**Author's Note:** PART 3 of this article is a continuation from PART 1 and PART 2, which were discussed in previous publications. Please see PART 1 of this article for a composition and valuation analysis of American Capital Agency Corp.'s (NASDAQ:AGNC) MBS portfolio. Please see PART 2 of this article for a composition analysis of AGNC's derivative portfolio. PART 1 and PART 2 helps lead to a better understanding of the topics and analysis that will be discussed in PART 3.

*The link to PART 1's analysis is provided below:*

*American Capital Agency's Mid-Q4 2013 Composition And Valuation Analysis - Part 1*

*The link to PART 2's analysis is provided below:*

*American Capital Agency Corp.'s Mid-Q4 2013 Composition And Valuation Analysis - Part 2*

This three-part article is a very detailed look at AGNC's MBS and derivative portfolios. I perform this detailed analysis for readers who anticipate/want such an analysis performed each quarter. For readers who just want the summarized conclusions/results, I would suggest to just scroll down to the *"Conclusions Drawn"* section at the bottom of each part of the article.

## Focus of PART 3 of Article:

The focus of PART 3 of this article is to provide a mid-fourth quarter update on AGNC regarding the valuation of its derivative portfolio. I feel this mid-quarter update will provide readers a general direction on how the first-half of the fourth quarter of 2013 has panned out regarding AGNC's derivative strategy. PART 3 of this article will mainly focus on a detailed valuation analysis through the week ending 11/15/2013 regarding AGNC's derivative portfolio. This will also help readers understand how the second-half of the current quarter could pan out as interest rates fluctuate.

In conjunction with PART 3's valuation analysis, specific swap rates and U.S. Treasury yields will be provided. This information will assist when projecting a valuation for AGNC's derivative portfolio through the week ending 11/15/2013.

At the end of this article, a mid-fourth quarter AGNC book value ('BV') per share projection will also be provided (BV as of 11/15/2013). BV per share projections as of 11/22/2013 and 11/29/2013 will also be provided to shed more light on the CURRENT BV.

## Side Note Regarding Using Prior Quarter's Ending Derivative Balances as an Appropriate Current Quarter Valuation Technique:

Prior to beginning the valuation analysis of AGNC's derivative portfolio, I feel a discussion regarding why I use this specific valuation technique should be addressed.

A few readers in the comments section of a few of my past mREIT articles (including other authors' articles) highlighted that an mREIT will have numerous quarterly activities, which will change a derivative portfolio within any stated quarter. Therefore, these specific readers feel it is basically useless to even try and value an mREIT's derivative portfolio (or any account for that matter) in a current quarter. As was the case in PART 1 of this article concerning an mREIT's MBS portfolio, I would have to firmly disagree with this notion.

The asset valuations of any account on the balance sheet that represent the prior quarter's ending balance are definitive in nature. Regarding an mREIT's derivative portfolio, even if some of these instruments are exercised, expired, settled, or terminated in the current quarter, they still must be accounted for. They simply are not *"washed away"* and disposed of without having an accounting treatment. Regarding this particular topic, all of AGNC's derivative instruments have quarterly fair market value ('FMV') adjustments (also known as *"mark-to-market"* adjustments) which will either be *"realized"* or *"unrealized"* by the end of the fourth quarter of 2013 (FMV as of 12/31/2013). As such, knowing the FMV and notional balances of these derivative instruments as of 9/30/2013 is fairly important.

Therefore, the only aspects left open to interpretation are the amount of derivative instrument additions, exercises, expirations, settlements, and terminations in a given quarter (including the stated underlying terms). This is where a level of *"projection"* based on certain *"assumptions"* must be taken into consideration. AGNC's entire quarterly derivative valuation change will be accounted for in the following income statement account: *1)* *gain (loss) on derivative instruments, net*. The following two balance sheet accounts are affected: *1)* *derivative assets, at fair value;* and *2) derivative liabilities, at fair value*. Through detailed research and data compilation, one can project (to a reasonable degree) how management "*should"* act within any given quarter regarding additions, exercises, expirations, settlements, and terminations. However, I stress beforehand this will not be an *"exact science"* each quarter. There will be some variances that occur in a quarter if more/less additions, exercises, expirations, settlements, and/or terminations actually occur than originally projected. Additionally, unanticipated quarterly changes in stated tenor (maturities)/underlying terms held within the derivative portfolio would cause a slight deviation in asset/liability valuations.

