MBS Market Strategies

Transcription

1 U.S. Housing & the MBS Market in Review and 2015 Outlook Contents Mortgage Activity and Housing - Refinance applications down 53% from the peak but recently increasing, a 500%+ surge in multifamily housing starts, and increasing rental demand despite soaring rents Financial Strategies Group Drew Simmons Vice President dsimmons@gobaker.com (866) Mortgage Credit - Credit availability still historically low, and implications of the FHFA s attempt to widen the credit box via FN/ FH 97% LTV loan programs New Trends in US Housing - A shift in household formation, homeownership still well below peak but near the long term average, and a slowdown in home price appreciation MBS Fundamentals - Spread tightening, low spread volatility from stable prepays in 2014, tight correlation of MBS and UST yields, MBS supply, and the Fed s impact on demand Rising Rates - A review of the tightening cycle in 2003 compared to a potential 2015 rising rate environment and the relative impact on MBS depreciation by coupon 2014 Prepayments - Overall slowdown in speeds with the main outlier being pools serviced by Quicken, the convergence of speeds in specified pools and a look at extension protection in specified pools FHA MIP Changes and the Impact on GNMA Speeds Are Prepayment Risks Behind Us? Recommendations Contact Information The Baker Group LP Headquarters 1601 NW Expressway, 20th Floor Oklahoma City, OK (800) Austin, TX 1700 Rio Grande, Suite 120 Austin, TX (888) Birmingham, AL 2204 Lakeshores Drive, Suite 213 Birmingham, AL (855) Indianapolis, IN 8365 Keystone Crossing, Suite 0 Indianapolis, IN (866) Salt Lake City, UT 2975 West Executive Pkwy, Suite 139 Lehi, UT (800) Springfield, IL 901 Community Drive Springfield, IL (888) Page 1

2 New-Normal for the U.S. Housing Market From , the housing market experienced the most expansive refinance wave in history as a result of record low rates and the success of HARP 2.0. Agency MBS prepayment speeds were far less volatile in 2014 as a result of subdued refinance activity and lackluster growth in home purchases. Through January 21st, 2015 the Mortgage Bankers Association s (MBA) refinance index was more than 50% below the levels of (Exhibit 1). The U.S. Census Bureau s gauge of total housing and building starts reached 1,089,000 through December 2015, slightly up on the year. The most intriguing development with housing starts has been the revival of multi-family construction. Single family (1-unit) starts have increased 92% from the 2009 lows while multi-family (2+ unit) starts surged more than 500% (Exhibit 2). Many economists feel this is indicative of the new normal for U.S. housing...more renters and fewer buyers. Exhibit 1 - MBA Refinance Index: 20-Today historically increase along with single-family units as the economy recovers from a previous recession (red bars). What s different this time is the precipitous decline in single-family units after nearly 20yrs of growth. According to data from the U.S. Census Bureau, the median asking rent and single family starts increased in tandem from 1991 through Fast forward to today and single family starts are down 62% while the median asking rent is 26% higher (Exhibit 4). Moreover, these two indices are moving in the opposite direction from their historical averages. Since 1959, single-family units are 34% lower while the median asking rent is 40% higher...hardly reverting to the mean. The catalyst has been tight lending standards coupled with tepid wage growth. Exhibit 5 from Bank of America shows the MBA s Mortgage Credit Availability Index (MCAI) is 86% lower than the Exhibit 3 - Single (Yellow) vs Multi-Family (Blue) Starts: 1963-Today Exhibit 2 - Single (Yellow) vs Multi-Family Starts (Red): 2008-Today Exhibit 4 - Single-Family Units (Yellow) vs Median Asking Rent (Blue): 1963-Today New Normal or Reversion to the Mean? An ever-increasing number of families and individuals are renting their homes, resulting in the largest percentage increase of multifamily starts since As shown in Exhibit 3, multi-family starts peak from 2Q It s no surprise that U.S. homeownership has fallen to the lowest since 1995 at 64.4%. However, the long term average is just 1% higher (Exhibit 6). Our December th publication highlighted the Federal Housing Finance Agency s Page 2

