Agency MBS: Generating Alpha without Credit Risk

Image

Stewart: Hey, welcome back to The Home of the World's Smartest Money. This is the InsuranceAUM Podcast, and my name is Stewart Foley, and I'll be your host.

And today we're here to talk about agency MBS. I love this topic because it is near and dear to my heart as a core fixed income nerd, geek, whatever you want to call me.

So we're going to talk about where the risks are, how investors get paid for taking them, and where active management may create opportunities for alpha.

And today our episode is entitled Agency MBS: Generating Alpha Without Credit Risk.

And today we're joined by Tyler Patla, who's a Senior Vice President, Managing Director of Core Fixed Income and Director of Agency Mortgages at Pioneer Investments, a Victory Capital Management franchise.

Tyler leads the agency mortgage team and helps oversee the firm's core fixed income investment professionals. Before moving to the asset management side, he traded agency mortgages at Mizuho Securities and RBC Capital Management, and earlier in his career managed mortgage positions at JPMorgan through the global financial crisis, which is the reason I'm gray.

He also has an interesting academic background for a mortgage investor. Tyler graduated summa cum laude from Duke with degrees in both economics and physics. So that is interesting. Structured securities and physics professionals seem to be good bedfellows, as it were. There's a decent amount of overlap.

Yeah, welcome to the InsuranceAUM Podcast, Tyler. We're glad to have you.

Tyler: Thank you very much, Stewart.

Stewart: So we always start them off the same way, which is where did you grow up and what's an interesting fun fact that your colleagues might be surprised to learn?

Tyler: So I grew up outside of New York City in a town called Westfield, New Jersey, where I was also the Blue Devil. So I was a Blue Devil in high school and in college. I guess I saved a few dollars on having to buy new gear.

Stewart: Oh, that's funny. Yeah, I never thought of that. We were the Owls. There's no chance of that happening for me.

Tyler: So I guess an interesting fact is that I was a big tap dancer growing up, and through my time when I lived in New York, I was a semi-professional tap dancer, which thanks to age and knees and location is not something I do too often these days except to teach my kids.

But that was something fun I did and isn't very closely connected to my current life.

Stewart: 400 podcasts, that's a first. That's a first. You know what? I got to be honest with you, I have always thought that is so fascinating.

And what a talented tap dancer can do, it is so interesting. And it was, I think, much more mainstream back in the day, but it's still super cool to do. And it's got to be very athletically challenging, right? I mean, you've got to have tremendous stamina to do that.

Tyler: Yeah, you can get pretty tired by the end of a difficult routine. And I liked being the worst person in class and learning from the people who can do those kinds of amazing things that you talked about, even if I always didn't do it perfectly.

Stewart: Yeah, no, I understand.

So let's talk about credit risk because there's an important distinction within the agency market. Ginnie Mae securities carry the full explicit guarantee and full faith and credit of the U.S. government.

While Fannie and Freddie securities don't carry that same explicit guarantee, you've made the case that agency MBS can nevertheless have credit risk comparable to U.S. Treasuries.

Can you walk us through your thinking there and why you think that's unlikely to change?

Tyler: Sure. So as you mentioned, Ginnies carry the full faith and credit of the U.S. government. Fannie and Freddie have that implicit guarantee from the U.S. government.

So really, when you invest in agency mortgages, your investment is backed by the homes that collateralize the loans in that pool.

You are backed by Fannie and Freddie's general insurance funds and the credit risk transfer that they've sold off to the securitized credit market since the global financial crisis, which effectively, you would need a housing crisis two to three times as bad as what we saw in 2007 through 2009 before Fannie and Freddie would be in anything close to what they were in at those times.

And even then, what we saw then and what we would see now is that the federal government realizes that the health of the mortgage market is essential to the health of the U.S. economy and really the global economy.

So even if we were in that kind of situation again, and we saw it coming out of COVID how one of the first things that the Federal Reserve bought were agency mortgages to stabilize the spreads in that asset class to help stabilize the housing finance industry because of how essential it is to the U.S. economy, it would be exceedingly unlikely that that support would not be there in the future if it's needed.

And because of structural reforms, it is exceedingly unlikely to be needed.

So I think there's never been a default on an agency mortgage security, not in the global financial crisis, not in COVID.

If a borrower defaults, unlike in securitized credit or investment-grade corporates, where that creates a loss for the investor, investors in agency mortgages get their money back at par.

