AGNC Investment Corp (AGNCO) 2016 Q4 法說會逐字稿

Good morning and welcome to the AGNC investment Corp fourth quarter 2016 shareholder call.

(Operator Instructions)

Please note this event is being recorded.

I would now like to turn the conference over to Katie Wisecarver in Investor Relations. Please go ahead.

Thank you, Chad, and thank you all for joining AGNC Investment Corp's fourth quarter 2016 earnings call.

Before we begin, I'd like to review the Safe Harbor Statement. This conference call and corresponding slide presentation contains statements that to the extent they are not recitations of historical fact, constitute forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. All such forward-looking statements are intended to be subject to the Safe Harbor protection provided by the Reform Act. Actual outcomes and results could differ materially from those forecast, due to the impact of many factors beyond the control of AGNC. All forward-looking statements included in this presentation are made only as of the date of this presentation and are subject to change without notice.

Certain factors that could cause actual results to differ materially from those contained in the forward-looking statements are included in the Risk Factors section of AGNC's periodic reports filed with the Securities and Exchange Commission. Copies are available on the SEC's website at www.SEC.gov. We disclaim any obligation to update our forward-looking statements unless required by law.

An archive of this presentation will be available on our website and the telephone recording can be accessed through February 16 by dialing 877-344-7529 or 412-317-0088 and the conference ID number is 10098994.

To view the slide presentation turn to our website, www.AGNC.com and click on the Q4 2016 earnings presentation link in the lower right corner. Select the Webcast option for both slides and audio, or click on the link in the Conference Call section to view the streaming slide presentation during the call.

Participants on the call today include Gary Kain, Chief Executive Officer, Peter Federico, Executive Vice President and Chief Financial Officer, Chris Kuehl, Executive Vice President, Aaron Post, Senior Vice President, Bernie Bell, Senior Vice President and Chief Accounting Officer.

With that I'll turn the call over to Gary Kain.

Thanks, Katie. Thanks to all of you for your interest in AGNC. The fourth quarter got off to a relatively benign start, but that changed very quickly on election night. President Trump's surprise victory kicked off a major equity market rally, and this risk on sentiment sparked a sizeable sell-off in the Treasury bond market that rivaled the severity and speed of the Taper Tantrum in Q2 2013.

In fact, the quarter-over-quarter move in the 10 year Treasury was slightly larger than what we witnessed in the second quarter of 2013. However, while agency MBS underperformed Treasuries, their performance was materially better than what we experienced during the taper tantrum.

Against the back drop of the large violent spike in interest rates and a measured widening in agency MBS spreads, AGNC's economic return was negative 5.2% for the quarter, largely offsetting the positive 5.6% return agency produced the prior quarter. Now before we begin a standard review of the quarter early results and the portfolio, I want to discuss our perspective on interest rates and the economy, given the changing political landscape along with some of the risk factors that could materially change that outlook.

First and foremost, the direction and even the magnitude of the moves in both rates and equities are reasonable, given the potential for both corporate and individual tax cuts, a more business-friendly regulatory environment and the possibility of other forms of fiscal stimulus such as increased infrastructure and defense spending.

The impacts of an America First protectionist-oriented agenda however, are less clear. Such actions could certainly be inflationary in the near term but they could also lead to unintended consequences and negatively impact global economic growth. The combination of Brexit and the Trump election have bolstered the confidence and appeal of nationalist parties throughout Europe.

If nationalist parties maintain this momentum and not significant election wins this year, the impact on the EU could be dramatic. Additionally, President Trump's stance on trade could be a material negative for Mexico, China, and other economies. In summary, while the baseline economic path for the US is stronger post-election, the risk to the global economic and political landscape remain biased to the downside and may have actually intensified.

Interestingly, with the exception of the taper tantrum, the global economic environment has generally been the dominant driver of larger changes in US interest rates since the European debt crisis of 2011. Moreover, while other sovereign rates have increased over the last several months, those changes have been considerably smaller than what we have seen in the US.

This is logical, given divergent monetary policy and other factors, but the growing interest rate differentials should provide some resistance to further increases in US rates. With that as the introduction, let me turn to slide 4 and quickly review our results for the quarter. Comprehensive income totaled a loss of $1.19 per share.

