MFA Financial, Inc. (NYSE:MFA)
2012 UBS Global Financial Services Conference Call
May 8, 2012 8:00 AM ET
Stewart Zimmerman – Chairman and CEO
Bill Gorin – President
Good morning. Thank you for joining us for the first presentation of the morning. Pleased to have MFA with us today. They have a long track record investing in both agency and non-agency RMBS and one of the first agency mortgage REIT to diversify the non-agency. They are also a little bit different than their peers and that they take less interest rate risk, both on the agency and the non-agency side.
Representing MFA from my left is Stewart Zimmerman, Chairman and CEO; Bill Gorin, President; and Craig Knutson, Executive Vice President, who runs their non-agency RMBS business. I will turn it over to Stewart. Thank you.
Thank you (inaudible). We appreciate the opportunity of speaking at the conference today and thank you all for attending. This will make my General Counsel very happy. So just a quick peek.
We are an internally managed and you can notice that that’s underlined. We are self advised and self managed real estate investment trust and we are positioned to benefit from our investment – I don’t know why that happened – in both agency and non-agency residential mortgage-backed securities.
So when folks look at us into the kind of – distinguish us from some of the peers – actually, I don’t think we have a lot of peers in this side, but we are specialist in residential mortgage-back securities, which comprise of both agency and non-agency securities. And again, being self advised and self managed we think kind of distinguishes us from some of the other folks.
An experienced management team and we’ve been doing this, we put the company on the map in the New York Stock Exchange in April of 1998 and we have gone through some of the most difficult of times as you all know. Investment opportunities certainly exist in the residential mortgage-back securities particularly in the non-agency sector. Again, still somewhat inefficient market as compared to the agency market. The agency MBS investment returns, they continue to benefit from a steep yield curve and despite governmental actions, refinancing rates are not spiking. So we can talk our HARP, HAMP and there will be other governmental programs, again I can just say to you that the CPR expectations, both prepays on the agency and non-agency side have come in well within our expectations.
In this period of economic uncertainty, our goal remains to generate double digit returns in equity with an appropriate level of leverage. And again, if you go back historically, I think you will find that we have been, if not the least levered, but certainly one of the least levered companies in our space.
Very proud of this slide, always enjoy talking about it. But since January of 2000, we’ve had about a 14.5% annual compounded return. I think that speaks volumes in terms of how the company is run, in terms of being prudent, in terms of being able to manage at risk.
I am going to turn over to Bill in just – to Bill Gorin, our President, in just a moment, but one of the things I would like you to notice and Bill will give you some greater detail. When you look at the next interest spread both on the agencies and you look at the non-agencies and again, Bill will you some detail, and when you look at the spreads and you look at the asset allocation, I think that’s a very important part of our entire story.
I am going to turn over to Bill who will give you the detail.
So with this page we try to illustrate for you how we allocate our equity, how we look at the various returns within the residential mortgage-backed securities sector. So the first column is agency MBS, which probably more of you are familiar with than the non-agency. There you see our debt-to-equity ratio is 6.87 times, which is probably similar to some of the mortgage REITs you may be familiar with. What I would like to point out again within the agency column is that the yield on our assets is 3.15% and we’ve achieved that without any 30-year fixed rate. So this is with hybrids, 75% hybrids, 25% 15-year generating a yield in excess of 3%. That’s a good thing.
Our cost of funds is high and that’s because we are more mature mortgage REIT, we put on swaps five years ago, which are running off this year in the main, and that cost of funds, 1.71 is expected to trend down this year. The spread of 144 is probably in the low side but it’s not because of the asset yield, it’s because of the cost of funds and we expect that cost of funds to trend down this year.
Next column is the non-agencies. There you see the debt-to-equity is 1.8 times and this is why we are very excited about the non-agency sector. The yield, the loss adjusted yield before any leverage is 6.9%. So 6.9% yielding asset with what we believe are conservative loss assumptions makes it a very attractive asset to us. Cost of funds is 2.16% and the spread is 4.76%. So hopefully these numbers speak to you as to why we are very interested in this sector.
All in all we are about 3.5 times levered as Stewart pointed out.
So, focusing for us on the agency side our average amortized cost 102.8%, which I think shows very well compared to other mortgage REITs. We have limited exposure to HARP 2.0 because the assets that are old enough to be impacted by HARP 2.0 for us are adjustables where the coupon is about to set down or interest only such that if you were to refi out of this interest-only hybrid into a fixed rate, your coupon would go up because of the – not your coupon, your payment would go up because of the amortization. As I mentioned, our funding cost of the agency should be trending down which I think is pretty unique in the sector.
On the non-agency side, as of the end of March, we owned about $4.5 billion market value. The average amortized cost was 73% of par. Loss adjusted unlevered yield of 6.92%. Now when you buy an asset that yields 6.92% when the tenure is close to 2%, how interest rate sensitive is that asset? We believe the asset is not interest rate sensitive because 2% of the yield is interest rate, the other 4% or 5% is for credit. So we think it’s a lot more credit sensitive. We don’t think they are correlated to treasuries. In fact what we’ve actually seen is they are inversely correlated to treasury so that when the economy appears to be doing well and the stock market goes up and bonds don’t do well, the non-agencies trade with the stock market, not with treasuries. So we think by adding non-agencies, we’ve decreased our interest rate risk. In addition, as I mentioned, our average cost is 73% of par versus agencies which typically trade at premiums. Therefore prepayments go up, as yield goes up. You don’t have to amortize premium, you have to accrete discounts. So prepayments are only a positive on the non-agency side when they are purchased at a discount.
This illustrates or actually shows you our top 20 holdings of non-agencies. So you see we have circled three numbers in red. First number circled is the average FICO for these top 20 holdings. Now these top 20 holdings are not very different than the entire portfolio. So the average FICO is 733. So that shows you in no way are these subprime assets, these are prime and near prime jumbos. The next number is not circled but I would like to point it out, the WALA, that means how many months old the asset is. So here the assets are in excess of five years old. So we have five years of monthly pay history of each of these underlying loans, which gives us a very good data. So more than five years into the life of these assets, how much are delinquent? 90% or 60+ delinquent after five and some odd years. But when we tell the asset is yielding 6.92%, what are we assuming? What we are assuming is that 39% actually default. So we are assuming nearly twice as many people will default as are currently delinquent after five years.
Another illustration of what loss adjusted yield means. So if you had bought a non-agency asset six years ago, you would have assumed no defaults because they were all AAA, in fact all our assets were AAA when they were originated and when you had bought a mortgage-backed security six or seven years ago, you would have assumed a prepayment speed of 15%. So you know our price is about 72, our actual coupon is about 4.5%, if there is no defaults, prepays is 15%, the yields on the assets would be 14%. What happens if we just slow the asset down, take the prepay speed down from 15% to 6%, that lowers the yield from 14% to 10%. And then if you use loss assumptions similar to the ones we have used were 44% default, 50% severity, keep the prepay speeds as low number of 6%, the asset yield is 7%. So that sort of shows you the assumptions we’ve used to calculate the 7% yield.
So housing fundamentals are still weak though they seem to get stronger every day. But we’ve built this weakness into our assumptions. So one, before there is any losses on our assets, there is yet 4.6% of subordinated interest that still exists beneath us, that’s credit enhancement, they bear the first losses. As I mentioned, we paid $0.73 on a dollar. So $0.27 of capital losses could occur before there is any loss of our principal. In addition we are assuming losses of about $1.4 billion against the $6 billion face of assets. So we’ve assumed $1.4 billion of losses occur within the $6 billion and that’s how we get the 7% yield. And hopefully this pie chart will explain that in a different way.
So the face amount is about $6 billion, we paid $4.4 billion or 73% of par. The purchase discount, so if you paid 73, that means of a discount about 27, we’re assuming we get back about $0.78 on a dollar. So we are accreting from $0.73 to $0.78, we are assuming there is losses of 22%, and those losses will occur after the 4.6% credit enhancement is wiped out. So I think I’ve gone into some of Craig’s pages, I think we will leave (inaudible) continue.
