Great Ajax Corp (AJX) 2021 Q4 法說會逐字稿

Ladies and gentlemen, thank you for standing by, and welcome to the Great Ajax Corp. Fourth Quarter Fiscal 2021 Earnings Call. (Operator Instructions)

Thank you. Larry Mendelsohn, CEO, you may begin your conference.

Thank you very much. Thank you, everybody, for joining us for the fourth quarter and fiscal 2021 investor call. I appreciate you joining us. Before we get started, I just want to point out on Page 2 the safe harbor disclosure about forward-looking statements.

If we jump to Page 3, 2021 was a very productive year, and Q4 2021 was another good quarter as well, although there's a little bit of noise in the Q4 2021 income statement numbers that we'll talk about on this call. A significant increase in loan performance and loan cash flow velocity continued and has also continued into the first quarter of 2022 so far. This continuing increase in the present value of loan cash flow and the resulting decrease in unallocated reserve discount in excess of modeled expectations led to an additional acceleration of income on loans during the fourth quarter of $4.2 million.

In Q4, we purchased a significant amount of loans, both NPLs and RPLs, with certain specific characteristics in good locations and at low percentages of underlying property value. At December 31, 2021, we had approximately $84 million of cash and more than $300 million of unencumbered bonds and loans. The significant cash balance does create some earnings drag. And while the significant cash flow velocity from our mortgage loans and our mortgage loan JV structures increases income acceleration, it also rapidly pays down our loan and securities portfolio leverage, which can reduce ROE.

Our manager's data science guides the analysis of loan characteristics and geographic market metrics for performance and resolution pathway probabilities and its ability to source these loans through long-standing relationships very much enables us to acquire loans that we believe have a material probability of long-term continuing reperformance. We've acquired loans in 352 different transactions since 2014, including 7 transactions in the fourth quarter. We own 19.8% of the equity of our manager at a close to 0 basis, and we do not mark to market our ownership interest on our balance sheet or through the income statement.

Additionally, our affiliated servicer provides a strategic advantage in non-performing and non-regular paying loan resolution processes and time lines and also a data feedback loop for our manager's analytics. In today's quite volatile environment, having our portfolio teams and our analytics group at the manager working closely with the servicer is essential to maximize reperformance probabilities loan by loan by loan. We've certainly seen the benefit of this during the COVID pandemic and in Q4 2021 and in Q1 2022 so far with significant ongoing increase in loan cash flow velocity and credit performance.

Like our 20% equity interest in our manager, we have a 20% economic interest in our servicer at a low basis as well. We don't mark to market our equity interest in the servicer on our balance sheet or our income statement either. The data analytics and sourcing relationships of our manager and the effectiveness of our affiliated servicer also enables us to broaden our investment reach through joint ventures with third-party institutional investors and thereby invest in much larger transactions as well.

The servicers loan expertise is definitely appreciated by our joint venture partners as several of them now pay our servicer for providing third-party due diligence services for other transactions that they may work on. We still have low leverage. Our December 31, 2021, corporate leverage ratio was 2.4x. Our Q4 2021 average asset-based leverage was 2.3x, even though we made significant acquisitions in Q3 and Q4. We keep trying to increase asset-based leverage, but the significant cash flow from our loan portfolio keeps offsetting this.

We also have $26 million invested in Gaea Real Estate Corp., a REIT that invests in triple-net lease veterinary clinic properties, multifamily properties and multifamily repositioning loans. Gaea is managed by a subsidiary of our manager. We think Gaea has a great deal of optionality and that Gaea can grow materially. Many of the Gaea-owned triple net lease veterinary clinics have annual rent increases based on uncapped CPI. Additionally, several of Gaea's mezzanine repositioning loans also have equity participations in the underlying collateral properties.

On Page 4, we'll talk about some highlights of Q4. Net interest income from loans and securities, including $4.2 million interest income from the increase in the present value of cash flow in excess of modeled and unallocated reserves, was approximately $18.5 million in Q4 2021. Our gross interest income, excluding the $4.2 million from the increase in cash flow to our reserve models, was similar to Q3, but net interest income was $200,000 lower, primarily due to calling our 2019 C securitization and accelerating the amortization of remaining costs and paying double interest on the financing of the underlying loans for a period.

Interest expense was up marginally after subtracting the double interest paid from calling 2019 C. However, our average asset-based debt balance was $50 million higher for the quarter. A GAAP item to keep in mind is that interest income for our portion of joint ventures shows up in income from securities, not interest income from loans. For these joint venture interests, servicing fees for securities are paid out of the securities waterfall. So our interest income from joint venture securities is net of servicing fees, unlike interest income from loans, which is gross of servicing fees.