Regarding the company's most recent quarter, AGNC decreased its net long TBA MBS and forward settling MBS position by ($21.7) billion during the third quarter of 2013. As such, AGNC went from a net long TBA MBS and forward settling MBS position of $14.8 as of 6/30/2013 to a net (short) TBA MBS and forward settling MBS position of ($7.3) billion as of 9/30/2013. Management continued to perform such a large, quick reversal of its net long TBA MBS and forward settling MBS position due to the rapid rise in mortgage interest rates/U.S. Treasury yields that occurred during the latter half of the second quarter of 2013 and most of the third quarter of 2013.

Management aggressively converted its entire net long position as of 6/30/2013 to further offset future valuation losses on its TBA MBS and forward settling MBS portfolio in a rising interest rate environment. Such an aggressive conversion early in the quarter was a deviation from past quarter activities. Furthermore, management performed a majority of its conversions when interest rates were near the highest point during the quarter thus increasing the amount of the company's TBA MBS and forward settling MBS realized losses. I personally felt this was poor timing on behalf of management.

When looking at AGNC's gain (loss) on derivative instruments, net account, I projected a valuation (loss) of ($30) million for the third quarter of 2013. My projected range for this account was ($330) - $270 million. AGNC reported a realized and unrealized valuation (loss) of ($339) million for the third quarter of 2013. This calculates to an over-projection of $309 million regarding AGNC's realized and unrealized derivatives accounts. As such, actual results were basically at the lowest end of my range. Therefore, several of my assumptions proved to be incorrect regarding the third quarter of 2013. Again, it was not the valuation technique itself that was flawed but several of the assumptions I used.

I correctly assumed management would continue to lower its net long TBA MBS and forward settling MBS position. However, I did not anticipate management converting a material proportion of the company's net long TBA MBS and forward settling MBS position so early in the quarter while interest rates were near the highest point of the quarter. Furthermore, as interest rates quickly declined during the last two weeks of the quarter, AGNC actually had a modest net (short) TBA MBS and forward settling MBS position. As such, additional valuation losses occurred during the last two weeks of the quarter. If AGNC held a portion of the company's net long TBA MBS and forward settling MBS position during the third quarter (and not converted to a net (short) position by the end of the quarter), a net realized valuation gain would have occurred within this specific portfolio.

Taking a step back, all income statement account valuation figures I projected were basically within my stated ranges for the quarter. However, most reported figures trended near the lower end of my ranges. As such, overall cumulative results were lower than I expected.

With that being said, one can still *"reasonably"* predict how management *"should"* perform regarding such activities in a specified quarter. This should be evidenced by my past articles regarding AGNC's derivative valuation projections for the second quarter of 2013. On a valuation basis, when including both realized and unrealized valuation losses regarding AGNC's derivative portfolio, I was basically *"spot-on"* with my projected $1.5 billion in derivative valuation gains. AGNC's actual derivative portfolio valuation gain was $1.44 billion, which led to a total net variance of $56 million or 4% of the account balance when compared to my projected results. Given the numerous projections and assumptions at play within this particular account, I feel my projection was pretty accurate.

AGNC's management took the appropriate steps in the second quarter of 2013 to try and minimize BV losses given its MBS portfolio as of 3/31/2013. The following are links to my second quarter of 2013 income statement and BV articles where such derivative valuations and assumptions were accurately projected:

*American Capital Agency Corp.'s Upcoming Q2 2013 Income Statement Projection (Part 1)**American Capital Agency Corp.'s Upcoming Q2 2013 Income Statement Projection (Part 2)**American Capital Agency Corp.'s Upcoming Q2 2013 Book Value Projection (As of June 30, 2013)*

To appropriately begin a current quarter's account valuation analysis, this specific technique has proved to be the best methodology in my years of research and data compilation. If a similar valuation analysis was attempted without using this specific methodology, material valuation variances would most likely occur in each instance.

## 2) AGNC's Derivative Portfolio - Valuation Analysis (Through the First-Half of the Fourth Quarter of 2013):

Using PART 2's composition analysis as a reference, we can now begin the valuation analysis of AGNC's derivative portfolio through the week ending 11/15/2013. This valuation analysis will be performed on the following four accounts that make up a majority of AGNC's derivative portfolio: *a) net TBA MBS and forward settling MBS; b) interest rate swaps; 3) interest rate swaptions;* and *4) net U.S. Treasury securities.*

* Side Note:* To better illustrate the nature and function of AGNC's derivative portfolio, I show each derivative instrument as being net

*"long"*or

*"(short)."*All net long positions have a

*"direct"*relationship with AGNC's MBS portfolio regarding valuation movements. As such, if AGNC's MBS portfolio has a valuation gain during a specified time period, AGNC's net long derivative position(s) will likely have a similar valuation gain (and vice versa). All (short) positions have an

*"inverse"*relationship with AGNC's MBS portfolio regarding valuation movements. As such, if AGNC's MBS portfolio has a valuation gain during a specified time period, AGNC's net (short) derivative position(s) will likely have an offsetting valuation loss (and vice versa). This can be confusing for some readers, but I feel presenting the information this way is the most effective when trying to truly show the nature of AGNC's derivative instruments. If a particular reader ever becomes confused while reading this article, I would suggest referring back to this side note to gain one's bearings.