3 (FHFA) attempt to widen the credit box with the revival of 95% LTV loan programs from Fannie Mae and Freddie Mac. It remains to be seen if these types of loan programs will be a healthy avenue to increase homeownership. What s undesirable is an increase in homeownership attributed to unconventional private lending practices that led to the housing crisis. Exhibit 7 - MBA Purchase Index: Today The recent activity from the MBA s purchase index is another example of a fundamental shift in household formation. Weekover-week purchases fell 6.87% from the December 17th report and down 1.84% year-over-year. More importantly, recent purchase volume has been hovering near 20yr lows and 34% lower than the average since 1990 (Exhibit 7). Virtually every measure Exhibit 5 - MBA Mortgage Credit Availability Index: 2004-Today Exhibit 8 - Case Shiller National Home Price Index Exhibit 6 - U.S. Homeownership: Today of housing activity shows lackluster growth with the exception of home prices. Exhibit 8 shows the seasonally adjusted national home price index from Case Shiller. The housing crisis contributed to a 26% decline in home prices but have since recovered to only % below the 2007 highs. While the recovery in home values has been positive, recent price gains have decelerated. From August 2013 to February 2014, year-over-year gains averaged.6% but have since slowed to 4.81% as of September MBS Spread Volatility Fannie 15yr current coupon spreads against the 5yr Treasury Note tightened by more than 50% to end 2014 at 43bps. Fannie 30yr spreads against a 5yr/yr Treasury blend fell to a lesser extent, 34bps tighter to 92bps (Exhibit 9). Lower spread volatility and a more stable prepayment environment have led to a higher correlation of MBS yields and Treasuries over the past year. As shown in Exhibit, the correlation of current coupon FN 15yrs and the 5yr Treasury Note has been 82-93% since October of Current coupon FN 30yrs showed an even higher correlation to the 5yr/yr Treasury blend, staying above 90% during the same period. Refi burnout was the main driver for stable prepays and therefore lower spread volatility in 2014 despite 30yr mortgage rates falling by more than 50bps. Spreads were also kept in check with the absence of policy shocks such as HARP 2.0 in 2012 or the taper tantrum of All things being equal, a continuation of less basis risk and stable prepayments that reduce convexity could contribute to similar spread volatility for Page 3

4 Exhibit 9 - MBS Spreads Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13 Mar-13 Apr MBS Spreads FN 15yr & 30yr Current Coupons May-13 Jun-13 Jul-13 Aug-13 Fannie 15yr Current Coupon - UST 5yr Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Fannie 30yr Current Coupon - UST 5yr/yr Blend Source: Bloomberg LP, The Baker Group LP Exhibit a - FN 15yrs to UST 5yr Correlation Jun-14 Jul-14 Aug-14 Sep-14 Oct-14 Nov-14 Dec-14 Exhibit b - FN 30yrs to UST yr/5yr Correlation Exhibit 11 - Agency MBS Holdings: 2000-Today Supply and the Fed s Impact on Demand The end of QE3 will require private investors to step up to the plate in the Fed s absence of MBS purchases. Exhibit 11 from J.P. Morgan provides a historical perspective of MBS holdings by the major buyers since In 2014, the Fed eclipsed the private sector to the tune of nearly $230bn of MBS purchases. Fortunately, the historically low supply of Agency MBS in 2014 is expected to continue in 2015 so the private sector will need to absorb less slack. Exhibit 12 from J.P Morgan shows that 2014 was the lowest Agency MBS gross issuance since 2006 with a similar projection for Exhibit 12 - Agency Gross Issuance: 2000-Today Source: J.P. Morgan Page 4