Net spread income, which includes dollar roll income, remains solid at $0.64 per share in Q4 when we exclude the $0.26 benefit from catch-up amortization. Total book value per share decreased $1.74 or 7.6% to $21.17 in Q4. Tangible book value, which excludes the goodwill and other intangible assets associated with the internalization, was $19.50 at year-end.

As I mentioned earlier, economic return was negative 5.2% for the quarter. At-risk leverage was essentially unchanged at 7.7 times tangible book, and our portfolio totaled $58 billion as shown on slide 5. I also want to mention that we did set up an at-the-market equity offering program, or ATM.

An ATM is a useful tool to have at your disposal to issue relatively small amounts of stock into the market over time. Given AGNC's current discount to tangible book, we would not expect to use it in the near term, but it is something that we have used periodically in the past and believed it was logical to set it up so that it was available in the future under the right conditions.

It is important to remember that AGNC is now internally managed, and therefore Management has no incentive to issue shares just to increase AUM and Management fees. Turning to slide 6. We can look at some full year 2016 highlights. Despite a challenging fourth quarter, total economic return was approximately 4% for the year.

Comprehensive income totaled $0.80 per share while net spread and dollar roll income equaled $2.36 per share exceeding our $2.30 in shareholder distributions. AGNC's total stock return was approximately 18% for 2016. If we turn to slide 7, I want to take a minute to discuss the return outlook for agency MBS in the current interest rate environment.

While the spike in interest rates and wider agency MBS spreads negatively impacted book value in Q4, they improved the returns on new purchases, thus potentially enhancing our franchise value over time. The two exhibits on the bottom of the page attempt to illustrate the potential ROE performance on new MBS purchases.

The table on the bottom left shows net interest margins on three different coupons in today's environment with hypothetical hedges and resulting duration gaps. As you can see from the illustration, net margins of between 100 and 150 basis points are achievable before accounting for potential rebalancing or convexity costs.

I think the more interesting thing for investors to digest is the table on the bottom right. The table shows an illustration of the relationship between leverage, net interest margin, and gross ROE. The simple math behind the table indicates that with a wide range of hypothetical net interest margins and with leverage ranging from 7.5 to 8.5 times, all other things being equal, our gross ROE would fall out between 11% and 16%.

The ROEs in this illustration are gross numbers and thus don't capture operating expenses. However, since AGNC's net operating expenses after accounting for the MTGE Management fees are only around 75 basis points, net numbers should not be significantly lower.

Changes in interest rates, spreads, repayments, funding relationships and other factors could materially impact actual results and eventual shareholder returns. That said, this table helps illustrate why we are constructive on AGNC's business as we enter 2017. With that I'll turn the call over to Chris to discuss the market and the portfolio.

Thanks, Gary. Turning to slide 8 I'll start with a review of the markets. As Gary mentioned interest rates sold off dramatically following the election in November with a yield on 10-year Treasury notes hitting a high of 2.6% in the middle of December, before recovering some of the move higher to end the year at 2.43%.

The move in rates was very rapid, with five-year and 10-year notes spiking approximately 50 basis points in the two weeks following the election, before ending the quarter higher at 77 basis points and 82 basis points respectively. Agency MBS underperformed interest rate hedges during the quarter with LIBOR option adjustment spreads widening approximately 15 basis points.

Other credit sensitive fixed income assets, such as investment grade corporates and CMBS, performed reasonably well during the fourth quarter as the risk on mind set lead to a more favorable outlook for economic conditions going forward. Let's turn to slide 9.

At-risk leverage was unchanged as of December 31 at 7.7 times tangible equity. However the market value of the investment portfolio declined to $57.7 billion, given lower prices on agency MBS, and a reduction in current face of our holdings of approximately $3.5 billion.

As we've discussed on prior calls, we have focused on maintaining a balanced asset composition with respect to both call and extension protection, despite 10 year swap rates reaching an all-time yield low of 1.25% seven months ago. While we continue to think interest rates will remain low by historical standards, our portfolio has a significant portion of seasoned MBS, and so if rates do move higher the extension risk would be contained.

To this point, and the tables on the bottom of slide 9 you can see the weighted average seasoning of our on-balance sheet holdings in 15 year MBS is five years, and in the case of 30 year MBS, it's more than three years. In aggregate, 75% of our on-balance sheet pools are seasoned more than three years. By comparison in 2013, less than 5% of our holdings were in this category.

