Lecturing Birds on Flying by Pablo Triana

Author:Pablo Triana
Language: eng
Format: epub
Publisher: Wiley
Published: 2010-05-11T04:00:00+00:00

One would tend to assume that a widely shared reaction to all the disappointments and disastrous performance during the crisis (which came on the heels of historical disappointments and disastrous performances, let alone plenty of publicly available warnings) would be to shy away from VaR-embracement and to discard the safety mechanism that won’t protect during turmoil (that in fact discards the very possibility of turmoil). But while these would surely be hard times to enhance the ranks of devoted VaR groupies, voiced reactions by those in the game seem to lean not towards the annihilation of VaR, but rather towards its reform. The goal seems to be to try to preserve the preponderance of the tool through a few technical twitches, not to humbly admit that a failed (and, once more, inevitably so) guide should not be kept around. By going that route, those non-annihilators contribute to increasing the risk of unfettered and misguided risk-taking and of the subsequent market catastrophes.

The technical amendments recently suggested for VaR range from the use of a larger data sample (go further back in history so that you can capture a larger myriad of events that may possibly result in a more reliable mirror on future events), to the use of a shorter data sample (go less far back in history so that recent events are “read” faster by the model), to the disproportionate weighting of more recent data (so that what happened yesterday counts more than what happened last year), to the more regular updating of the data sets (shift from monthly or quarterly to weekly or even daily as a way to capture sudden changes in market environment). But this is like our fictional car dealer trying to make up for the nonfunctioning of the air bag during accidents by offering an upgrade in its texture and manufacturing material. Those technical niceties won’t help a bit as long as the damn thing continues to refuse to inflate as we crash against the concrete wall. The main issue simply would continue to be avoided and stubbornly left in place, unaltered, ever dangerous.

The problem with VaR is not that the historical data is too short or too long, but that it is entirely based on historical data. This makes it a hopeless task, and not just because the past is not a wonderfully accurate reference. Readers should keep firmly in mind that one would obtain an entirely different VaR figure depending on the historical data sample chosen. And that such choice is entirely arbitrary (i.e., there is no set-in-stone golden rule; it’s up to you and your intrinsically intrinsic self). Basel II constrains personal selection to a minimum of one-year observation period, but beyond that it’s your call (by the way, what’s so great about one year of data?). And whether you go for a two-year or five-year window will make a big difference to your VaR number, with direct implications on your risk-taking and your capital charges, and on how exposed you are deemed to be by others.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.