As a result, since our joint venture investments have been growing faster than our direct loan investments until Q4, GAAP interest income would grow more slowly than if we directly purchase loans outside of joint ventures by the amount of servicing fees, and GAAP servicing fee expense will decrease by the corresponding offsetting amount. Since we called our 2019 C securitization in November and have not yet resecuritized it, servicing fees increased in Q4 since we now hold all of the 2019 C assets as loans and related interest income will increase by the amount of the servicing fees.

An important part of discussing interest income is the payment performance of our loan portfolio. At December 31, approximately 72.3% of our loan portfolio by UPB made at least 12 of the last 12 payments as compared to only 13% at the time we purchased the loans. This is strong, even though in June, August and November of 2021 we purchased a significant number of NPLs in loans and in joint venture bond structures.

In our first quarter of 2020 investor call, we mentioned that we expected the COVID-19-related economic environment could negatively impact the percentage of 12 of 12 borrowers in our portfolio. Thus far, the impact on our regular payment performance has been much less than expected and the percentage of our portfolio that is 12 of 12 has been quite stable. Additionally, we have seen significant prepayment increases from a subset of impacted borrowers that experienced material increases in absolute dollars of equity and were in specific geographic markets.

These patterns, along with increases in housing prices and price stability in certain markets that we have concentrations in, helps maintain these payment and prepayment patterns and leads to increases in the present value of borrower payments in excess of modeled expectations and the related income recognition of $4.2 million of unallocated loan purchase discount reserves in Q4. Approximately 27% of our full loan payoffs in the fourth quarter were from loans over 180 days delinquent.

While regular paying loans produce higher total cash flows over the life of the loans on average, they can extend duration. And because we purchase loans at discounts, this can reduce percentage yield on the loan portfolio. However, regular paying loans generally increase our net asset value. They enable financing of lower cost of funds and provide regular predictable cash flow. Loans that are not regular monthly pay status tend to have a shorter duration. However, we generally expected that this duration reduction would be less than typical due to the impact of certain COVID-19 resolution extension requirements.

But as I mentioned earlier, most of our loans are purchased as nonregular paying loans and the borrowers, our servicer, and the portfolio team at our manager have worked together over time to reestablish these loans as regular paying. We also expect that given stability of housing prices so far that higher prepayments from property sales will likely continue for both regular paying and non-regularly paying loans. We do, however, expect that prepayment from rate-term refinancing will slow sometime in second quarter of 2022.

Prepayment shortens duration and increases the present value of collectability of the portion of the discount reserves in excess of modeled expectations. As I mentioned earlier, however, it also means less total cash flow and net interest income to be collected over the life of the loan, so a lower pool yield. Our cost of funds in the fourth quarter was effectively flat versus the third quarter. Given inflation and Fed rate increase expectations, we'd expect our cost of funds on adjustable-rate repurchase agreements to increase over time. However, we also expect to call several of our 2019 securitizations and resecuritize them at a lower cost of funds in the next few months. We've already called our 2019 C transaction.

Net income attributable to common stockholders was $7.4 million or $0.32 per share after subtracting out $1.95 million of preferred dividends. There are a couple of other things to note. Our acceleration of discount allowance related to credit performance in the fourth quarter was $4.2 million. Cash flow in excess of expectations continued to increase in the fourth quarter and so far in the first quarter of 2022. We expensed approximately $2.9 million related to the GAAP required fair value accrual of the warrant put rights from our second quarter 2020 issuances of preferred stock and warrants versus $2.5 million in Q3 of 2021.

We had about $100,000 in onetime reduction in net interest income due to the timing gap between buying the 2019 C loans out of the trust and calling the 2019 C bond structure. We also had approximately $367,000 of loss on debt extinguishment from the acceleration of amortization of remaining deferred issuance costs related to calling 2019 C as well as the repurchase of $1.3 million of convertible notes. We had a onetime expense of approximately $500,000 of tax consulting fees related to several acquisitions, new financing structures and transactions that we have closed or are currently working on.

We had a $200,000 increase in real estate operating expenses, primarily from some REO impairments. Absent transaction-related onetime items and the tax fees, earnings would have been approximately $0.37 per share. Book value was $15.92 at December 31 versus $16 per share at September 30. The difference in book value comes from -- primarily from the declaration of our $0.10 special dividend on December 31.

