By the Numbers
Shopping a $14 billion portfolio
Mary Beth Fisher, PhD | April 21, 2023
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
This article has been corrected and updated to reflect that the Silicon Valley and Signature Bank securities portfolios are not eligible collateral for financing at the Fed’s Bank Term Funding Program.
Speculation has built around the potential market impact of the $14 billion liquidation of agency CMBS formerly held by Silicon Valley Bank and Signature Bank. The holdings are overwhelmingly comprised of recent vintage, low coupon Ginnie Mae project loans whose indicative market values are substantially below par, but not low enough to make the assets positive carry. The size of the portfolio relative to outstanding issuance, and low liquidity in secondary trading of Ginnie Mae project loans also make such a liquidation challenging. There are three potential outcomes: a deep price cut to attract either a broker-dealer or levered buyers; a strong bank buyer who would likely negotiate long-term financing similar to that offered by the Fed’s bank term funding facility; or a Fed sponsored Maiden Lane type facility which would take ownership of the assets and manage them over time.
There are some idiosyncrasies about the Ginnie Mae project loan market and the characteristics of the bonds that are likely to make the sale of the portfolio challenging:
- Ginnie Mae securities are fully guaranteed by the US government and have a 0% risk weighting when calculating capital requirements. The natural buyer of the principal and interest classes are banks, insurance companies, pension funds and Federal Home Loan Banks. The securities are typically classified as held-to-maturity assets.
- The secondary market for Ginnie Mae project loans is thin in normal circumstances. The current environment where held-to-maturity portfolios are under intense scrutiny has significantly curtailed the potential buyer base.
- The GNPL bonds held by Signature Valley Bank (SVB) were almost entirely from the 2020 and 2021 vintages, which have an average coupon of 1.62%, an average WAC of 2.72% and WAL of 4.5 years assuming a prepayment speed of 15 CPJ (Exhibit 1). Although 15 CPJ often used for par coupon bonds at new issue, it’s an unrealistic speed assumption for current pricing of these securities.
Exhibit 1: Selection of SVB GNPL holdings
Even priced at 15 CPJ the average discount dollar price across the portfolio is 83.75; that’s still too high given that project loans deep out-of-the-money to refinance historically prepay at 0 to 5 CPJ. As an example, one of the bonds in the portfolio is the GNR 2021-151 AB security (Exhibit 2). It has a 1.75% coupon and a 2.95% WAC. Current project loan commitment rates are 4.75% to 5.00%, meaning the borrowers are about 200 bp out-of-the-money to refinance.
Exhibit 2: Example yield analysis using adjusted pricing
The yield table analysis shows two potential prices for this security: one with a 78 handle and one with a 75 handle. At a flat 78 dollar price and 2 CPJ the yield rises to 4.15% with a J-spread of 48 bp. At 5 CPJ the yield is 5.04% and the J-spread is 148 bp. A rise in prepayment speeds over time, which will almost certainly happen but could take several years, would result in even higher yields.
In the short term, with prepayment speeds slow and the front end of the curve steeply inverted, the bond is likely to remain persistently negative carry at speeds below 5 CPJ because the yield on the bond is lower than financing cost. There are two important caveats:
- The carry is not nearly as negative as it looks even at 0 CPJ because of the deep discount dollar price
- Layering on a pay-fixed, receive-floating swap with the same notional as the market value of the bond will not only hedge the interest rate risk, but will also result in a portfolio that is positive carry.
A note about carry
The shortcut for estimating carry is subtract the financing cost, typically the overnight or term repo rate in percent, from the yield of the bond. This estimate works great for fixed-rate bonds, with principal that cannot be prepaid, trading roughly at par. For example, the 5-year on-the-run Treasury security (T 3.625 3/31/28) has a yield of 3.626% and is trading about a quarter tick below par. Assuming a repo rate of 4.83%, the estimated carry in percent would be 3.626% – 4.83% = -1.20% annually, in a static rate environment.
This shortcut is problematic for any bonds trading at a significant discount or premium from par, and for mortgages where prepayments can dramatically alter the cash flows and impact the yield. The actual carry on the bond can be higher than estimated when securities are at a deep discount, because the coupon and principal is earned on the face amount while the repo interest is paid on the market value. The reverse is also true for bonds trading at a premium.
In the case of the GNPL portfolio that is being liquidated, although the coupon interest rates on the securities are rather low, the principal and interest securities in the portfolio benefit from principal payments due to scheduled amortization. These payments can about double the monthly cash flow received, even in the early years when prepayments tend to be low. For example, the GNR 2021-151 AB has a 1.75% interest coupon with an amortization schedule that is adding about 1.71% in principal payments each month, for a combined cash flow of 3.46% of the face amount for the current month (Exhibit 3). To date, this bond has at lifetime CPR of 3, but has had zero prepayments for the last year. At a 78.00 price, the negative carry is only -0.47%, even at 0 CPR. A market price with a 71 handle to make the carry flat at 0 CPR.
Exhibit 3: Projected carry calculation for GNPL assuming 78.00 price, 0 CPR, 0 CPD
Current overnight repo rates are about 4.83% and heading higher, assuming the FOMC raises the short rate to 5.25% at the May meeting. Prepayments in recently issued Ginnie Mae project loans deep out-of-the-money to refinance will likely stay low until the Fed eventually starts to ease and/or borrowers build up enough property price appreciation to make a sale attractive.
Finding an interested buyer
Hedge funds and private equity investors might be willing to buy the portfolio and lever it up given a 5% haircut on financing, since they tend to hedge the interest rate risk and the pay-fixed swap is positive carry. The required discount could be less for total return investors looking at longer holding periods.
A strong bank could buy the portfolio on the condition that the Fed or another counterparty is willing to finance the securities on a non-mark-to-market basis at below-market rates, similar to the terms recently established by the Fed’s Bank Term Funding Program. The facility lends money against the par value of the securities, not the market value, at an advance rate of 1-year OIS plus 10 bp. A bank could currently renew that funding for at least two years, through March 2025.
All that being said, there is still a lot of duration, prepayment and financing risk for either buyer. The securities are also fairly illiquid, and relative to the modest amount of secondary trading in the market, the portfolio the FDIC is trying to sell is gigantic (Exhibit 4). Prior to the pandemic, total issuance in Ginnie Mae project loans was $15 to $20 billion per year. There is currently $65 billion outstanding of Ginnie Mae project loans of 2020 and 2021 vintages, and the SVB portfolio holds about $12 billion of that, or 18.5%. Any buyer is probably going to have to plan to hold the portfolio to maturity.
Exhibit 4: GNPL issuance and outstanding
The Maiden Lane option
In reality, it seems unlikely that the FDIC and BlackRock will be willing to cut the price of the portfolio enough to attract a leveraged investor. Bank investors could eventually emerge, but given the enormous scrutiny on earnings amid commercial real estate rumblings, it seems unlikely that the portfolio could be sold off in a timely fashion.
That leaves the strategy that many have been suggesting since the collapse – that the Fed should set up a fund similar to the Maiden Lane LLCs that took over the risky assets of Bear Stearns that JP Morgan and other banks shunned at the time. It’s unclear what the political risk might be from taking over the the portfolio, but it potentially presents the least risk to the financial markets, which would struggle to digest it.