So really, it's just about getting that money back and having the reinvestment risk, not the risk of getting any loss of principal to your initial investment.

Stewart: Yeah. And it's also like, I don't know, somehow in the fabric of the U.S. that everybody ought to own a home. And all of this is designed to make that more attainable by lowering borrowing rates.

But if you take credit risk off the table, there are still plenty of other risks that are there.

Mortgages are unusual because the borrower has an option to refinance, and borrowers refinance for a variety of reasons. It's not always interest rates that make that decision. It very well may be a relocation, or you might've bought a 40-acre goat ranch. Oh, that was me, sorry.

And that's interesting. So what risks are investors actually taking in agency MBS and how are they being compensated for those risks?

Tyler: Sure. So again, as you mentioned, the borrowers have that option to refinance, and that's somewhat unique to the U.S. mortgage market as opposed to mortgages in the rest of the world, is that they have a fixed-rate liability that they can either pay for the next 15 or 30 years, whatever the term of the mortgage is, or they can pay it all back next month.

And they're most likely to exercise that option in the event that interest rates fall.

So essentially, the investors are short a call option on the rates market to the borrower.

But what is interesting is that unlike in an equity options market or callable investment-grade corporates, that option is not optimally exercised.

So if you are drawing out the probability of exercise on an Apple call option when Apple has gone up a lot, you can be confident that it's 100% if the spot price is above the strike price and 0% if it's below.

But you don't see that, to your point, in agency mortgages because borrowers do not refinance or do not exercise that option efficiently in either scenario.

Even in the case where interest rates go down significantly, they've got a lot of incentive to refinance their mortgage, you usually don't see more than about a 50% annualized rate of exercise of that option.

Conversely, and this has become more prominent in the last few years as interest rates have risen, leaving the average loan well below the current mortgage rate, they will choose to repay that loan even though there is a disincentive to do so because they need to move for a new job or a growing family.

Or they maybe want to do a cash-out refinance, and resetting to a higher mortgage rate is still cheaper than having credit card loans at a 20% interest rate that they're paying month to month. To instead roll that into their mortgage ends up being the more efficient option.

So our job as investors is to capture that premium, that yield premium that is in the base case in the market when mortgages get sold to investors, and to optimize those returns, try to find the borrowers that are least efficient at exercising that option relative to their market price.

Stewart: Yeah, I think it's true. It's like you and I look at an OAS and you think about that call, and you know that you and I'd be, or you'd be more likely to be efficient with that call, but most people don't monitor mortgage rates and call their mortgage broker and, "Hey, can I refi today?"

So I think you're making a really, I mean, it's not only a great point, but I think it's how people behave.

And you can see it in the data, but you also know I'm not spending every day wondering if my mortgage is refinanceable, so I'm sure others aren't either.

But let's talk a little bit about the portfolio level. I mean, an insurance CIO has a lot of assets competing for space in the general account, and Treasuries, investment-grade corporates, structured, other structured products, and everything in between.

Can you talk a little bit about what role agency MBS plays in a multi-sector portfolio, and how differently can you construct that portfolio based on what the investor is trying to accomplish?

And let me tack on, given the discussion around liquidity, this is, I think, one of the biggest benefits of this asset class, if there was such a thing as option-adjusted yield on liquidity. It's kind of a strange concept.

Tyler: It's on the efficient frontier of not just risk-adjusted returns, but liquidity-adjusted returns.

So agency mortgages are the second most liquid asset class in the world by many measures, to U.S. Treasuries.

And even in some cases, the on-the-run agency mortgage, you could probably find more liquidity at any given time relative to different Treasury issues.

So the ability to move in and out at a low transaction cost is a great calling card for the asset class. Tens, if not hundreds, of billions trade on any given day.

So rebalancing a portfolio is really not going to have much of an impact the way it might in securitized credit or corporate bonds.

So we often see it as a bit of a Treasury surrogate in a lot of multi-sector portfolios where your base case yield is usually in the 100 to 125 basis points over Treasuries.

Because of that option cost, the long-term average is that agency mortgages return about 50 basis points over equivalent-duration Treasuries.

So we kind of see it as that 50 is your initial compensation for that complexity risk and that convexity risk, that interest rate volatility risk that you're willing to take to be in the asset class that will very likely accrue over a market cycle, although maybe not in any given month or even any given year.

And then what we do is try to use security selection to offset as much as possible of that, call it 50 to 75 basis points of additional option cost, to capture as much of that yield as possible to be a more efficient use of capital.