5 While 2015 will be a transition year for MBS demand, the Fed isn t completely out of the game with roughly $20bn in monthly MBS paydowns to be reinvested through December When the Fed stops their reinvestments, the biggest impact will be in the lower coupon TBA market. As shown in Exhibit 13, 84% of 30yr purchases were in 3-4% coupons while 82% of 15yr purchases were in 2.5-3% coupons. Exhibit 13 - Fed MBS Holdings Federal Reserve MBS Holdings by Coupon Distribution $1.729 Trillion 15yrs - $206 Billion 82% in 2.5-3% Coupons 30yrs - $1.5 Trillion 84% in 3-4% Coupons Exhibit 15 - Fannie Mae 15yrs by Coupon: GN II 30yrs 0% 0% 31% 39% 24% 5% 1% 0% GN I 30yrs 0% 0% 29% 23% 12% 23% 12% 0% FN 30yrs 0% 1% 28% 26% 29% % 4% 2% FH 30yrs 0% 0% 27% 27% 26% 12% 6% 1% FN 15yrs 7% 48% 32% 8% 3% 1% 0% 0% FH 15yrs 5% 52% 35% 7% 1% 1% 0% 0% Source: The Federal Reserve, The Baker Group LP Impact of Rising Rates by Coupon Current expectations have the Fed beginning to raise rates in June or September 2015 in incremental steps. A good proxy to gauge the impact to MBS in a rising rate environment is most likely the tightening cycle from Exhibit 14 shows the yr Treasury compared to Fed Funds during this period. The funds rate was incrementally increased from 1% to 5.25% while the yr treasury sold-off in 3 distinct periods. Exhibit 14 - yr Treasury (Blue) vs Fed Funds (Yellow): Exhibit 15 shows the impact on Fannie 15yrs during these periods with lower coupons experiencing the worst price depreciation. At the time, the lowest coupons in the Fannie 15yr stack were 4s which experienced a -7.3% price decline during the first sell off, -6.8% in the second and -6.5% in the third. However, price declines within Fannie 15yr 6s were far more defensive. The 6s depreciated by -1.7%, -2.5% and -3.4% in the 3 respective periods of rising rates. Period Coupon % Decline Px Decline FN 15yr % -1.7pts FN 15yr % -2.9pts #1 FN 15yr % -4.3pts FN 15yr % -5.8pts FN 15yr % -7.4pts FN 15yr % -2.5pts FN 15yr % -3.5pts #2 FN 15yr % -4.8pts FN 15yr % -6pts FN 15yr % -6.8pts FN 15yr % -3.6pts FN 15yr % -4.8pts #3 FN 15yr % -5.2pts FN 15yr % -5.7pts FN 15yr % -6.4pts By 2007, the 2003 tightening cycle resulted in an inverted yield curve that began much steeper than today with higher nominal yield levels. The economy was also heating up at a much more rapid pace, especially in the housing sector. The old cliché of this time is different holds true as the Fed will be more patient in the face of low-inflation. Regardless of the pace of the next tightening cycle, the current environment suggests the Fed won t push back rate hikes much further. Investors of high coupon MBS pools had to endure some of the fastest prepayments on record during the refi wave. However, as rates begin to rise, higher coupon MBS pools should have less price depreciation. This is especially true with the absence of the Fed s demand in lower coupon pools. High coupon, seasoned specified pools with attractive loan level attributes such as low loan balances will provide the best relative value. Page 5

6 2014 MBS Prepayments Prepayments over the past year were relatively benign for Agency MBS pools, a welcomed change from the warp-speeds of Over the past 12 months, aggregate Fannie and Freddie 15yrs paid 7-11 CPR with Ginnie 15yrs at a slightly faster clip, topping out at 13.7 CPR (Exhibit 16). Thirty year pools paid in the low teens for Fannie and Freddie while Ginnies reached 17.9 CPR in July (Exhibit 17). Exhibit 16-15yr Agency MBS Speeds mo CPR Year Fixed MBS Historical Prepayment Speeds 6 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 Jul-14 Aug-14 Sep-14 Oct-14 Nov-14 FNMA FHLMC GNMA I GNMA II Exhibit 17-30yr Agency MBS Speeds mo CPR Year Fixed MBS Historical Prepayment Speeds 8 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 Jul-14 Aug-14 Sep-14 Oct-14 Nov-14 FNMA FHLMC GNMA I GNMA II Outlier - Servicer Speeds Although aggregate speeds were slower and less volatile compared to , a few themes played out in 2014 that may continue to impact prepayments in The predominate theme for the year was the prepay disparity between non-bank servicers and traditional/ banking servicers. Over the past few years, many of the larger banks were forced to sell their mortgage servicing rights due to a reduction in capacity. As shown in Exhibit 18 from Bank of America, this opened the door to a larger share of servicing rights by non-bank servicers. Servicing rights by non-banks have nearly doubled since May 2013 to 30%. The mortgage market was extremely competitive in 2014 amid an appetite for loan volume as refi demand plummeted. This, coupled with stagnant demand for home purchases, created an environment where non-bank servicers were highly aggressive by either competing on rate or by targeting lower credit borrowers. For more than a year, the most notorious of which was Quicken Loans. Exhibit 18 - Non-Bank Servicer s Share of MBS Exhibit 19 shows the average 12mo CPR for the top 5 servicers of Fannie 30yr 4.5s. Quicken s 12mo prepay ratio relative to the cohort was 1.67x, 55% more than the next fastest payer, BofA which paid just a touch faster than the cohort at 1.08x. The rest of the top 5 servicers (all banks) have paid pretty much in-line with the cohort over the past 12 months. Part of Quicken s strategy appears to target slightly lower credit borrowers. Within Fannie 30yr 4.5s, Quicken s average FICO is 721 compared to a 744 average FICO for the cohort. Quicken also has a higher average loan-to-value at 80% compared to the cohort of 75% (Exhibit 20). While the difference is admittedly small, reaching just slightly down the FICO range for top tier credit pricing (i.e. the best available note rate) could easily pull business away from the major banks with more stringent underwriting criteria. Exhibit 19 - FNCL 4.5% Top 5 Servicer Speeds US Bank -> 0.97x COHORT -> 1.00x Wells Fargo -> 1.02x Citi -> 1.03x Bank of America -> 1.08x Quicken -> 1.67x FNMA 30yr 4.5% - 12mo CPRs Exhibit 20 - FNCL 4.5% Top 5 Servicer FICO/LTV Servicer Name 12mo CPR FICO LTV Quicken Bank of America Citi Wells Fargo US Bank COHORT Page 6

7 Specified Pools and the Impact of Rising Rates The low level of prepay speeds from the lack of refi activity in 2014 led to a convergence of speeds within loan level attributes. Fannie 30yr 4.5s of 20 began 2013 with <85K loan balance pools paying as much as 20 CPR slower than those with loan balances greater than $225K. By January 2014, that disparity had tightened to just 4 CPR (Exhibits 21 & 22). As of the most recent November print, with 30yr mortgage rates 50bps lower, the gap slightly widened to just under 7 CPR, reminding investors of the value of low-loan balance collateral even as the market prepares to raise rates. Exhibit Speeds by Loan Balance CPR FNCL s by Loan Size CPR slower than the 400K+ loan balance group. The disparity of speeds by loan balance isn t nearly as drastic as the rate refi wave of , but any form of extension protection as rates rise will be valuable. Exhibit 23 - FNCL 4s vs LLB Payups <85K 85K-1K 1K-150K 150K-175K 175K-225K 225K+ Exhibit Speeds by Loan Balance CPR FNCL s by Loan Size Exhibit 24 - Loan Balance Speeds When Rates Rise FNMA 30yrs of 2003 by Max Loan Size 2006 Average CPRs <85K 85K-1K 1K-150K 150K-175K 175K-225K 225K+ The convergence in speeds by loan balance have naturally led to payup compression for low-loan balance pools. Exhibit 23 from J.P. Morgan shows that while prices for generic Fannie Mae 30yr (FNCL) 4s are near early 2013 levels, the payups for LLB FNCL 4s are nearly 0 ticks lower. This suggests that LLB pools are relatively cheap as demand for prepay protection has steadily declined. However, LLB pools also offer slightly more extension protection as rates rise. In the same way that lower loan balance pools have less incentive to refinance as rates fall, there s also less disincentive to refi (or move/relocate) as rates rise. Exhibit 24 shows the average CPR on Fannie Mae 30yrs of 2003 when 30yr mortgage rates increased about 0bps during the first six months of Pools with balances of $99K or less paid about K 0-199K K K 400K+ Other specified pools such as those with high concentrations in investor and/or vacation properties have also converged relative to their respective cohorts. As shown in Exhibit 25, using Fannie Mae 15yr 20s as an example, pools with more than a 50% concentration in investor properties paid an average 19.7 CPR from compared to 25.3 CPR for the cohort. However, this relationship flipped in Investor pools averaged 14.8 CPR over the past 12 months with the cohort averaging 13.2 CPR. This has led to a similar reduction in demand and therefore lower payups. Page 7

