our work three factors for bonds. The term factor, which you can think of as an ICAPM market factor. High-grade bonds that have just maturity or duration-related term risk. That risk you can do away with if you're duration matching in an ICAPM portfolio, your goal or liability profile. There is also credit risk in bond markets. We'll call it a default factor. Turns out though that the default factor is completely redundant with the market factor for equities. It's definitely there in the bond market. It's just as redundant.
But remember what I said earlier that one of the criterion of a genuine common risk factor is that it's not redundant with other factors. And so, although we might use a default factor in a fixed income model, we also have to be aware that it's redundant with the market factor for equity. So, as you get back to ICAPM portfolio theory, we don't want redundant factors. We can drop the default factor off there. There's another factor there in fixed income. It's a payment risk factor. Potentially, it has a small effect in our empirical test, but it's there. But also, more recently, it's not even statistically significant anymore. It might not even actually be there.
Again, if you take these asset pricing tests and we apply them to real-world portfolios, what you're left with is a term factor for fixed income and an ICAPM factor for return-seeking risky assets, not just equities. And we can take those two factors together and those are our two risk factors that we can use an asset allocation for portfolio theory.
That term factor essentially is our liability hedging asset. And we can make it risk-free relative to the goal, which is cool. And then we have our return-seeking risky asset portfolio, where the risk is defined relative to the liability hedging asset, where you're defining the risk there. We can take, again, this zoo of factors and boil them down to really just to that matter for asset allocation.
If our application is more than asset allocation, it's manager research or understanding the returns of a particular hedge fund or particular equity strategy or bond strategy, we may use more factors than that because we may have these short windows of time that we're also evaluating. But as you loop them back to portfolios for long-term investors, asset allocations for long-term investors, really all you need are those two factors; term factor and an ICAPM-based market factor.
Ben Felix: That's pretty crazy to think about considering all the factors that are out there.
Peter Mladina: Well, exactly. And what's cool here is now we're getting close to the holy grail of financial economics through the ICAPM potentially. And again, there's still more work to do, particularly in pricing assets at the security level. That work hasn't been done yet. But at least at the asset class level and the portfolio level, we have a unification of portfolio theory and asset pricing with the ICAPM, where theory and empirically, we can say that portfolios of assets and asset classes are essentially explained by simply two factors properly constructed.
Ben Felix: We use dimensional funds, which tilt towards small-cap and value and all that kind of stuff. What are the implications of what you just said for investors that are tilting toward those other factors?
Peter Mladina: With all fair disclosure here, I think Dimensional is an outstanding manager when you're trying to capture a factor-based premia. What's an approved management art platform here? I want to just step back and say, when you talk about fulfillment, there are really three approaches at the highest level. You can take a passive approach on the risky asset side. That's what we're talking about specifically here. If you believe markets are very efficient, you're going to take a very low-cost passive approach. It's expense-efficient. It's tax-efficient. You can do a good job aligning it with your goal profile. There's a lot of sense to do it that way and maybe that ought to be best practice.
At the other end of the spectrum, there's an active approach. It's going to cost you more money. It's going to be less tax-efficient. We didn't really get into this. But based on all our empirical tests, we've been able to show that, after expenses, there's not much evidence for alpha. Alphas are very random. The returns are well explained by some mix of factor betas. There's something like a factor-based approach. Maybe it's semi-active or semi-passive. To me, those are all confidence levels.
I think that if you step back, like I said for a moment there, in real-world portfolios, all you need is the ICAPM market factor to explain away all alphas. Now, size and value in particular still showed some return variation that explains some systematic risk. And tests of non-live portfolios, but just historical returns, we certainly see return premia and things like value stocks. Again, those are hypothetical portfolios. And what my research tries to do is bring it to the real world with real-world portfolios because there are real-world constraints.
And then there are other anomalies, what I'll call instead of risk factors anomalies, things like profitability and momentum. I'm not here to say that a very efficient quant manager can't pick up a return premium associated with value or momentum or profitability even. I'm just saying those things aren't rising to the level of a common risk factor that defines it the way we think it ought to be defined by meeting those four criteria.
