Exam Feedback May 2019 Part 2 Exam Feedback

[nikic]

Anyone remember the answer for capital planning?? I dont quite remeber the answer choices
And what is the answer for BCVA??I blind guessed it as c)10000 or 1 mil?
and for the AI and machine learning
I picked Ai and machine learning reduces trading costs but doesnt do well for price recovery??is that answer choice right?

I can't recall the first two. For capital planning, it should have been something rather obvious I'd presume.

For BCVA...I don't recall such a computation.

As for the AI/ML one...I believe I marked the same answer (reduce trading cost, but doesn't do well for price discovery (other answers were just wrong).

Was there another AI/ML question where one of the choices was "credit risk is still not able to be driven by AI/ML?

 

[nikic]

for the captial planning, essentially they were asking about the CCAR exercise on US banks. So, its the min. regulatory capital ratios.

Didn't the answer say something like "it's sufficient to put into effect the contingency plan when the capital levels breach the minimum regulatory capital ratios"? In which case, it would be wrong since the contingency plan should be put into effect at some point before that?

 

[tirerasalim]

Found the exam to be a bit tricky- especially at the start with the first few questions since I was not sure of any of the answers. I grew more in confidence around questions 20-30. It's weird cause during part 1 that I took in November 18, it was the opposite, felt good right away since I was sure about my first few answers.
also thought there were a lot of theories, especially on book 4, was anticipating that I'd use my calculator way more than this.
good luck to all for the results!!

 

[shivsom1718]

Didn't the answer say something like "it's sufficient to put into effect the contingency plan when the capital levels breach the minimum regulatory capital ratios"? In which case, it would be wrong since the contingency plan should be put into effect at some point before that?

I don't exactly remember if something related to the contingency plan is included in the answers. if the capital levels exceed the reg capital ratios. Its a good thing right?

 

[nikic]

I don't exactly remember if something related to the contingency plan is included in the answers. if the capital levels exceed the reg capital ratios. Its a good thing right?

Yes but my understanding is the minimum capital ratios are the bare minimum. The capital contingency plan of each financial institution should come into effect at a level above the minimum capital ratios, as by then it'd been too late. For example in the insurance industry, the minimum capital ratio could be 130% but internally the company's capital contingency plan should kick into action when the capital ratio drops below, say, 160%. Just my two cents.

I can't recall the answer I marked, but it jived with my logic at least, being involved in the ICAAP process of an insurance company IRL.

 

[Amarnadh D]

1)quadratic programming is the right answer since model is the best one and requires more data inputs
2)Questions on Expected loss: Inputs are Credit spread is 480 bps, EAD is 735,000, RR is 0.4. EL is 35,200, but this is not part of the choice. I double checked, one of the option is 33,200..seems it's a typo. Pleas confirm if anyone remember the question.
3) Question on SOFR: Answer is Overnight treasury repo rate
4) for General Perato distribution - does it measure size or frequency of losses ?
6) from RMIM, I choose Modigliani is the best measure for ranking the portfolios
7) Operational risk loss due to floods is loss of business network in the affected area

 

[nikic]

1)quadratic programming is the right answer since model is the best one and requires more data inputs
2)Questions on Expected loss: Inputs are Credit spread is 480 bps, EAD is 735,000, RR is 0.4. EL is 35,200, but this is not part of the choice. I double checked, one of the option is 33,200..seems it's a typo. Pleas confirm if anyone remember the question.
3) Question on SOFR: Answer is Overnight treasury repo rate
4) for General Perato distribution - does it measure size or frequency of losses ?
6) from RMIM, I choose Modigliani is the best measure for ranking the portfolios
7) Operational risk loss due to floods is loss of business network in the affected area

2) Don't you have to solve for lambda (i.e. 480bps / (1 - 0.6))? And then use that to find the PD, and use the PD * LGD * EAD to find the EL?
4) It's the size of the loss, no? My Answer was Pareto / size of loss beyond the threshold.
7) I also selected ops risk was provision due to the loss of business...but I have doubts that it is correct as the Basel Ops Risk category for natural disaster DOES NOT account for loss of business.

 

[christylee]

Anyone can guess what the pass score would be?
Would it be anything higher than 52/80?

 

[nikic]

Anyone can guess what the pass score would be?
Would it be anything higher than 52/80?

Fairly confident the passing score would be no more 48/80 (i.e. 60%). Remember, on average, 55% of candidates will pass. I have a hard time believing 55% of candidates can score some two thirds of the paper correct. In fact I won't be surprised the eventual passing score is in the 42-46/80 range.

My anecdotal experience from Part 1, where I scored no more than 65/100 but got a rather comfortable 2-2-1-1. In fact for FMP I knew for certain I had 11 mistakes yet I still got a 1st quartile for it.

All I can say is if you get 48/80 you should already be feeling good!

 

[Amarnadh D]

2) Don't you have to solve for lambda (i.e. 480bps / (1 - 0.6))? And then use that to find the PD, and use the PD * LGD * EAD to find the EL?
4) It's the size of the loss, no? My Answer was Pareto / size of loss beyond the threshold.
7) I also selected ops risk was provision due to the loss of business...but I have doubts that it is correct as the Basel Ops Risk category for natural disaster DOES NOT account for loss of business.

RR is 0.4. As per formula, denominator s 1-RR. So it is 1-0.4 = 0.6 which is LGD.
Even I selected size of loss, but as per the definition, I'm suspecting it should be frequency of loss.
Ops. risk can caused from either internal or external events. Floods is external event which damage the business network. Hence this is right answer.

 

[Jaskarn]

On the LTCM question, I admittedly can't remember what the options were and I also can't exactly recall what I put down for sure.

On the portfolio construction...I was totally confident Quadratic Programming was correctly described. I didn't even bother reading the rest! What does it matter if it says "multiple variables" or "3 or 4 variables"? Shouldn't they mean the same? I could be wrong here and if so one mark lost for something I was confident about!

On the manager with best performance, why was it peer alpha? I believe the justification that was given just didn't make sense, so I picked the other alpha where the justification made sense, as it was being compared to the benchmark performance and saw if alpha was significant or otherwise.

Yes, the answer "declaring cash dividend will reduce Tier 1 dividend" didn't relate to the question, but I am confident none of the other answers were correct. Lest I'm mistaken and there was something that was "more correct" than it.

On the specials spread, I believe I answered the same but can't recall the verbiage exactly.

On the KMV/Merton/CreditMetrics...I actually picked CreditMetrics. Admittedly this was a shot in the dark for me.

Reagrding Alpha question peer alha is important beacsue question said that Fund manager wants to review other managers that comes below him it;s like yearly appraisal thing that happens in our companies yearly. So in that that manager need to compared performance of those manager moung themselves to know which manager actually performed best and which performed worst based on that he will decide their bonus.

Regarding Quadratic Programming, stratification was exactly true as compared to Quadratic Programming because in wording there was clarity and that what's exactly mentioned in notes.

 

[Amarnadh D]

Anyone can guess what the pass score would be?
Would it be anything higher than 52/80?

It can be in between 42-48 marks, but again depends on the criticality of the paper.

 

[sofiamits]

I can only agree with the majority that I found the exam tricky and of course was surprised by the relationship between wordy Vs quantitative questions. After leaving the exam room I felt quite disappointed since I was well prepared (at least my feeling).
As the time goes by and after reading other opinions I can say that the feeling is getting a bit better cause I can see that there at least the general feeling more or less. Regarding some of the questions - my answers:
-> General Perato distribution - size of losses conditional that the thresold has been reached
-> I also resolved this this way: solve for lambda (i.e. 480bps / (1 - 0.6)) and then use that to find the PD, and use the PD * LGD * EAD to find the EL
-> AI same here -> reduced costs but doesnt work for pricing
-> Risk Budgeting and which asset to include in portfolio: US Bonds (it was also the one with the higher negative correlation so the effect was reducing the overall VaR)
-> BCVA I chose the 40000 but I thing I made a mistake... It should be normaly CVA+DVA but I couldnt find any of the answers.
-> Filtered VaR was the one that was correct described compared to the others
-> Incremental - Margin - Component VaR
-> With regards to RAROC do you remember the number?