In addition to having shorter durations, seasoned pools generally have a more stable prepayment profile and therefore have better convexity. In other words, the duration does not change as much with changes in interest rates and therefore doesn't require as much hedge portfolio rebalancing when interest rates move.

Lastly with respect to prepayments, you'll notice there are forecasted lifetime CPRs, declines at seven CPR and 10 CPR for our on-balance sheet 30-year and 15-year holdings respectively, as of December 31. I'll turn the call over to Peter to discuss funding and risk management.

Thanks, Chris. I'll begin with our financing summary on slide 10. Our repo funding cost at quarter end increased to 98 basis points, up from 83 basis points the prior quarter. This increase was due to the combination of the Fed's 25 basis point rate hike in December, and normal year-end funding pressure.

A significant portion of the increase in our repo cost was offset by a corresponding decrease in the cost associated with our pay fix swaps. Specifically, the majority of our short-term borrowings are hedged with pay fix swaps. As short-term rates rise, the floating rate that we receive on our swaps also rises, thereby lowering the overall cost of these hedges.

On average during the quarter, the cost of our pay fix swaps was 57 basis points, down from 74 basis points the prior quarter, or a 17 basis point improvement. Given the offsetting nature of our pay fix swaps, our overall average cost of funds including the implied funding cost associated with our TVA assets increased only slightly during the quarter to 115 basis points from 113 basis points last quarter.

Lastly, we continued to expand our funding activities through our captive broker-dealer Bethesda Securities. At quarter end our repo position at Bethesda securities totaled $4.7 billion. Funding through this vehicle is attractively priced and as such we will continue to move a greater share of our funding to our broker dealer over the next several quarters.

Turning to slide 11, I will briefly review our hedging activity during the quarter. Given the move in interest rates and interest rate volatility during the quarter, we added to our pay fix swap and short Treasury hedge positions. These incremental hedges were generally longer term and thus give us greater protection against higher long term interest rates.

As a result of these actions, our aggregate bench ratio increased to 91% of our funding liabilities at quarter end, up meaningfully from 75% the prior quarter. Finally on slide 12 we provide a summary of our interest rate risk position. The spike in interest rates during the quarter caused our asset duration to extend to 4.9 years, an increase of nearly two years from the three year average asset duration that we reported last quarter.

Due to the changes in our hedge portfolio that I already mentioned, the duration of our hedge portfolio extended by about a year to 3.6 years at quarter end. As a result, our net duration gap remained well contained at 1.3 years at quarter end. And with that I'll turn the call back over to Gary.

Thanks, Peter, and at this point we would like to open up the call to questions.

Thank you, sir.

(Operator Instructions)

The first question comes from Bose George with KBW.

Hi guys, good morning. First, just in terms of where your leverage and your duration gap stood at quarter end, are those good places where you'd like to be or should we expect any change there?

So with respect to duration gap, I think that it's ballpark where we would like to be given the interest rate environment. Obviously, if interest rates go up our duration gap will naturally lengthen. I think we would be willing to let the duration gap get longer if rates were to sell off and we probably let it contract some but into a rally as well.

So I think what I would say is we would continue to offset some of the changes that would naturally occur if rates moved, but not certainly not all of them as we have done in the past. Around leverage, mortgages are a little wider this quarter and so our leverage is probably a little higher as we sit here today than it was at quarter end or year-end, but not substantially.

I think that if mortgage spreads were to widen more, we would be comfortable increasing leverage. When you look at the performance of mortgages versus other fixed income products, credit is sort of getting close to multi year tights at this point. Or is at multi year tights in many categories.

So because of the increase in volatility and the rate move, agency MBS have lagged, and that's logical given the move. But on the other hand, on a relative basis, we feel mortgages are pretty reasonably priced and so I think that under the right circumstances, we would be more inclined to raise leverage than lower it.

Okay that's helpful thanks and actually the comment on credit leads to my next question. You guys didn't really mention CRT in the prepared remarks? What are the latest thoughts there in terms of the potential growth outlook?

Still as we said last quarter, and when we began the initiative or when we opened AGNC up to credit investments, we're very interested in CRT and conforming mortgage credit over the long run. But spreads are at multi year tights in CRT as well right now, and so we've added some. But we are going to add it at a slower pace, given where we are today.

