Comment Detail

Comment

As part of the overall review of Single-Family Mortgage Pricing, University Bank believes that it is imperative that FHFA review and restructure the pricing structure for MSRs created out of single-family loans sold to FHFA supervised agencies. The value of MSRs has a direct impact on the pricing of loans. The current structure has several material negative impacts on financial institutions and consumers. The safety and soundness of financial institutions and their long term viability is negatively impacted by the current structure.

Why We Should Restructure How MSRs are Created

Problem Statement

With the rise in long term interest rates, Mortgage Servicing Rights (MSRs) are rapidly rising in value, creating two problems for banks active in mortgage loan origination:
1) The tax-effected value of MSR assets above 25% of Tier 1 Capital is subtracted from Bank Tier 1 Capital [79% of the value of MSRs above 25% of Tier 1 Capital is deducted].
2) As the value of MSRs held on the books of banks is rises, it causes increased haircuts to regulatory capital and potentially bumping up against the hard limit (banks are not allowed to hold MSRs above 100% of Tier 1 Capital).
In addition, the current structure of the MSRs causes key policy issues:
3) Banks are encouraged to sell MSRs. The main buyer is Wall Street Hedge Funds. Severing the customer relationship and throwing customers literally to the wolves is not good public policy, nor is it good for banks.
4) MSRs are a depleting asset. When MSRs are added to the balance sheet with a value, ultimately, that value must at some point in the future be written off, lowering future earnings. When MSRs rise in value, this is shifting income from future years into the present.
5) MSRs cause volatility in bank earnings. Wall Street hates earnings volatility and this means that the ability of banks active in the mortgage origination business to raise capital is somewhat constrained and the cost of that capital is increased.
6) The purchase market for GNMA MSRs is thin. We have seen at times in the past when putting GNMA MSRs out to bid that only two serious non-bank bidders showed up and no banks showed up. Why? This is driven by the high capital requirements posed on banks which make ownership of MSRs onerous and the fact that not many non-banks want to buy GNMA MSRs because the potential draw on liquidity can be high.
7) The current structure with its extraordinarily high servicing fees for adjustable rate single-family mortgage loans (37.5 b.p.), which costs bear no relationship to actual cost of servicing which is much lower, completely blocks the creation of a forward TBA MBS market from developing. Consumers desire 7-year and 10-year ARMs however these loans must currently all be held in portfolio and as a direct consequence of this poor public policy, and therefore the quantity of these loans available to consumers is limited.

Solution Statement

Give banks the option of creating Mortgage-Backed Securities (MBS) with servicing less than the current high minimum required levels, and in line with actual costs of annual servicing using a subservicing contract structure. If this were allowed, the net present value of the MSRs for the banks that opt to go down this path would be close to zero. Therefore, there would not be a capital charge for holding MSRs. Banks could hold the MSRs they create and wouldn’t ever need to sell. Sales of MSRs would drop and the value of MSRs sold would rise. Non-banks would no longer be advantaged over banks. Banks and non-banks that want to create MBS MSRs with traditional levels of large servicing fees would continue to be free to do that. The other banks and non-banks that want to take advantage of the new flexibility would be free to set the minimum at any level above the new minimum level of servicing fees.

Idealized Structure

The good news is that this idealized structure already exists. Flagstar Bank has created 21 large private label jumbo mortgage securitizations using this type of structure. Flagstar’s idea behind building in a subservicing structure to its deals was to eliminate the MSR that had to be booked and the capital “charge” (and also maximize additional cash flows by selling additional IO bonds or a more valuable unitary bond). Flagstar did not book any MSR on these deals. 100% of the gain on sale went to the revenue line and generated profit.

When securitizations are done currently, a minimum MSR servicing spread is required by each of FNMA, FHLMC and GNMA. Originally these servicing spread minimums were established decades ago when all the work was done manually and before the widespread adoption of computers, at a time when costs were much higher. Despite the rapid advance of computing technology which has substantially reduced the cost of servicing mortgage loans, minimum servicing spreads have never been touched over the decades. Current minimums are 0.25% for conventional and 0.19% for GNMA, however, GNMA requires that servicers, at all times, maintain a “Minimum Portfolio Servicing Spread” of 0.25% (Chapter 3, Section C of the GNMA guide). This is an absolute minimum and cannot be achieved by rounding.

