Investor enthusiasm for Ginnie Mae pass-throughs has faced two obstacles in recent years: supply-limited liquidity and lumpy, often unpredictably fast prepayments in premium priced pools.

The latter problem – tradable supply – is being repaired in the current environment. FHA lending and its corollary, Ginnie Mae production are currently going gangbusters. FHA’s share of mortgage lending has revived from a moribund 3% in 2006 to about 25% currently. Ginnie production, as low as 5% to 10% of monthly agency pass-through issuance 2005-7, has popped above 40% in recent months and is currently running at 20 to 25% of monthly supply. (The revival of Ginnie Mae is also discussed in my column in this month’s HousingWire Magazine “Ginnie Mae: The Once and Future Queen?”)

The prepayment problem is more intractable and dominates consideration of Ginnie Mae long-term investment values. It can also impact pricing in the short run, as the latest round of prepayment reports proved. The market had been expecting prepayments during September on Ginnie premiums to be fast, but not as fast as materialized. By contrast GSE prepayments were slow (and slow by historical standards as well). Moreover, it was not expecting Ginnie 2s to accelerate significantly more than Ginnie 1s.

The surprise infected market perception Ginnie 1 and 2 premiums alike, causing Ginnie 6s and 6.5s to weaken versus comparable Fannies, higher coupon 30-year Ginnies to weaken versus comparable 15-years and dollar rolls in affected coupons to cheapen.

For example, in August Ginnie 1s originated in 2007 prepaid at 36.1 CPR (an annualized measure of the percentage of principal prepaid in a month). That is, if the monthly rate were sustained for a year, 36% of the principal outstanding August 1 2009 would be prepaid by July 31, 2010. In September the same Ginnie 1 “cohort” prepaid at 41.9 CPR. By comparison, 2007 origination Ginnie 2s prepaid at 32.2 CPR in August, but hit 60.6 CPR in September. Similar wide disparities were exhibited across other recent vintages of 6s and 6.5s.

A little background: Ginnie 1s are the oldest and most homogenous pass-through program, backed by loans from a single issuer, all paying an interest rate 50bp above the pass-through coupon. Ginnie 2 pools may be either multi-issuer or single-issuer and permit wider interest rate dispersion within the pool. For example, mortgages in fixed rate Ginnie 2 pools issued on or after July 1, 2003 must bear an interest rate at least 25 but not more than 75 basis points higher than security interest rate. All else equal, the presence of higher rate mortgages in Ginnie 2 pools should result in modestly faster prepayments in the 2s compared to same coupon 2s. Not surprisingly, all is not equal. In general, the programs are accessed differently by the various segments of the MBS market: Ginnie 2’s smaller pool size, greater flexibility and multi-issuer option favors smaller lenders, while the Ginnie 1 program has superior liquidity (e.g., “better execution” for loans meeting pooling requirements). Similarly, Ginnie 2 pools containing loans from many issuers might be more diverse, geographically and with regard to the servicers’ lending and servicing business practices. That diversity should encourage a more stable prepayment profile over time than in single issuer Ginnie 1 pools.

The sharp acceleration in Ginnie 2 prepayments in September can’t be explained by broader latitude in the rates on pooled mortgages. For example, Ginnie 1 6s are backed by 6.5% mortgages, while the loans backing 2007 Ginnie 2 6s have a weighted average rate of 6.47% (!). The explanation is not a fundamental, interest-rate response. As analysts at Barclays Capital put it, “many investors were spooked by a sharp increase in GN 2 prepayments, which look to have been the result of servicer buyouts.” this case in November.)

Servicer buyouts result from the valuable option granted to servicers to buy delinquent loans at par out of Ginnie pools under certain conditions. In general, for pools issued on or after Jan. 1, 2003, the borrower must miss three consecutive monthly payments. For pools issued before that date, the borrower must be delinquent for three consecutive months OR fail to make up a missed payment over four consecutive payments (in other words, a rolling delinquency). Note that servicers of GSE securities do not have this right; instead, the GSEs will buy delinquent loans out of pools (this, and the impact of buyout on modification, are topics for another discussion).

This option can be quite valuable if the loan can be made to reperform, a feat somewhat easier to perform before the housing/foreclosure crisis and much easier to perform with rolling delinquencies. Beginning in the late ’90s, some servicers began buying delinquent loans out of predominantly premium pools, curing them and reselling at a premium, increasingly into private securitizations. By 2002 it was a very popular practice among servicers and raised serious objections among investors (over the foregone above market interest).

