EP 24: Behind the headlines: Private credit defaults, recoveries, and AI risk in 2026

For DLD, we have five events that we call a default. One is restructurings. Two would be filing for bankruptcy. Three would be missing your interest payment, but not curing it within the grace period.

Four would be a distressed exit, which is kind of another form of a restructuring. Right. And the fifth way would be if the fair value mark is actually at zero. That essentially is a default.

Okay. So what we do not include is covenant violations, nor do we include having a PIC component to it. Okay. So that we do not include.

Because in our view, if a borrower opts to pick a small portion of interest and pay the rest in cash, that's not a default. Welcome to Private Capital Call, the podcast with leaders in asset management, investing, and capital markets. I'm your host, Randy Schwimmer. This week, we're joined by Bill Cox and Eric Rosenthal from KBRA.

Bill serves as Chief Rating Officer, leading KBRA's Global Ratings Platform, and is widely recognized as the driving force behind the firm's private credit and private markets initiatives, establishing KBRA as a true market leader in private credit research and analysis. Eric is Head of Default Research at KBRA DLD, where he focuses on default analysis across private credit, leveraged loans, and high-yield markets. Many of our listeners will know Eric well. We had the pleasure of working alongside him at the Lead Left newsletter and have long relied on his monthly direct lending default report, the most comprehensive default publication in the private credit space.

We are very excited to have both of you here today on Private Capital Call. Welcome. Thank you very much. So maybe just talk a little bit about your different approaches, Bill, coming from the ratings perspective, Eric, from the research perspective in terms of how you work so that we can then frame that for our listeners.

So the data and research that we provide is in support of our credit ratings platform. We rate hundreds of middle market CLOs, credit facilities, BDCs, feeder notes and other structures. And our ratings are primarily for the purpose of institutional investors. And in order to arrive at those ratings, we are looking at loan performance, underlying borrower performance that serve as collateral inside those transactions.

So that's where our data is derived directly from the deals. And mine's a little bit different. So unlike Bill, I do not have the financials to back that up. Instead, I'm looking at the public BDCs, really focusing on the fair value marks to see and kind of coming up with our own index based upon purely direct lending names.

Fantastic. Fantastic. So we'll talk about how those approaches converge or not. But just let's take a step back.

We're here in late March, 2026, having almost since the first part of the year been inundated with all sorts of warning signs about private credit up and down the market. it's under great scrutiny. As I like to say, it's being held under suspicion of everything. And investors are looking for signs of cracks.

And we've seen, unfortunately, several research agencies and large banks coming out with very high default predictions for the year, up to 15% in some cases. And I always suggest that if that is the case, then the equities market is going to be hurting. So first of all, what are you seeing at KBRA? And maybe we can talk about your 2026 default rate forecast at the same time.

Yeah, sure. So I don't dispute the emerging cracks. The DLD group at KBRA, we have a watch list, which we call our default radar. And we highlight companies on default radar that are showing signs of stress.

Now, there are two tiers. We have our red tier and we have our orange tier. And the red names will show the deeper signs of weakness than, say the orange names. Now, the red list, it is actually at an all-time high of 157 borrowers.

Meanwhile, the default radar itself, it's up 16% over the past year, and the dollar volume has nearly doubled. But despite all that, we still don't see the 2026 default rate going much above 2%. We actually forecasted 1.5% for 2025, and that's exactly where we landed.

And out of how many total? Loans on there, we have 2,900. Wow, almost 3,000. Yeah, almost 3,000 names in our index.

And these, again, are direct lending names. Now, as you mentioned, Randy, other outlets are calling for much higher rates, some even saying the double digits, like you said. But if I can quote my ninth grade daughter, the math, it ain't mathing. OK, I like that.

Even if the whole list of the default rate are defaulted this year, and we're talking 232 companies, the rate would still only get to 5.5%. Wow, OK. And that's about as likely as, let's say, the Jets winning the Super Bowl next year.