The agencies talk about servicers having “skin in the game” and servicing that is worth more than what servicing costs is what they think gives servicers an incentive to keep up their performance. FNMA could always take their servicing if they were not performing, and the servicer would lose an opportunity. GNMA is even more sensitive as they must advance interest on a timely basis and they don’t have other cash flows beyond very limited Guarantee Fees. Moreover, most of their servicers are non-banks and this worries them greatly (no balance sheet to support the advances in a downturn).
Flagstar, although the first, is no longer the only issuer doing this type of a structure although there are different twists. Chase now uses a very similar structure. UWM uses a similar subservicing type structure as well. Bayview uses a 0.15% b.p. servicing fee. Goldman uses a two-part structure to divide up its 0.25% fee, 6 b.p. for regular servicing and 19 b.p. into a special I/O support bond which is sold but which depletes if a successor servicer has a higher fee. There are others too, but this should give you an idea that different servicing fees besides the standard 0.25% are out there and in widespread use on conventional prime loans. Flagstar also uses this on its agency investment property PLS deals.

Issues to Consider

A. ABA has long expressed concern about the capital treatment of MSRs driving servicing out of the banking system, but pushing MSR values toward zero entails a broader policy discussion that must be vetted with ABA committees.
→ If MSR values were zero on the balance sheet, this would have the following benefits:
1. The volatility of earnings generated by changes in interest rates and the quarterly mark to market of MSRs would be eliminated. Today the MSRs have the same risk profile as IOs. IOs are highly volatile in price over time. Bank stock investors don’t like volatility of earnings, so this proposal would also increase the value of banks by dampening the volatility of their earnings stream.
2. The incentive for banks to sell customer relationships would disappear. Customers would be serviced by the bank or non-bank that originated the loan unless the MSRs were sold.
3. While the balance sheet value of the MSRs would be zero, there is still substantial value in the net present value of the customer relationship because of the revenue generated from cross-selling additional products or services to the same customer. Banks and non-banks that were not able to cross-sell or not good at it and wanted to sell their MSRs, could still book a further gain by doing so.
4. There would be no effective limit on the amount of MSRs that a bank could originate and hold. Banks would gain market share.
5. This new flexibility would be optional. Banks and non-banks could opt to continue to create MSRs with substantial value.
6. MSRs would flow from non-banks to banks which can maximize the value of the customer relationship via extensive product cross-selling.

B. Under a subservicing contract structure for MSRs, who funds the cash flow for alternative (delinquent or successor) servicing fees to be paid?
→ In Flagstar’s case, it would just be treated as a cash flow shortage which would reduce the principal remaining in reverse of the order of seniority of the bonds in the cash flow waterfall. Goldman creates a special I/O support bond. Other solutions are possible.

C. Sales into a security with servicing fees to be assumed that are less than market rates (even if still making a profit) would cause servicing losses to be recognized at the time of securitization. This loss/gain recognition raises complexities and uncertainties regarding servicing rates.
→ Under the idealized structure, the cash flows are always above costs, no matter what the scenario, but are so small and with a level of uncertainty, so that no value is ascribed to the MSR and no loss or gain is taken upon its creation. 100% of the cash revenue generated from selling the loan at a premium to origination cost is run through a gain on sale revenue account and falls to the bottom line contributing towards profit.
→ The concept that we believe would be ideal, would result in a larger gain on sale and no value being ascribed to the MSR.

D. There is concern that banks would risk examiner criticism if they bid servicing fees down to levels that are too low. Examiners are likely to be skeptical that banks really are recovering their costs. As experienced during the financial crisis, assumptions about servicing costs are closely related to assumptions about loan performance, foreclosure rates, and foreclosure timeframes, especially when servicing includes liquidity obligations. Encouraging assumptions that are too optimistic so that a servicer can win the bid (with a lower fee) is eventually de-stabilizing.
→ Under the idealized structure, servicing fees would rise if delinquencies rise. Fees are in line with a typical subservicing fee structure. There is always the revenue to pay for the cost of servicing in an amount slightly larger than what is required, unless the servicer’s costs were out of line.

E. The implicit value of the cash flows that would be diverted from the servicing fee under the proposal have to be accounted for somewhere. Depending on what alternative execution is allowed and what execution method is preferred by the MBS issuer, that may increase incentives/pressure to bid the fees down (which would create bigger excess-servicing strip). Any ABA advocacy would have to have ready solutions to this.
→ In a typical transaction, under this scenario the gain on sale would be typically 0.80% to 1% higher relative to the original loan amount and the investment in MSRs would be 0.80% to 1% lower. MSRs would flow from non-banks to banks which can maximize the value of the customer relationship via extensive product cross-selling. Under the preferred scenario, no excess servicing strip would be produced, the bond holders would earn higher rates of interest during good times, and slightly lower levels of interest during difficult economic conditions, when interest rates are generally lower anyway. Even though yields were slightly lower, the value of the bonds would rise as interest rates fell materially.