Uncategorized
Servicer impact on prepay speeds in Ginnie Mae project loans
admin | March 15, 2019
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.
It is well established that the prepayment profile of a residential mortgage loan can be impacted by the underlying servicer. Residential MBS investors often prefer pools with either a wide dispersion or particular concentration of servicers in order to dilute or enhance the servicers influence on speeds. Commercial mortgage speeds have often been considered less likely to be influenced by choice of servicer since borrowers are subject to significant prepayment penalties, more sensitive to property price appreciation, and quick to identify and take advantage of opportunities to refinance. Despite this conventional wisdom, analysis of historical speeds of Ginnie Mae project loans shows significant variation among servicers, with prepay speeds of some servicers being up to 50% faster or slower than reference cohorts. The differential impact on project loan prepay speeds may be particularly relevant in deals with high concentrations of particular servicers.
Evaluating prepayments across servicers
Agency servicer prepayment rankings are produced by comparing historical prepayment speeds of each servicer’s pools against the universe of comparable pools while controlling for most other factors that drive prepayments – including loan size, loan age, coupon, vintage, construction or non-construction loan, and prepayment penalties. The speeds of each servicer’s loans are compared to a reference cohort – that is, the universe of loans with the same characteristics. The difference between the historical CPR of the servicers pools and the CPR of the reference cohort is calculated across several horizons, and the servicers are ranked from fastest to slowest. The most recent ranking for Ginnie Mae project loan servicers across a 12-month horizon is shown in Exhibit 1.
Exhibit 1: Historical servicer speed ranking – Ginnie Mae project loans
Note: March 2019, reflects February speeds. Servicer rankings are produced monthly. Links to the most recent top 40 and top 200 servicer rankings appear in the APS Portfolio Strategy Library each week. Source: Ginnie Mae, eMBS, Amherst Pierpont Securities
The column highlighted in blue indicates that the servicers are sorted from those with the highest speeds on a percentage basis relative to the underlying reference cohort, to those servicers who have the slowest speeds. The highlighted row, which in this instance corresponds to the servicer Jones Lang LaSalle, identifies the servicer whose historical prepay speeds are currently closest to those of the universe.
Applying servicer speed dispersion to project loan deals
The number of servicers represented in Ginnie Mae project loan deals varies depending on the number of loans and the size of the deal. Deals from 2010 vintage often contained 30 to 40 loans, but recent deals have become larger and more diverse, averaging 75 loans since 2016. There are 43 servicers tracked in our historical prepay table in Exhibit 1. The concentration of those servicers across deals can vary considerably. Three recent deals are compared in Exhibit 2; the three deals are reasonably similar in size, number of loans and number of loan servicers. The split between multifamily and health care loans in each deal is also very close – this is relevant because it is one factor which the historical prepay speed comparison does not control for when comparing cohorts.
Exhibit 2: Comparison of three recent Ginnie Mae project loan deals
Source: Bloomberg, Amherst Pierpont Securities
The column on the far right illustrates how the weighted average servicer contribution to the deal might impact prepayment speeds. An example of the calculation for the GNR 2018-162 deal is shown in Exhibit 3. The underlying loans are aggregated by servicer, and the percentage of the deal represented by the servicer is in the second column. The third column uses the historical prepayment data, shown in Exhibit 2, that indicates over a 12-month horizon how much faster or slower has that servicers loans prepaid compared to the reference cohort. The number at the bottom is the deal weighted average of the differences in speeds based on current loan balance. It indicates that based purely on the distribution of servicers, the GNR 2018-162 deal might pay 11.5% faster than a comparable cohort of loans.
Exhibit 3: Servicer impact on prepayment speeds for GNR 2018-162
Source: Bloomberg, Amherst Pierpont Securities
The same analysis was applied to the other two deals: the GNR 2018-169 deal would be expected to prepay 4.3% slower than a reference cohort, and the GNR 2018-130 falls between the other two at 6.4% faster potential prepayment speeds.
Conclusion
The influence of servicer on Ginnie Mae prepayment speeds is material. Investors should evaluate deals for particularly high concentrations of servicers who, across multiple horizons, consistently tend to have loans which prepay at faster or slower speeds than materially equivalent collateral. The future is not necessarily going to look like the past, but it’s an additional factor that can be incorporated when evaluating project loan deals and particularly prepayment sensitive tranches, such as the IO.