So as of the end of January, we had right around $250 million in CRT, and we would expect that the pace of growth to be relatively slow in the near term, given current pricing. The ROEs on CRT are give or take 9% levered ROEs, and we think the agency space has better potential at this point.

Okay that's helpful, thanks, and actually one macro question. There's been a few articles now just about what the Fed might do with their MBS portfolio, what that might look like as it normalizes? Just wanted to get your take on what you think the Fed might or might not do over there?

So there has been a lot of discussion about the Fed ending reinvestment, and it's obviously a topic in the agency mortgage market that gets quite a bit of focus. What I think, there's a couple of things to keep in mind. First off I'll answer the question directly which is, we don't see the Fed ending reinvestments as a 2017 kind of event.

I think realistically, we think that the -- there are that most likely, there may be some extreme cases, but we would say the most likely scenario would be maybe mid-2018, maybe second quarter 2018 if the economy continues to perform kind of well. But you have to be cognizant that -- [Bernagie] put out a piece that was interesting related to his opinion -- and again that's not the Fed's -- on ending reinvestments.

But I think that the Fed is a ways away from ending reinvestments. And then they will most likely do it in kind of a tapered fashion. So I think it will continue to get discussion and I think that might lead to some volatility in mortgage spreads. But on the other hand, from our perspective there's another piece that's important in this equation, which is where should mortgages be priced if the Fed were not involved.

And yes, the Fed's been involved for a while, but the mortgage market has performed fine even post crisis in periods without the Fed. And if you look at where spreads are at this point, we think even without the Fed, you're not talking about them being 50 wider. There may be 10 to 20 basis points wider in an environment without the Fed, especially if that occurs over time.

So I think from our perspective, we understand it's going to be a big driver of short-term moves in spreads expectations on the Fed, but we're not overly concerned by it.

That's helpful, thank you.

The next question will come from Steve Delaney with JMP Securities.

Good morning, thanks for taking the question. Wanted to talk a little bit about repo rates? Looking at just the short repo, sort of one to three months, you were 91 basis points at year-end, up 16 versus 75 at December?

Now that makes sense with the December Fed hike, but guys, we got some color last week that short repo was down closer to the low 80 basis point range? Which obviously, given the comparable rates were more like in the 70s at September. It's almost as if the repo market, is not at this point, if the color was accurate, is not really pricing in that full December hike?

So just curious if you can offer any comments on that and if you think the repo market is still benefiting from the money-market reform in October or maybe there's another dynamic? Thank you.

Good question, Steve. Good morning, thank you.

Hi, Peter.

You're absolutely right. I'm glad you asked the question. There was really a dramatic shift in pricing, really came close to year-end maybe a week before year-end and we started to see the pressure in repo rates subside significantly, and since then they've continued to improve.

For example, three month funding through our broker-dealer is probably more like 75 basis points. And three-month two-party repo outside our broker-dealer is more like 85 to 90 basis points. When you think about that relative to three-month LIBOR rates which are now closer to 105 basis points, you're talking somewhere between 15 and 30 basis points through three-month LIBOR.

Which will have a very meaningful impact on our overall cost of funds when you think about our repo being linked up with a pay fix swap where we are receiving that three-month rate, so you're absolutely right. There's been significant improvement in funding. There's ample liquidity in the marketplace.

Part of it is the money-market reform. But I think part of it is just, in the current environment, when you go over quarter end, but in particular over this last year-end, there was particular attention being paid to the composition of balance sheets, and that really made the funding fairly tight for most of the quarter. But again, it's subsided a lot in the markets, very good right now.

Thanks, Peter and two clarifications on that? On the receive side of your swaps, is the majority of that versus three-month LIBOR, or is some of it also one-month LIBOR received?

No, at this point it's almost 100% three month LIBOR.

All right excellent, thank you for clarifying that for me. Then on the benefit of Bethesda, which the pricing you laid out here was, let's just say at least 10 basis points. If you come in and that was on the three-month, is that a good rule of thumb, or do you see that like on the shorter repo as well that you're able to reduce it by about 10 basis points?

Yes, and so let me make two points on that. Yes, we see that benefit fairly consistent, so call it 10 to15 basis points from anywhere from one month out to one year. So that's one point. The second point is though that our funding, we talked about our three-month funding being about 15 to 30 basis points better than LIBOR.