It should also be noted not all of AGNC's derivative instruments are held for the sole purpose of mitigating a valuation loss on the company's MBS portfolio. AGNC holds different types of derivative instruments for different purposes. For example, AGNC has hedging instruments in place (interest rate swaps and swaptions) which help offset both a valuation loss on the company's MBS portfolio and the rising interest costs of repurchase ('repo') loans. Other derivative instruments (U.S. Treasury securities, *"to-be-announced"* ('TBA') MBS, and forward settling MBS) can have varying uses depending on whether each derivative instrument is net long (short). AGNC's U.S. Treasury securities, TBA MBS, and forward settling MBS typically have a direct relationship to the company's MBS portfolio when a net long position exists (non-hedging position). However, when these derivative instruments are in a net (short) position, an inverse relationship is typically developed and these instruments become hedges per se regarding valuation changes on AGNC's MBS portfolio.

However, AGNC's derivative portfolio is not implicitly designed to protect the company's book value ('BV') from *"spread/basis risk."* Spread/basis risk is the risk of an increase of the market spread between the yield on AGNC's MBS portfolio and the rates/yields on the company's derivative portfolio. The spread/basis risk between AGNC's MBS and derivative portfolios can cause short-term FMV fluctuations that are independent of general market interest rates. As such, spread/basis risk could relate to other factors which impact the current and future mortgage/fixed income markets. One recent example is the current and/or anticipated future monetary policies by the Federal Reserve ('FED'). Therefore, while a net long (short) derivative position should move in a direct (inverse) correlation to AGNC's MBS portfolio under normal market conditions (regarding general valuations), there are times when spread/basis risk causes a dysfunction between the two portfolios and their typical correlated valuation movements. Readers should be aware of such risk and understand this ultimately hinders management's strategies regarding the mitigation of BV erosion.

**a)** **Net TBA MBS and Forward Settling MBS (Net (Short) Position as of 9/30/2013):**

Having established the composition of AGNC's TBA MBS and forward settling MBS portfolio in PART 2 of this article, we can now begin this account's valuation analysis. Both quarterly realized and unrealized valuation changes and dollar roll income (expense) associated with AGNC's TBA MBS and forward settling MBS portfolio are recognized in the gain (loss) on derivative instruments and other securities, net account within the income statement.

For the sake of reducing redundancy within PART 3 of this article, I would refer readers to PART 1 of this article regarding specific fixed-rate agency MBS price movements for 15-year (Table 3) and 30-year (Table 4) holdings across the various coupon rates. Table 3 and Table 4 break out the MBS price movements by government sponsored entity ('GSE'). This includes both Freddie Mac (OTCQB:FMCC) and Fannie Mae (OTCQB:FNMA) MBS. As was the case with AGNC's MBS portfolio in PART 1 of this article, these same two tables are the *"cornerstone"* of providing precise valuation projections for AGNC's TBA MBS and forward settling MBS portfolio.

Using Table 3 and Table 4 from PART 1 of this article as a reference, along with making several assumptions regarding AGNC's TBA MBS and forward settling MBS net changes during the current quarter (helped by providing Table 9 in PART 2 of this article as a reference), Table 14 below shows AGNC's weekly and cumulative quarterly projected valuation gain (loss) on the company's TBA MBS and forward settling MBS portfolio through the week ending 11/15/2013.

## Table 14 - AGNC Weekly and Cumulative Quarterly Valuation Gain (Loss) on its TBA MBS and Forward Settling MBS Portfolio (Through the Week Ending 11/15/2013)

Regarding AGNC's net (short) TBA MBS and forward settling MBS portfolio as of 9/30/2013, I have projected a quarterly net valuation gain of approximately $35 million (rounded) through the week ending 11/15/2013 (see first blue reference *"A"* in Table 14 above). This is either through realized or unrealized net valuation gains. As stated above, since mortgage interest rates/U.S. Treasury yields have increased when compared to rates/yields as of 9/30/2013, a net (short) position should incur valuation gains.

* Side Note:* I feel when projecting a company's intra-quarter accounting figures, all typical quarterly income and expenses within an account should be projected over the entire quarter versus a proportional quarterly breakout. However, regarding asset valuations, certain conversions/settlements or purchases/sales only occur during a specified

*"point-in-time"*during the quarter. As such, when specifically interpreting this methodology to AGNC's TBA MBS and forward settling MBS portfolio, I project AGNC's quarterly dollar-roll income (expense) as the total income (expense) through the entire quarter versus a proportional share. This quarterly balance only changes when my underlying assumptions regarding the composition of AGNC's TBA MBS and forward settling MBS portfolio change. In contrast, I project the quarterly valuation gain (loss) of AGNC's TBA MBS and forward settling MBS portfolio at a specified point in time (typically on a weekly basis). This way, when projecting the remainder of the quarter, only specified point-in-time valuation adjustments are accounted for and updated (as long as my underlying assumptions do not change). If I feel a need to change the company's quarterly dollar-roll income (expense) account (due to a material change to the company's TBA MBS and forward settling MBS portfolio), then I make a

*"true-up"*or

*"true-down"*adjustment accordingly. When this methodology is used, I feel intra-quarter account figures are not

*"skewed"*during any point of the quarter.

As such, I have projected a quarterly net dollar-roll (expense) of ($43) million for the fourth quarter of 2013. Even though AGNC suggested it may once again reverse the company's net (short) TBA MBS and forward settling MBS position to a net long position by the end of the fourth quarter of 2013, I feel the recent upward movement in mortgage interest rates/U.S. Treasury yields has prompted management to either maintain, or reestablish, the company's net (short) position. As mentioned in PART 2 of this article, as overall mortgage interest rates/U.S. Treasury yields increase, AGNC should maintain a net (short) position within this account to incur valuation gains within the company's TBA MBS and forward settling MBS portfolio (thus offsetting MBS portfolio valuation losses). When AGNC is in a net (short) TBA MBS and forward settling MBS position, it generates a dollar-roll (expense) as opposed to dollar-roll income when in a net long position. As such, AGNC's projected quarterly dollar-roll (expense) should modestly increase during the fourth quarter of 2013 when compared to the prior quarter. This quarterly dollar-roll (expense) is partially offset by a projected net (short) valuation gain of $20 million through the week ending 11/15/2013. Since I have projected AGNC currently has a net (short) TBA MBS and forward settling MBS position, all intra-quarter (short) TBA contracts would have a valuation gain as mortgage interest rates/U.S. Treasury yields moved higher during the first-half of the fourth quarter of 2013 (after an initial minor movement lower). These valuation gains would be partially offset by all intra-quarter long TBA contracts, which would have a valuation (loss). Using Table 14 above as a reference, these two combined figures result in a net valuation (loss) of ($23) million through the week ending 11/15/2013 (see second blue reference *"A"* in Table 14 above).

Therefore, when combining these two blue references together, I have projected a quarterly total net valuation gain of approximately $15 million (rounded) regarding AGNC's net (short) TBA MBS and forward settling MBS account through the week ending 11/15/2013. This incorporated all long (short) TBA MBS and forward settling MBS positions as of 9/30/2013 including various assumptions regarding quarterly *"re-rolls"*, conversions/settlements, and sales (if applicable).

**b)** **Interest Rate Swaps (Net (Short) Position as of 9/30/2013):**

Having established the composition of AGNC's interest rate swaps in PART 2 of this article, we can now begin this account's valuation analysis. As was the case with AGNC's TBA MBS and forward settling MBS portfolio, quarterly realized and unrealized interest rate swap valuation changes are recognized in the gain (loss) on derivative instruments and other securities, net account within the income statement.

AGNC's interest rate swaps are typically derived from the *"over-the-counter"* ('OTC') market and may be *"centrally cleared"* through a registered commodities exchange. AGNC values these centrally cleared interest rate swaps using the daily settlement prices determined by the corresponding OTC market/exchange. The OTC market/exchange typically uses a pricing model that incorporates underlying swap rates, overnight index swap rates, and forward LIBOR percentages to produce daily settlement prices. In some instances, AGNC may also have *"non-centrally cleared"* interest rate swaps. AGNC values the company's non-centrally cleared swaps using a combination of inputs from counterparties and independent pricing models to estimate the net present value ('NPV') of the future cash flows of the swap using a forward interest rate yield curve in effect as of the end of the measurement period. This includes any adjustments that may need to be accounted for due to the potential *"non-performance"* of a particular counterparty holding the interest rate swaps (if applicable; non-performance risk).

