PDA

View Full Version : Part 5: Krakowski


Avi
12-29-2003, 11:04 AM
Just have to vent a bit here about the exhibits. Somehow I can't follow them. E.g. Exhibit 2 purports to be a linear weighting. How? It's not a linear Least squares regression, and it's not a straight line.

Exhibit 3 seems a bit circular. How can the Adjusted Damage ratio be both the result and the precursor of the damage adjustment?

:shake:

I can't wait to finish this paper.

Edit: Nevermind. I got exhibits 2 & 3. It's those darned Balance adjustment exhibits where the chosen process variance's derivation is conveinently left out etc. that have me romoving whatever hair I have left.

If this is 5, what is 7 like - nevermind, don't answer the question :wall:

carenthir_dm
01-15-2004, 01:53 PM
I work with Mr. Krakowski, and will be happy to get answers to any further questions you have on the paper.

Anita
01-28-2004, 08:45 PM
I think I'm going to need a lot of help with these exhibits. Would someone please explain how to get Column 5 (Fitted Catastrophe Ratio) on Exhibit 2?

I assume the projected catastrophe ratio of 0.585 is calculated from fitting a line to Column 4 and using an average loss date of 2001.5.

I appreciate any help you can provide.

J.T.
01-29-2004, 08:48 AM
What do you guys think is going to be tested about this paper? I had a hard time getting through it, and so far it is the most confusing paper I have come across. I really feel like I have no idea what they are going to ask about it (and I know they are)!

statzman
01-30-2004, 01:31 AM
Carenthir_dm,

Ask Mr. Krakowski if he would be kind enough to provide some new exhibits that are a little easier to read. Or maybe, a little more explanation on the exhibits in the paper. This of course, won't be part of the syllabus, but it would certainly be helpful in understanding the exhibits that are on the paper - they can be somewhat confusing.

If he would do that, it'd be a tremendous help in studying for the exam.

Thanks.

GA Peach
02-02-2004, 12:40 PM
Anita,

If you look at page 291 in the article, he states:
...calculate a countrywide (linear) trend in damage ratios, weighted by amount of insurance." Continue reading and it should be somewhat helpful.

I'd have to say, CAS isn't going to ask us to trend on the exam. Since this is a new article, just get the basics and don't worry about the details - there are just too many questions they can ask without nit-picking - IMO!

Wigmeister General
02-02-2004, 12:44 PM
Just have to vent a bit here about the exhibits. Somehow I can't follow them. E.g. Exhibit 2 purports to be a linear weighting. How? It's not a linear Least squares regression, and it's not a straight line.

Exhibit 3 seems a bit circular. How can the Adjusted Damage ratio be both the result and the precursor of the damage adjustment?

:shake:

I can't wait to finish this paper.

Edit: Nevermind. I got exhibits 2 & 3. It's those darned Balance adjustment exhibits where the chosen process variance's derivation is conveinently left out etc. that have me romoving whatever hair I have left.

If this is 5, what is 7 like - nevermind, don't answer the question :wall:

Just focus on 5. Let 7 be the problem it will be when it's time to sit for it. (But know it gets worse. Really, it does.)

Avi
02-02-2004, 02:03 PM
I'm a little over halfway through the syllabus, and I feel like throwing myself off of a precipice at times.

Yes, I set an extremely aggresive schedule, and I know I'm hard on myself in general, but my boss put it very pithily. 1-4 is Undergrad, 5-9 is Grad school (he's got a PhD in Statistical Mechanics, so I believe he knows whereof he speaks).

:wall:

I know I'm pessimistic - but now I'm downright depressed :yikes:

Ah well, back to begin Feldblum's masterpiece on Pers Auto asset share pricing. :wall:

ed999
02-02-2004, 06:23 PM
Avi,

I know you might be mad at me (afterall, I was one of those who was unable to provide any help in your Yahoo study group), but I just want to tell you not to give up. I personally think that of all the people here taking 5, you are one of the best prepared, devoted candidate with the maximum likelihood(no pun intended) chance of passing. And you really deserve to pass. So please don't be so hard on yourself.

I am also extremely frustrated, and I find that my frustration has been cyclical because there are times when I finally (begin) to understand a topic, and it gives m e muc hhope that I can pass this sucker in the end!

Anyway, I dont know where this will lead me to so let me stop here. But best of luck to both of us! I have been doing speed reading and am almost midway thru feldblum's WC ratemaking page. I have been really busy at work and busy with "going out" on the weekends that I dont' have ANYtime to do problems, nor write them. I hope you will understand.

Best,

Anita
02-04-2004, 04:47 PM
I know we don't need it for the exam, but I'm having trouble moving on from my question above.

I run a linear regression in Excel with Calendar Year the independent variable (x) and Col 4 Cat Ratio as the dependent variable (y).

The result is y = 0.015x - 29.5

Using 2001.5 as the average accident date in the projections, you get a projected Cat Ratio of 0.585. So, this is easy. Thought I was done. However using the above formula, you do not get column 5 (Fitted Cat Ratio). The formula used to get this column is y = 0.01368x - 26.8. Where does this come from?

I know we don't need this for the exam, but will someone please tell me how to get column 5? It is driving me crazy!!!!

statzman
02-20-2004, 12:55 AM
CAS has posted a link to new exhibits from Krakowski.(pdf) (http://casact.org/pubs/forum/03wforum/errata.pdf) I haven't looked at the exhibits much, but it should be an improvement over the first set.

Other errata updates (http://casact.org/admissions/syllabus/2004/update.htm) are here.

Anita
02-25-2004, 01:05 PM
On page 296 he states, "Further, it is clear (see below) that the standard Buhlman-Straub formulation is not appropriate in the present context."

Two questions: why is it clear? and where below (page and paragraph)?

Thank you for any help. It has been a while since I took 4B.

exa5
03-05-2004, 12:14 AM
I also run a linear regression in Exel, with all the same to you(Avi),except that the independent variable is not Col 1, but numbers from1 to 30. That is where you make a mistake ,I think. the result is y = 0.015x + 0.1113. and at June, 2001 projectiong is 0.584. and the Col5 is y = 0.0137x + 0.1409.And at June,2001 projection is 0.572
:P