MFA Financial, Inc. (MFA) Q1 2015 Earnings Report, Transcript and Summary

Ladies and gentlemen, thank you for standing by, and welcome to the MFA Financial, Inc. first quarter earnings conference call. [Operator Instructions] And as a reminder, today's conference call is being recorded. I would now like to turn the conference call over to your first speaker, Danielle Rosatelli. Please go ahead.

Good morning. The information discussed on this conference call today may contain or refer to forward-looking statements regarding MFA Financial, Inc., which reflect management's belief, expectation and assumption as to MFA's future performance and operation. When used, statements that are not historical in nature, including those containing words, such as will, believe, expect, anticipate, estimate, plan, continue, intend, should, could, would, may or similar expressions, are intended to identify forward-looking statements. All forward-looking statements speak only as of the date on which they are made. These types of statements are subject to various known and unknown risks, uncertainties, assumptions and other factors, including those described in MFA's Annual Report on Form 10-K for the year ended December 31, 2014, and other reports that it may file from time to time with the Securities and Exchange Commission. These risks, uncertainties and other factors could cause MFA's actual results to differ materially from those projected, expressed or implied in any forward-looking statements it makes. For additional information regarding MFA's use of forward-looking statements, please see the relevant disclosure in the press release announcing MFA's first quarter 2015 financial results. The discussion today also contains certain non-GAAP financial measures. Information relating to comparable GAAP financial measures may be found in the first quarter 2015 earnings presentation slide, which has been filed with the SEC and are posted on our website at mfafinancial.com. Thank you for your time. I would now like to turn this call over to Bill Gorin, MFA's Chief Executive Officer.

Thanks, Danielle. I'd like to welcome everyone to MFA's First Quarter 2015 Financial Results Webcast. With me today are Craig Knutson, MFA's President and Chief Operating Officer; Gudmundur Kristjansson, Senior Vice President; Steve Yarad, CFO; and other members of senior management. In the past 12 months, we continued to build a more robust business strategy. MFA converted to a holding company structure, which increased our investment flexibility. We built out a team and the analytics for whole loan investments. As a result, we're able to expand our investment focus across a wide range of credit-sensitive residential mortgage assets. In the first quarter of '15, we've generated net income of $78 million or $0.21 per common share. The dividend was again $0.20 per share. Book value per common share was $8.13, and based on continued good performance within our credit-sensitive Legacy Non-Agency assets, we again transfer significant amounts, approximately $22 million from credit reserve to accretable discount. Now turning to Page 3. Despite the low interest rate environment, we continue to identify and acquire attractive credit-sensitive residential mortgage assets. It's been approximately 9 years since the last federal funds rate increase. The unemployment rate has declined and may continue to decline in 2015, yet the labor force participation rate remains low, inflation remains low in the U.S. and it borders on deflation in Europe and Japan. Interest rates remain low across the yield curve on a global basis. Commodity prices are weak and the strong dollar is impacting U.S. companies. As a result, future U.S. Federal Reserve actions will remain dependent on incoming data but we remain positioned for more flexible monetary policy by the Federal Reserve that will be responsive to measures of labor market, indicators of inflation, international developments and other economic data. We continue to limit the interest rate sensitivity of our portfolio. Our net duration as of the end of March was 0.53, our leverage ratio was 3.3:1 and 72% of our mortgage-backed securities are adjustable rate, hybrids or step-up. Turning to Page 4. In the first quarter, we continue to identify attractive investment opportunities across the residential mortgage asset universe. We increased our holdings to securities backed by reperforming/nonperforming loans to approximately $2.3 billion while increasing our holdings and credit-sensitive residential whole loans to $387 million. In addition, we increased our holdings of credit risk transfer securities to $127 million. On the other side of the ledger, we opportunistically sold $11 million of Non-Agency MBS issued prior to 2008, realizing gain of $6.4 million. This is the 11th consecutive quarter we've realized gains through selected sales of Non-Agency MBS based on our projections of future cash flows relative to market pricing. Again, we did not acquire any Agency MBS in the quarter. Now turning to Page 5. As you can see, MFA's yields and spreads remain attractive. Despite the interest rate environment, our net interest rate spread actually trended up. Now we've broken this slide into comparison of the first quarter '15 to a fourth quarter of 2014 on both the GAAP and non-GAAP basis. As you may recall, prior to 2015 when we acquired and repo-ed Non-Agency assets with the same counterparties, we account for these assets and related repurchase agreements as Linked Transactions. This effectively reduced MFA's assets and liabilities as what caused the difference between our GAAP and non-GAAP numbers for the fourth quarter of '14. Starting in 2015, due to the adoption of updated accounting standards for repurchased agreements, Linked Transaction accounting will no longer apply. No more Linked Transactions. And accordingly, an additional $1.5 billion of Non-Agency assets that we own are now reflected on our balance sheet. Just like to point out that if you look at non-GAAP fourth quarter and the first quarter '15, the debt to equity ratio remains the same, 3.3:1 Turning to Page 6. We present the yields and spreads for a more significant holdings. Given the leverage we are currently utilizing or may utilize in the future, each of these asset types are generating attractive returns to MFA shareholders. On Page 7, we illustrate how our holdings have evolved through 2014 and through the first quarter of '15. We've incrementally increased our holdings of more credit-sensitive but less interest rate-sensitive residential mortgage assets over this time period. Turning to Page 8. Our book value was basically flat during the quarter. Our net income of $0.21 per share exceeded our dividend of $0.20 per share by $0.01. Now turning to Page 9, Gudmundur Kristjansson will give an update on MFA's interest rate sensitivity.

Thanks, Bill. On Slide 9, we see the interest rate sensitivity of MFA's assets and liabilities as measured by interest rate duration. The duration of our assets declined by 15 basis points from 1.45 at the end of the quarter, primarily because of acquisitions of RPL and NPL MBS, which exhibit very little sensitivity to changes in interest rates and to continuing seasoning of the rest of our MBS portfolio. On the hedging side, $410 million of short swaps matured in the quarter, resulting in our lengthening of our interest rate hedges by 20 basis points to minus 3.8 average duration at quarter end. In aggregate, the net duration of our portfolio was 53 basis points at the end of the first quarter, almost unchanged from 56 basis points at the end of last quarter. Given the increased interest rate volatility we have observed in the last couple of quarters as well as the increased uncertainty regarding monetary policy at the Federal Reserve is now fully data-dependent, we are happy to be operating our company with a very low interest rate duration as well as very little sensitivity to the long end of the curve. With that, I will turn the call over to Craig, who will discuss our Non-Agency investments.

Thanks, Gudmundur. On Page 10, we purchased approximately $500 million of RPL NPL mortgage-backed securities in the first quarter while experiencing paydowns of approximately $200 million, thus, growing its portfolio by a little over $300 million. We particularly like these assets due to their low sensitivity to interest rates, what we believe to be low credit risk while at the same time, providing low double-digit ROEs. Their low duration is due primarily to the 300-basis-point coupon step-up after 3 years, which gives us confidence that they will have, at most, a 3-year final. In fact, most similar deals have been called well prior to their 3-year soft final. Now while the underlying loans certainly have credit risk, the combination of significant credit support, approximately 50%, and the locked down nature of the cash flows on the subordinate piece ensure that the senior bonds are well protected, and this credit protection increases as the deal seasons. Turning to Page 11. The credit metrics on the loans underlying our Legacy Non-Agency portfolio are very solid. Home price appreciation and principal amortization, approximately 60% of the underlying loans are now amortizing, both of these combined to reduce the LTVs. Delinquencies also continue to decline. 60-plus day delinquencies as of March 31 for the portfolio have declined to 14.7%. The loans are, on average, 108 months or 9 years seasoned. And as indicated in our press release and as Bill mentioned earlier, we again lowered our estimates of future losses in the portfolio and we transferred over $22 million from our credit reserves to accretable discount. All else equal, this increases the yield and we will recognize over the remaining life of these bonds. Turning to Page 12. On Page 12, we illustrate the LTV distribution of current loans in the portfolio. In particular, we focused on the at-risk loans, where the homeowner owes more on the mortgage than the property is worth. Although these mortgage loans are current, these borrowers are not delinquent at this time. These are the loans that we worry most about transitioning to delinquent in the future because of the fact that the borrowers are under water. As of March 31, less than $200 million face amount of current loans had LTVs over 110%. This is only about 4% of the current loans. On Page 13. We show realized losses experienced on the portfolio over the last 3 calendar years and for the first quarter of 2015. Keep in mind that these realized losses are fully expected and precisely why we have established the credit reserve. Note that after realized losses of $164 million in both 2012 and 2013, losses in 2014 decreased to $90 million. During the first quarter of 2015, realized losses were $20 million. Now the $874 million credit reserve has already been reduced by all actual losses already realized. At one point, these credit reserve was as high as $1.5 billion. These realized losses occur when the property securing mortgage loans are liquidated for less than the outstanding loan