« Risk Management and Economic Development |
Main
| Turmoil in the US Pension Funds Sector »
September 18, 2008
Uncertainties about Credit Portfolio Models
There is a huge loss of confidence in the financial marketplace. Who would otherwise think that despite the massive US $ 85 billion bailout of AIG by the FED, the markets will continue to slide. The intervention is real, not talking the market up. The faith in the market is completely destroyed, at least for the time being, and the faith is the central bank has dented. The impact of US $ 180 billion is yet to be seen and one hopes that the Central Banks together can overpower this crisis of monster proportion. ( Hooray, the Dow is in the green as I write, so it may be that we still believe our central banks)
One silent casualty has been the Credit portfolio Model that churned out default probabilities and expected loss numbers. Quite clearly, the present group of credit portfolio models will be reassessed and one is hopeful that the soul searching will lead to better modeling in future in arriving at the economic capital.
Once we come out of the current state of the market, and start thinking about the nightmare with some poise ( hopefully, sooner than later)Models that depend upon PD, LGD, EAD and Default correlations will undergo a post mortem. Let's think of a few issues that may come up with credit portfolio modeling.
First, the class of models need to pay more attention to the derivatives snd off balance sheet items that goes in the bucket of the unknown and the risk information destroyed in the process.
Second, the correlations in the credit portfolio models suffer from the generic problems of subjectivity in time horizon, in many cases such correlations are based not on a global dataset, but only on US data, which is far from ideal.
The normality assumption acts as another constraint, that needs to be thoroughly examined, with the need for backtesting capabilities for model validation process.
these constraints are no reason to throw the model away, but to strengthen it for future.
As Rene descates once observed, our reason need not be clouded by blind faith in black box models. The search for better method is as relevant today as it was in the early seventeenth century, when Descartes observed, in his own flowing language:
" observing many things which, however extravagant and ridiculous to our apprehension, are yet by common consent received and approved by other great nations,I learned to entertain too decided a belief in regard to nothing of the truth of which I had been persuaded merely by example and custom; and thus I gradually extricated myself from many errors powerful enough to darken our natural intelligence, and incapacitate us in great measure from listening to reason."
Rene Descartes, Discourse on Method, 1628.
Posted by sunandoroy at September 18, 2008 05:26 PM
The present problem is not due to inadequacy of credit portfolio models that are used by banks. I understand are doing what they are supposed to do pretty much OK.
The crisis has hit investment banks much more than deposit taking banks, other than Wamu and not Wachovia (which was largely due to account freeze by JPmorgan), which has a huge concentration on mortgage portfolio for a long time. Even so for banks, problem stuck in investment portfolios and not banking book portfolios. In my view, the core problem was leverage, lack of transparency in accounting (such as ambinugity in fair value reporting) aggreavated by mortgage crisis, that lead to severe liquidity crunch, across the board downgrades and massive writedowns.
Posted by: Sumit at October 6, 2008 03:12 PM
Thank you sunandoroy!. it was quite an eye opning for me. but after analysing the current global creit crisis, I CAN ONLY CONCLUDE THAT it is not the sysytem or models BUT THE un scruplous peaple with dishonet intention, the SICK MBO DRIVEN managment, focusing only on fals accounting profits to REAP MORE IN SAHPE OF BONUSES AND PERKS than the sustaianable RETURN TO shareholders, the ultimate GREED, (a shorter pay back period, in that case for the HR capital).
i am particulalalrly mentioning the Lehman BRos. case, the non-provisioning of losses and accounting manipulation by the CFO.(please see BNEt.com).
i beleive models can fairlry predit OUTPUT what inputs are put into it. BUT when greedy poeple wants it than IT IS A GIGO.
ALSO A 100% REALINCE TO MODELS IS ALSO NOT A FAIR GAME. human judegment is more important in many cases IF THE PEOPLE ARE NOT GREEDY.
Posted by: MUHAMMAD SABIR at October 14, 2008 05:03 AM
Post a comment