Now initially when we started buying these non-agency assets, we didn’t use any leverage. But overtime we have added various forms of leverage which included repo, includes securitized debt and it includes a multi-year collateralized financing. So we’ve diversified our funding sources. Also because we are only two time levered, that means about one third of our assets are funded with equity. So we are not highly dependent on short-term repo, which actually is ironic, even people think that agency assets are very safe, while typically those are funded with short-term repo. They have to be, the asset yield is not high enough to pay for the longer term financing. Here because the asset yield is close to 7%, we could afford to pay up, pay for longer term financing. So we actually have a lot less roll over risk on these assets than you would on the agency book.
Let’s go back to the beginning, as Stewart mentioned, we are internally managed, our incentives are directly in line with shareholders. Our goal and we would have told you this goal 10 years ago and 12 years ago is to generate double digits ROE. How has that been working out? Well, since 2000, if you reinvested dividends, you would have generated a return of 14.5% per annum. Now well, how the fundamentals aren’t yet strong? We believe we have more than factored this into our cash flow projections and credit reserve estimates. In addition, for our non-agency assets, we’ve diversified our funding sources including securitized debt, structured financing and equity in addition to regular repo.
So with that, if there’s any questions, we would be glad to take them.
You mentioned that prepayments (inaudible) can you talk about why that is both on the agency side and the non-agency either for on context your portfolio or the market in general?
Great, let me start on the agency side. We don’t buy (inaudible) general. So if you buy specific pools that are less prevalent to prepays general.
And on the non-agency side, I think we’ve modeled prepayments pretty slow from the beginning and the primary reason for that is when we started proposing non-agencies, the dollar prices were in the 50s and some cases even 40s. So the most dangerous thing you can do is to overstate the expected prepayment speeds. We don’t see the same prepayment speeds on non-agency that you see on agencies because as I am sure you know, it’s a lot more difficult to get mortgages to get jumbo mortgages. So whether it’s because credit scores are banged up or whether it’s because homeowners don’t have equity in their homes, you typically don’t see the same prepayment speeds.
I would just like to reiterate. Craig’s models and the folks that he worked with, we don’t look non-agency securities with the idea that you can kind of model prepays at a number to come back with the yield, we don’t do that. If anything, we go (inaudible) we go and come in the other direction in terms of trying to be very moderate in terms of our assumptions on prepays, on non-agencies.
By focusing on all forms of residential mortgage-backed securities, not just agencies or non-agencies, it works our very well for the portfolio because as we mentioned for agencies, you pay a premium. And when there is a prepayment, you have to amortize that premium. In the non-agencies because we purchase them at a discount, we actually accrete that discount and in the last couple of quarters, the amount of discount we have accreted and the amount of premium we amortized have been about the same. So prepayments are neutral to us. In many cases, we can see prepayments becoming a positive to us because 27% discount gives you a lot more upside than a 3% premium on the agencies.
A couple of questions on the assumptions the non-agency slide that you had. The first on the severity losses, I was wondering what sorts of assumptions are you making in terms of housing, underlying collateral and is that changing, as housing appears to be improving somewhat? And then secondly, on the default rates, using an assumption of two times what is 60 days delinquent in a five-year average life seems really conservative, now I was just wondering how you came up with that assumption.
Sure. So on the severity side, I know our slide says approximately 50% and I think overall across the whole portfolio and we are talking about probably 450 line items give or take, I think our average default rate in the future, our average severity is somewhere around 50% but it really varies by bond and actually by loan. So when we look at a security, we drill down to loan level details. So for instance, if we have a lot of delinquent loans in a particular bond delivered in Florida, we know that the foreclosure process and can drag easily out for three years in Florida. So our severity assumption on loans like that will be significantly higher than they would be a loan, say in California, where the process is pretty quick.
So again, it may average 50%, but there could be bonds where we are using as high as 70% loss severities and there could be cases where maybe it’s as low as 35%. But it will really depend on underlying loans where they are located how long they have been delinquent.
And then you said we revised those assumptions. Not really, we do review bonds periodically, we have two people do spend all their time basically doing surveillance on the bonds that we own. So on occasion, we may revise assumptions, and we have moved some money out of credit reserve to accretable discount. I think we tend to be a little bit slow to do. I hate to look at three months of data that looks promising and change assumptions and then, six months later change it back again.