amount. In addition, for many of the fixed rate bonds in the portfolio, unrealized losses are generated when mortgage loans are modified through coupon reductions to troubled homeowners. While the loan modification reduces the interest rate paid by the borrower, the bond that we own has a contractual fixed rate coupon. So the interest collected from the borrower may be less than interest owed to us to the bondholder. In order to cure this interest shortfall, the trustee uses principal receipts to pay interest on our bond. This use of principal to pay interest effectively undercollateralizes our bond as the underlying principal balance of the loan is now less than the principal balance of the bonds that we own. In some rare cases, this loss is recognized in the period in which it occurs and pass through as a realized loss. But in most cases, the loss is not realized until the loan balance is reduced to 0 and yet, we still have a bond balance outstanding, which is very likely many years from now. At that point, the unrealized loss will become a realized loss. Turning to Page 14. We have become increasingly active in the credit-sensitive residential whole loan space, growing this asset class threefold since September 30 to $387 million. We're excited about this asset class for several reasons, one, investments in this asset class utilize much the same residential mortgage credit expertise that we have effectively deployed in the legacy non-agency space since 2008. Two, supply dynamics suggest that we will have significant opportunity to purchase these assets as current holders are obliged to shed them. Total supply is estimated at anywhere between $500 billion and $1 trillion, with $30 billion to $50 billion of expected annual sales in 2015 and 2016. Compare this to legacy non-agency, which is the shrinking market with fewer and fewer opportunities to purchase bonds. Three, we're excited to have the ability to oversee servicing decisions on troubled loans. For instance, by offering modifications to borrowers with such a modification will produce a better NPV outcome than a foreclosure and liquidation. Again, compared to legacy non-agency MBS, where we have no seat at that table, no visibility into the decision and often see losses pass through on these bonds that look excessive. Four, residential whole loans are good assets for us. That is the qualifying interests for purposes of the REIT qualification, which most RPL NPL mortgage-backed securities are not. And residential whole loans are qualifying interest for the purpose of the 1940 Act exemption, which legacy non-agency MBS are not. And finally, we believe that credit-sensitive residential whole loans further round out MFA's focus on credit versus pure interest rate risk. Turning to Slide 15. We are buying these credit-sensitive whole loans at material discounts to both their unpaid principal balance and also to the underlying property value. Our $387 million portfolio comprises 2,700 loans acquired through 13 transactions with 11 different counterparties. We've established relationships with dozens of key market participants, most of whom are different than our traditional trading partners for agency and non-agency MBS. We added some leverage to this asset class in the fourth quarter, and we're in discussions with additional lenders to add more financing in 2015. Leverage ratios on this asset class could be 2 to 3x debt-to-equity. And with that, I'd like to turn the call back over to Bill.

Thanks, Craig. So to summarize, we continue to execute on our strategy of acquiring attractive credit-sensitive residential mortgage assets. Not a lot of change, not a lot of variance based on our guess of what the fed is going to do with its next meeting. This is a more consistent strategy that works over time. You've seen over the last year that we've substantially grown our holdings of RPL NPL securities and loans. The credit-sensitive assets we own continue to perform well. Future Federal Reserve actions will depend on incoming data, but whatever their decision is, we believe we are well positioned for both changes in interest rates and the [indiscernible]. That completes today's presentation. Allan, could you please open up the lines for questions?