It should be noted a projected quarterly net valuation change in AGNC's interest rate swaps account can also be manually performed if one has the expertise to perform such a calculation and certain variables and interest rate swap rates are known. These manual calculations are a great tool to understand when comparing results to the various model projections. One key piece of information to obtain to begin a proper valuation of an interest rate swap is the weekly and cumulative quarterly net change of the fixed rate payer side of the swap.

## Table 15 - Weekly and Cumulative Quarterly Interest Rate Swap Rates (Fixed Rate Payer Side of Swap) (Through the Week Ending 11/15/2013)

* Side Note:* Since the weekly and cumulative quarterly net change to 3-month LIBOR had been immaterial through the week ending 11/15/2013 (what the receiver side of the swap is currently based on), a discussion on the net change of the receiver side of the interest rate swap will be omitted from this analysis. This includes the omission of a LIBOR table since 3-month LIBOR had only moved 1 basis point ('bp') during the first-half of the fourth quarter of 2013.

Using Table 15 above as a reference, one can see as the tenor (or maturity) of an interest rate swap increased, the rate paid by the fixed rate payer increased as well. For instance, as of 9/30/2013, the rate paid by the fixed rate payer on a 7-year interest rate swap was 2.19% while the rate increases to 2.79% on a 10-year interest rate swap.

Through the week ending 11/15/2013, the cumulative quarterly rate paid by the fixed rate payer had generally decreased regardless of tenor (except for the 10-year interest rate swap). For instance, the rate paid by the fixed rate payer on a 3-year interest rate swap had decreased 12 bps through the week ending 11/15/2013 while the rate paid by the fixed rate payer on a 7-year interest rate swap had decreased 5 bps. However, the rate paid by the fixed rate payer on a 10-year interest rate swap had no net change through the week ending 11/15/2013. These subtle changes may not seem like a big difference. However, since the rate paid by the fixed rate payer on a 10-year swap had not changed while the rate paid by the fixed year payer on a 3-year swap decreased 12 bps, the 10-year swap would have no quarterly valuation change while the 3-year swap would have a modest quarterly valuation loss. Since AGNC had a net (short) interest rate swaps position of ($50.2) billion as of 9/30/2013, even subtle weekly and cumulative quarterly basis point movements cause material shifts to valuations.

Table 16 below shows AGNC's weekly and cumulative quarterly projected valuation gain (loss) on the company's interest rate swaps through the week ending 11/15/2013. These weekly and cumulative quarterly projected valuation changes include making several assumptions regarding AGNC's interest rate swap additions, expirations, and terminations during the current quarter.

## Table 16 - AGNC Weekly and Cumulative Quarterly Valuation Gain (Loss) on its Interest Rate Swaps (Through the Week Ending 11/15/2013)

Using Table 16 above as a reference, I have projected a quarterly net interest rate swaps valuation (loss) of approximately ($190) million (rounded) through the week ending 11/15/2013 (see first blue reference *"B"* in Table 16 above). However, as mentioned in PART 2 of this article, AGNC also needs to account for the company's quarterly net periodic costs on interest rate swaps. These costs are the net pay rate (fixed rate paid less floating/variable rate received) AGNC is currently paying on all of the company's interest rate swaps. Through a valuation table not shown within this analysis, I have projected AGNC will record quarterly net periodic costs on its interest rate swaps of ($115) million (see second blue reference *"B"* in Table 16 above). I have also projected a slight quarterly valuation (loss) on AGNC's interest rate swaps within the Markit IOS total return swap index of ($5) million (net of all interest rate swap termination fees; see third blue reference *"B"* in Table 16 above).

Therefore, when combining these three blue references together, I have projected a quarterly total net valuation (loss) of ($310) million regarding AGNC's net (short) interest rate swaps account through the week ending 11/15/2013.

## c) Interest Rate Swaptions (Net (Short) Position as of 9/30/2013):

Having established the composition of AGNC's interest rate swaptions in PART 2 of this article, we can now begin this account's valuation analysis. As was the case with AGNC's TBA MBS and forward settling MBS portfolio and interest rate swaps, quarterly realized and unrealized interest rate swaption valuation changes are recognized in the gain (loss) on derivative instruments and other securities, net account within the income statement.

As a reminder, interest rate swaptions are basically options to enter into underlying interest rate swap contracts. Whereas interest rate swap contracts have no initial upfront costs (gains and losses are incurred as interest rates fluctuate over the life of the swaps), interest rate swaptions have implicit upfront costs (similar to an option contract; generally speaking).

