Fred, I like the way you are thinking. I wish more people on this board thought critically. But I think you are going down the wrong path. The MBS mortgage didn't blow up solely because of verification of income (The Credit Card industry is still chugging along, without verifying most people's income). The MBS market blew up because loans were being made with the source of repayment being collateral and not cash flow. If you aren't involved in finance/banking, basically what this means is that everyone forecasted the collateral values would at least remain constant (most forecasted they'd increase forever). They weren't worried about incomes because if the borrower didn't pay, the bank could foreclose on the asset and sell it. Since they were increasing Y/Y, not much risk to the bank (we do have selling costs and such, but those are helped by the gain in market value). But what happened? The loans were made to bubble-level prices and the floor came out from under them. Borrowers don't want the houses anymore (they are underwater and mortgages have lots of various laws (state and federal) which limits the recourse. These precipitous drops in collateral values were accompanied with spikes in unemployment as well. Basically, what I'm saying is income wasn't the sole purpose of the problem. It did play a part, but there were a lot of moving parts.
Well put, rawraw. I agree that there were a lot of moving parts.
Full disclosure: I was a "quant" in 2007-2008 working on RMBS (Residential MBS) pricing engine for a Wall Street bank. RMBS, CDO, CDS are hot potatos now. However, I do believe some lessons learned on credit risks are still applicable to P2P consumer loans.
There are almost 100 attributes in downloadable LC loan data, somebody can analyze which are (statistically) significant factors for default.
I do give statistical preference in my model to loans whose "is_inc_v" is TRUE. Note, this is just a statistical preference, not a binary yes-no decision.