Taxable income was $0.40 per share. Taxable income in fourth quarter of '21 was primarily driven by continuing lower financing costs, increases in prepayment, especially for nonperforming loans and from continuing high cash flow velocity on performing loans. We saw many delinquent loans prepay in full and generate tax gains and 27% of all prepayments were in nonperforming loans. In Q4, we completed one joint venture structure in late November 2021, totaling $330 million in UPB of reperforming and nonperforming loans with approximately $760 million of underlying property value.

We retained approximately $55 million UPB in the form of debt securities and beneficial interest in this joint venture. We also purchased $148 million of residential reperforming loans, including approximately $100 million UPB from our JV partners, 66% share of the 2019 C structure, with UPB of $150 million and total owing balance of approximately $153 million at 54.1% of the underlying property value. Previously, our 33% interest in 2019 C was held as securities. We purchased another $10 million of NPLs and small balance commercial mortgage loans at approximately 50% of property value.

The loans we purchased in Q4, however, were on our balance sheet for only 52 days on average. Cash collections and cash on hand at December 31, we had approximately $84 million of cash. And for the second -- and for the fourth quarter, we had an average daily cash and cash equivalent balance of approximately $79 million. We had $86.6 million of cash collections in the fourth quarter, which is a 5% increase over the third quarter which was a 7% increase over the second quarter which was an 11% increase over the first quarter.

Our surplus cash tempers earnings and ROE, but provides us with significant optionality in today's volatile environment. As I mentioned earlier in this call, at December 31, we also had more than $300 million face amount of unencumbered securities from our securitizations and joint ventures and unencumbered mortgage loans. As of February 28, 2022, we had $76 million of cash.

Approximately 72.3% of our portfolio by UPB made at least 12 of their last 12 payments compared to only 13% at the time of loan acquisition. This difference creates material embedded net asset value versus our loan purchase cost basis. It also enables us to continue reducing our cost of funds and increase credit enhancement advance rates through rated securitization structures. This percentage is marginally lower than September 30, but that is not performance related is because we purchased a material number of NPLs in late Q3 and Q4.

On Page 5, purchased reperforming loans represent approximately 89% of our loan portfolio at December 31. Purchased RPLs represented 96% at June 30. We primarily purchase RPLs that have made less than 7 consecutive payments and NPLs that have certain loan-level underlying property specifications that our analytics suggest lead to positive payment migration. The positive payment migration of these purchased RPLs and NPLs regularly results in an increase in fair market value of the loans and a decrease in cost of funding over time.

On Page 6, we continue to buy and own lower LTV loans. Our overall RPL purchase price is approximately 46% of property value and 88.4% of UPB. We have always been focused on loans with lower LTVs and certain threshold levels of absolute dollars of equity and target geographic locations. In the current times of rising rates and the potential for market disruptions, this becomes even more important for RPLs and NPLs.

Purchased NPLs increased in the fourth quarter of 2021. For NPLs on our balance sheet, our overall purchase price is 90% of UPB and 55% of property value. Purchase price represents approximately 84% of the total owing balance, including any arrears. As a result of the low loan-to-value and higher absolute dollars of equity on average for our NPL portfolio we've seen that rising home prices have significantly accelerated prepayment and regular payment velocity on our NPLs as borrowers can capture a significant and growing equity. Under CECL, this leads to greater interest income from the acceleration of unallocated reserve loan purchase discount due to the increase in present value of collected unanticipated cash flow.

On Page 8, California continues to represent the largest segment of our loan portfolio. Our California mortgage loans are primarily in Los Angeles, Orange, and San Diego counties. We've seen consistent payment and performance patterns from loans in these markets. Performance in Southern California has far outperformed expectation during the COVID-19 pandemic. We've also seen consistently strong prepayment patterns. In Q4, California prepayments represent nearly 36% of all prepayments by UPB, even though it is less than 28% of our portfolio.

Florida prepayments have also increased significantly. We purchased a nonperforming loan portfolio of approximately $85 million UPB in late Q3 2021. All of the loans are in Miami-Dade, Broward, and Palm Beach counties in Florida. In Q4 and so far in Q1 of '22, these loans have far outperformed expectations. We've also seen demand in prices for homes and rentals increase materially in Phoenix, Dallas, Charlotte, Atlanta, and several other metro areas where we have concentrations. Until Q3 of 2021, we had seen this appreciation primarily in single-family homes and less so for condominiums. But in late Q3 and in Q4, we have started to see this materially in condominiums as well.

