New York | Reading the financial press over long holidays in essential. In the days immediately before or after a holiday, there are inevitably important news items that will be missed. Exhibit 1 is the festering situation in the world of non-bank mortgage finance, where a combination of excessive regulation and interest rate manipulation by the Federal Open Market Committee has set the industry on a collision course with reality in 2019.
Eddie Small at The Real Deal in New York summarized the situation:
“Lenders are trying to navigate the new landscape using tactics like selling their mortgage-servicing rights or lending to borrowers they would have previously overlooked. Dan Gilbert, chairman of the largest nonbank lender Quicken, told the Journal that purchase mortgages are becoming more central to the company’s business.”
Let’s set the stage. Back in 2008, the FOMC opened the proverbial floodgates, pushing interest rates down to near zero and ushering in a bull market in both commercial and residential real estate starting in 2012. To give you a sense of just how far up the FOMC has manipulated home prices, the chart below shows loss given default (LGD) for the $2.5 trillion in bank-owned 1-4 family mortgages.
In Q3 ’18, the LGD for bank owned 1-4s was negative 15%, meaning that recoveries on foreclosed homes actually exceeded the gross amount of defaulted loans. The long-term average loss rate for bank-owned 1-4s is 65% going back to 1984. And the level of both recoveries and defaults for the portfolio is very low, below $1 billion. But both the negative level of credit loss and the low default rates are outliers that will be reversed.
By engineering an artificial sellers market in real estate after 2012, the FOMC also created a huge opportunity for adept mortgage firms to make piles of money, both on mortgage refinance transactions and also by managing the vast flow of defaulted mortgages coming out of the crisis. At the start of 2012, by comparison, the LGD on bank owned 1-4 family mortgages was 94%, meaning that banks were loosing $0.94 per dollar of face amount of loan every time a mortgage defaulted. Today, with home prices now above 2008 levels, mortgage servicers are actually making $0.15 profit per dollar of the original loan amount in those rare cases where a mortgage actually goes through foreclosure.
With interest rates rising and the accumulated backlog of defaulted mortgages largely (but not entirely) resolved, mortgage firms are now facing the worst of all possible worlds. On the one hand, mortgage servicing is no longer profitable for many non-banks because of the Dodd-Frank, the regulations imposed by the Bureau of Consumer Financial Protection and the 50 states. While mortgage firms are able to generate decent margins on mortgage refinancing, the cost of originating new purchase loans is over $8,000. Kroll Bond Ratings put the market into context in a recent report:
“[N]on-bank lenders are experiencing [gain on sale] GoS margin compression and falling origination volumes as the industry transitions to a higher rate, purchase-focused market. For large banks, 3Q18 mortgage banking results largely mirrored industry trends with slightly better GoS margins and declining origination volumes. While linked quarter margins improved for some (JPM, WFC, HTH), margins compressed further for Flagstar (FBC), HomeStreet (HMST) and PennyMac (PFSI). More notably, GoS margins year-over-year (sometimes a better proxy given seasonality associated with the market) for all companies in KBRA’s FI Mortgage Panel were down an average of 33 bps (24%).”
According to the Mortgage Bankers Association, the average pre-tax production profit for non-bank mortgage lenders was 21 basis points (bps) in the second quarter of 2018, up from an average net production loss of eight bps in the first quarter of 2018, but down 24 bps from the second quarter of 2017. Keep in mind that residential loan officers typically make 1% or more in commissions on new loans, thus the industry is still operating deep in the red on every loan originated.
As lending volumes have slowed over the past several years and purchase mortgages have become the dominant loan type, the profitability of many non-bank lenders has disappeared. Unlike banks which are able to earn money from the custodial deposits related to mortgage payments, non-banks must survive on gain-on-sale of new loans and servicing fees. With the cost of servicing loans up 200-300% since 2008, however, many independent mortgage banks no longer can count on a profitable servicing book to see them through periods of a sellers market in housing, when new home purchase mortgage lending is typically unprofitable.