You look out a year for example, that funding benefit is actually more like 60. So 50 to 60 basis points, so both through our broker-dealer and direct, when you are looking out beyond three months, our funding through LIBOR, that spread is actually significantly lower than 30 basis points so it's attractive funding really across the curve.

And Steve we talked extensively on the last call about the importance of this. Especially given the fact that we have a relatively high hedge ratio, but where repo funding is versus three-month LIBOR in particular is a big deal for the profitability of our business.

Understand. Thanks for the comments, guys.

Thanks for the question.

The next question will come from Rick Shane of JPMorgan.

Hi guys, thanks for taking my questions. The charts on slide 7 are incredibly helpful. The thing that is actually somewhat surprising to me though when I look at them, is the relationship that you illustrate here between duration gap and spread.

And in my mind it seems like taking the duration from a spread perspective, taking it down is pretty cheap. Especially given the current environment. So I'm curious, given the risks of letting the duration run north of 1 in the current environment, what the rationale is because it only looks like you're picking up 10 or 15 basis points a spread?

Look it is a good point and I think that does show up in this illustration. Remember that -- we'll call it one year duration gap -- is not a large duration gap. You're talking about base durations on mortgages on our portfolio as being five. So you've hedged out a big portion of the interest rate risk.

I think that is definitely a factor in why we -- while the duration as Peter mentioned of our portfolio extended by two years give or take, our net duration gap only extended by a year, because we felt it was worth it to keep our duration in check. But obviously the -- how much you get paid for taking duration is a function of, very much so a function of the yield curve.

And it also depends on, in this illustration, we didn't use the various mix of hedges so there are other ways that you can maybe increase that benefit by using, let's say long, a bar bell of hedges or using longer term swaps that this simplistic example doesn't capture. But look, your main point is a good one, which is we don't believe like this is an environment where we need to or should be running a two- or two and a half-year duration gap.

We think the value from the mortgage portfolio is more on a hedged basis and is fine on a hedged basis and so I think it's that reason that we're keeping the duration gap kind of in check, so to speak.

Got it, and look, I think it would be illustrative and the math is pretty simple to also put the book value risk associated with this, as a third chart on this table, or on this page. The second thing I want to talk about is, look, I think what you are saying is in the current environment mortgage securities are reasonably priced with some risk from Fed exit.

But the reality is that when you discuss buybacks in the past, the two conditions that you've looked for are, a discount to book value, stock trading discount to book value which exists now -- and also relatively cheap MBS. I'm sort of, I mean I understand that the ATM is a defensive strategy to have in place in case conditions change, but I'm a little bit surprised in the current environment that one of the better investment opportunities isn't to buyback some shares?

So what I would say about the, first off on the ATM, is just to reiterate what I said. We don't plan on using it right now, being that we're at a noticeable discount to book. What I would say is, and I'll talk to kind of the driver of both repurchase decisions, and if they were to come up, issuance decisions.

It is important to -- the key measure that we will look at, is our price to tangible book. And that's an important thing for investors to think about. I think that's pretty straightforward, that's the right measure to look at, and it's especially true given the particular nature of our intangible asset. Our intangible asset, essentially is the reduced cost structure associated with the internalization.

And that cost structure advantage, if you are buying shares back, and you're shrinking the size of a portfolio, you're actually hurting your intangible asset as well. And conversely, to the extent that we were to issue shares, there's a material benefit to both the Company and to the intangible asset, which is that your incremental management fees, or your incremental operating costs, are not going up like they are for an externally-managed Company.

So if the Company were to be growing both the intangible asset is going to increase in value, and you're getting benefits in terms of a kind of proportionately lower operating expense ratio. So I think that's kind of way we're going to look at things. Again, if the shares were to get weaker we would be very aggressive on share repurchases.

I think we have been historically, and we certainly will be going forward. But generally, you need that price to book discount to certainly be in the 80s and generally lower 80s, before that gets compelling. Hopefully that helps.

Got it, thank you very much, Gary.

The next question will come from Doug Harter of Credit Suisse.

Thanks, going back to Bethesda, on the funding, can you talk about whether that current 10 basis points or 10ish basis points you're saving, whether that you're at a level now where you're offsetting the overhead there or how much more you would need to grow those balances to make it truly accretive when thinking about the costs?

