Task 2

[John Bashunov]

[John Bashunov]

You can approach this one in many number of ways. Most people are saying to use CTE, but the percentile value will vary depending on your risk tolerance. 

Regardless of what your approach is, though, this article is not to be overlooked. 

[Christie Ritten]

Was anyone else confused about the wording of the first bullet point? At first I thought that it meant 100,000,000 was the maximum, the amount of air miles earned if everyone chose to earn air miles. In other words, the spending of all CDL customers directly translated in miles. But looking at the spreadsheet (100,000,000 entered into the "Assumptions" tab), I think this number already incorporates the fact that some people will choose cash back instead. I will be going with this interpretation unless anyone disagrees.

On Thursday, July 25, 2013 9:27:56 PM UTC-5, John Bashunov wrote:

[John Bashunov]

Christine,

I believe that the 100,000,000 is bound to change, based on the unknown distribution of people who will take up the cash-back program. Nevertheless, for this task, we are assuming that all of the assumptions have already been agreed upon and we are only recommending at which the use of CTE over VaR, the percentile, and the value to which we set CTE(x) or VaR(x) to. 

Task 3 will ask you to do sensitivity tests, and Task 4 will ask for variables not incorporated into the model. For both, you can talk about the percentage of people that choose cash over miles.

Let me know if you agree.

[Christie Ritten]

Yes, I agree. Thanks for your thoughts!

[John Bashunov]

ngram, David, “Getting to Know CTE,” Risk Management Newsletter, Jul 01, 2004.

[daki...@gmail.com]

I plan on using my performance metric as the profit being greater than 10% that is mentioned in scenario background.  For a risk metric I will select either CTE or VaR measure that I find resonable. Still need to think that through.  Then using my performance and risk metric I would derive a price that BBA should sell an Air Mile.  Are you all thinking along the same general line? 

[daki...@gmail.com]

I am planning a similar process that was mentioned by a prior group (https://groups.google.com/forum/#!topic/final-assessment-jan31-feb4/3pEvDG3fVZc)

The approach makes sense. 

[John Bashunov]

Make sure you explain the differences between the 2 metrics and keep in mind the relationship between CTE and VaR. Then, based on your/management's risk tolerance argue why 75, 90, or 95 is superior. 

[AInul Shafiqah Shafie]

Is it worth looking at the distribution of the profits and losses? i created a quick chart of my data and it showed that the profits just dropped off somewhere in the 98th percentile. I'm thinking this makes a stronger case of using the CTE as the main risk metric to price the miles. 

[Christie Ritten]

Seems like we have no choice but to make PVP/Revenue > 10% our performance metric. What do we really need to say to justify it? So far I just have that it's what management uses, so it's good for comparisons and decision-making. (Especially because we are using the same discount rate.) Any other thoughts?

[John Bashunov]

This was my approach for this problem:

1. Explain Risk Metrics 

2. Based on the paper by David Ingram, there is a direct correlation between VaR and CTE for normally distributed gains/losses

3. Prove that VaR and CTE is correlated by plotting PVP/P at various price points...hence Normally Distributed  

4. Therefore, you can use CTE and VaR interchangeably

5. Another way of looking at CTE is by dividing simulation into 2 groups: Group A and Group B (Group A is the worst case scenarios, the outlier group where you lose $, and Group B is where you make $)

6. Therefore you want expected value of Group A to equal 0 (CTE (X) = 0)

7. But we don't just want to make money, we want to ensure that our best case scenarios make us at least 10% in profit (PVP/R)

8. Another way of looking at this is my asking, at which point will every point in our best case scenario group make 10% or more? At VaR(X) = 0.1 obviously!

9. And since CTE and VaR are related, you know you'll be killing 2 birds with 1 stone (having sure that CTE(x) >0 and VaR(x) =0.1), so you'll always be making your goal and making a profit, if you land in Group B

10. Now which percentile do you pick? Well if you want to be super conservative, you'd go with the 95th percentile. You're 19 times more likely to be in Group B than Group A...that's a lot of padding room that will cost you in price and competitiveness. 

11. How about picking the 90th percentile where you're winning 9/10 but you'll have a strong case at staying competitive and have management happy?

