Guest Post: A Repo Puzzle - What Happened in July-August 2007?

Tyler Durden's picture

A Repo Puzzle:  WTF Happened in July-August 2007?

By JM

Nothing moves a subject forward like exploring open problems.  Think of this as a first (actually second) entry in a collection of puzzles, hopefully added by more than just myself.  In the tradition of the Scottish Book, each entry will include an attribution/dedication, a problem statement, data that provide some motivation, and associated hypotheses, and conjectures.    

Since anonymous attribution (JM) is nonsense, the problem is dedicated to Lisa Murphy of Bloomberg.  The problem may or may not be significant.  It is not necessarily an indication of financial apocalypse.  In fact, there may be a simple, trivial answer, which would be stupid.  It’s Lisa Murphy’s problem.

Problem:  The repo market for corporate bonds experienced a six sigma+ delivery fail event.  Why?

A part of the puzzle is that this event was isolated to corporate bonds delivery.  MBS, Treasuries, and agencies saw no such spike in July-August of 2007.

The exact dates of these corporate bond delivery fail events were the weeks of 7/4/2007 (98,608); 8/15/2007 (80,505); 8/22/2007 (92,605); and 8/29/2007 (100,965). 
 
Hypothesis 1:  It’s a thin market issue.

Even now, corporate securities are not a big part of repo financing.  Treasuries, followed by MBS, dominate the repo financing market.

Amount outstanding as of December 29, 2010.       
 
When supply is tight, then repo security delivery is naturally more difficult.  When supply is thin in the first place, you have the potential for those bottle-rocket moves.  However, this doesn’t explain that it was essentially a unique occurred for 28 days within a 3,833 day period.   

Hypothesis 2:  Delivery fail spikes in Treasury markets clearly coincide with these securities going special, which is in turn associated with firesales in risk assets. Can this connection between repo funding and delivery fails explain risk premia across asset classes?

Conjecture 1:  Given Hypothesis 2, are these increasing delivery fail event an indicator of dwindling issuance impacting a large modality of security financing? Check out the sustained delivery fails in MBS security repos throughout 2010.