Good question, Doug. Yes, we're right now, we're at a point where any incremental activity, I would say, would be net accretive given the cost of that enterprise. We expect, as I said, to grow that position. Currently it's close to $5 billion but that position could easily be $10 billion to $15 billion.

We'll likely make progress toward that over the next several quarters, and the costs there are generally fixed. There is volume-based transaction costs associated with it, but the benefit will far -- the funding benefit will far exceed the fixed cost and the variable cost of that enterprise. So we're going to continue to grow that business.

Great, and then just thinking about the spreads you show on slide 7, can you just talk about the different extension risks of those different securities and how you might be relative return of a three versus a four, at a stated stagnant duration gap, and how that looks in various rate scenarios?

Sure and it's a good question. Obviously, threes have largely extended, as much as there's only a little extension left. So in a sense, they are an easier instrument to hedge and they have less interest rate risk with a comparable duration gap than fours do.

Fours have more extension and actually and obviously more contraction risk as well, as we call it, more negative convexity. So you get this higher spread, but you do take more convexity risk. And over time, you will have more rebalancing costs associated with a four than a three, so you don't just immediately go and look at this table and say, okay let's put our whole portfolio in fours and forget those other coupons.

But we wanted to list all three to give kind of an illustration. So in the case of threes, you're going to have kind of the convexity component that's actually pretty small and in the case of fours, you're getting paid to take more convexity risk. So we think those tradeoffs are reasonable in today's environment.

Got it, and do you have a -- when you're looking at kind of portfolio or asset selection today, are you favoring kind of the lower convexity, slightly lower returned today, or are you comfortable taking some of that convexity risk and getting paid for it?

I think what's important is you have to look at that from an aggregate portfolio risk perspective. And kind of given where our overall portfolio is, we're actually comfortable adding more in fours than in threes, or 3.5. So our incremental activity has been biased, we'll say, up in coupon, because we're comfortable with the aggregate risk in the portfolio in terms of the convexity exposure.

And part of that relates to the seasoning in our portfolio, the coupon distribution. And if you look even up 100 from here, we give you the duration extension table, the portfolio goes from 1.3 years to 2.1 years, and compare that to what happened over the last, with the last 80 basis point shock. We went from, we basically had two years of extension, not one, less than one. So when you look at that, I think we have, we certainly have the capacity to take on a little bit more convexity exposure in this environment.

Makes sense, thanks, Gary.

The next question is from Joel Houck with Wells Fargo.

Thanks and good morning. I wonder if you could go back to early comments on your prepared remarks? You talked about how mortgages performed better this time than they did in the taper tantrum, which is obviously true. But talk about why that was? Was the relative performance better than you would have thought, heading into the surprise election?

Or was it not surprising? Because it looks like your performance in the fourth quarter on balance was a little bit better than kind of where people had modeled you guys in terms of book value decline.

So what I would say is the performance was kind of -- if you look at it from a big picture perspective, about where, within reason, where we would have expected. How we would have expected mortgages to behave. They widened, but the widening was much more contained, as you said, versus the taper tantrum.

And we've had discussions on calls and with investors over the past two or three years, and said that if we got a spike in interest rates we wouldn't expect a repeat of the taper tantrum and there are a couple of really good reasons. The number one reason is that when we went into the taper tantrum, MBS were at very tight levels because of the Fed's QE program that had been announced.

Option-adjusted spreads were negative 20 or thereabouts, and there was a lot of room for things to widen. In addition, you had tremendous origination volumes coming through. We had just gotten to brand new lows, refinancing volumes were -- pipelines were full.

Investors, the Fed had just kicked off the program, investors were so comfortable being long for the duration, and mortgages. And you also had other extenuating circumstances. Such as bank capital was changing at the time, and the AFS rules were adjusting, and the impact on prices on capital) So a lot of moving parts back then, of which many of those things are non-existent today, or much smaller.

So that being said, I think generally speaking, performance was somewhat logical given a big move in rates. What I would say though, that is interesting was, against the back drop of the very strong performance of other parts of the fixed income market, it gives us more comfort with respect to mortgage performance going forward.