And to your other question, we can modify that to the duration or average life needs of a portfolio.

We can do that in the portfolios that allow interest rate hedging. We've executed a lot of strategies over the years for investors that have very short-duration expectations or ones with longer timelines and are looking more for long-duration surrogates.

And that can be either via active hedging or via structuring within the agency mortgage market.

So on the short end, that can be buying adjustable-rate mortgages, ARMs, or for short or long, it could be buying what are called collateralized mortgage obligations, which is where you take agency mortgage securities and then Wall Street cuts up those cash flows either on the front end or on the long end, a floating rate and an inverse floating rate, whatever would best benefit the client and give them the targeted return timeline and yield that they're looking for with as much or as little variability in that timeline that they're willing to stomach to perhaps get additional yield on top.

Stewart: It's interesting. There's a line in your materials that caught my attention: "Passive agency MBS doesn't exist."

And it's somewhat provocative because some investors certainly buy strategies that are marketed as passive agency mortgage exposure.

Can you explain why you think that it doesn't exist?

Tyler: It is technically impossible.

So if you wanted to buy a passive S&P 500 fund, that's easy. The fund just buys a pro rata share of 500 securities and you're going to be left with something that is substantially the same as the index.

For corporate bonds, it might be a little more difficult. There's a few thousand securities that are represented in the investment-grade corporate index.

For agency mortgages, there are over one million different unique securities in the agency mortgage market that the index tries to represent via about 1,000 synthetic instruments that you can't go out and buy.

So many securities that we own in our portfolios, we're the only one that owns them. There's no one else in the world that can have those securities.

And within a cohort that are represented by those thousand instruments, individual securities can vary in return by 1%, 2%, 3% in a given year.

So when asset managers promise passive, what they're really saying is we are not attempting to exceed the returns of the index, but tracking error in practice tends to be pretty similar to an active low-tracking-error strategy like the one we try to execute that doesn't rely on macro bets and is really trying to use finding the borrowers that are least efficient at exercising that option to generate excess return.

So that passive, really what you're getting is about 30 to 40 basis points of annualized tracking error historically. You're just locking in slightly below-index returns.

So we think for a couple more basis points of implicit fee, and what is really a very competitive product from the asset management perspective, seeking to produce excess returns over a longer period by realizing we can't own a pro rata share of every security, so we might as well buy a representative sample, but ones we think that are better on the margin for the investor.

Ones that are going to be less likely to give your money back at par when interest rates fall and prices rise, and are more likely to give your money back at par when interest rates rise and prices fall below par, are going to be a better return for the investor over a market cycle.

Stewart: The reason I'm smiling is because this is something that I've never... Well, I mean, listen, not that, who am I, but when we ran mortgages, this security selection idea wasn't something that we discussed.

And you describe agency MBS as the consummate, and I'm quoting, "consummate security selection asset class," which is so interesting to me, and it does show you how uninteresting I am.

I'm like, wow, I never gave thought to that because for those who don't know, each pool of mortgages is unique.

Those mortgages, like my mortgage is sitting in a pool someplace, but it's not sitting in 22 of them, it's sitting in one.

And so can you walk us through how you approach security selection from both fundamentals and technicals?

Tyler: Sure. So when you go to take out your mortgage, whether that's with an online financial technology company or you go to your local credit union or your local bank to take out a mortgage, most likely that loan will end up in a pool either solely of loans from that originator, that bank or that fintech company, or mixed in with ones from other originators that has then been compiled by Fannie Mae or Freddie Mac or Ginnie Mae.

And the agencies publish statistics, average statistics and loan-level statistics about those borrowers.

Not your address, not your age, not explicitly your income, but things like the loan-to-value ratio, what state you're in, what your loan purpose was.

Was it a purchase loan? Was it a refinance? Was it a cash-out refinance? Is it your primary residence or is it a vacation property?

Obviously, we know how old the loan is based on how long it's been since origination, the size of the loan, which is often one of the biggest determinants about how efficient you are.

Someone with a much bigger loan has more incentive to be efficient with their refinancing or their prepaying decisions than someone with a much smaller loan.

So we end up having, for these million securities, for these 40 million underlying loans, we have a massive database that has about 50 independent variables, which are the characteristics of the borrowers, and then the level of prepayments, the dependent variable.