Really, the way common risk factors were really originally defined going back to the 1990s or even the CAPM when people started thinking about trying to unify portfolio theory of asset pricing. I think what it takes is an efficient quantitative manager. You may be able to harvest these premia. Maybe these premias should be thought of as anomalies. Maybe they're there. Maybe they're not. So to me, it gets to confidence levels. The genuine common risk factor meets the set of criteria that I talked about. It's far more reliable going forward. It prices assets and portfolios of assets in a systematic way. It's not to say there aren't other return anomalies out there that an efficient quant manager can't capture. I just would argue they don't rise to the level of a common risk factor that systematically explains asset returns and portfolio asset returns that aren't genuine sources of compensated risk, so to speak. Confidence levels. It could very well be that these anomalies – a genuine common risk factor is far more likely to persist in the future than a return anomaly. But it's not to say that a return anomaly won't persist.
Ben Felix: At the most extreme end of active, how well are hedge fund returns explained by factor models?
Peter Mladina: Understanding the return and risk of hedge funds is the most complicated simply because the best hedge funds are different. If we look at hedge funds as an asset class, HFRI composite or something like that is a proxy for hedge funds as an asset class, it's very simple. Essentially, factor is doing a very good job of explaining the return and risk, whether you use a CAPM market factor, an ICAPM market factor, you will see that the average hedge fund return is nothing more than a low-beta exposure to the market factor, and that's it. That's why hedge funds have a 0.9 correlation to the market factor. It has a lower volatility than market return simply because it has a lower beta.
You can simply say, on average, the hedge fund return is 30% equities and 70% Treasury bills. Because you can run a simple regression against the market factor. You'll have no alpha. You'll have a 0.3 or so beta. Your portfolio is mostly cash and equities. Now, why you would pay in 2 and 20 for that? I don't know. That's just redundant. You already own equities and some optimal allocation to cash in your portfolio. There's nothing special about it.
As you get into other main categories of hedge funds, event-driven, macro, you might need a broader set of common risk factors to get there. In a prior work I did, you can even use quantitative factors. What hedge funds are doing kind of changes through time. In the earlier history, we saw more evidence of alpha since the tech bubble where, frankly, the data set of hedge funds is much cleaner than it used to be. There are biases in the data set and the index returns. And so, we don't really see on average or across the hedge fund categories really any alpha.
At that level, the category level, again, you might need a broader set of factors. But still, factors pretty much explain all the compensated return and risk. There may be some unexplained risk, but it's not coming with a verifiable alpha. It's uncompensated risk. You don't want to bear risk in your portfolio and not be compensated for it, which is why what we care about is bearing compensated risk, risk that comes with a return premium. People like to say risk and return are related. All risk comes with a return premium. It's uncompensated risk.
When we solve for exposure to compensated risks, common risk factors that we can otherwise easily own in the portfolio, if we want, is there evidence of managers with alpha? In particular with hedge funds, if we can find strategies. Because they can use leverage and go along short. And they're really isolating their alpha so that it's 90% of the risk contribution and return contribution to strategy, that's a valuable hedge fund strategy because it's kind of like a completely different asset class then, where you have a robust return premium, that's the alpha, that's statistically significant. It contributes 90% of the return and risk.
I think the hedge fund space, there's still opportunity there, but you better know what you're doing. And very few people do, in terms of finding those managers and strategies. You have to be very careful. And it's less and less as we move through time. But they are there done right, even for taxable investors. If it's this type of strategy that really isolates its alpha and that alpha is reliable. And I'm saying factor-adjusted alphas. Pretty every potential factor left is the alpha. That can be an interesting strategy, but they're very rare. Hard to find. They had to deteriorate.
Ben Felix: You kind of touched on this just now. Is there adverse selection? If a hedge fund is willing to take your money, is that a bad sign?
Peter Mladina: In the real-world, markets seem to be getting more and more competitive. As we look at our factor models and we move through time, our suares are going up across asset classes. Alphas seem to be random almost regardless of what factor model you use. And in hedge funds, we're using the techniques I described; factor-based analytics, and decomposing the risk. In the old days, we could kind of attribute it to bad data.
We're still finding a very small handful of hedge fund strategies that are interesting. Most of them are closed or give you sort of a limited allocation. They're out there, but accessing them might be a big, big challenge for the average investor. As they have opened up and gotten bigger, that performance isn't necessarily persisting. Even in many cases, it's not persisting anyway. It may still be there, but in a much muted version of what it was previously. I'm just saying, eyes wide open. It's a very difficult asset class, but it's doable. And done well, it can actually improve a portfolio, even a taxable portfolio. So I don't want to throw it out completely.
