Nothing to do with airlines, we speak here of validating expected loss against actual loss.
A point made by Bruce M in recent comments is that there needs to be consistency in the modelling methodology behind the suite of models for the risk components PD, EAD and LGD. One task that should bring this point to the fore is the validation of EL against AL.
The PD (and EAD) models can be easily validated because their predicted outcomes become certain after 12 months. LGD is hard because
- the observation period starts later: if an account defaults in the 11th month of the 12-month outcome window, observation of the actual LGD outcome (i.e. actual loss) can only begin at that point, which is already 11 months later than the sample cohort.
- the observation period may be long
- ideally one needs to wait for the longest AL to resolve, but one can’t know in advance how long this will be
This means that ELs can only be reliably validated against ALs if the sample cohorts are quite far back in time – perhaps 2-3 years depending on product.
Nevertheless an adequate job can be done on more recent cohorts, considering that even on recent cohorts, at least some of the ALs will be known. I recommend a graphic approach showing EL vs AL for many quarterly cohorts simultaneously, with certain ALs in a bold colour, and as-yet-unresolved defaults shown on a possible – probable – worst case basis via suitable graphic clues (e.g. colours, hatching, error bars). Such a display will show a ‘fan’ effect, whereby older cohorts have a more certain EL-AL reconciliation, whereas for more recent cohorts the zone for AL fans out. (EL is a historic fact and is always known exactly)
Carrying out an EL-AL validation is a good way to review the consistency of model approaches and to detect those situations that fall between the cracks.
5 comments
3 July, 2008 at 11:20
Andrew
Clive,
It may also be useful to note that economic conditions are likely to have changed more over the 2-3 year (or longer) horizon that the LGD takes to resolve than the 12 month horizon for the others, meaning that any variation between EL and AL may have more to do with economic factors than modelling error.
OTOH – it could be that the modelling error was high, but the economic condition change acted to mask this, meaning that although the models were bad, they appeared correct. Possibly a worse outcome long term for the institution.
3 July, 2008 at 13:14
Chris
Andrew,
Can economic factors significantly alter LGD, especially over time horizon of a few years? I would think that product/collateral characteristics should not change much, less perhaps some change in related laws.
3 July, 2008 at 17:10
Andrew
I was thinking more on the unsecured lending side than the secured, but even there housing prices (for example) are more likely to change in 3 years than 1. For example in the US, a housing loan written, say, 2 1/2 years ago may have had (correctly) a low LGD assigned against it if it went into default after 12 months good, but the movement in housing prices since then would mean that the LGD would have dropped substantially, particularly if the loan is without recourse.
The PD and EAD would have been validated at 12 months – the LGD (if the house could be sold at all) would have shifted rapidly in the last 12 months.
4 July, 2008 at 14:57
Paul
This is particularly difficult for non retail esp non property assets where there is a real paucity of data and each workout is unique. It is possible to buttress internal experience with external data from liquidators / auction houses, but caution should be exercised as circumstances of liquidation may be diffenet to assumptions of internal LGD estimate
8 July, 2008 at 02:06
Validation « ozrisk.net
[…] simplest setting is validation of an individual component, especially PD. Last week’s post touched on the more difficult context of validating that the chain of models PD-EAD-LGD work […]