Ginnie audited a number of originators whose volumes of buyout and resale were high, but did not find any apparent abuses. The one outcome was to tighten the rules to eliminate rolling delinquencies.

Profiting on the sale of reperforming loans is just one of servicers’ incentives to buy delinquent loans out of pools before foreclosure. Ginnie Mae requires servicers to maintain delinquencies at levels defined by three ratios: the percentage of loans in the Ginnie Mae servicing portfolio that are 90+ delinquent or in foreclosure, the percentage 60+ or in foreclosure and the ratio of cumulative delinquent P&I payments to total P&I due to the issuer. (For servicers with over 1000 loans those ratios are 5%, 7.5% and 60%, respectively. Smaller servicers have higher thresholds.) Analysts at several firms attributed some of the September jump to rising delinquencies. Analysts at Deutsche Bank, for instance, suggest servicers bought delinquent loans out on a large scale and in many cases may have been able to avoid foreclosures by modifying the loans under the Home Affordable Modification program.

Another factor, noted by analysts at BofA (this research reaches us courtesy an investor who closely follow this research group’s comments), is a requirement specific to FHA loans (Ginnie collateral includes VA, Farm Administration and other government residential mortgage loans). That is, servicers are required to advance scheduled principal and interest payments to investors in securities backed by the mortgage. In turn, the FHA reimburses 100% of principal and interest at a rate set by the FHA (currently thought to be close to a constant maturity 10-year Treasury rate). As a result, in the current yield environment, the servicer must advance several hundred basis points of interest to investors in Ginnie premiums out of their own pockets. This requirement is extinguished when a loan is removed from a pool, a strong incentive to buy the loan out.

Some analysts attributed the prepayment surge in Ginnies to a new factor announced mid-September – impending tighter restrictions for FHA streamline. (Analysts at Deutsche Bank commented that the changes “may well have led some servicers to speed FHA borrowers through the streamline process before the rules are tightened. This latter effect may continue for another month or two until the rules become effective.”

Although analysts understand the general impact of Ginnie and FHA program guidelines on the prepayment behavior of premiums, the disparity between 1s and 2s was less explicable. Barclay’s analysts noted that buying loans out of Ginnie 2 pools “ … is more discrete because GNMA does not release buyout data by issuers for GN2 pools. Of course, these are only hypotheses and the exact timing of servicer buouts is large idiosyncratic as we have witnessed many times before. A large servicer can single-handedly move aggregate speeds on a GNMA cohort by more than 20 CPR in a month.”

(Buyouts can be confirmed for Ginnie 1s in supplementary pool data published with a one-month lag – in November for September activity. Again, by cohort, MBS analysts mean a specific issuance vintage within a coupon within a specific pass-through program. Across the agencies, there are dozens of programs, but the 30- and 15-year fixed rate programs are by far the most important.)

Barclay’s analysts do expect the Ginnie 1s to catch up. The 2s are unlikely to maintain their “advantage” because 1) they generally have lower delinquencies than the 1s, and 2) a spike in buyouts in September would mean fewer loans in the delinquency pipeline going forward. More generally, they anticipate prepayments to trend up for Ginnies for the next couple of months given the recent rally in mortgage rates since mid-August, the sharp build up in delinquencies and higher dollar prices for premium loans.

By cheapening Ginnie 30-year premiums across the board, the market seems to have agreed. The market may also have been reacting to the unappealing uncertainty and volatility imparted to prepayments by Ginnie and FHA program design, underlined by the unintuitive discrepancy in behavior between the two programs. The uncertainty and volatility have long been a factor otherwise limiting the potential audience for a security backed by the full faith and credit of the U.S. government.

If you’re thinking this is just a problem for MBS investors and traders, think again. Demand for Ginnie Mae pass-throughs sets the price at which lenders can sell their FHA, VA and other government loans as securities, and that price determines the interest rate they charge borrowers. The FHA program in particular has become one of the government’s lifelines to housing markets, particularly for borrowers with less-than-perfect credit. It also has taken on quite a bit of water itself in the foreclosure-housing storm, an issue that could – if program adjustments such as those recently announced can’t bolster capital – come directly back to taxpayers.

[Linda Lowell is a columnist for HousingWire. Her in-depth feature, Kitchen Sink, is published monthly in the magazine.]