Observations on the Mortgage Market: 2011 HMDA Data

Yesterday the financial regulators released data on the 2011 mortgage market, as collected under the Home Mortgage Disclosure Act (HMDA).  Setting aside my belief that HMDA should be repealed and the data collection ended, the release, summarized here by the Federal Reserve, does offer a number of insights into both the mortgage markets and mortgage regulation.

The headline take-away is mortgage loans have fallen to 16 year lows, particularly among home purchase mortgages.  One of the reasons, not so much discussed, is that lenders are a whole lot less willing to hold mortgages on their own books.  In 2006, lenders actually held about 39% of purchase mortgages on their balance sheets.  By 2011, that percentage had fallen almost in half to 21%.  Now there have been a few exceptions.  Wells Fargo, for instance, is actually holding more purchase mortgages on their books in 2011 than in 2006.  Most lenders, however, have greatly reduced their balance sheet exposure.  Take at one extreme, Citibank, which went from holding over half (52%) in 2006, to just 8% in 2011.  While lenders are clearly, and understandably, trying to minimize their interest rate risk, I believe another fact is trying to control both their credit and legal (not to mention political) risk of holding mortgages.

If you aren't holding mortgages, then the other option is to sell them, either to other banks or to Fannie/Freddie.  Here we see the impact of the GSEs increasing their credit standards (appropriately in my opinion).  Go back to 2001 and lenders were selling almost half their "unconventional" mortgage originations to the GSEs, by 2011 this had fallen to just over a third.  One does not see a similar trend in GSEs purchases of conventional mortgages, which remains just above a third in both 2001 and 2011.

Despite the increased concentration in banking, in general, share purchased by the Top 10 bank originators changed little between 2006 and 2011, increasing slightly from 35.2% to 36.9%.  The other trend to catch my eye was that most of the decline in mortgage activity from 2010 to 2011 was among low or moderate income households.  Purchase lending for high income was almost flat.

These are just some initial thoughts.  Still digesting what is a pretty large data dump.