See, unlike talking to your colleague, I don't respond to that because I'm in agreement with you. It's a lot of suffering, 50 plus years of watching this. So the big question is, why is that? And really, there's two things.

One, direct lending defaults, they take longer than syndicated loans or high yield. We have some names on our default radar that have been there for two plus years. And the second reason would be the direct lending universe it huge more than double the number of syndicated loans But most of the direct lending sphere is made up of these middle market companies And that means smaller defaults Now sure you going to have some big name defaults think Pluralsight But that not the majority of the market So why is there such a wide range?

Was it methodologies? You're existing, and maybe we sort of talk about this, but you're existing 2025 year end and then even going into the first quarter, it's a little bit higher than 25, but still versus 8%, I think Fitch was saying 8%. And then others, not even in the ratings business who are up much higher. Why is that?

I'll say one thing about that before Eric answers, and that is a rigorous methodology that's based on data was likely going to end up in the same place or a similar place as to where Eric ended up. And on the rating side, we have evidence of that. So we look at not even the same exact companies, although there is overlap, but a similar number, about 2,500 companies. They account for a trillion dollars of debt from direct lenders, private credit debt.

So it's a pretty large sampling across size, across sector. And our forward monitor, based upon looking at their financial statements that we received directly from their lenders and or from their sponsors is essentially a very similar number, 2% by par and about 3.5% by count. Wow.

Exactly the same number I'm forecasting for count. So independently, you come up. Same thing, yeah. Data-driven.

And I think the difference is a lot of times the research that we have seen where there are these much higher numbers, it's one of three things going on. They don't have data, but they want to be relevant. So they're extrapolating from some other place, maybe the BSL markets or some other. Second, the mechanics of the private credit market are just so different than the mechanics, especially in a situation of stress than the BSL market.

And again, extrapolation of that company missed this covenant, therefore it's at the fault is something that we're seeing a lot. And then, unfortunately, I think the media then reports on those. And then you have other research shops reporting on the media reporting on those. So you have this search.

Multiple conclusions that are inaccurate. Yeah. So that's the compounding that we're seeing. But I feel pretty confident that if you're looking at the actual data that these numbers that Eric referenced and that we're referencing on the rating side are the more accurate view of what default levels are actually occurring.

So I think even to the inexperienced observer, I think it occurs to one that methodology, must be behind some of these differences. So what you characterize as a default, for example, and I know that, I think it was Prostkauer who has big database and so forth, and there were others. I think Lincoln has historically had this as well, where defaults really when they refer to that or they're referring to financial covenant defaults, not payment defaults. And the media will pick up on a number and say, look at the defaults when they're actually just referring to a technical issue in the document.

So maybe if you could, how does KBRA define defaults? And then how do you see that different in some of the other folks that are reporting? Yeah, sure. So for us, for DLD, we have five events that we call a default.

One is restructurings. Two would be filing for bankruptcy. Three would be missing your interest payment, but not curing it within the grace period. Four would be a distressed exit, which is kind of another form of a restructuring.

And the fifth way would be if the fair value mark is actually at zero. That essentially is a default. So what we do not include is covenant violations, which you mentioned for like Lincoln, for example, nor do we include having a PIC component to it. So that we do not include because in our view, if a borrower opts to PIC a small portion of interest and pay the rest in cash, that's not a default.

Okay. And do you distinguish between good PIC and bad PIC? No, it's something I'm looking to do more of, but the reality is I've seen some names that are theoretically bad PIC just being added in the prior quarter. And yet they're still being marked at 98 par.

So the idea that that's a default would seem to me quite punitive. Others are just jumping to the bad pick definition and automatically assuming it's a default when you're going to kind of work yourself into trouble doing that. Okay. So that's super helpful.

Are there specific sectors? Because clearly AI and software this year is getting a lot of attention, but we forgot about retail last year and the auto space and so forth. I assume there are more cyclical sectors that you're also watching. Maybe you could talk about kind of industries that you're watching more closely than others.

