Revisiting the Net Benefits of Freddie Mac and Fannie Mae
Benefits to GSEs ($Billions/Year)
Exhibit 4 Annual Benefits to GSEs
Balances Outstanding ($ Billions)
Spread (Basis Points)
For the jumbo-conforming spread, we used a range of 24-28 basis points in our 2001 study. Aside from Passmore, Sherland, and Burgess (2003), most studies since 2001 have produced estimates within that range.55 While we keep the lower end of our range unchanged, we do raise the upper end of the range slightly in light of the analysis by McManus and Ramagopal (2005) that suggests an allowance for bias of two to four basis points is warranted. To err on the conser- vative side, we add two basis points to the top of our previous range, making the current range 24-30 basis points.56
For the effects on conforming adjustable- rate mortgages (ARMs) and jumbo loans, we previously assumed that GSE activities reduce rates by five basis points. Here we use ranges to express a level of uncertainty about the specific magnitude, but continue to use five basis points as the center of the ranges. For the effect on ARMs, we assume four to six basis points, reflecting the availability of information from regular interest-rate surveys. We regard the effect of the GSEs on jumbos as subject to greater uncertainty, so the range we adopt is three to seven basis points.
Those that do not, such as McKenzie (2002) and Torregrosa (2001), used Monthly Interest Rate Survey (MIRS) data and included observations prior to 1997, when the Federal Housing Finance Board first began to edit the data more attentively to exclude erroneously classified observations. Despite their use of the less-reliable data, those studies estimate the differential to be only slightly below 24 basis points.
They point out that nearly all researchers using the MIRS data continue to trim the data for outliers and misclassified loans. In Section 3 of their study, they present a thorough analysis of the effect of this trimming. They conclude that although trimming does eliminate bias that would occur if ARMs misclassified as fixed-rate mortgages (FRM) were left in the data, some downward bias to the estimates of the jumbo-conforming spread remains, with the extent of the bias depending on the true spread and the rate of misclassified ARMs.
Estimating the Direct Benefits of Freddie Mac and Fannie Mae