On Page 9, our portfolio migration. At December 31, approximately 72.3% of our loan portfolio made at least 12 of the last 12 payments, including approximately 64% that made at least 24 of the last 24. This compares to approximately 13% at the time of purchase and 74.5% pre-COVID in 2019. Non-paying loans, which usually have shorter durations than paying loans, received significant time line extensions as a result of COVID-19 moratoriums. This typically negatively affects yield on true nonperforming loans as extended resolution time lines can lead to more property tax, insurance expense, legal and other repairs. However, in the past 4 quarters and continuing so far into the first quarter of '22, we've seen the increase in prepayment of nonperforming loans shorten duration on average rather than extend duration.

Since we purchase most of our loans when they were less than 12 for 12 payment history, our servicer has worked with most of our borrowers over time. While it's too soon to understand the full impacts of COVID-19 on home prices and mortgage loan performance, so far, the impact on our portfolio has been positive. We've seen demand in prices for homes in our target markets increase, cash flow velocity on the loans increase and prepayment in full also increase. 12 for 12 loans in today's loan market still trade at materially higher prices than our cost basis. As a result, the fair market value of our loan portfolio and our implied corporate net asset value estimates are materially higher than our GAAP book value, which presents loans at the lower of market or amortized cost.

On Page 10, some subsequent events. We have agreed to purchase approximately $13 million UPB of NPLs and RPLs in 8 transactions subject to due diligence. The purchase price for the loans is approximately 99% of UPB and approximately 94% of the owing balance, including arrears, and approximately 46% of the underlying property values. On March 4, we declared a cash dividend of $0.26 per share, a $0.02 per share increase over our Q3 dividend paid on March 31 to holders of record of March 18, 2022. And in January, we exercised our pro rata right to acquire our percentage interest in Gaea Real Estate Corp.'s second private round of equity. The second round was done at an 18% premium to round 1. We do not mark-to-market our Gaea ownership but carry it based on cost with adjustments from equity accounting. We own 22.2% of Gaea Real Estate Corp.

Some financial metrics on Page 11. Average loan yields, excluding the accelerated income from unallocated discount due to the present value of cash flows in excess of modeled expectations, declined marginally by approximately 0.2%. Income that gets accelerated under CECL can reduce yield on the affected loans in the future as that unallocated discount gets captured earlier under CECL. For debt securities and beneficial interest, remember that yield is net of servicing fees and yield on loans is gross of servicing fees. Debt securities and beneficial interest is how our interest in our joint ventures are presented under GAAP.

As our JVs increase, as they did in 2020 and 2021 relative to loans, the GAAP reporting will show a lower average asset yield by the amount of the servicing fees. Leverage continues to be low, especially for companies in our sector. We ended Q4 2021 with asset-level debt of 2.2x and average asset-level debt for the quarter was 2.2x. Our asset-level debt cost of funds was 0.1% higher in Q4 versus Q3. This is primarily the result of double-paying interest as a result of a timing gap between the purchase of the loans and the calling of 2019 C structure.

As I mentioned earlier on the call, we have increased asset-based debt levels, but this has been offset by increased asset-based cash flow velocity that pays down asset-based debt. We are working on some new securitization structures that will result in lower cost of funds than some of our existing outstanding securitizations and intend to call and resecuritize several outstanding structures.

On Page 12, our total repurchase agreement related debt at December 31 was approximately $546 million, of which $228 million was non-mark-to-market mortgage loan financing and $233 million was financing on Class A1 senior bonds in our joint ventures. Repurchase agreement debt is our only floating rate debt. We expect our unencumbered securitization structures to return approximately $275 million of this into fixed-rate debt in the next few months. At December 31, we had $142 million face of unencumbered bonds as well as $139 million UPB of unencumbered equity beneficial interest certificates and $45 million UPB of unencumbered mortgage loans. Combined with $84 million of cash at December 31, we have significant resources for being on offense and defense in this unusual and volatile environment.

That's all the notes I wanted to bring up on this call. If anybody has any questions, we're perfectly happy to answer any questions you might have.

(Operator Instructions) Your first question comes from Kevin Barker with Piper Sandler.

So there's a lot of moving parts here around the funding side and the right side of balance sheet. Your funding costs have come down to roughly below 3% here, but I would expect slightly a little bit more rapid decline in interest expense. Could you talk about some of the moving parts with the calling of securitizations in these securitization structure? And then the movement in interest rates recently and how that's going to impact your expectation for interest expense going into, let's say, the first half of next year?

