Shifting rates strain Street models of MBS
admin | March 13, 2020
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.
The recent sharp move to historically low interest rates has put surprising strain on some commonly used Street models of MBS. The FTSE Russell Yield Book, widely used by investors in pass-throughs, CMOs and MBS derivatives, has shown a sharp drop in estimated MBS duration in recent weeks based apparently on Yield Book projections for the primary 30-year mortgage rate available to borrowers. The mortgage market sell-off on March 10 has temporarily resolved this issue by increasing the projected mortgage rate, but the extreme model volatility highlights the challenges modeling MBS in this market. Models become more important in times of market stress, but at the same time may become less reliable.
Currently almost the entire pass-through market is trading at a premium and is exposed to the risk of faster prepayments. Refinancing is the most volatile component of mortgage prepayments and is the most susceptible to modeling errors. Certain sectors of the mortgage market are particularly exposed. MBS derivatives are heavily levered to prepayment speeds and models play a central role in valuing and hedging these bonds. Specified pools also depend on models to value and hedge price premiums to TBA.
Interest rate modeling also plays a fundamental role in valuing MBS, although investors usually consider prepayments models as the primary source of MBS modeling error. This assumption may be flawed with interest rates at records lows and close to 0%. Many investors use classic interest rate models that don’t permit negative interest rates. Not allowing negative interest rates means these rate models may understate the negative convexity and option cost of mortgages.
Other market participants have moved to models that permit negative interest rates, but there are different approaches. For example, some models use a negative rate floor, while others might permit rates to fall without bound.
Yield Book users started to notice the drop in MBS durations as rates fell in late February and through the first week of March. Among others, the duration on Fannie Mae 30-year 2.5% TBA contracts dropped from more than 3.0 on February 21 to 0.3 by March 6 and moved negative on March 9 (Exhibit 1). Amherst Pierpont models on March 6 saw the 2.5% duration at 3.08 years. The Credit Suisse trading desk posted an effective duration for the 2.5% contract on March 6 of 3.07 years, and the JPMorgan trading desk posted 2.47 years. On March 10 Yield Book’s numbers jumped back to levels comparable to late February.
Exhibit 1: Yield Book OADs for Fannie Mae 30-year pass-throughs fell sharply
Users quickly focused on Yield Book projections of primary mortgage rates. Based on Amherst Pierpont estimates, on March 6 Yield Book assumed an initial primary 30-year rate of more than 3.20% but projected a decline of more than 60 bp even with rates constant (Exhibit 2). Declining primary rates drive refinancing higher and shorten MBS duration. Yield Book published a note on March 9 describing its method for projecting primary rates, apparently in response to user inquiry. The company also said it working to make projected primary rates visible to users in future releases.
Secondary mortgage rates increased almost 50 bp on March 10, which pushed projected primary rates above 3% and caused Yield Book’s projected durations to lengthen significantly. At this point they are much closer to durations used by trading desks and durations from other models.
Exhibit 2: Primary rates drop quickly in Yield Book even with UST rates constant
Originators handling heavy refinancing often keep rates elevated both to manage loan processing volume and make profits by selling loans at premium prices into the secondary market. Mortgage rates lately have lagged the drop in Treasury rates and MBS yields, with the spread between primary and secondary mortgage rates reaching more than 160 bp (Exhibit 3).
Exhibit 3: Primary mortgage rates have lagged the sharp drop in secondary rates
Originators have steadily added personnel in the last decade, including a burst of hiring for loan brokers and supporting staff in late 2019 (Exhibit 4). But the latest drop in rates has stretched capacity.
Exhibit 4: Originators have added staff lately, but not enough
The Mortgage Bankers Association’s mortgage refinance index, which measures weekly applications to refinance, has surged to the highest level since early 2013 and has more than doubled since the beginning of 2020 (Exhibit 5). The index has room to rise further. It reached much higher levels in 2012 and 2013 when a similar portion of loans were deeply in-the-money to refinance. The MBA reports the index weekly for the preceding week’s activity, so it does not yet incorporate data the week ending March 6.
Exhibit 5: Mortgage refinance applications are the highest since 2013
The increase in applications translates almost directly into a larger workload for loan brokers. That workload has also jumped to the highest level since early 2013 (Exhibit 6). Workload is measured by dividing the refinance index by the number of loan brokers and supporting staff employed by the industry. The burst of hiring at the end of 2019 had little effect on the average workload.
Originators can have trouble hiring staff quickly. Before the financial crisis, originators could hire temporary workers to staff call centers and help push loans through the application and closing process. However, since the financial crisis underwriting has become more complex, which increases the time and money needed to train new brokers. Originators may find that hiring temporary workers is ineffective and may be reluctant to hire personnel when there is no guarantee that rates will remain low long enough to recoup the investment.
Exhibit 6: Loan brokers’ case load has increased sharply in 2020
Low capacity makes it profitable for originators to keep rates high. Originators can maximize the profit they make on each loan while keeping origination volumes as high as their staff can handle. Originators can always lower mortgage rates later to bring in additional borrowers if volumes begin to fall, as long as interest rates remain low. If secondary rates remain low long enough, competition between originators should drive primary rates down to more typical levels—roughly 90 bp above the par yield on 30-year MBS.