And what our task is to try to sift through all that data, separate the signal from the noise, understand that the mortgage market is constantly changing and evolving as home prices accelerate or decelerate or even go down in some markets, as the origination process changes, as rules change with the agencies, the amount of fees that have to change, and try to isolate the unappreciated variables in the market that aren't reflected in market pricing that we think are going to lead to bonds that will perform better in the ways that we alluded to before, because the market gets more and more efficient.

Forty years ago, when our predecessor firm was founded, those academics were some of the first to discover what we now think are basic things like large loan sizes prepay more efficiently than small loan sizes.

And as smart folks have sifted through this data for the past 40 years, the market's become more and more efficient.

Originators will separate their loans by borrower characteristics to get paid more by investors.

So it's our job to find the characteristics or the combination of characteristics that are underappreciated by the market.

Stewart: That's interesting. I did not know that. That is really interesting.

So here comes the physics part of your education. You spent some of your career building mortgage pricing and valuation models, which is not something that a lot of folks can say they've done.

So I can't let you out of here without talking about AI. So you've got 50 variables that help you separate the red ones from the green ones.

Agency MBS has an extraordinary amount of data, and you mentioned some of it: borrower characteristics, prepayment behavior, rate volatility, markets, technicals, valuation, all of that.

Can you talk a little bit about how you think AI can change or has changed the MBS market and the way you manage money? Because I suspect that the kind of analysis you can do has just gotten more interesting.

Tyler: I think it works from both sides. I'll touch upon the asset management side first and the type of analysis that we do, and then I'll also touch upon the potential evolution of the borrower-facing mortgage origination market.

So for the type of analysis we do, AI has really expanded the amount of analysis that can be done.

Instead of running regressions and having them potentially come out with interesting solutions or not on an ad hoc basis, I can subscribe our AI agents to go through iteratively hundreds if not thousands of combinations of variables and subsectors and cohorts and maybe help me rank the ones that are most interesting and worthy of further analysis.

So that has iteratively allowed us to dig a lot deeper than we might've been able to a few years ago.

We also developed an in-house model using machine learning, so a little bit pre-AI, but some of the same techniques created a random forest model that would allow us to predict the bonds that would have high social scores.

So Fannie and Freddie published social scores that use some of the already published data, like the geography and first-time homebuyer, but some things that aren't published, like whether they live in an underserved community or the race of the borrower if they're an underserved race.

And they don't publish it for smaller securities because that would threaten the anonymity of borrowers.

So there's no way that I could determine exactly which pool the 40-acre goat ranch that you took a mortgage out on would be in as a way to protect their anonymity.

So part of that led into not publishing these social scores, which, regardless of how you feel about ESG investing, there's some of our clients that care a lot about that, there's some that don't care at all.

It is a predictive measure because higher social scores tend to be less, or tend to be slower prepayers, both in the low-interest-rate and the high-interest-rate environments.

So we are able to use that random forest model to use the data that we had available to accurately predict the social score for these smaller pools that did not have social scores published.

And we used in-sample and out-of-sample data on the larger pools to confirm that. And then we used the prepayment behavior for the ones that didn't have the scores published to confirm this as well.

So it's just an example of one way that we just vetted a bunch of data and we allowed it to select the combination of variables that would be most predictive, and then we found that it worked, testing it out of sample in two different ways.

And then what is getting a little more notice in the news media these days, and that you may read in some Bloomberg articles or even Financial Times, is the expected overhauling of the mortgage origination market because of AI.

We saw some new efficiencies in COVID. Some things were allowed to be done in an automated or model-based fashion that hadn't been done before, perhaps confirming that your home value hasn't depreciated so much that you couldn't refinance your loan when similar homes had sold in the area.

But using AI at these originators, they've all talked about how it's made them more efficient.

It's reduced their closing timelines from six weeks from the time you apply for a loan to the time it can close, to four or even three for some of the more efficient ones.

And this has made some investors concerned that this will increase the efficiency of borrowers in the mortgage market.

And then additionally, that it will be reflected in market pricing.

So all the borrowers may be more efficient at refinancing that loan, but because they're more efficient, investors will demand more compensation, which may lead to higher mortgage rates for the borrower.

So instead of paying 7%, you may now be paying 7.25% or 7.5% because investors know the day you move in the money, you're going to get a push notification on your phone.

It's going to be able to automatically pull your pay stub from ADP and use a model to confirm the value of your house, and a week later you have a new mortgage at a much lower fee level than you might've had a few years ago.

So that's something that's going to happen over time.