[Operator Instructions] Our first telephone line question comes from the line of Dan Altscher with FBR Capital Markets. Cole Allen - FBR Capital Markets & Co., Research Division: This is actually Cole Allen on for Dan Altscher. I guess, I wanted to first touch on your residential whole loans and actually, just your whole credit-sensitive portfolio. You've talked a little bit about the servicing decisions that you guys are -- you get to make on the new whole loans versus the legacy non-agency that you guys have? I guess, how are you monitoring those servicing decisions? And can you walk us through, I guess, that whole process and how they've been performing the services thus far?

So again, we are involved in overseeing those servicing decisions. So we have an asset management process that's pretty much works loan by loan. So while the servicer obviously has a contact with the borrower, we work very closely with the servicer, communicate multiple times per day. As far as how those decisions get made, they're really made in the best interest of the best net present value solution. So -- and when I refer to legacy, because those are bonds and we face the trustee and the trustee faces the servicer, we don't really have any knowledge of how those decisions get made. All we see are results. We see a liquidation, typically or we see a coupon reduction. So we think that we can make much more meaningful decisions in real time that, hopefully, produces better net present value outcomes. Cole Allen - FBR Capital Markets & Co., Research Division: Okay. And then, I guess, this quarter, you've -- we actually saw credit risk transfers kick up again. I'm just wondering what are you guys seeing in this market. And I guess, are you going to continue to expand it? And will this eventually maybe get broken out as its own segment?

Well, $100 million is still a very small amount but we just put it in because we know people are very focused on this segment. We are opportunistic. When the spreads are very wide, we're a buyer. And when the spreads are not as wide, we may not be a buyer. So it's hard to forecast whether this portion will grow. We do own some, and we didn't want to point it out but $100 million is a very small portion of our whole portfolio. Cole Allen - FBR Capital Markets & Co., Research Division: That's fair. That's fair. And then I guess, one last quick one. You mentioned that some of the RPLs and NPLs have been called and there's a good amount of them called over the next year or so. What are your guys' expectations for that? And I guess, is there a time frame that you expect? Or is it just going to be kind of lumpy and you don't have great insight into it?

Well, as you know, call is the option of the issuer, not the investor. But based on historical precedent because, as Craig mentioned, all the cash flows go to the senior portion. They delever very quickly for the issuer, so they are incented to call them probably before the end of third year but it's their option, so it's hard for us to forecast.

Just to add to that, most of the time that deals get called, not all the time, but most of the time that a deal gets called, the issuer reissues the deal within collateral. So a deal gets called, but there's usually an opportunity to participate in the new deal that follows on very quickly. Cole Allen - FBR Capital Markets & Co., Research Division: Okay. And you guys have access-- that you usually get to look at that deal?

Of course.

We'll move now to the line of Steve Delaney with JMP Securities.

Looking at Page 14 and hearing Craig's comments about the significant expected future supply. I mean, you guys have definitely been able to find access but I'm just curious if you see the possibility of some kind of mega trades coming down. We noticed in late March, Freddie sold about $1 billion in UPV of nonperformers. I think that was their first trade, and we haven't seen that Fannie has announced anything yet. But I'm just curious if you're hearing any chatter in the trading community. If this $1 billion is settling here in May, are you expecting that there might be a large securitization coming out this summer from the buyer of those lines? Or what is your thought about the GSE role in contributing to the NPL/RPL opportunity? Just generally, if that's more appropriate.

Steve, this is Bryan Wulfsohn. Yes, I guess, we would expect significant supply to come from the GSE. We have seen Freddie sell some loans. We wouldn't be surprised if Fannie might come behind that and sell some loans, and we would expect sort of a steady continuous supply then of nonperforming loans coming out from the GSEs. They have somewhat, I would say, in around $100 billion of nonperforming loans. So that can be significant supply coming out from there.

And Steve, you're right. Typically, home loan trades of this type turn into securitizations a couple of months later. So you're correct.

