Comments on: G,B, and the rest http://ozrisk.net/2008/04/27/gb-and-the-rest/ Risk Management in Australia Wed, 23 Jul 2008 19:42:15 +0000 http://wordpress.org/?v=MU By: Clive http://ozrisk.net/2008/04/27/gb-and-the-rest/#comment-26262 Clive Mon, 28 Apr 2008 14:22:48 +0000 http://ozrisk.wordpress.com/?p=319#comment-26262 Guy, thanks for your contribution and insights on two interesting points. #1: 'bad survivors': presumably any extent to which a defaulted account can continue servicing debt adds to profit, even if eventual write-off is not avoided. IIUC you're pointing out that well-managed collections/recovery procedures are a profit centre for banks. Perhaps there's a danger of Basel language tempting analysts into linear thinking PD->default->EAD->Loss as one-way traffic. #2: We're on the same page about closed goods and I'm hoping to illustrate with the benefit of a worked example - watch this space in a week or two and let's resume the discussion. Cheers, Clive Guy, thanks for your contribution and insights on two interesting points.

#1: ‘bad survivors’: presumably any extent to which a defaulted account can continue servicing debt adds to profit, even if eventual write-off is not avoided. IIUC you’re pointing out that well-managed collections/recovery procedures are a profit centre for banks. Perhaps there’s a danger of Basel language tempting analysts into linear thinking PD->default->EAD->Loss as one-way traffic.

#2: We’re on the same page about closed goods and I’m hoping to illustrate with the benefit of a worked example - watch this space in a week or two and let’s resume the discussion.

Cheers,

Clive

]]> By: Guy http://ozrisk.net/2008/04/27/gb-and-the-rest/#comment-26260 Guy Sun, 27 Apr 2008 23:55:12 +0000 http://ozrisk.wordpress.com/?p=319#comment-26260 Clive, great coverage - commended - I'm now an avid reader. Just on your G,B and the rest article, you discussed an important and relevant issue which is what I call 'Bad Survivors'. The reason these are important is that they can be quite profitable accounts, particularly with credit cards (assuming operational unit costs held constant across risk bands.... intuitively this is not the case, but exponents of the 'contribution to fixed costs' argument would discard as not relevant anyway). Another piece I think is important is the treatment of Closed Goods or Attrited Goods. It seems a dilemma for scorecard builders as to whether these should be included in the build or not (attribute correlations are impacted, but degree is debatable as it is varied) - either way though, in my opinion, it is critically important (in operationalising Basel II) that the business understands whether Good Closers were in or out. Reason being, if a cut off was set to drive a RAROC outcome by shooting for say a cumulative G:B of 12:1 on a scorecard built with Good Closers included (treated as Goods), then Attrition (or changes to) would impact the RAROC outcome (up or down). If the model was built with Good Closers excluded (ie. smaller denominator), then PDs would be more or less 'worst case' in the sense that the profit generated by the Good Closures would be unaccounted cream for the taking (assuming out acquisition costs). In my mind, to do RAROC right, Good Closers should be included in PD model build, but an Attrition emergence model should be augmented to the calculation, AND risk based scalars applied to all fixed unit costs (ABCs). Interested to read your feedback on this. Cheers GM Clive, great coverage - commended - I’m now an avid reader. Just on your G,B and the rest article, you discussed an important and relevant issue which is what I call ‘Bad Survivors’. The reason these are important is that they can be quite profitable accounts, particularly with credit cards (assuming operational unit costs held constant across risk bands…. intuitively this is not the case, but exponents of the ‘contribution to fixed costs’ argument would discard as not relevant anyway). Another piece I think is important is the treatment of Closed Goods or Attrited Goods. It seems a dilemma for scorecard builders as to whether these should be included in the build or not (attribute correlations are impacted, but degree is debatable as it is varied) - either way though, in my opinion, it is critically important (in operationalising Basel II) that the business understands whether Good Closers were in or out. Reason being, if a cut off was set to drive a RAROC outcome by shooting for say a cumulative G:B of 12:1 on a scorecard built with Good Closers included (treated as Goods), then Attrition (or changes to) would impact the RAROC outcome (up or down). If the model was built with Good Closers excluded (ie. smaller denominator), then PDs would be more or less ‘worst case’ in the sense that the profit generated by the Good Closures would be unaccounted cream for the taking (assuming out acquisition costs). In my mind, to do RAROC right, Good Closers should be included in PD model build, but an Attrition emergence model should be augmented to the calculation, AND risk based scalars applied to all fixed unit costs (ABCs). Interested to read your feedback on this.

Cheers

GM

]]>