AGNC's interest rate swaptions are valued using a combination of inputs from counterparties and independent pricing models based on the FMV of the underlying interest rate swaps. An additional calculation based on the remaining length of the option needs to be performed as well. This includes any adjustments that may need to be accounted for due to the potential *"non-performance"* of a particular counterparty holding the option (if applicable; non-performance risk). If a swaption expires unexercised, the realized (loss) on the interest rate swaption would be the upfront cost paid on the creation/addition of the swaption. If AGNC sells or exercises an interest rate swaption, the realized gain (loss) would be the difference between the FMV of the underlying interest rate swap and the upfront cost paid on the creation/addition of the swaption.

It should be noted a projected quarterly net valuation change in AGNC's interest rate swaptions account can also be manually performed if one has the expertise to perform such a calculation and certain variables and interest rate swap rates are known. These manual calculations are a great tool to understand when comparing results to the various model projections. As was the case with AGNC's interest rate swaps, one key piece of information to obtain to begin a proper valuation of an interest rate swaption is the weekly and cumulative quarterly net change of the fixed rate payer side of the underlying interest rate swap

Table 17 below shows AGNC's weekly and cumulative quarterly projected valuation gain (loss) on the company's interest rate swaptions through the week ending 11/15/2013. These weekly and cumulative quarterly projected valuation changes include making several assumptions regarding AGNC's interest rate swaption additions, exercises, expirations, and terminations during the current quarter.

## Table 17 - AGNC Weekly and Cumulative Quarterly Valuation Gain (Loss) on its Interest Rate Swaptions (Through the Week Ending 11/15/2013)

Using Table 17 above as a reference, I have projected a quarterly net interest rate swaptions valuation (loss) of approximately ($75) million (rounded) through the week ending 11/15/2013 (see first blue reference *"C"* in Table 17 above).

However, as mentioned in PART 2 of this article, AGNC also needs to account for the company's quarterly costs in relation to all interest rate swaption additions, exercises, expirations, and terminations during the current quarter. Since AGNC had ($13.2) billion of interest rate swaptions set to expire in 1 year or less, I have projected the company acquired ($3.5) billion of interest rate swaptions (based on the notional value of the underlying interest rate swaps) at an initial cost of ($110) million. I have also projected the company exercised, had expired, or terminated $2.5 billion of interest rate swaptions for a net valuation gain of $40 million. Using Table 14 above as a reference, these two combined figures result in a net valuation (loss) of ($70) million through the week ending 11/15/2013 (see second blue reference *"C"* in Table 17 above).

Therefore, when combining these two blue references together, I have projected a quarterly total net valuation (loss) of ($145) million regarding AGNC's net (short) interest rate swaptions account through the week ending 11/15/2013.

## d) U.S. Treasury Securities (Net Long Position as of 9/30/2013):

Having established the composition of AGNC's U.S. Treasury securities in PART 2 of this article, we can now begin this account's valuation analysis. As was the case with AGNC's TBA MBS and forward settling MBS portfolio, interest rate swaps, and interest rate swaptions, quarterly realized and unrealized U.S. Treasury security and U.S. Treasury security futures valuation changes are recognized in the gain (loss) on derivative instruments and other securities, net account within the income statement.

AGNC purchases (or sells short) U.S. Treasury securities and U.S. Treasury security futures to help mitigate the potential impact of changes in mortgage interest rates (hence the valuation of the company's MBS portfolio). AGNC borrows securities to cover (short sales) of U.S. Treasury securities under reverse repo agreements. AGNC accounts for these instruments as *"security borrowing transactions"* and recognizes an obligation to return the borrowed securities at FMV based on the current value of the underlying borrowed securities.

It should be noted a projected quarterly net valuation change in AGNC's U.S. Treasury securities and U.S. Treasury security futures account can also be manually performed if one has the expertise to perform such a calculation and certain variables, U.S. Treasury yields, and short-term repo loan rates are known. One key piece of information to obtain to begin a proper valuation of a U.S. Treasury security is the weekly and cumulative quarterly net change of the yield on the various U.S. Treasury security maturities.

## Table 18 - Weekly and Cumulative Quarterly U.S. Treasury Security Yields (Through the Week Ending 11/15/2013)

Using Table 18 above as a reference, one can see as the maturity of a U.S. Treasury security increased, the yield increased as well. For instance, as of 9/30/2013, the yield obtained on a 5-year U.S. Treasury security was 1.39% while the yield increased to 2.64% on a 10-year U.S. Treasury security.