Mark McGrath: What explains the performance of the Yale endowment?
Peter Mladina:I haven't looked at that in 15 years. Frankly, it would be interesting to do it again with the evolving tool set we've developed over time, which is more sophisticated than what we did back then. I mean, what I can say back then is what we were able to do was largely replicate Yale's return using synthetic leverage, factor attribution, and most importantly, replicating the private investment portfolio, particularly the venture capital portfolio that Yale had employed over the course of its history. At least under Swensen.
Going back 15 years anyway, what we could say was Yale's outperformance is significantly explained by its exposure to private assets and venture capital in particular. Give them credit for being early investors in venture in particular, private investments and utilizing that methodology. As everyone tried to replicate Yale using the endowment technique, the success hasn't been what it was back in the day. But in a nutshell, I think we can attribute it to venture, particularly venture when it had that run up until that point.
Again, like I said today, it'd probably be interesting if I had the time to kind of utilize maybe more sophisticated techniques we've developed over the years since then to kind of revisit that question. I'm pretty sure, particularly with our ability to be able to decompose private asset classes into their underlying factor betas and the residual alphas, we could probably make even more progress in understanding the other terms today.
Ben Felix: I would read that paper for whatever that's worth.
Peter Mladina: It'd be interesting. I'm not sure we moved the needle in terms of research that gives us better solutions going forward. How can we improve retirement outcomes and wealth outcomes? Bridging academic research with practitioners has sort of been what I've been trying to do. And that type of research, that ultimately improves those outcomes, I think.
Ben Felix: We talked about your research on traded portfolios. Can you talk about this in the context of jointly testing factor models and manager performance, which is what that research aimed to do?
Peter Mladina: There are a number of ways to do these so-called asset pricing testing. All of these tests suffer from what's called the joint hypothesis problem. Any test of market efficiency depends on the validity of the asset pricing model, so that also any test of a valid asset pricing model depends on the assumption of market efficiency. So how do we overcome this so-called joint hypothesis problem? That all asset pricing tests, all tests of market efficiency or asset pricing have?
And the technique that we sort of innovated and used in the and the research that got to my claim on what the genuine common risk factors are was to essentially say that we're going to basically test this world or universe of real-world portfolios. We're going to test both the manager returns, the manager performance and various factor models. The distribution of alpha should be random in the real-world.
We're actually mitigating this joint hypothesis problem by simply saying that whatever asset pricing model gives us closer to the attributes of market efficiency, i.e. random alphas, that is the best factor model. And so, we're able to jointly test factor models and manager performance using this methodology. Whichever factor model gives us this attribute.
Our primary test metric was distribution of alphas. And maybe this alpha is manager skill. Or maybe it's missing factors. If it's a missing factor that we find alpha and we add a new factor and the alpha goes away, we know that that source of alpha was a missing factor. And then that's a genuine common risk factor. If we add a factor and the alpha doesn't go away, there's a problem to that factor because it's not showing up as a genuine market premium.
And so that's kind of the theoretical concept that underpins the technique we used to identify what the genuine common risk factors are. That's where I said in a nutshell, when we looked at evaluating fund returns, and we took a look at the ICAPM market factor, as we added other factors, the size factor, the value factor, the momentum factor, we did not explain away more alpha. We didn't see alphas being reduced as we added more factors.
And in fact, what we found was that we already had random alphas by just using the ICAPM market factor. We didn't need to add any more factors to explain away alphas. So that told us right there that it was sufficient to have an ICAPM market factor to actually have the attributes of an efficient market where there was no alphas. There was no unexplained return premia, whether those return premia was factors or a manager's skill. And ICAPM market factor was sufficient to explain away all alphas. Done. We don't need to add any more alphas.
Like I said, adding size and value helped explain some systematic risk. That was a secondary test statistic. So that's where I might say size and value are secondary factors. But they weren't necessary to explain away alphas. Contrast that to fixed income, because I think it's a good way to think about it, we did similar tests in a similar paper evaluating bond returns, real-world portfolios. Again, we started with a term factor and there was alpha everywhere. When we added a credit factor, the alpha significantly disappeared.