So another thing that's different is at the beginning of the taper tantrum, other parts of the fixed income market were widening as well. Whereas in this move, that was not the case. So we do feel like against the back drop of where the rest of the fixed income space is priced, mortgages are certainly more attractive.

And hence your willingness to let duration gap move out a little more, we'd see a little bit more of a rate sell off this year?

Or leverage, or if the market gets spooked by Fed reinvestment plans and so forth, I think we would be more comfortable taking leverage up, and potentially, as we discussed earlier, maybe the tradeoff would be shorten duration gap and take leverage up.

Okay, and my last one is back on Bethesda Securities. I know Peter mentioned perhaps $10 billion to $15 billion in the next couple quarters? Is there a practical limit to the size of that, given that you probably want to maintain relationships across the street?

But how do you, longer term, how do you think about the appropriate size of your own repo through your own broker-dealer, as opposed to what you do through the rest of the street?

Joel, you're absolutely right again and this is Peter. I said $10 billion to $15 billion in dollars. I've also said before that I would expect it to be something less than maybe 30% of our funding. So that percent really gets to your point, which is we are cognizant of making sure we maintain liquidity with all our counterparts.

We have 36 repo relationships outside Bethesda Securities. We want to maintain those relationships and keep them all active, so there's that sort of practical limit. And I suspect somewhere in the neighborhood of 30%. We also can run repo on a two party basis through Bethesda Security as well.

It's going to sort of materialize over time but yes, we're cognizant of making sure we maintain active relationships with all of our counterparties.

Okay, thank you guys very much.

Thank you, Joel.

Our last question will come from the line of Brock Vandervliet with Nomura.

Thanks, I feel like I always get to ask the broken record question about dollar roll. It looked like the TVA balances fell materially, at least over the course of the quarter? But the specialness looked very special, if you could just talk about those dollar roll dynamics in the quarter, it would be helpful?

Sure, so you're right. Our TVA position was down a fair amount from our 9/30 balance, in part because the 9/30 balance had included a $2.6 billion forward settling specified pool. Pools that were brought on balance sheet during the quarter. And then the balance of the decline was just due to the fact the majority of our sales in the fourth quarter were out of a TVA position primarily in lower coupon threes.

Dollar roll financing generally traded weaker during the fourth quarter, particularly in the second half of the Fourth Quarter into year-end. It did recover a fair amount in the first week or two of the year-end to the January settlement cycle in certain coupons, but has since languished a bit since then.

Currently, rolls are trading flattish to repo, to as much as 10 or 15 basis points through in certain coupons, basically higher coupon production 30 year, 3.5 and fours, with lower coupons trading worse.

And I guess as a follow-up, I know this is always a tactical trade, but what is the long term view on specialness, especially given that the overlay of what Gary was talking about with respect to the Fed and the end of reinvestment?

I think it's logical that lower coupon 30 year threes for example, from a roll implied financing rate perspective has cheapened up, just given that coupon is no longer being produced in any size. Productions moved to 3.5 and fours. I think the other thing I'd say is, at this point there's very little downside for roll financing or roll levels.

I think one, as soon as production starts to pick up in a month or two, we're in the middle of winter seasonals right now, that should help support roll valuations. I also think once the effects of the Fed having moved up in coupon into three and a halfs and fours and cleaning up that flow, starts to become more evident that should also be supportive for rolls.

So I think it's reasonable to assume roll financing doesn't have a lot of downside from where it is today, and probably picks up or improved a bit once we get into the spring, as origination ramps up.

Just one thing to the big picture question, which is a part of that. Which is, rolls have been special well before the Fed started buying mortgages as well. So the biggest dynamic that creates some roll specialness is just the origination process. The fact that an originator is trying to hedge their production two or three months forward and selling forward and then that, in a sense is the number one core driver of why there's generally roll specialness over time in the mortgage market.

So that dynamic obviously is not going to change even if the Fed were to stop participating. So I think again we do think there will be opportunities over time in the roll market, but the Fed certainly helps the process by cleaning up the float and by kind of having a current month bid.

Okay, thank you.

Thanks a lot.

Ladies and gentlemen we have completed our question and answer session. I'd like to turn the call back over to Gary Kain for concluding remarks.

I'd like to thank everyone for their interest in AGNC and we'll speak to you again next quarter. Thank you.

Thank you. The conference has now concluded.

(Operator Instructions)

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