Through the week ending 11/15/2013, the cumulative quarterly yield change on a U.S. Treasury security with a 5-year maturity decreased by 3 bps while a U.S. Treasury security with a 10-year maturity increased 7 bps. These subtle changes may not seem like a big difference. However, these cumulative quarterly yield changes can materially impact valuations through the various U.S. Treasury security maturities.

When compared to the prior quarter, these bps movements had the exact opposite net effect during the current quarter because AGNC had a net (short) U.S. Treasury security position of ($9.2) billion as of 6/30/2013 and a net long U.S. Treasury security position of $1.2 billion as of 9/30/2013. Even though AGNC's net long U.S. Treasury security position was only $1.2 billion as of 9/30/2013, even subtle weekly and cumulative quarterly basis point movements cause modest shifts to valuations.

Table 18 below shows AGNC's weekly and cumulative quarterly projected valuation gain (loss) on the company's U.S. Treasury securities and U.S. Treasury security futures through the week ending 11/15/2013. These weekly and cumulative quarterly projected valuation changes include making several assumptions regarding AGNC's U.S. Treasury security and U.S. Treasury security futures additions and settlements during the current quarter.

## Table 19 - AGNC Weekly and Cumulative Quarterly Valuation Gain (Loss) on its U.S. Treasury Securities and U.S. Treasury Security Futures (Through the Week Ending 11/15/2013)

Using Table 19 above as a reference, I have projected a quarterly net U.S. Treasury securities and U.S. Treasury security futures valuation gain of approximately $20 million (rounded) through the week ending 11/15/2013 (see first blue reference *"D"* in Table 19 above). However, AGNC also needs to account for the company's quarterly interest income (expense) on all long (short) positions. I have projected quarterly net interest income of approximately $10 million (rounded) through the week ending 11/15/2013 (see first blue reference *"D"* in Table 19 above). Both figures take into account all assumed U.S. Treasury security and U.S. Treasury security futures quarterly additions and settlements through the week ending 11/15/2013.

Therefore, when combining these two blue references together, I have projected a quarterly total net valuation gain of $30 million regarding AGNC's net long U.S. Treasury securities and U.S. Treasury security futures account through the week ending 11/15/2013.

## Brief Discussion of MTGE's Derivative Portfolio - Valuation Analysis:

AGNC and American Capital Mortgage Investment Corp. (NASDAQ:MTGE) had a slightly different hedging portfolio as of 9/30/2013. This was displayed within Table 13 in PART 2 of this article. As of 9/30/2013, AGNC had a hedging coverage rate of 93% while MTGE had a hedging coverage rate of only 77%. As the percentages indicate, MTGE had a less hedged portfolio as of 9/30/2013 when compared to AGNC. However, it should be noted MTGE's hedging portfolio had a higher overall weighted average duration (model projection of interest rate sensitivity in years) when compared to AGNC. As of 9/30/2013, MTGE's hedging portfolio had a weighted average duration of (5.7) years while AGNC's hedging portfolio had a weighted average duration of only (3.6) years. These subtle differences will cause slight valuation differences between the two derivative portfolios for the current quarter (proportionally speaking).

I have similar weekly and cumulative quarterly valuation gain (loss) tables for MTGE regarding the company's TBA MBS and forward settling MBS portfolio, interest rate swaps, interest rate swaptions, and U.S. Treasury securities. However, this article's main focus is on AGNC (not MTGE). As such, I have projected the following quarterly total net valuation gains (losses) regarding MTGE's derivative portfolio through the week ending 11/15/2013:

**a)***TBA MBS and forward settling MBS total net valuation gain of $6 million;*

**b)***Interest rate swaps total net valuation (loss) of ($39) million;*

**c)***Interest rate swaptions total net valuation (loss) of ($17) million; and*

**d)***U.S. Treasury securities and U.S. Treasury security futures total net valuation gain of $6 million*

Therefore, when combining these four accounts together, I have projected a quarterly total net valuation (loss) of ($44) million regarding MTGE's derivative portfolio through the week ending 11/15/2013.

## Conclusions Drawn (AGNC's Derivative Portfolio - Valuation Analysis):

I have projected the following quarterly total net valuation gains (losses) regarding AGNC's derivative portfolio through the week ending 11/15/2013:

**a)***TBA MBS and forward settling MBS total net valuation gain of $12 million;*

**b)***Interest rate swaps total net valuation (loss) of ($310) million;*

**c)***Interest rate swaptions total net valuation (loss) of ($145) million; and*

**d)***U.S. Treasury securities and U.S. Treasury security futures total net valuation gain of $30 million*

When combining these four accounts together, I have projected a quarterly total net valuation (loss) of approximately ($413) million regarding AGNC's derivative portfolio through the week ending 11/15/2013.

