risk adjusted ILF

[ed999]

what did everyone get for this? i remember getting something like 4.91??

anyone? lambda in the problem was somehting like 191? it was 5% of the Exp value @ BL

[Ullico]

i struggled to interpret whate they wanted here. they gave risk load as 5% of pure premium, but i think they still wanted it to be variance based. i calculated the basic limit variance to get an equivalent risk load to the 5% of pure premium and then applied that risk load to the variance on the higher layer. don't remember the answer though .. sorry.

[Jack Divine]

I believe the risk load should still be variance-based. They gave the risk load as a percent of pure premium for the lower layer to calibrate the lambda parameter.

[ed999]

yes i knwo it's still variance based... i jsut calc the given risk load first and then back out a lambda based on variance.

[Avi]

Rsik load was aid to be equivalent to 5% of PP at 100K limit which was about 913.65 or so. Using the given E(g(x);l)², this gave rise to a λ of about 7.57*10^-7.

Then the ILF at 500K was this λ times the expected second moment (5*10^10 or something?) added to the PP divided by the original 100K PP plus the 913.

Or at least, that's what I did.

[ed999]

yes. me too.

[berad]

I knew that that's what they wanted but I read the question too literaly and applied the 5% of pure premium. My bad for not assuming what they wanted. :swear:

[islas_del_maiz]

oops, i used std. on the other hand they didn't say anything about what to use i think.

[HankMorgan]

Rsik load was aid to be equivalent to 5% of PP at 100K limit which was about 913.65 or so. Using the given E(g(x);l)², this gave rise to a λ of about 7.57*10^-7.

Then the ILF at 500K was this λ times the expected second moment (5*10^10 or something?) added to the PP divided by the original 100K PP plus the 913.

Or at least, that's what I did.

I took the same approach, but got λ = 7.57 * 10^-8 ....rounding error?:wink:

[carenthir_dm]

FWIW, mine was 7.57*10^-7 as well.

[KindGrind]

I think I got 7.57 x 10^-8, too

[A-Head]

I think I got 7.57 x 10^-8, too

The claim frequency was .10 - are you including this in your calculation?

[by234]

the frequency would be cancelled to derive Lamda, I think.

[KindGrind]

oooh I think I got it

it is ^-7 if you used E(g(x;k)) and ^-8 if you used PP instead

I think bith would yield the same answer but the lambda might be erroneous at ^-8 since it's lambda*0.1

I prolly used 0.1E(g(x;k)) + 7.57 x 10^-8 * E(g(x;k)²) / ...

which would too yield the good answer (I think? :P )

[ed999]

i just remember the answer being 4.193?

[tommie frazier]

I remember gettin 4+ on that. I'll know when I get the exam back, as I wrote down answers and approaches in the booklet.

[Avi]

I also remember the first digit being 4.

[ramanujan]

I did the calculations just now and I got 4.17 which sounds familiar.

[ramanujan]

What was the answer to part b? I think I wrote something like ruin theory, reinsurance method, marginal surplus and something I don't remeber.

[KindGrind]

i wrote like

Ruin theory
CAPM (in bault's discussion there was something about some sort of CAPM)
Marginal surplus
Marginal variance

quite random answers :p

[gambate]

oooh I think I got it

it is ^-7 if you used E(g(x;k)) and ^-8 if you used PP instead

I think bith would yield the same answer but the lambda might be erroneous at ^-8 since it's lambda*0.1

I prolly used 0.1E(g(x;k)) + 7.57 x 10^-8 * E(g(x;k)²) / ...

which would too yield the good answer (I think? :P )

I really hate myself so much right now...during the exam, i solved the problem using 7.57*10^-8 ....then when I looked through it the second time, decided to include E(n)...which then I use 7.57*10^-9....

[asilem]

i wrote like

Ruin theory
CAPM (in bault's discussion there was something about some sort of CAPM)
Marginal surplus
Marginal variance

quite random answers :p

They asked for 4 other than the MS and MV methods. I think I ended up writing your first two, reinsurance, and utility theory. Am I totally making up that last one?

[Bobby Savoy]

Looking at the chicken-scratch on my booklet, I got 4.174 for part a. I had a lambda of 7.57x10^-7. I had a brain freeze on part b. Oh well..that seems to be the case for me...get the calculation right (or mostly right) and then miss the one-point list.

[ed999]

come to think of it, for part b, wouldn't the shapely value and covariance share be legitimate methods?