3 Credit Risk Capital questions

[ajsa]

Hi David,

1. Is the estimate of correlation provided by regulator under Advanced IRB?

2. Does IRB also account for Credit Risk Migration (CRM)?

3. What is the purpose of Ratings-Based Approach (RBA), Supervisory Formula (SF), and Internal Assessment Approach (IAA)? Are they for the treatment of securitisation exposures?

many thanks!

 

[David Harper CFA FRM]

Hi asja,

1. (see here, e.g.). Per our other thread, correlation (like granularity) is hard-wired into the IRB (so, i think we can say yes, supervisor provided, but in the case of correlation & granularity, I prefer "hard wired" or "predetermined"): it is a function of the exposure type (e.g., corporate). As previously discussed, we are referring to the rho in the IRB. As a totally secondary matter, within the exposure type, the rho is informed by PD; e.g., corporate rho is 12% to 24% (type) and varying within that range based on PD.

Bottom line: the bank does NOT set the rho internally itself.

2. See graphic below from page 20 of tutorial 7d

Yes, definitely. CRM may only seem less explicit in IRB b/c it is treated (naturally) via the LGD component, as IRB is a function (PD*EAD*LGD*f(M)) rather than a "lookup" weight mutliplied by the exposure. It is treated under "Risk Components" of IRB.

(and just a note: if you were forced to guess about this on the exam, what is a good guess? I think "yes" is a good guess because we'd expect the more advanced approaches to incorporate CRM)

3. Yes, as below, these are a "hierarchy" of methods under IRB securitization. From Basel:
"609. The Ratings-Based Approach (RBA) must be applied to securitisation exposures that are rated, or where a rating can be inferred as described in paragraph 617. Where an external or an inferred rating is not available, either the Supervisory Formula (SF) or the Internal Assessment Approach (IAA) must be applied. The IAA is only available to exposures (e.g. liquidity facilities and credit enhancements) that banks (including third-party banks) extend to ABCP programmes.

David

From tutorial 75, page 20:
http://learn.bionicturtle.com/images/forum/nov18_crm.png

 

[ajsa]

Hi David,

Thanks for your detailed explanations!

2. Is the risk weight function actually the ASRF formula? I understand ASRF is used in both FIRB and AIRB, right? And can bank choose to use another model?

3. Are RBA, SF, and IAA part of ASRF, or they are parallel to ASRF? If they are part of ASRF, how are they reflected in ASRF?

4. Is IRC actually for market risk capital, even though it measures credit risk? How does IRC interact with market risk capital?

Thanks!

 

[David Harper CFA FRM]

Hi asja,

2. ASRF is the framework that justifies the IRB risk weight function. Superficially: ASRF = IRB risk weight function.
Can they choose another model? Absolutely not. Bank supplies PD (FIRB) or PD/LGD/EAD/M (AIRB) but they are stuck with the IRB formula

3. Are RBA, SF, and IAA part of ASRF, or they are parallel to ASRF?
No, they are not "integrated" into the IRB risk weight function; they have their own set of rule for securtization (e.g., RBA basically amounts to a lookup table of risk weights for the securitization exposures ).

4. Yes, IRC was lately "tacked on" to the market risk; I have some coverage in the tutorial and here:
http://www.bionicturtle.com/learn/article/proposed_reviews_to_basel_ii_market_risk_frm_in_the_news/
...really briefly, they are supposed to add a VaR-type estimate for migration and default to the market risk

David

 

[ajsa]

Hi David,

3. So under IRB, the credit risk RWA = RWA from IRB + RWA from securitisation (using RBA, SF, or IAA)?

4. I can partially understand why migration risk can go to market risk capital.. but for default risk, isn't it already captured in credit risk capital? will including in IRC(market risk) create double-counting?

5. so how does IRC contribute to (or work with) market risk capital? just to add on the top of it? But IRC uses 99.9% 1 year horizon, is it still consistent?

Thank you!

 

[David Harper CFA FRM]

Hi asja,

3. No, RBA/SF/IAA are specifically for securitization exposures (i.e., they get their own special rules)
4. IRC is for the these "incremental risks" in the trading book, explicitly to compensate for deficiencies in market risk VaR. i think there have been criticisms related to potential double counting. But (IMO) the way to view the IRC is: Basel's way to try and address the fact that "credit risks" are increasingly (obviously) part of market risk. Securitization blurs the line by trading credit risks
5. yes an add-on (incremental). Re: "But IRC uses 99.9% 1 year horizon, is it still consistent?" Very astute! NO, many many practitioners have objected to precisely this inconsistency, i will spare you Basel's convoluted defense. Clearly, it's inconsistent!
...i personally view the IRC and stressed VaR as troublesome b/c they add two layers of complication (I mean, all-in, B2 is getting to be unweildy) AND introduce inconsistencies...i think they are both veritable PROOF that Basel has acknowledged the framework has real problems

David

 

[ajsa]

Hi David,

Hopefully, this is my last question on this topic.

Is IRC for specific risk only?

Thanks.

 

[David Harper CFA FRM]

Hi asja,

I am not sure about that. Clearly, IRC means to include "specific" default & migration; I am not recalling a limitation that such risks be "specific." Not sure.

David

 

[ajsa]

Hi David,

http://learn.bionicturtle.com/images/forum/nov18_crm.png

In your figure above it seems FIRB and AIRB treat derivative differently.. could you elaborate?

Thank you!

 

[ajsa]

Hi David,

5. "yes an add-on (incremental)."
And Stress VaR will also need to add on Market Risk capital at the same time, right?

Thanks.

 

[David Harper CFA FRM]

Hi asja,

In regard to IRB: Derivatives (please note I use a little calculator to denote "internal" and a building to denote "supervisor"):
Basel para 300. "There are two approaches for recognition of CRM in the form of guarantees and credit derivatives in the IRB approach: a foundation approach for banks using supervisory values of LGD, and an advanced approach for those banks using their own internal estimates of LGD." (put another way: just like the typical treatment of components)

...more exam relevant is the treatment of collateral under FIRB, note the LGD is supervisor (not bank) supplied: "Under the foundation approach, senior claims on corporates, sovereigns and banks not secured by recognised collateral will be assigned a 45% LGD. 288. All subordinated claims on corporates, sovereigns and banks will be assigned a 75% LGD."

Re: stressed VaR: yes, this formula was assigned in the AIMs. Recall the proposed charge is:
k = max(yesterday's VaR, 60-day average VaR*Factor) + max(yesterday's stressed VaR, 60-day average stressed VaR*Factor)
...so it is a TOTAL add-on of the stressed VaR
...which, btw, makes absolutely no sense to me (the stressed VaR overwhelms/subsumes the regular VaR)...totally nonsensical

David