When comparing MBS, interest rate swap, interest rate swaption, and U.S. Treasury security quarterly total net valuation changes through the week ending 11/15/2013, it appears AGNC's net long MBS portfolio, net (short) interest rate swaps, and net (short) interest rate swaptions experienced minor to modest valuation losses. AGNC's net (short) TBA MBS and forward settling MBS portfolio and net long U.S. Treasury securities had minor valuation gains. However, as shown above, the quarterly total net valuation (loss) trumps the quarterly total net valuation gain.

As stated above, these results exhibit what is known as spread/basis risk. Spread/basis risk is the risk of an increase of the market spread between the yield on AGNC's MBS portfolio and the rates/yields on the company's derivative portfolio. The spread/basis risk between AGNC's MBS and derivative portfolios can cause short-term FMV fluctuations that are independent of general market interest rates.

Therefore, while a net long (short) derivative position should move in a direct (inverse) correlation to AGNC's MBS portfolio under normal market conditions (regarding general valuations), there are at times when spread/basis risk causes a dysfunction between the two portfolios and their typical correlated valuation movements. I personally feel spread/basis risk has risen due to the *"uncertain"* nature of future economic policy regarding the Federal Reserve's ('FED') Quantitative Easing ('QE3') bond purchasing program (monthly MBS and U.S. Treasury security purchases) and its eventual exit of the mortgage/bond markets via *"tapering."* As such, spread/basis risk may continue to remain elevated over the next several quarters while the markets continue to speculate how interest rates will react in the future.

## BV Projection (For the Weeks Ending 11/15/2013, 11/22/2013, and 11/29/2013):

Finally, we have arrived at AGNC's projected BV per share based on PART 1's MBS portfolio valuation analysis and PART 3's derivative portfolio valuation analysis (including all other accounts not discussed within this three-part article).

* Side Note to BV Projection:* Prior to providing three weekly BV per share projections, it should be noted all remaining revenue and expense accounts are already projected out for the entire fourth quarter of 2013 (as stated earlier). I project my BV per share amounts like this because eventually AGNC will incur quarterly activities within the company's revenue and expense accounts. As such, I feel a more precise mid-quarter BV will be obtained when performing the BV projection based on this methodology. Instead of prorating quarterly interest income for only half of the quarter, the full quarter's worth of interest income is taken into consideration for the mid-quarter BV per share projection. One example would be quarterly interest income from MBS holdings. The same holds true for all expense accounts not already discussed within this three-part article. One example would be quarterly repo loan interest expense. In my opinion, this provides a more consistent and precise BV projection when ultimately performing this same BV per share projection as of 12/31/2013.

## The following are my AGNC BV per share projections:

**BV as of 11/15/2013: $25.10 per share (Range $24.60 - $25.60 per share)**

**BV as of 11/22/2013: $24.70 per share (Range $24.20- $25.20 per share)**

**BV as of 11/29/2013: $24.25 per share (Range $23.75- $24.75 per share)**

These BV per share projections EXCLUDE the soon to be declared dividend for the fourth quarter of 2013 because AGNC's stock price has yet to *"reset"* lower. This occurs when the stock reaches its *"ex dividend"* date.

I would caution readers that AGNC's CURRENT BV per share amount changes daily. One quick, sharp move in mortgage interest rates, MBS prices, interest rate swap rates, and U.S. Treasury yields can materially change AGNC's CURRENT BV per share by material amounts. This is why I feel it is imperative readers understand how AGNC values the company's MBS and derivative portfolios and show how modest changes in overall rates, prices, and/or yields have a direct impact on AGNC's asset valuations (hence BV per share amount). This is why I provide readers with three weeks of BV per share amounts to illustrate how AGNC's CURRENT BV frequently changes thus causing material BV per share movements.

Since AGNC's stock price continues to trade at a material discount to CURRENT BV, I would **recommend a HOLD** position due to the current volatile and uncertain interest rate environment. However, some attractive opportunities to purchase initial/additional shares could continue as markets speculate future economic conditions/FED monetary policies.

* Final Note:* An article about AGNC's fourth quarter of 2013 dividend range scenarios will be provided prior to the company's quarterly dividend declaration in the second or third week of December 2013.

Comments()