Taking the politics of housing reform out of the equation, a central question being considered among policymakers and analysts is what the size should be of any private first-loss credit enhancement in supporting the mortgage market. The Senate's Corker-Warner proposal has drawn considerable attention to this debate by establishing that level at 10%. Recent analysis by the Urban Institute suggests 4% to 5%.

However, securities issued by individual lenders could pose considerable credit risk, judging from actual defaults on loans guaranteed by the government-sponsored enterprises. Ignoring the effects of seller origination, sourcing, and servicing processes in a post-GSE secondary market could generate losses for taxpayers. Seller screening, differential credit enhancement, and diversified lender pool requirements could address this issue.

The Urban Institute study's authors apply reasonable assumptions regarding loss severity and default rates to arrive at a 4% to 5% loss rate. That would, according to their calculations, provide ample loss coverage above and beyond what is needed for today's originations, which are of much higher quality than the vintages the Urban Institute studied.

The authors hit on a key point in their study regarding the need for diversification across pools being securitized. However, diversification is usually thought of in terms of geographic terms rather than lender concentrations – which in the past have been shown to exhibit significant variation in loan performance, controlling for other factors that explain defaults.

I conducted a statistical analysis of loan default using that same GSE loan-level data as the Urban Institute study. I included in the default model such individual borrower risk attributes as borrower credit score; loan-to-value-ratio; debt-to-income ratio; property type; occupancy status; property location; and a variable representing particular sellers to Fannie Mae and Freddie Mac from 1999 to 2011. A number of sellers had abnormally high default rates during this period, ranging from 1.5 to three times higher than other sellers. The variable distinguished these seller effects from all others while controlling for differences in borrower quality, geography and other factors that might affect differences in defaults across sellers. The statistical results showed that depending on the origination year, a number of sellers were associated with default rates that were 1.2 to1.5 times greater than other sellers, controlling for other risk attributes.

Applying a 1.5 seller effect factor to the Urban Institute's estimate for Fannie Mae's 2007 book of business loss rate of 5.44% yields a loss rate of 8.2%, which suggests credit enhancement well above 5% is required. And while the newer vintages acquired by the GSEs are clearly of higher credit quality, stressing those books as if they were originated in 2007 rather than in 2001, as assumed in the Urban Institute's study (where those vintages had the benefit of years of home price appreciation and low interest rates before the crash) could lead to losses appreciably higher than the 2% to 3% losses estimated in the study.

So what does this all mean? First, the lender loan manufacturing process matters. Although origination and sourcing practices have improved since the crisis, there will always be variability among lenders. Since any catastrophic federal guarantee would cover losses at the security level, a pool of loans originated by a single issuer could pose significant risk to the taxpayer and thus require substantially more credit enhancement than a pool compiled from many lenders. Thus, whatever entity is created to aggregate loan pools and issue securities should be required to establish multilender pools that diversify lender process effects in addition to geographic or other risk factors. Seller and servicer screening activities must be strengthened as part of this new securitization process.

More broadly, the impact of lender effects on credit losses is just one of many issues policymakers need to address in establishing private-public mortgage credit risk sharing and underscores the complexity in the details of GSE reform. This is why Congress should establish a bipartisan and broad-based panel of experts to produce a set of options on setting the level of private credit enhancement for a post-GSE secondary market. This group would take into consideration the scale of the resulting mortgage market; effects on borrower access to credit markets and associated costs; taxpayer impact; and the potential effects on Federal Housing Administration mortgages.

A number of proposals have been put forward already on the form and substance of a private credit enhancement. However, a Congressionally-sanctioned body of housing and mortgage capital markets experts providing sound and balanced analysis could be a catalyst for reform.

Clifford Rossi is the Professor-of-the-Practice at the Robert H. Smith School of Business at the University of Maryland.