 

[nikic]

RR is 0.4. As per formula, denominator s 1-RR. So it is 1-0.4 = 0.6 which is LGD.
Even I selected size of loss, but as per the definition, I'm suspecting it should be frequency of loss.
Ops. risk can caused from either internal or external events. Floods is external event which damage the business network. Hence this is right answer.

My mistake, the hazard rate should be 0.048 / (1 - 0.4) indeed as you say.

Hmmm, if it indeed refers to the frequency of loss, then welp...one more mistake!

Well, I marked that as well as the ops risk event, so if it indeed is the answer, that's great!

 

[kylian.mbappe]

I can only agree with the majority that I found the exam tricky and of course was surprised by the relationship between wordy Vs quantitative questions. After leaving the exam room I felt quite disappointed since I was well prepared (at least my feeling).
As the time goes by and after reading other opinions I can say that the feeling is getting a bit better cause I can see that there at least the general feeling more or less. Regarding some of the questions - my answers:
-> General Perato distribution - size of losses conditional that the thresold has been reached
-> I also resolved this this way: solve for lambda (i.e. 480bps / (1 - 0.6)) and then use that to find the PD, and use the PD * LGD * EAD to find the EL
-> AI same here -> reduced costs but doesnt work for pricing
-> Risk Budgeting and which asset to include in portfolio: US Bonds (it was also the one with the higher negative correlation so the effect was reducing the overall VaR)
-> BCVA I chose the 40000 but I thing I made a mistake... It should be normaly CVA+DVA but I couldnt find any of the answers.
-> Filtered VaR was the one that was correct described compared to the others
-> Incremental - Margin - Component VaR
-> With regards to RAROC do you remember the number?

BCVA = CVA - DVA so it was 40K to me

 

[kylian.mbappe]

My mistake, the hazard rate should be 0.048 / (1 - 0.4) indeed as you say.

Hmmm, if it indeed refers to the frequency of loss, then welp...one more mistake!

Well, I marked that as well as the ops risk event, so if it indeed is the answer, that's great!

it is size of loss.

 

[kylian.mbappe]

My mistake, the hazard rate should be 0.048 / (1 - 0.4) indeed as you say.

Hmmm, if it indeed refers to the frequency of loss, then welp...one more mistake!

Well, I marked that as well as the ops risk event, so if it indeed is the answer, that's great!

PD = 1 - exp(-s/(1-RR))

 

[Jaskarn]

@ nikic:
a) for the SOFR question the answer would be that its derived from repo trades. I guess its B.
b) For the Stressed loss question, i guess it would be stressed loss = stressed pd * EAD * LGD and then stressed EL - pre-stress EL
c) For collateral, yes it is high threshold.
d) For question 12 in credit risk, its that there was a requirement in OTC regulation for the trades to be posted to a central repository.
e) For the distressed company / high volatility, the impact of asset volatility on subordinated debt is always unclear but the senior debt value reduces?

This was my second attempt after failing the Nov'18 exam. Comparatively, IMO, May'19 was easier than the Nov'18 exam. Well, one can say that's because I'm more prepared this time but from my perspective, ~80% of the questions were really concise and tested the basics of FRM II learning objectives. If you have a good grasp of the concepts and formulae you can easily find the answer. I left the room feeling that I'll pass this time but will have to wait for 6 weeks to get to know if I did or not !!

For a company in distress subordinate bond behave like equity hence it increases

Below reference is from notes.

 

[ziminli1228]

it is size of loss.

yes, it's the size of loss, frequency of loss will use binomial, neg binomial or poisson distriution.
op risk i chose legal costs.

 

[nikic]

it is size of loss.

Do you happen to recall the question / answer in more detail? The distribution as I recall models the severity of losses, but the frequency of losses plays a role in setting the appropriate threshold. I'm getting all mixed up now as to what was tested.

BCVA = CVA - DVA so it was 40K to me

For some reason I cannot seem to recall this question. Any additional details?

PD = 1 - exp(-s/(1-RR))

Yes, precisely this.