Yeah, and spot on, Randy. So last year, we saw retail, obviously certain parts of manufacturing-related sectors, but those tend to be small in the landscape of private credit. We did see some problems in certain parts of the healthcare industry having to do typically with healthcare roll-up strategies, which went sideways because of changes in regulation But again relatively no large impact on the overall default rate We have isolated software in a number of recent research pieces including one that we published last week where we looked at 495 software companies, two things out of the gate.

One, we define what a software company is on a consistent basis. So it's not a matter of, as some media reports have said, well, this BDC is calling it a software company, this BDC is calling it something else. Our ratings analysts have a process for saying that business model is X sector. And so we know that these are consistently software companies.

Second, one thing that we thought was unfair is the concerns about AI being talked about as if they're suddenly just discovered or that all software companies are going to be negatively impacted by AI. The reality is when we looked at these 495 software companies, many of them are benefiting from AI. Their margins are improving, their product mix is different. So we developed a framework for high AI exposure, meaning risk versus low AI exposure, meaning potential opportunity.

And when we did a deep dive on those companies that had high AI risk, we found that it was no surprise to their sponsors or lenders that they were struggling already. Their financial metrics were already showing what you'd expect to be shown. And many of them were already either receiving additional investment or were being traded out of the portfolios that we look at. So when you guys put out the AI paper, which I heartily recommend to our listeners, are there things in there that make you cautious or more concerned about 2026 and 27 as we roll forward just given technological obsolescence or even just background radiation in the form of layoffs and revenue challenges for some of these companies that may be impacted more than others?

Yeah, for sure. The realization of a business model that's struggling in this environment that's most imminent is a maturity. So for a subset of those 160 or so companies that were high exposure, high risk to AI. We looked at those with imminent maturities, and it turns out about 41 of those were clearly companies that had not been able to refinance in the past several years because of ongoing struggles.

But the way private credit typically works in these environments, we saw that more than when we called the sponsors and or the lenders to get an update on those 41 companies with imminent majorities with high risk, more than half of them had already been moved out of the portfolio. The sponsors had already taken out the debt and that every single one of those roughly 21 companies so far, the lender experienced no losses in the restructuring, no economic loss in the restructuring or in the exit.

So that's one thing that was encouraging. On the other hand, AI as an impact on companies, not just in the software sector, but broadly, is going to be mixed. Some will benefit. We're seeing margin improvements in certain kinds of companies, but we're also seeing some companies struggle in a variety of ways.

And the last thing I'll say is that the current environment with higher rates, potential impact on inflation, impact of energy prices on some of these sectors, and the general economic uncertainty are much more ominous situations than AI is on the horizon. So we'll have the research that you put out and some of the information available to our listeners as well. But we have published for a long time this notion that the middle market is actually a better risk than the broadly syndicated market.

That data is not new. It's been out there for 30 years. I think it was Fitch who originally put it out. S&P followed.

You guys, in our view, my view, have now the most scaled and comprehensive data in direct lending. 2025, again, not a surprise, but the direct lending defaults were better than the syndicated loan market by a wide margin. You point to the reasons why, which we agree with, which is that the middle market lenders tend to be more collaborative working together than the broadly syndicated lenders who are not often the same ones in a more troubled situation. from the ones that started investing in that business.

And that's part of the challenges for liquid loan market. We see it again and again. In the liquid loan market and in other parts of even the bank market, interests diverge in a stress scenario. And in the private credit landscape, time and again, it seems that the sponsor and the lenders recognize what they each have at stake and work to an optimal solution.

And typically, the senior lender is in the most advantageous position and often walks away with the best part of what's remaining of the assets. And interestingly, the liquid loans in your forecast are doing worse than even high-yield bonds. Why is that? And part of that's the rating composition.