Certainly. So in early November, our 2019 C was a structure where we owned 1/3, and we had a number of partners own the other 2/3. In early November, we bought out their 2/3 interest, but we couldn't call the structure until December 25. So we bought their interest and bought out the underlying loans and then the cash sat in the trust's account to pay down the call on December 25. So as a result, we paid interest on buying out our partner's interest and interest on the securitization for that gap period of time, which makes it look like our cost of funds is higher because we paid more interest on a similar average balance, but we, in fact, double paid on -- for a period of time for that call because we bought out their interest 6 weeks before we called the deal. And the securitization itself had a much higher coupon than the rate that we financed the loans at. The loans -- we're financing the loans non-mark-to-market at 2.5% and the securitization at a 3.9% coupon.

And it's about 4.5% all on yield.

Yes. So we -- so double paying the interest is what really caused our cost of funds to not look lower in the fourth quarter. The -- going forward, our cost of funds on our repo has increased a hair so far. I mean, very little, if at all. We do expect that as Fed increases, it will go up on the asset repo basis point for basis point with Fed hikes. But we also expect that within the next 6 weeks, a chunk of that will be turned into fixed-rated debt. So we'll effectively lock in that financing for a long period of time. So the -- on the floating rate side, we should be down to sub-$300 million, about $250 million of floating-rate debt sometime in the next 6 weeks or so assuming markets don't come undone.

Okay. So taking that all into account, so what was your...

And by the way, that will actually lower -- and that would actually -- for the securitizations we call, our 2019 securitizations, the lowest coupon is $390 million. And for our -- we have a couple 2018s that are in the force. So our expectation is our new securitizations will actually lower cost of funds on our -- the deals -- the securitizations we call and effectively lock in fixed rates not too different than our current repo rates on loans for a long period of time. So it's definitely helpful to get these done.

Okay. So what was your run rate cost of funds, let's just say, at the end of the quarter?

At the end of...

End of 4Q. Because you were at 2.94% during the quarter, but obviously, there was a lot noise there. So like what's the exit rate basically on the year? And then can we put the pieces together to figure out change in repo costs combined with the change in securitization costs?

Right. So if you take out the double payment, our cost of funds in Q4 was the low 2.9%. So a little bit lower than Q3. And I would anticipate that when we -- the securitizations that we call and resecuritize will lower our cost of funds further. But to the extent we still have, say, about $250 million or $270 million of floating-rate debt, that would be subject to Fed rate increases.

So do you think...

So calling our 2019 securitizations probably lowers our cost of funds related to those by close to 100 basis points following the resecuritizing. We still would have $250 million or so of floating-rate debt that would increase with each Fed increase, assuming that doesn't get securitized or turned into a structured at some point.

So in total, how much -- how many -- how much of your securitizations do you expect to call and resecuritize in the next 2 or 3 quarters?

All of our 2019 securitizations that are on balance sheet, so -- other than 2019 D and 2019 F, which are rated long-term structures. Those are low cost...

Can you remind how much -- can you remind us how much the cumulative is?

So 2019 C is $150 million. We have another $100 million that we're adding to that. So our next securitization will be a little under $250 million. So call it about $400 million will be called between, say, now and June.

Okay. So $400 million will be called, but you're probably going to increase the balance, right, just given...

That will -- it will actually increase leverage. There'll be -- the securitizations will be effectively cashed out refis, but at lower cost of funds, securitization to securitization of about 90 to 100 basis points.

Okay. So despite Fed -- so putting that together, it seems like even if we have a 100 basis point move in Fed rates on the $250 million repo balance and the resecuritizations here should more than offset that, right? So you should see incremental declines in interest expense throughout 2022? Am I thinking of that right?

Yes. Our goal is -- yes, our goal is to be at least an offset. And the question is, can we get the -- get them all called, rerated -- or rated and resecuritized quick enough. So that's the goal, and the goal is to call them all by June 30.

Okay. And then on the interest income, what's your expectations for the slowing of prepayment and the impact on the yield that you're going to have?

Sure. So prepayment in January was actually higher than it was in Q4. Prepayment in February was approximately the same as it was in Q4. Our expectation is that the increase in mortgage rates won't really show up in rate turn refis until April or May as people rush to refi right now or they're using what they locked in January or February to close on their refis right now. So we would expect a portion of the prepayment to slow come April, May, but we're not anticipating that much slowdown in prepayment of nonperforming loans.