So I think we are perhaps a little bit slow to make those adjustments but it is something that that we look at and we look at it all the time.
The two times number, so the defaults versus delinquency, again it’s just coincidence on that page that it’s approximately two times but the way that we think about default is first of all, loans that are delinquent for the most we assume that those will default. We don’t see a whole lot of loans cure after they go delinquent. So that part is really easy. The challenge in non-agencies is as you look at these pools today, as Bill pointed out, they are 5.5 year seasoned. As you look at these pools today, how many of the loans that are current today will default in the future and that’s really where it becomes a little more challenging. So I would say the thing that we look at most in making those determinations are what we think are, what we know are the LTVs today on those properties.
So the loans that are current that have very high mark-to-market LTVs are obviously the ones that we are more concerned with. And so, it’s that bucket that we will look at, we will look at credit score trends of borrowers to try to determine the likelihood of default. We will look at what the LTV is on those properties and so – and we will look at pay histories as well. And as every year goes by and we get another 12 months of pay history, that’s a very important statistic for us because if I told you that someone has a 140% mark-to-market LTV but I also told you that mortgage is 6.5 years old and the borrower has never missed the payment, they haven’t missed one payment in 6.5 years. So that tells you something. Could he default tomorrow? Well, sure he could, he could get divorced, he could lose his job, he could have a family illness. Those are traditionally by the way the reason that people used to default on mortgages, right? If you go back 10 years, that’s the only reason people defaulted on mortgage.
So it’s this notion of strategic default which was the really scariest factor two years ago, three years ago when we started this. And the folks that had a very bad mark-to-market LTVs were the ones that you really worried about the strategic default. I can’t tell you that they won’t strategically default now but if they haven’t done so in the last couple of years and in many cases, now these loans are actually amortizing, so they are hybrids, the coupons reset down, but the loan started to amortize. So they are actually paying off principal now.
So again, I can’t tell you that they won’t default but we do feel better about that.
Sure. We are certainly seeing supply, I will say and we said on the earnings call last week, the supply is a little bit more spotty, it’s a little bit more lumpy, if you will, this year. The well publicized trades is a Maiden Lane, so it was a Maiden Lane II, which was a big overhang last year, it’s actually largely responsible for the price declines that we saw last year because the market really, the appetite that the market had was not sufficient to digest that supply. And again, I will point out most of those Maiden Lane II efforts were subprime, which we actually don’t traffic in but the whole sector sort of traded off with that. And in January, they got off basically all the rest of the Maiden Lane II sales, it was three different sales, very large transactions, and they went exceedingly well, better than anybody expected. And most of those bonds, if not all those bonds, were put away into firm hands. There is now a Maiden Lane III that’s actually out this week, those are mostly CDOs, although there are some residential bonds underneath those CDOs but they will trade as CDOs.
So we expect those will trade pretty well. It’s somewhat ironic but the larger trades tend to trade better sometimes than the smaller trades because it’s a little bit of a food pipe for those assets. So supply is good albeit somewhat spotty, I think again we said in our earnings call last week, we put almost $400 million to work in the first quarter in non-agencies and we put about $200 million to work since the end of the first quarter. So we are certainly finding good value, but we can go buy and we might not find anything and then the next week we might buy three or four bonds.
Can you talk about what goes into the decision process between allocating capital in agency and non-agency?
We always look for value. And again, as you know, we started as an agency REIT and in 2008 we made the determination that the two classes meaning both agency and non-agency is very complementary and will create better value for our shareholders. So it’s a matter of value. Now the piece puzzle, of course, is something called (inaudible) and to be compliant, we have to have at least 55% of our assets – real estate assets and that’s defined for us as whole pool agency assets. Having said that, we will be above the 55% level irrespective and we continue to like the asset class. So you have two complementary asset classes in terms of agency and non-agency, one working with the other to produce I think some very, very handsome results for our shareholders.
Great. If there are no further questions, we can wrap it up. Then we will have a breakout and will lead to 14th room which is on the fourth floor up at the stairway to your right outside.
Thank you, everybody.