You know, high-yield, about half of it is double B-rated versus syndicated loans, which is much more B-minus or lower. So better risk. Yeah it is This looks to be the third straight year we going to have an under 2 I say default rate for high yields Wow Okay No discussion defaults would be complete without a discussion of losses and recovery rates And one of the challenges that I've seen maybe since COVID, but certainly in recent years, has been the decline in general of recovery rates on sort of leverage loans, but even in direct lending.

Why is that? And maybe talk about how investors in this current market, where there is a lot of misinformation out there, how should investors be thinking about the asset class in terms of not just defaults and so forth, but also the recoveries? Because that's ultimately what matters is how much money you get back. Yeah, sure.

So for me, the fault rate gets all the attention, but really the main focus and my biggest worry been the implied recovery rates. The days of 80% recoveries, those are long gone. implied recoveries in our index declined to 47% last year. That's down from 55% in 2024.

And unfortunately, I don't see much improvement this year. We're calling for about 50%, a little lower on actually an average basis versus a par-weighted basis. But I guess the good news is because the default rate is still relatively low, that the ultimate losses, loss given default rate, that's going to be low again. We're talking about 1%, that's what we're forecasting, versus the 0.

8% that we saw in 2025. And just for the benefit of our listeners, loss given default defined that for us. Yes. Default rate times 1 minus the implied recovery is the fun formula for that.

But it's basically taking those two aspects, default rate, implied recovery rates, and giving you a final number. So if your default rate was 10%, but the implied recovery was 100%, who cares? We'd be great, but that's obviously not the case. And why is it that recoveries have declined in the last five years or so?

Is it poor underwriting? Is it higher leverage? What's the... Well, the evidence is anecdotal so far, but clearly there was a vintage of valuations during a low interest rate environment that are probably going to be among the weakest recoveries on average.

We're also seeing because of that scenario, although there were some very, obviously, some very good companies that otherwise got into difficulty because of the leverage level that they got to in that environment. We're seeing a barbelling. We're actually seeing some strong recoveries for companies whose business models were actually not that bad, but the leverage was just too much in that valuation environment versus others where we're seeing large and some, a lot of small companies, particularly in the softer asset areas like software, they can switch to close to zero pretty quickly without much of a business model to lean on.

So as we wrap up our conversation, we are talking to, in these broadcasts, a mix of institutional and retail investors. It feels like the institutional investor in general has been through cycles before with private credit. Retail money is relatively new, just given the cycle. As we end this, if you could both give investors some hope and guidance in terms of how to look at information that comes out when it's dramatic headlines and so forth and it's not coming from KBRA, how should they, in their minds, screen those headlines and think about doing the homework themselves as to how to differentiate some of the better firms such as yourselves from some of the noise?

Happy start. Sure. So maybe it's the way I look at my forecast. It's a bottom-up approach in terms of going through information rather than doing the top-down approach.

Because if you start looking at all the headlines, there's a lot of fear going on there. But if you do the actual research and work, I'm looking at my forecast, I'm going through the 2,900-plus names in our index, one by one, trying to figure out will it default and when. Again, this is more art than science, but it's worked well with forecasts. And I think that for an investor would be the same approach as I would look to see what's maybe being held by the BDCs, specific names, and you'll get the right answers here.

And that's probably one of the reasons why the default rates are actually low. Modest. Yeah. I'd say remember to differentiate the various factors that are being discussed.

The reality is default rates are increasing, but they're still relatively low and so much within the capacity of funds and their various investment vehicles to absorb. They're designed to absorb defaults at much higher rates than we're experiencing, even in this relatively slightly elevated environment. Second is manager performance is going to become increasingly differentiated. So lean on your manager.

If your manager has a long track record, has been through cycles, expect to see their performance out. Do those who maybe were new to the game in recent years. So that is going to be, in our view, the primary transmission of the conditions that we're seeing. Manager performance will become increasingly differentiated, and that is important to look at as you're thinking about making additional investment decisions.

Bill Cox, Chief Rating Officer and Eric Rosenthal, Head of Default Research at KBRA. Thank you both for joining us on Private Capital Call. Have a nice day. Bye.