Ezra Klein links to an interesting paper by Gary Gorton of Yale University. In that paper, Gorton describes how the subprime crisis is linked to runs on other structured products through the repo market. A sale and repurchase agreement ('repo' for short) involves a short term swap of collateral (t-bills, etc) for cash - the value of the transaction is discounted by some margin known as a haircut.
For example, Bank A wants to raise cash and pledges an asset worth $100 as collateral. Bank B takes the asset and gives Bank A $98 in cash. The $2 difference is the haircut and depends on the credit risk of the counterparty.
Here is where Hiram Minsky enters the picture - look at the graph below from Gorton's paper:
Klein makes the point that it is not the spike in haircuts demanded that is troubling, it is the period of extremely low haircuts just prior to the crisis. Minsky warned that such periods of calm betray extreme underestimation, and hence underpricing, of risk that eventually leads to crisis, asset fire-sales and flight to quality - a scenario that Paul McCulley of PIMCO termed a "Minsky Moment". A smaller scale example of a Minsky Moment is the collapse of the hedge fund LTCM in 1998. Followers of that story may recall that LTCM convinced its counterparties to allow them to borrow without taking a haircut on collateral. We all know how that ended.
I haven't seen any research exploiting the potential predictive information in haircuts, credit spreads, etc in detecting a rising probability of these Minsky Moments. Given the recent rediscovery and surging popularity of Minsky (I certainly had never heard his name in all my years studying economics) perhaps some enterprising economist is already working on it.