The same market manipulation by the FOMC that has caused interest rates to fall and real estate prices to soar has also encouraged non-bank mortgage firms, which are already struggling with profitability, to sell a portion of future loan servicing fees at premium prices. New production mortgage servicing rights (MSRs) for conventional Fannie Mae and Freddie Mac loans are currently going at between 5.5 and 6x annual cash flow, record price multiples for MSRs that again illustrate the huge distortions introduced into the world of housing finance by the FOMC.
Which brings us to that little pre-holiday data point. On November 15, 2018, the Government National Mortgage Association (aka “GNMA”) published a bulletin to issuers operating in that market that makes a number of changes to how MSRs are financed and sold. A few members of the industry press commented on the rule before T-Day, but by and large the mortgage bankers still don't get the joke. The memo from GNMA COO Michael Bright states:
“Effective immediately, Ginnie Mae is implementing new notification requirements for Issuers engaged in certain subservicer advance or servicing income agreements, which do not require prior Ginnie Mae approval, but can impact an Issuer’s ongoing liquidity position and financial obligations. While Ginnie Mae currently permits subservicers to advance funds on behalf of an Issuer to pay security holders under the MBS Program, subservicers will now be required, upon request, to notify Ginnie Mae about such advances, including details about the frequency, amount, and purpose. Similarly, Issuers that enter into pledges of servicing income, or other transactions that encumber an Issuer’s Servicing Income, that are not subject to an Acknowledgment Agreement, must notify their Account Executive via email no later than 15 business days after the date that the transaction agreement is executed. Upon notification, Ginnie Mae may require the Issuer to provide the specific terms of the transaction, relevant documentation, or updated financial information.”
The new GNMA regulations require all banks and non-banks operating in that market to disclose all past financings, sales and participations of MSRs. Why is this important? Because the ability of an issuer in the $2 trillion GNMA market to 1) pay bond holders and 2) purchase bad loans and conduct loss mitigation is crucially dependent upon the solvency of the issuer/servicer. When an independent mortgage bank “sells” part of their future servicing income to help offset current losses on lending, for example, we create a scenario where the mortgage bank is more likely to fail when 1) interest rates rise and/or 2) default rates increase.
The Economics of Servicing
Let’s quickly dive into the economics of loan servicing in the Ginnie Mae market, which we addressed in depth last summer in a working paper entitled “Increasing Capital & Liquidity for GNMA Mortgage Servicing Rights” last summer. Let’s assume that we have a hypothetical mortgage with an unpaid principal balance of $300,000 and a loan coupon of 5%. The typical GNMA issuer gets a net servicing fee of 32bp annually. Take 32 bps on $300k and you get $960 per year in gross servicing.
Most servicers can administer a performing mortgage for $7 per month, so again 12 x $7 or $84 per year. This means that, in theory, the non-bank could “sell” the other $880 or so of the servicing income or "strip" to a financial investor like Black Rock or Apollo. But this works only if there are no problems with the loan, no need to speak to the obligor, call them or send them mail, etc. Sending out a piece of first class mail costs at least $1, for example, vs an electronic mortgage statement.
The moment a GNMA loan goes delinquent, the cost of servicing skyrockets, 10x the normal cost or more, as do the fees the servicer eventually receives for fixing the loan. But all of the excess fees for servicing a distressed loan are on the back end and must be financed by the non-bank, along with advances of interest, principal, taxes and insurance required to protect the value of the home.
The only cash the GNMA issuer/servicer receives each month is the 32 bps servicing fee. Thus the reason why GNMA wants to know how much of a non-bank’s 32 bps of servicing has been sold away. Remember that the full 32 bp strip is capital meant to finance operations during periods of peak defaults. Under the current practice in the mortgage industry, many non-banks have sold away more than half of their gross spread from their MSR. We hear in the channel that GNMA is considering in the near future imposing a 25bp minimum for retention of income from the MSR.
