Revisiting the Net Benefits of Freddie Mac and Fannie Mae
effect on conforming loan rates (7 basis points). However, prominent scholars have shown the PSB estimate to be biased downward.52 Thus, Passmore’s study understates the contributions of Freddie Mac and Fannie Mae to the housing market.
For mortgage-backed securities, we used a range of 10-30 basis points, the same as in 2001. While the funding advantage on MBS is difficult to estimate, we believe it should be between the advantages for short- and long- term debt securities.
The GSE Funding Advantage
GSE funding advantage estimates are needed for three types of securities—short-term, long-term, and mortgage-backed.
For short-term securities, we used a range of 10-20 basis points in our 2001 study, and we believe that range remains valid.
Exhibit 4 presents our estimates of the annual funding benefits to the GSEs from federal sponsorship. The estimates reflect the spreads discussed above, as well as balances outstand- ing for GSE debt and MBS as of year-end 2005. Total funding benefits to the GSEs are in the $4.7-13.1 billion range, up from $2.3-7.0 billion in our previous study based on data from September 2000.
For long-term securities, our previous study estimated the range at 10-40 basis points.53 Comparing recent estimates to what was avail- able for our 2001 study, we believe our previous range was about right, but that a small increase to the lower bound is now warranted. Thus, for long-term debt we now apply a range of 15-40 basis points.54
Direct Benefits to Homeowners
In a parallel analysis, we estimated the savings to borrowers from the GSEs’ activities by apply- ing the interest-rate reductions to the relevant outstanding mortgage balances.
See Blinder et al. (2004), Greene (2004), Flannery and Lockhart (2005), Roll (2005) and McManus and Ramagopal (2005). PSB redefine, rather than re-estimate, the jumbo-conforming spread. Their measure is the difference between the average rate on jumbo loans and the average rate on loans over $100,000. This will always be a much smaller difference than the conventional measure, which is the amount by which the functional relationship between interest rates and loan amounts shifts up at the conforming loan limit. (See Exhibit 1 in our 2001 report.) In a sensitivity analysis PSB report that their estimate with a conventional specification is 28 basis points, squarely in the range of conventional estimates.
We believed then, as now, that the appropriate comparison for this purpose is between financial firms with ratings of AA or better. Nothaft, Pearce and Stephanovic (2002) estimate a range of 27-30 basis points for spreads between the GSEs’ securities and AA- debt; this is likely an overestimate because it does not control for the greater liquidity of GSE securities. Their estimate is based on the same bond-level data analyzed by Ambrose and Warga (2000) in their update for the CBO, and it is very close to the estimated spread for AA securities reported in that study.
Although none of the recent estimates is below 24 basis points, we are not confident that any of these estimates fully capture the effect of the liquidity advantage imbedded in GSE issues. Furthermore, Hubbard (2004) argues that the low GSE failure risk would justify GSE yields well below AA yields even without federal sponsorship. Thus, we believe the GSE advantage in the issuance of long-tem debt could be as low as 15 basis points.
Estimating the Direct Benefits of Freddie Mac and Fannie Mae