Yes, that's where I was kind of going. And obviously, as we saw with CRT, right? I mean, CRT got so tight, you couldn't even buy it. And as the volume picked up, the spreads came out. I think you found it attractive again because you weren't involved so much early but you kind of got involved after spreads widened and it -- just kind of trying to get a sense for whether the NPL or RPL securitization game, whether it might even become a better rate, if the volume actually picks up?

So Steve, you sort of highlighting our strategy, we -- the residential mortgage universe is not a small one and we can pick and choose our spots, basically reinvesting our runoffs. So -- which you point out is correct. If the spread is wide, we're there. Otherwise, we will find another track of investments from the same asset class.

And if the whole loans pick up, obviously, you've got a little more control over the servicing, et cetera, there. Then you found some repo financing but could -- as you're volume picks up and maybe you've got $0.5 billion instead of $300 million, is that something where you could actually -- have you guys considered actually being an issuer of securities yourself to finance your whole loans?

Absolutely. Although, you sort of go home and now you think, "Gee, am I a buyer or seller?"

Yes. No, no, I hear you. I hear you. But of course, you're still -- even if you're a seller, you're really in financing because you still got that inherent credit-sensitive subordinate piece that you're really looking for your return on. So one -- just one final -- I appreciate the comments on it, guys, because I think that's -- looking at where you're positioned, this is probably the biggest clear and present opportunity that you have. Just, I guess, for Steve Yarad. The accounting distinction on your NPL/RPL whole loans, you -- the 40% are fair value. Should I just assume that when you put something at fair value, there's a serious level of delinquency and you're actually at that point looking to the fair value of the underlying real estate rather than the performance of a loan?

Steve, that's correct. When we acquired these packages and loans, we go through them and we segregate them into what we consider to be more in the nature of reperforming loans and nonperforming loans based on the delinquency at the acquisition date. And we, under the accounting standards, we make an election to either account for them as fair value or to account for them at a carrying value basis. And the ones that we account for fair value, you're correct, are the ones that are more delinquent and more in the nature of nonperforming loans.

We have a question in queue from the line of Jason Weaver with Sterne Agee L. Jason Price Weaver - Sterne Agee & Leach Inc., Research Division: I'm going to dovetail a little bit on Steve's comments here. On the NPL and RPL assets you purchased during the quarter, what would you say the average cost to underlying principle is on each of those segments?

There are the purchases in the first quarters?

Securities and loans.

You term up your loans. Correct, not the RPL/NPL securities. Jason Price Weaver - Sterne Agee & Leach Inc., Research Division: RPL and NPL securities.

The securities are about pars. Jason Price Weaver - Sterne Agee & Leach Inc., Research Division: About par. Okay. Fair enough. And the other thing, am I right to assume, your credit reserve at this point is almost exclusively dedicated to the non-agency legacy MBS portfolio?

Yes. Jason Price Weaver - Sterne Agee & Leach Inc., Research Division: So with -- I didn't catch Craig's comment on the percent that we're amortizing but I did hear that only about, I think, 4% were significantly above 100% of LTV? How just -- can you walk me through how you get to your 24% implied credit reserve on that? Is that really just the function of more modifications going through the pipeline? And you touched on this during your comments but I'm trying to make sure I'm understanding correctly.

So it's 2 things. One, we presently have about 15%. I think it's 14.7% that are 60-plus days. And then, in addition, our assumptions assume that loans in the future will also default. Loans that are current today will also default. And the page, I think the page you referenced, I said that about 4% of the principal balance of the loans that are current today have LTVs over 110%. So as far as how those assumptions all interrelate, remember that it's not just loans defaulting, it's also prepayment assumptions that drive that. It's loss severities on loans that default. So it's a pretty rigorous process that we go through every quarter to review that credit reserve. And as you point out, modifications also cause losses, right? Although those losses might not be realized for many years but modifications will also cause losses to the extent that principal is used pay interest because every dollar or principal that's used to pay interest, can never pay principal back on the bond. Jason Price Weaver - Sterne Agee & Leach Inc., Research Division: Fair enough. Okay. And one final. Aside from your multiyear set repurchase agreement facility, any other opportunities on the horizon to expand financing beyond the traditional repo space?