Note that the annual float on the mortgage example above is in excess of $30,000 per year in a high tax state. This float can generate twice the fees that a servicer receives for actually administering the loan, but only a bank can fully capture this benefit. Note too that larger loans generate bigger servicing fees, one reason that big banks like Wells Fargo and Bank American will pay up for jumbo loans, which they often hold in portfolio.
The trouble for investors in MSRs arises from the optionality regarding termination of servicing rights by GNMA, an issue that is fundamentally connected to the solvency of the issuer/servicer. Thus the new GNMA regulations provide a stark warning and also a roadmap to future contagion in the non-bank sector. Keep in mind well more than half of the non-banks in the US residential space have probably blown through their bank credit covenants due to impaired capital, poor profitability or both.
The other issue for investors in non-bank mortgage firms and MSRs is participations. When a non-bank sells or finances part of their 32 bp GNMA servicing strip to a financial investor, this is most often done via a participation agreement. There is no “sale” and the mortgage bank remains the owner of record of the MSR. In the event of default, a bankruptcy trustee for a non-bank or the FDIC acting as receiver for a failed bank will likely try to reject the participation agreements that have not been fully isolated.
As we noted in American Banker, some lawyers like to pretend that the current practice on Wall Street regarding loan participations is safe and sound, but in fact unless the asset is legally isolated from the failed bank or non-bank, the purchaser of participations in loans or MSRs stand at risk of total loss. It happened before in a little mess called Penn Square Bank. The FDIC repudiated all of the dead bank’s loan participations, gutting five large banks in the process and causing four (Chase Manhattan, Continental Illinois, Seafirst and Michigan National) to subsequently fail and be sold. Of the five banks most impacted, today only Northern Trust Co (NTRS) survives that default event as an independent bank.
The Road to Contagion
The first step toward contagion in the world of non-bank finance, as we’ve noted before, will be a liquidity squeeze that is ongoing and forcing many smaller non-bank firms out of business. Even were the FOMC to do just one more rate hike in 2018 and then wait awhile for the System Portfolio to unwind, it is unlikely that loan spreads in residential loans or other asset classes are likely to recover in the near term. In fact, so great is the concern about profitability that GNMA is meeting with the largest issuers, industry maven Rob Chrisman wrote just before the holiday:
“Ginnie Mae sent a "liquidity letter" to its 14 largest issuer/servicers in late October, telling them to come up with contingency plans as the profitability picture worsens. The management teams of these 14 shops will be sitting down with Ginnie officials in early January to discuss the matter further. Ginnie also issued an all-participants memo, dictating new standards for firms seeking to become issuers, including the stipulation that applicants submit to a corporate credit evaluation similar to what the rating agencies put them through.”
The industry currently has capacity to do at least $2 trillion in new mortgages annually, but volumes in 2018 will be closer to $1.4 trillion and declining – and with little profitability. We easily could see a 10% reduction in the number of non-bank mortgage firms this year and a larger downward headcount adjustment in 2019, a grim figure that suggests thousands of job losses in the world of residential mortgage finance.
Step two in the Merry-go-Round of non-bank risk will be credit. The real fun is some two to three years out. This is when years of low-interest rate lending starts to mature and throw off supra-normal loss rates. We suspect that some of the REITs and Buy Side shops that have been hungrily acquiring MSRs at single digit returns may become sellers when the true cost of credit is revealed. As Mike Lau told us a few weeks back in The Institutional Risk Analyst ("The Interview: Michael Lau on the State of Mortgage Finance"), MSRs are a mid-single digit internal rate of return asset (IRR) through the credit cycle and w/o leverage.
Just as credit spreads have galloped 20% in the past 30 days, we suspect that indicators such as LGDs and loan default rates for bank owned loans will also start to move pretty quickly as and when movement occurs. Whether or not home prices start to soften as quickly as credit turns across the US is really the big question. If we see loss rates for residential and commercial assets start to rise in a sustained way, then we’ll know that overheated home prices are headed for a correction. Watch those LGDs over the next year. We’ll be addressing this point in the next edition of The IRA Bank Book.
The New York Times
By Emily Flitter, Matthew Goldstein and Kate Kelly
By R.C. Whalen