It's more likely to be a factor for loans that have not yet been created and mortgage pools that have not yet been created than for ones that are already in portfolios for investors.

So it's something that we are staying on top of and something that we think will be more incremental than overnight.

But it does lead into a lot of the analysis we do and will potentially increase the value of security selection because you have the mortgage originators and mortgage servicers that are early adopters and efficient adopters of this new technology relative to the borrowers that go to their local credit union that fill out the form with the little blue thing underneath so it automatically makes a copy underneath.

Loans that we can identify that were originated in that way are already less efficient at exercising that option, and that difference will probably expand over time.

Stewart: That's been a great education, man. You really delivered on the title of this podcast.

There's so much more than credit risk, and you really did a great job of going over it and explaining why it is, which makes perfect sense.

So really appreciate you being on.

Tyler: Thank you, Stewart.

Stewart: Fun one for you on the way out the door. It's a modified version of what we'd asked before, which is this one goes, describe your best real or fantasy dinner.

The guests, and what are we having? Where are we going?

Everybody's kind of been naming the guests, but nobody's kind of gone into the rest of it.

So who's coming to dinner with you?

Tyler: So I'm sure I could have a good time if I brought back some inspiring creative minds from the past, and I could signal my sophistication by saying I'd love a dinner with Leonardo and Van Gogh and Stephen Sondheim.

But I think my honest answer is the sappy one in that my father passed away about 15 years ago, so he never got to meet my two children who are eight and five, and they've been asking a lot of questions about him.

And I think it would be really cool if the three other people at the table were my two kids and my father, and they could get to meet each other and I could get to thank my dad for being a role model in ways that I hadn't been able to act on while we were still together.

But what we would eat—

Stewart: Yeah, it's very touching. I can relate to that one. That's really nice.

Tyler: I don't really care what we eat at that point.

Stewart: Are you cooking?

Tyler: Maybe since we are in North Carolina, we'll get some nice Carolina barbecue and some mac and cheese, so even if the kids don't want the barbecue, I'm sure they'll eat the mac and cheese.

Stewart: Are you down with that Carolina mustard-based sauce or are you more of a—

Tyler: Yeah, I'm more of a mustard or vinegar. I appreciate it. I do enjoy my Texas or Kansas City barbecue from time to time as well. I will admit that.

But if you're not down with the vinegar-mustard-based, then you probably aren't allowed to live here, so I'll say I can appreciate both.

Stewart: Yeah, there you go. Texas is all about the brisket. I lived in Chicago, love it, there's great barbecue places there, but holy smokes, Texas is unbelievable.

So listen, thanks so much for being on. We've been joined today by Tyler Patla, Senior Vice President, Managing Director of Core Fixed Income and Director of Agency Mortgages at Pioneer Investments, a Victory Capital Management franchise.

Tyler, thanks so much for being on.

Tyler: Thanks so much. I appreciate it, Stewart.

Stewart: My pleasure. Thanks for listening.

If you like what we're doing, please rate us and review us wherever you listen to your favorite shows.

And then we've got a YouTube channel at InsuranceAUM Community. Check us out, let somebody know if you like what we're doing.

We'll see you next time on the InsuranceAUM Podcast.

The InsuranceAUM podcast is available on

Share this post

Contacts

---

Pioneer Investments

Image

 

Pioneer Investments manages $147 billion in assets and has a long-standing history of innovation with deep expertise managing fixed income portfolios and creating customized solutions within the more opportunistic areas of the securitized market.

Pioneer Investments’ culture of innovation, in the securitized market, originated at Smith Breeden, where its founders developed early option-adjusted spread modeling techniques for MBS valuation. The innovative approach continues under Victory Capital, which manages over $10.7 billion for insurance companies. We are focused on delivering competitive risk-adjusted returns, while considering the accounting, regulatory, and capital management needs of our insurance clients to create long-term partnerships.  We understand the unique needs of insurers, and we provide customized and efficient risk-based capital solutions that align with insurers' risk tolerances and investment objectives.

Source: *Pioneer Investments, a Victory Capital Investment Franchise, as of June 30, 2026
 

Jay Alexander, CFA, CAIA
Managing Director, Institutional Markets
jalexander@vcm.com
+1 (612) 965-5426

 
Emma White
Director, Institutional Markets
ewhite@vcm.com
+1 (617) 422-4569

Marko Komarynsky
Director, Institutional Markets
mkomarynsky@vcm.com
+1 (210) 697-3613
 

View the contributor page