8 Two factors generally elevate investor prepayments relative to generics; rising rates and lenient credit standards. Rising rates are generally coupled with a healthy economy, easier credit and increasing asset values, namely home prices. As HPI increases, owners of investment properties are incentivized to capture their equity for reinvestment or debt reduction. Lenient credit standards provide more cash-out refi options for investment properties as well as widened flexibility with LTV and income limits. Exhibit 25 - FNMA 15yrs of 20 - Investor Prepays Average CPR FNCI 20s - Average CPRs Cohort A look at the same 2006 period from Exhibit 24 supports the notion of faster prepays within investor collateral as rates rise. Here, we review Fannie Mae 15yrs of Before the Fed started raising rates in 2004, 15yr mortgage rates hovered around 4.5-5%. The biggest spike came in the first 6 months of 2006 when the 15yr mortgage rate climbed about 75bps to 6%. Exhibit 26 shows that pools with at least 50% investor collateral paid an average 11.3 CPR in 2006 compared to 9.1 CPR for the cohort. However, when rates started to fall, investor pools began to pay slower than the cohort and by 2009 their roles completely reversed with the cohort paying 2.8 CPR faster. Exhibit 26 - FNMA 15yrs of Investor Prepays Average CPR % Investor FNCI 2003s - Average CPRs Cohort 50-0% Investor GNMA Speeds On January 7th, 2015 the U.S. Department of Housing and Urban Development (HUD) announced a reduction in the annual Mortgage Insurance Premiums (MIP) for FHA loans. After nearly 5 years of increasing MIPs to offset losses in the FHA s insurance fund, the annual MIP will be reduced to 85bps. The 50bp reduction marks the lowest since April 2011 and applies to both purchase and refinance borrowers for terms greater than 15yrs. Exhibit 27 shows that annual MIPs have incrementally increased since June 20 when it stood at 50bps. The biggest impact is expected within recent vintage Ginnie Mae 20yr and 30yr pools (paying the highest annual MIPs) with high concentrations of FHA loans. These vintages consist of originations with coupons of %. According to Citi s Yieldbook projections, these cohorts could experience a 3-6 CPR increase (Exhibit 28). Exhibit 27 - FHA s Annual MIP Changes Since 20 History of FHA MIPs Period Annual Upfront Prior to June June 20 - Oct Nov 20 - Apr May Apr May Mar Apr Dec New Changes Exhibit 28 - CPR Projections: Select GNMA Cohorts Are Prepayment Risks Behind Us? Despite talk from the Fed for rate hikes this year, the FHA s MIP changes present future prepayment risks within GNMA pools. And although burnout was the main theme in 2014 with refinance activity down significantly from 2013, prepayment risks have taken center stage with mortgage rates at 18mo lows. The MBA s report of mortgage applications from the second week Page 8

9 in January this year showed the biggest week-over-week gain on record, up nearly 50% with refinance applications reaching the highest volume since July A deeper dive into the refi data reveals the mortgage industry s primary target - high balance/jumbo loan borrowers. The recent surge in refis along with the Oct 17th report produced the two highest readings of the average loan size for refinance applications since the series began in As shown in Exhibit 29, the average refi loan size increased to an impressive $306K on October 17th, 2014 and most recently to $286K. Exhibit 29 - MBA Average Refinance Loan Size An increase in prepayment speeds is expected this March from February prepayments. Due to the abnormally high average refi loan size, jumbo and high loan balance collateral should experience the fastest paydowns while low loan balance pools will be the most defensive. For the short term, pools with the biggest prepayment risks are jumbo loan collateral and/or high loan balance pools, recent vintage GNMA 30yrs with high concentrations of FHA loans, and recent vintage 30yrs and 15yrs serviced by non-bank servicers. On a longer term outlook, investors should prepare to add more extension protection through seasoned 20yr-30yrs and shorter term 15yr and yr specified pools. Fortunately for LLB and investor pools, speeds are generally faster than their respective cohorts as rates rise. Recent vintage Ginnie Mae 30yrs have an incentive to reduce their annual MIPs, and will experience faster speeds over the next few months. However, Ginnie pools offer good relative value as rates rise. FHA and VA borrowers tend to experience higher overall turnover due to a combination of a more mobile demographic and credit curing. Within CMOs, anchoring cashflows on the short-end with PAC structures and high coupon collateral will help improve liquidity for the institution s funding capacity. The current level of rates will increase prepayments in the interim but is unlikely to translate into the hyper-fast speeds witnessed in The remaining borrowers who refi this year will only exacerbate refi burnout going forward. This year will be defined yet again by the Fed s direction, and current expectations are that rate hikes will begin by year-end. Regardless, investors should stress test the portfolio under multiple rate scenarios on an ongoing basis. Recommendations The US economy continues to improve but weak global growth has led to a strong dollar and a flight to US Treasuries that has driven rates to 18mo lows. The uncertain rate environment for 2015 suggests that MBS investors should seek a balance of pools that offer prepayment and extension protection. With the potential exit of the Fed s MBS purchases from paydowns, lower coupon TBA pools are expected to be impacted the most on the demand side. Investors should focus on high coupon collateral in 2015 to improve the portfolio s defensive position in the event of rising rates. In light of the aforementioned disparity in servicer speeds, concentrations in pools serviced by the major, tier-one servicers such as Wells, J.P. Morgan Chase, Citi, and BofA are preferred under both rising and falling rate scenarios. Along the same vein, retail pools will help mitigate prepay risks for in-themoney coupons compared to pools with high concentrations of third-party originated loans. The compression in prepay speeds of specified pools relative to generics in 2014 led to suppressed demand for collateral such as LLB and investor concentrations. However, with the recent drop in rates and increase in refinance activity, demand for prepayment protection has returned. Page 9