Wednesday, November 19, 2008

Diagnosing Freddie and Fannie

FIX THE FORMULA, SAVE THE WORLD

We know that Freddie and Fannie dropped their market share (h/t Krugman) at the time when the worst vintage loans were being originated. Because of Fannie's fancy-accounting penalties and because they did not participate in a lot of the "affordability products" dreamt up, they were not the drivers of the worst excesses of the markets.

Well, despite consistently applied loan standards, it looks like their economic model for underwriting was not designed with the notion that a housing price bubble could develop. Their formula appears to have been tied to house prices, rather than to "economic fundamentals".
If they certainly didn't create "the bubble", they did get pushed around by it, based on my first look at the figures.

Why did they, if it wasn't some exceptionally bad credit underwriting standards (the bulk of their credit losses appear to be in one part of the portfolio)? Well, despite consistently applied loan standards, it looks like their economic model for underwriting was not designed with the notion that a housing price bubble could develop and that could be a factor. Their formula appears to have been tied to house prices, rather than to "economic fundamentals".

BUBBLE PROOFING FANNIE MAE

FHFA/OFHEO, their regulator, sets a "conforming loan limit", among a number of other conforming characteristics. This is key to segmenting the market, for policy purposes, I'd suppose. Now, this limit periodically gets adjusted. How? It looks like it is roughly tied to the OFHEO house price index, with various degrees of lags and discretion.

Of course, all lending does not take place at the limit, so to speak. However, the average loan amount, the new business for a given year, tracks that year's loan limit pretty closely. The two are in a ratio of near 2, for the priod over which I can get data.

Chart 1. The progression of Fannie's conforming loan limit and the average loan size, shown as a ratio (green line). In 2006, the limit was raised to $417,000.



How closely did the limit progression track home price rises, during the go-go days? It lags. They were not being super aggressive. Had they followed the price index lock-step, the conforming limit would have risen faster. In fact, before the big step-up in 2006, to $417K, the limit was over $45,804 below where it might have been, if it had been raised lock-step since 1991 with home price increase. $45K is a material 'undershoot'. Today, with a fall in home prices and the emergency measures to raise the cap even more, the difference between the actual and implied has been wiped out and the limit is actually above where it would be under a lock-step price-formula only.

Chart 2: Green line tracks how much home prices are rising faster than the conforming loan limit is rising. Red line tracks cumulative dollar impact implied by the difference.
What if they had used an "economic underwriting formula", one that was based on economy-wide fundamentals, like median or average household income?

...it makes a case that it is not to difficult to "fix it", so that today's problems are less likely to recur, and the mission of the two GSE's is not cast aside, unthinkingly.
Well, for one thing, they would have cut themselves out of the marketplace, even more. That may seem trivial, but it's an important consideration, when you think of organization re-deisgn. In a bubble regime, one is intentionally, slowly withdrawing "liquidity", which is something that may be counter-intuitive to an organization whose mandate has always been to provide liquidity.

The two charts below show that there was stability in the ratios of the conforming loan limit to either mean or average income, until the mid 2000s, when it rose sharply.

These charts are, of course, bad news, now that we know the outcome of the rise in home prices and how badly this Administration has been in dealing with the aftermath, especially foreclosure mitigation efforts. Between 2001 and 2007, the average Fannie loan size rose 45%, but average household income rose only 16% (by Census Bureau estimates).

Nevertheless,
it makes a case that it is not to difficult to "fix it", so that today's problems are less likely to recur, and the mission of the two GSE's is not cast aside, unthinkingly.

Charts 3 & 4: Census data on average and median household income related to OFHEO's conforming loan limit (the conforming loan limit tracks average loan size, i.e. the actual values done by the GSEs, fairly closely). No adjustments are made for changes in such financial variables as inflation or general real-estate affordability factors.





No comments: