Who would have thought a new accounting standard might end up increasing prepayment speeds on US mortgages?
The two US GSEs will be, like other US financial companies, adopting FAS 166/167 from 2010. The new rule is aimed at bringing off-balance sheet vehicles back on balance sheet — but it could end up having a rather interesting effect on US mortgages by way of Fannie and Freddie.
Here’s Deutsche Bank’s Arthur Frank to explain:
Under current Fannie and Freddie policies, a seriously delinquent loan must be bought out of an MBS pool at par plus accrued interest if any of the following three events occur: (1) the loan has been seriously delinquent for 24 months; (2) the loan has been permanently modified, whether through the Home Affordable Modification Program (HAMP), or otherwise; or (3) a foreclosure or short sale has been completed or the property deed has been conveyed by the borrower to the GSE. However, Fannie and Freddie have the right, but not the obligation, to purchase any loan out of their mortgage pools if the loan is at least 90 days delinquent, but until now the GSEs have generally chosen to leave such loans in their pools until one of events (1), (2), or (3) above have occurred.
The major reason that the GSEs have not bought out the large number of delinquent loans in MBS pools is a capital constraint; when a non-performing loan is bought out at par, it is immediately written down on the balance sheet to about 40% of face value, which is approximately the market bid side price for pools of seriously delinquent non-performing loans. The capital for this loss must be obtained from the Treasury Department, and such capital is expensive; the GSEs must pay a 10% dividend to the Treasury on such preferred stock. For the GSEs, buying out a 6.5% loan at par, then borrowing 60% of the face amount and paying 10% on the loan is not, of course, a profitable business model. So for the most part, the GSEs have chosen to leave such seriously delinquent loans in MBS pools and continue to pay investors the coupon month after month.
But those incentives will of course change come next year.
Both of the GSEs have concluded that they’ll have to consolidate the outstanding loans that they guarantee — about $4,500,bn according to DB — onto their balance sheets. Which means that suddenly buying out those delinquent loans won’t require additional Treasury-obtained capital:
Based on information from their latest financial reports, these loans will come onto their balance sheets at par, not at 40% of face amount, where they currently buy out delinquent loans, triggering a large incremental capital requirement . . . In other words, as of January 1, 2010, all GSE loans, delinquent and performing, are already on the balance sheet at par, so buying loans out of pools will have no impact on the GSEs’ balance sheets anymore. The delinquent loans will already be on Fannie and Freddie’s balance sheet, and so there is no more writedown once they are bought out. Of course, the GSEs will maintain loan loss reserves against the delinquent loans on their balance sheet, typically for two years of expected losses, but this is substantially less than reserving the entire present value of the loss. In addition, reserves have already been held against the guarantee fee business outstanding, so the incremental amount of reserves held for a buyout will not be very large.
The GSEs will still have to obtain the cash to buyout loans, which they can do via additional debt issuance or by selling MBS pools from their retained portfolios — but the capital disincentive for delinquent buyouts will have largely disappeared. That would have an effect on Constant (or Conditional) Prepayment Rate (CPR)– which represents the proportion of underlying mortgages that would be paid off in a year at the current pace.
For instance, in the unlikely event that Freddie Mac bought out all the seriously delinquent 30-year 2007-vintage 6.5 per cent coupon loans, it would result in a one-month speed of 77 CPR — so about 77 per cent of the mortgages would be expected to be paid off within a year. Spreading the buyout over three months would end up with a CPR of 37. For context, Fannie Mae prepaid at 15.7 CPR across its MBS universe in October.
We’re not sure what sort of impact a higher prepayment rate might have (mortgage investors generally don’t like higher prepayment since it means faster payment of their principal — often in a low interest rate environment which makes it harder to reinvest) — but it’s all very interesting nonetheless.
File this under `Accounting consequences: unexpected.’
More on maximum negative convexity – FT Alphaville
Prepayment Speeds Show Signs of Life in Streamlined Refis – Structured Finance News
Behind Freddie’s profit, rising NPAs – Rolfe Winkler