Task 3

[John Bashunov]

[ancl...@gmail.com]

Are you only calculating the impact on the performance and risk metric or are you also determining a new price for each sensitivity test?

On Thursday, July 25, 2013 9:28:05 PM UTC-5, John Bashunov wrote:

[John Bashunov]

I'm just calculating new prices (running a bunch of goal-seeks) under the condition that CTE 95 = 0 still holds

[John Bashunov]

Big book of Risks

[ancl...@gmail.com]

I decided to first determine impact on performance and risk metric and only recalculate a new price for "important" assumptions...

I think both approaches will be fine :) 

[John Bashunov]

Want to share some of your findings?

- Did you do common shocks to "per mile price" and "equip/maintain"?

- Did you talk about interdependence among inflation and the 3 variables it was linked to?

- I recommend looking at the formulas in AT:BC in order to gain an understanding of the potential effects to be testings, then comment on immateriality, etc.

- Did you use any examples of scenarios (such as effects of a strike, etc.). I think this was the approach taken by the 3 worlds scenarios in one of the EOM exercises.

[John Bashunov]

Sweeting, P., 2011. Financial Enterprise Risk Management. Cambridge, UK: Cambridge University Press.

On Thursday, July 25, 2013 7:28:05 PM UTC-7, John Bashunov wrote:

[ancl...@gmail.com]

I am following the other groups advice and am not shocking "per mile price" and "equip/maintain". Do you not agree with the other groups' dicsussion?

I talk about some interdependencies but not all. My sens. testing is done on stand-alone assumptions.

I did not use any elaborate shock (like strikes). I did 1 basic shock (in each direction) for each assumption. Next I did some more shocks on the two most sensitive assumptions. I think you cannot perform a FULL sens testing here because you would need 10 pages to discuss ;)

[daki...@gmail.com]

I agree that you can go on and on with sensitivity testing as there are so many combinations you could test.  At this point I have just performed +-10% shocks to the assumptions one at a time.  The assumptions that I have done the shocks on are: miles purchased, baggage and misc revenue, annual program expenses, labor, discount rate, inflation assumption redeemed rate and forfeited rate (assumed either +-10% for all years with the redemption and forfeiture rates).  The assumptions that appear to be most critical are purchased miles, baggage and redemption rate.  With that said I am not sure if we test any of the three assumptions related to jet fuel (price, price per mile, and equip/maint per mile) as the jet fuel price is stochastically modeled and we know for sure the current price.  Also, I think we need to discuss interdependence as has been mentioned in earlier posts.  For example, I believe that the purchased miles, redemption rate, baggage/maint, labor and equip/maint are all interdependent because they all are impacted by miles that are redeemed. Am I thinking to simply here with just doing the sensitivity testing one variable at a time?   Look forward to your thoughts. 

[Christie Ritten]

I am going with the assumption that we don't have to test fuel price per gallon or per mile. I did test equip/maint per mile, though. Remember that it serves as a floor for each year of the model.

I tested each assumption one at a time.

You mentioned baggage as one of the most critical, but I would point out that labor is just as sensitive to absolute changes (like +$0.001) because they are both per mile and grow with inflation.

I see what you're saying that redemption/forfeiture rates, baggage, labor, and equip/maint all depend on miles purchased. But I don't think that makes them interdependent assumptions, exactly. I wouldn't think that increasing miles purchased by 10%, for example, would change our baggage input. It will sure change our baggage income stream, but not the per-mile input, in my view.

[John Bashunov]

I approached this problem with some more thought after reading through feedback. Here's what I did:

- 3 Shock levels ( +/- 1%, 10%, 50%) for incremental, moderate, and catastrophic changes

- All of the variables except for current fuel price (since isn't even used in the model, except to give you a % change multiplier for the current fuel price per mile

- Made a prices table

- Made a percentage table

- Redemption Value, Current Fuel Price per Mile, and Equipment and Maintenance per Mile were most sensitive

- Inflation, labor, and forfeiture were least

- Ended with some comments about known unknowns and % of CC holders choosing miles; taxes/fees; and effects of ticket prices

[Christie Ritten]

I don't think you want to test current fuel price per mile, since current fuel price per gallon is known/fixed. The two are related by fuel efficiency (miles per gallon). If efficiency doesn't change, there's no reason price per mile should. However, I did comment that improvements in fuel efficiency could change fuel costs over the projection.

[John Bashunov]

I agree that the current fuel price is fixed, but I don't necessarily agree that the current fuel price per mile is necessarily fixed as this is based on the assumption of fuel efficiency. Say for instance, they started using more fuel efficient planes, or the government mandated it. Then this estimate would change.

It all depends on how you argue it ;)

[John Bashunov]

Also, if you argued about fuel efficiency changing, why didn't you just sensitivity test the fuel per mile part and then mention that in there?

[Christie Ritten]

Okay, we are basically on the same page. I guess I was thinking that we know current efficiency, so why sensitivity-test it? It might change over the course of the 10 years, but we wouldn't want to change it in like year 1. So I commented in my memo that we might enhance the model to handle a yearly improvement factor or something like that. But I see your point that we would get similar information by changing the efficiency level that will apply to all years, even if it's not strictly realistic. I'm going to consider doing that sensitivity test. Thanks!

[John Bashunov]

Just trying to get you to cover all of your bases, so they can't mark you down on anything.