Well, as we've talked about, if we did grow our loan portfolio. We would look at the securitization market.

And we'll now move to the line of Joel Houck with Wells Fargo.

I'm just wondering if you can do a compare and contrast, risk and reward trade-off between the CIP bonds and reperforming NPL that you're buying to replace runoff. What is that in terms of IRR versus the overall credit risk profile?

I think -- so what we sort of did -- do this. We've sort of tried to do the unified theory of quantum physics and -- it all has to tie together, right? Because you can trade off any one of these assets. We can try to solve it all, and I think the ROEs on the CRT bond looked very good. But you have to remember, there's a low cushion there. So we think depending on the wider range of potential events, if nothing extreme happens, the CRTs look very good, especially if you're down towards the bottom pieces. But the variability of the returns changes based on your assumptions. That's how we do look at it. You're exactly right, Joel. And they all look relatively comparable but I think, historically, we felt that we could deal with more uncertainty not investing in the CRTs because the pieces are so thin at the bottom.

Right. And how does the volatility in these spreads allow you to be more opportunistic or less opportunistic in each one of the asset classes? Because I know the CRT bonds have moved around quite a bit. And people, I think including yourself, have tried to take advantage of wider spreads at times. I think they've come in here this year but maybe if you could wrap it up by commenting on that, that aspect of it.

So there's 2 decisions are whether you're buying or whether you're selling. So we've only bought when they have widened. But they haven't tightened to the extent we did decide to sell. On our legacy non-agencies, we bought when they're very wide. And if a particular asset does tighten, we'll opportunistically take those gains, which we've done 11 quarters in a row. So those are both, really, the 2 decision points.

Right. Is one more powerful than the other? Or is it -- are they kind of independent -- move independently of one another?

Well, the CRT prices did change pretty rapidly. So I would say since the CRTs have been just dated, they've probably been more volatile.

And we now have a question in queue from the line of Mike Widner and he's with KBW. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Let me follow up a little more with some kind of basic questions on the RPL and NPL whole loans. Just first, I think I know the answer to this, but when you make that designation at purchase, you split them out into the 2 buckets, loans of carrying value versus fair value. I assume that's a designation that kind of sticks around for life -- for the life of the loans? Or is there any movement back and forth, I guess, over time?

Well -- it's Steve Yarad. Once you make the designation to account from the fair value, that's irrevocable. So once they're a fair value, they stay at fair value. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: And how about the other way? I mean because, I mean, I would think the main distinction is if it's a nonperforming loan, there's no interest income on it, obviously. And so I mean, can you move it from the carrying value bucket into the fair value bucket if you want to if it goes NPL or whatever?

Once -- no, that doesn't happen either. So once we account for it a carrying value, to the extent that they do become a nonperforming loan later in life, we still account for them using a yield-based approach. But as you might appreciate, if enough of them in the various buckets that we bought go non-performing, then you would take impairment later down the road if those cash flows deteriorate. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Yes. Okay. And then on the ones that are in the fair value bucket, I just wanted to understand a little bit. I don't need great detail but what's the basis for what you're using as fair value? And I mean, is it a lifetime cash flow NPV sort of thing? Or is it a value of the underlying property? Or is it basically your assessment of the market value if you were to liquidate those today?

So it's really all of those things. So right, you're looking at the cash -- you're looking at the projection of cash flows by taking into account the underlying property value. If you're -- if it's a nonperforming loan, right? There's the time to liquidate and the cost associated with liquidating. And then basically, you're discounting that back at some market discount rate and that's how you come up with the market value. And we use sort of a third-party valuation service to assist us in that marking process. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: So I guess, I mean, specifically the question there is because we see this with other companies that are holding stuff at fair value. If the market decides that, hey, these things are actually great because the market's implying either increasing property values or lower discount rate or whatever. I mean, the market can choose to pay a lot more for these things 2 or 3 quarters from now. I guess, what I'm trying to figure out is what ends up dominating the equation? The underlying fundamentals or sort of-- the market -- because what you guys think about credit losses and discount rates may be very different than what the market chooses to pay at some given time.