We certainly -- for that, we think you'd need to see negative HPA to really slow down materially. Yes. So the -- we do expect, however, that -- because CECL accelerates some of this that though -- as the loans keep paying, keep paying, keep paying, they have less discount now. So theoretically, that decreases the long-term remaining yield on a loan that's been affected by the CECL recapture. But we're also -- where we're buying loans now is cheaper than we bought them 6 months ago also.

(Operator Instructions) Your next question comes from Eric Hagen with BTIG.

Maybe a follow-up on the prepay question. Can you confirm that slower prepays would also probably support lower taxable earnings and, therefore, potentially support less need to pay a special dividend at the end of this year, all things equal? And then...

Sure. Slower prepays would decrease taxable earnings if the loans also stop paying regularly. So if they went from they were kind of not really paying and didn't prepay, that would decrease taxable earnings. But the taxable earnings from -- if they just went from not paying to paying every month, it would decrease it by a lot less. So one of the things we've seen is so much of the prepay has come from nonperforming loans. So if they stop prepaying because they started paying, that would have a much more limited effect than if they stopped prepaying and just said, come after me for the next 5 years. You know what I mean?

So it's -- it really depends on the pattern. Right now, what we're seeing is nonperforming loans or delinquent loans are either reinstating and coming current and paying or they're selling their homes and paying off. And it's really driven more by how much equity they have. We've seen that when they have between about $50,000 and $130,000 of equity, they're more likely to reinstate and pay regularly. And when they're above $130,000 of equity, they tend to be more likely to sell and put a couple hundred thousand dollars in the bank.

Interesting. I guess a couple more. I mean the line, I think, you hear from a lot of credit investors is that "everything is priced to perfection." And so I guess what would you identify as maybe the one or two things that offer upside to an investor in RPLs, which isn't already placed into the asset?

Sure. We -- our whole approach is very different. It's not about maximizing leverage and driving ROE and taking risk and playing the momentum game. Ours is much more about buying things in smaller transactions from many, many, many different sellers. I mean we've done transactions of $300,000. We've done transactions of $800 million with joint venture partners. So for us, it's all about being a value investor in assets that meet specific criteria in terms of location, geography, coupon, LTV, absolute dollars of equity and all kinds of other asset facts and collateral-related facts.

So as a result, the other thing is we buy assets at discounts, not premiums. So it's a kind of different mentality. We're much more, I would say, HPA affected than we are borrower FICO affected. So the risk in our portfolio is not a 10% decline in HPA or a 15% decline of negative HPA. It would be home prices go down 30% and that could have an impact on the credit risk. But for us, it's not about being price protection. It's being -- we are big believers in that offense is easy, defense is hard, so focus on defense and offense happens.

And that's kind of been our mentality, which is why if you look at our stock over the last 5 years for non-agency REITs, it's pretty much the #1 performer for 1 year, 2 year, 3 year, 4 year and 5 year. And a lot of it is our book value is higher than it was prepandemic unlike most of other people in our sector whose book values are lower than prepandemic. It's just a different approach.

Yes. That's helpful. I guess one more just on the REO impairments that you mentioned. Can you go into just a little bit more detail on that? Like what would drive an impairment in a period where home price appreciation is really robust?

Sure. So one of the things you see -- and some of it is GAAP accounting. So if you foreclose on a house, okay, if you foreclose on a nonpaying mortgage loan and it sells at the foreclosure sale to a third party, that's a loan payoff, not an REO sale, okay? So we have a much lower percentage of REOs that we take back. Our REO total has gone down dramatically. But to the extent that the REO, when you get it back, it needs more repairs than you modeled by a significant amount, you would have to make those repairs and then sell the property, and you could have a loss on that REO because of the damage that you encounter inside that property, okay? So it's really very REO by REO. We haven't had an REO impairment in a very long time. And most of this impairment came from a couple of properties that had significant internal damage from -- which you come across occasionally where a borrower just does it on purpose.

Yes. And it looks like -- just compare period over period, it looks like it's a bigger number. But in the fourth quarter of 2021, we actually had a recovery of structured impairments. So it looks like the change is bigger than the actual impairment...

So it's $200,000 different than the previous quarter, but not $200,000 in total.

There are no further questions at this time. I'll turn the call back to Larry Mendelsohn for closing remarks.

Thank you, everybody, for joining us on our fourth quarter 2021 and year-end 2021 investor call. Feel free to reach out to us if you have any additional questions. We're always happy to talk about our company and our business and glad to answer any questions you might have.

This concludes today's conference call. Thank you for joining. You may now disconnect.