What looked like alpha, which is the term factor, was explained away when we added a default or credit factor. It turned out that that alpha, which is the term factor, was not manager-skilled rather. It was missing credit risk. Already, we know that the term factor by itself for fixed income is incomplete. We need a default factor to help explain away some of the alpha.
It turns out also, on the margin prepayment also helped explain away alphas, which is why we have a more complete factor model and fixed income with term default and prepayment. However, like I said, the default factor is redundant with the CAPM or the ICAPM market factors. We don't really need it because it's redundant with the market factor for equities. We're back to the ICAPM again.
So that's the interesting thing. We want to see factors explain away alphas in real world portfolios. If we can use market efficiency as a benchmark, all we need to find is the simplest factor model that explains away all alphas where alphas are just random. Because then we know that there's no unexplained return premia. No missing factors. No manager skill. We've basically achieved that ideal of an efficient market.
Ben Felix: Really, really interesting.
Mark McGrath: Okay. Last question for you, Peter. How do you define success in your life?
Peter Mladina: That probably deserves its own podcast, right? Professionally, the way I define success is making a difference. I hope my research makes a difference in investor outcomes. I truly believe the research that I've done, the ICAPM asset allocation methodology, the prevalence of compensated risk in alphas and portfolios is really good information that if used right, and hopefully with time, there'll be more and more tools to allow investors to be able to do this. Gives them better retirement outcomes.
In the world I live in, in wealth management, similarly, better wealth outcomes. Optimizing assets and portfolios to fund lifetime goals so they can do more with their assets in a more efficient way and that's more aligned with our goal. So to me, success professionally is making a difference in the world. Adding the body of knowledge in financial economics and improving retirement and wealth outcomes materially relative to maybe what they otherwise would have done.
I think personally, to me, success is living a life where I'm happy and healthy. At the end of the day, that's all I really care about is being happy and healthy. And then maybe a little bit of that rubbing off on friends and family, people I love. But I think that's how I define success both professionally and personally in a nutshell anyway.
Ben Felix: Great answer, Peter. And this has been a great conversation. We really appreciate you coming on the podcast.
Peter Mladina: Thanks, Ben and Mark. Lots of really great questions.
Ben Felix: Thanks, Peter.
Is there an error in the transcript? Let us know! Email us at info@rationalreminder.ca.
Be sure to add the episode number for reference.
Participate in our Community Discussion about this Episode:
Papers From Today’s Episode:
‘Real Estate Betas and the Implications for Asset Allocation’ — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3134732
‘An ICAPM Framework for Asset Allocation’ — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4319731
‘An ICAPM for Goals-Based Investing’ — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4943241
'Portfolios for Long-Term Investors' — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3790823
‘Yale's Endowment Returns: Manager Skill or Risk Exposure?’ — https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2959074
Links From Today’s Episode:
Meet with PWL Capital: https://calendly.com/d/3vm-t2j-h3p
Rational Reminder on iTunes — https://itunes.apple.com/ca/podcast/the-rational-reminder-podcast/id1426530582.
Rational Reminder Website — https://rationalreminder.ca/
Rational Reminder on Instagram — https://www.instagram.com/rationalreminder/
Rational Reminder on X — https://x.com/RationalRemind
Rational Reminder on TikTok — www.tiktok.com/@rationalreminder
Rational Reminder on YouTube — https://www.youtube.com/channel/
Rational Reminder Email — info@rationalreminder.ca
Benjamin Felix — https://pwlcapital.com/our-team/
Benjamin on X — https://x.com/benjaminwfelix
Benjamin on LinkedIn — https://www.linkedin.com/in/benjaminwfelix/
Cameron Passmore — https://pwlcapital.com/our-team/
Cameron on X — https://x.com/CameronPassmore
Cameron on LinkedIn — https://www.linkedin.com/in/cameronpassmore/
Mark McGrath on LinkedIn — https://www.linkedin.com/in/markmcgrathcfp/
Mark McGrath on X — https://x.com/MarkMcGrathCFP
Peter Mladina on LinkedIn — https://www.linkedin.com/in/peter-mladina-177194125/
Peter Mladina on SSRN — https://papers.ssrn.com/sol3/cf_dev/AbsByAuth.cfm?per_id=890472
Northern Trust — https://www.northerntrust.com/