12. I think we've got a winner

[Christie Ritten]

Thanks, John. Great point about competitiveness -- BBA isn't the only airline out there.

Something to consider is that the percentile level doesn't have to be the same for the 10% criterion as for the 0% (or $0) criterion. I think I'm going with VaR(90) > 10% and CTE(95) > 0%. The 90th percentile isn't overly conservative, but we've got a strong guard against negative profitability.

[daki...@gmail.com]

Are we sure that these are normally distributed gains/losses?  I do not think that we can cleary assume that. 

[John Bashunov]

You don't need to establish that they are normally distributed, all I'm getting at here is that VaR and CTE are positively correlated (which emplies they are normal). This is to drive my point home of killing 2 birds with one stone:

VaR (90) = 0.1 and CTE90 = 0

[John Bashunov]

CTE90 >0 sorry

[Christie Ritten]

VaR (90) = 0.1 and CTE90 > 0

John, this seems like two stones to me... I don't understand what you gain by arguing that VaR and CTE are positively correlated.

[Salman Shah]

BBA Management is concerned over the volatility in their base operations and aims for capital preservation.  So do we need to be over cautious with this deal? Further, the profit fro BBA is also undesirable from current operation so this deal might help them to revive their business/profitability.

 

The profitability chosen here is as required by the Management "present value of net profits as a percentage of initial revenues to be greater than 10%."  We do not have any other option.  This will be our performance metrics.

 

For Risk Metrics, we need to have Both VaR (90)>0 & CTE (90)>0.  This is conservative & competitive.

 

I want to have your opinion over the other options for risk metrics.  Should we consider Mean as performance Metrics? Should we take into account St. Dev as Mgt is concerned about the volatility of profits.

 

[Salman Shah]

But the problem is that Mean was required to be at least 10% but it coming out to be 23% with risk metrics (VaR(90)=8% and CTE(90)=0% ).

 

I feel I'm charging more price for the Air Miles as profitability is too high.  CDL Credit might walk away at this price.

 

Your opinion?

[Allan Costa]

I had the same thought so I lowered price a lot.  To even get to a mean of 20%, your CTE values start to go negative.  There is a lot of risk to take on, but high potential reward.  You need to decide how much risk you are really willing to tolerate.  But chances are that mean is going to remain high.

[Salman Shah]

Just worked on it a bit more.

 

Mean will remain high. Agreed.

Performance Metrics: I will show Mean and %>target as PM. %>target should be < 10%.

Risk Metics: CTE(90) and Var(90) both > 0.

 

[Christie Ritten]

Keep in mind that "%<target should be < 10%" is the same thing as "VaR(90)>10%." You probably don't even want to mention "VaR(90) > 0." (It's automatically satisfied whenever CTE(90) > 0, anyway.)

[Salman Shah]

I'm using Mean (PVP/Revenue) > 10% as performance metrics which is the target of BBA Management. I will justify it like BBA will be making profit on average.

For Risk Metrics, CTE (95).  CTE is better than VaR in long term contracts.  This is a 10 years deal so choosing CTE is better than VaR.  Percentile 95 is picked as BBA is overly conservative on this deal.  This is also first time BBA is planning this venture with CDL so conservatism is the strategy.  CTE(95) = 0% shows BBA will be in no profit and no loss in 5% worst case scenarios.

 

 

Whats your opinion over this?  Will it okay to just choose 2 parameters to make up decision and ignore VaR. One thing more.  If we keep CTE(95)=0 then that's will result in Mean ((PVP/Revenue) > 10%) substantially higher than 10% which is desired by the Management.  i.e BBA will be making more money then desired.  And that means the price is high for Air Miles. 

 

Any thought on it?

[Christie Ritten]

I think you are the right track, Salman. I would say no need to use VaR if you don't want to and can justify something else. Although, one thing...

"CTE(95) = 0% shows BBA will be in no profit and no loss in 5% worst case scenarios."

I think you just misspoke here. Better to say, "CTE(95) = 0% shows BBA will be in no profit and no loss in the average of the 5% worst case scenarios."

[Salman Shah]

I missed it out.  Have you submitted yours?  I will submit it tomorrow night.  So will be grateful if you can express your opinions.

[Salman Shah]

I missed to mention "average" in my post.   Good Luck!