So yes. I mean, the market discount rate would definitely affect the value of the nonperforming loans. Just, right, if someone was paying using a 7% discount rate initially and then the market moved to a 6%, that would definitely impact the price. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Yes. And so what I'm trying to figure out is, again, to what degree is the market's assessment of -- it doesn't really matter if it's credit spreads or whatever but, I mean, as we saw with the GSE, CRTs, right, those prices can be very volatile. And since you're not -- well, and so the question would be since you're carrying them at a fair value is, is this going to get swung around quarter-to-quarter by the market's assessment of what they're worth? Or is it more -- I know it's not black-and-white but is it more your assessment of underlying fundamentals or is it more just whatever the market feels like paying for these things today? Because it is a pretty big asset class and we're seeing more people in it. And so there is an argument that you can get a market value rather than an economic value, if you will.

Mark, I understand the question. And the way that -- when you account for -- you talk of assets at fair value and maybe Level 3 assets, you do have to make your best estimate of what the market values them at. You can't sort of value them necessarily at your own estimate of fundamental value. You have to use market participant approach. And that's the approach that we're taking in valuing these assets from quarter-to-quarter. It's hard to say whether that's going to mean they're going to be responding very volatility or not. It's going to depend on how the market perceives the value as we move forward.

I also think you see more volatility in the credit risk transfer deal simply due to technicals. I think the last sale that was done, there were some functions. There were 17x oversubscribed or something like that. So I think you'll see a lot more volatility because those spreads can get whipsawed by technicals. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Yes -- no, that makes sense. I just wanted to make sure -- I mean, the answer you gave is what I thought the answer was. So I appreciate it. I guess, one more little minor thing on this, and just want to make sure I know what I'm looking at in all these tables. In the slide presentation, I think it was Page 6. Yes, Page 6. You show a yield. So actually return on the RPLs of 5.84%. And then over on the table, the first -- well, the second table, I guess, in the press release, you list the yield on average earning assets for that bucket at 6.475. And I'm just wondering, is that -- well, what's the difference there? And should I be thinking of those as the same thing? Is one a quarter average versus a quarter end or...?

No, no. The simple answer. The difference, I believe, is servicing cost. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Okay. So the 5.84% showing, strips out, would do you say, 60-something basis points for servicing?

Correct. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Yes, the 63 basis points for servicing. Okay. Perfect. And then -- okay. So I've beaten that stuff. Quick question, I mean, any thoughts on FHLB membership? It seems to be all the rage these days.

So thanks for bringing it up. Yes, we've look at the strategy. We believe it has merits. And historically, it hasn't really matched with our types of assets. We -- our agency book really hasn't grown. Typically, legacy non-agencies, which no longer of high ratings would not be good assets to borrow against, but it's something we look at. We believe it absolutely has merit and it's something that might impact MFA in the future. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Got you. That makes perfect sense. And then this is the last one. You mentioned switching to a holding company structure. I hadn't really thought much about that but, I mean, is that a new thing. And sort of what's the -- is there anything in particular we should think about as to why that matters or why it happens?

Well, why it matters, it give us more flexibility. And actually, I believe we first reported this change in the first quarter of 2014. So it's been about 12 months. Michael R. Widner - Keefe, Bruyette, & Woods, Inc., Research Division: Yes. Actually, I thought it had been awhile. But I mean, you guys brought it up again. And I mean, is there anything particular about where you're placing assets? I mean, it came up again.

No, thank you. Part of the reason that we made the change was to allow us to set up entities that would be licensed to own loans. So you're exactly right. There's movement towards loans is the genesis of this new structure.

[Operator Instructions] We have a question in queue from the line of Doug Harter with Crédit Suisse.

This is Sam Choe filling in for Doug Harter. Our questions have been asked and answered.

Operator, I guess, that's all the questions?

No further questions in queue at this time.

All right. Well, I want to thank everyone for joining us today and for the thoughtful questions, and we sure look forward to speaking talking to you next quarter.