I'm going to work on Question 5 a bit...let me know if you have any thoughts on that :)

[Allan Costa]

What are people's thoughts on sensitivity testing the miles purchased by CDL?  I was under the impression that amount was set in the contract and couldn't change, especially since it is a one time, immediate occurrence,  Thoughts?

[Salman Shah]

Just started working on Task-3.  My opinion on this:

 

1. In sensitivity Testing, we are not supposed to do shock testing.  Just +-10% would be sufficient.   So I don't think, we need to do +-25%.

2. All the assumptions are to be changed by +-10% so that we can check the impact on all.

 

3. Do we need to combine interdependent assumptions?

 

I couldn't make it in Can-Do due to sensitivity testing. So really need your feedback on this.

[Allan Costa]

I did three interdependent sensitivity tests.  I chose them by looking at all the assumptions that are related.  For example, Inflation and Discount Rate are both economic, if one moves, the other is likely to move.  However, Discount Rate and the amount of miles that CDL purchases are not related, so sensitivity testing them together doesn't make much sense.

As for the single sensitivity tests, I tested each one, except fuel price, and number of miles purchased by CDL (Which I am still questioning), and I just shocked each one up and each one down once.

[daki...@gmail.com]

The directions at the end of task 3 (FYI also at the end of tasks 4-5) state the following:  "Include tables and/or graphs in your analysis."  I am thinking all we really can do with this exercise is a table showing the sensitivity tests.  Any ideas on what types of graphs they may be looking for?  The directions do not state that you must do both but if we do not do either I think it would mean there is a good chance you would fail the question and potentially the entire assignment. Also, for task 4 I think all we could do is a RCD tool table which should satisfy the requirement for that question.  I have not looked at task 5 closely enough to know if we can do both tables and graphs. Look forward to your thoughts. 

[Allan Costa]

I think it is more important to include what is relevant to the question and your response than to make sure you have both.  I don't have any graphs for 1-4, only tables. 

[Salman Shah]

During sensitivity testing price is considered to be fixed as determined in task2.  We have to see the impact on Profitability and Risk Metrics. Is it right?

[daki...@gmail.com]

I kept the price I recommended in task 2 fixed throughout the sensitivity testing process. 

[Salman Shah]

& you analysed the impact on Mean? Im doing it like that.  Just Checking the impact on Mean.

[John Bashunov]

I'm pretty sure it says to keep your mothodology from Task 2, but to allow price to vary.

For instance, how much does price change is interest rate is increased 10% under VaR90 = 10% 

I made a macro that would run a goal seek to save time on this one. 

[Salman Shah]

Metrics is supposed to be same but we need to investigate the impact on the matrics.  Reading the question again, I think its the price that will remain fix with the same matrics and value of the change is to seen on the matics.

[Gmail]

I agree. I do not think SOA is expecting us to make macros in a 96 hour time frame. I plan on discussing the change to mean with each sensitivity test. I also think I will include the values of the selected risk metrics. Will be putting this info into a table. 

Daniel Akier

(203)214-5105

Sent from my iPhone

On Jul 28, 2013, at 11:38 AM, Salman Shah <salman....@gmail.com> wrote:

Metrics is supposed to be same but we need to investigate the impact on the matrics.  Reading the question again, I think its the price that will remain fix with the same matrics and value of the change is to seen on the matics.

--
You received this message because you are subscribed to the Google Groups "Final Assessment July 25 - July 29" group.
To unsubscribe from this group and stop receiving emails from it, send an email to final-assessment-july-2...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Allan Costa]

I believe the price should remain fixed.  What does varying the price tell us?  Once we agree to a price with CDL Credit, we cannot change it, until the end of the contract, so if the future differs from our assumptions, our metrics will change, not our price.

[Christie Ritten]

As far as showing impact on price vs. impact on the profit metrics, I think it's just two different ways of looking at it. The first says, What price would prepare us if this unexpected assumption were to happen? The second says, Given that we're going with the price from Task 2 (which may or may not be a foregone conclusion), what would happen to profitability? For what it's worth, I take the second perspective.

Sounds like I did the same as Daniel: fix price, and show table of scenarios with the impact on profitability %. My table shows mean (just for informational purposes, it wasn't a part of my risk metric), and then my risk metrics (which happen to be VaR(90) and CTE(95)). I have no graphs for this task.

I am also confused about whether CDL is locked into purchasing 100M or not. To be safe, I'm doing sensitivity testing. Profitability is very sensitive to this assumption. So this gives us more to talk about in Task 4, risk mitigation strategies (like someone else mentioned, lock CDL into some minimum number of miles).

[Salman Shah]

I too have kept prices constant.  But even we can check the impact on price.  And check sensitivity on it.  Wont make much difference.

 

I have check the impact on the following assumptions:

1. Miles purchased

2. Disc-inflation Combined

3. Annual Prog Exp

4. Baggage & Misc Revenue

5. Labour

6. Redemption

7. Forfeiture

All increased & decreased by 10%

 

[ancl...@gmail.com]

I just added a sensitivity test on the "fuel efficiency". The model assumes that you can fly 100 miles with 1 gallon. When I perform a sensitivity test on this "assumption", it ends up being my most sensitive assumption!
Any thoughts?

[Christie Ritten]

I agree, very sensitive. The point I made in my memo was that we know current fuel efficiency, but that it is likely to increase in the future. So ideally we would add a fuel-efficiency improvement factor to the model.