Is there anything in Brehm 1 on dependencies/correlations that won’t be covered in future chapters? Struggling to identify any useful information here for flash cards and notes.

I’ve just been screenshotting the problems from the PDF and pasting those screenshots in Excel. It’s not ideal since you can’t do cell references to the specific numbers, but it’s close enough for me.

Also, I don’t think you need to worry about the correlation stuff in ERA Ch1. Correlations are covered in a lot of detail in chapter 3

Yeah I’ve also been screenshotting the questions. I also copy and paste the tables into excel using Acrobate Pro.

30 days out for me now. I’m going to work through the 2014 exam tomorrow. So far I’ve done 2011-2013. There’s a few small things I’m still a bit fuzzy on but I’m pretty comfortable with the bulk of the material. I’ve spent some extra time with Taylor since it’s such a new paper and there haven’t been almost any past questions on it, but other than that I’m still mainly just focusing on past questions.

How’s 9 going for you?

So much of Brehm seems like fluff material that can’t really be tested. Maybe that’s just because I took 9 and do a little capital modeling at my job but idk what I should even be making notes on for the non-quantitative stuff so far.

Pretty well actually. The last week or two I’ve shifted from being really anxious about doubling up to getting more and more confident that I’ll actually pass both. I’m a bit behind on 9 compared to 7, but I’m not taking 9 until the end of the testing window so I’ll have plenty of time.

I’ve made flashcards on Brehm based on the questions from past exams. There’s quite a few things they’ve asked about that would be good to memorize: 5 methods of traditional risk management, evolution of corporate decision making, 3 ways reinsurance adds value, risk types for modeling (projection, estimation, model), different copulas, definition of operational vs strategic risk, agency theory, bridging model, aspects of effective cycle management, 5 steps for managing operational risk, and soft vs technical vs behavioral modeling.

I’ve now worked the 2011-2015 past exams under exam conditions and been well above the pass mark for all of them. I’m going to spend later this week reviewing all the problems I got wrong on those sittings, then do 2016/2017 next week and 2018/2019 the week after. Feeling really good right now.

Do people plan on digging into the theory for Brosius’ linear approximation/Bayesian estimates? I have a ton of flash cards from my first pass on this but don’t really understand anything about this stuff. May need to just dig into the paper.

hmm…I don’t really have any flashcards except for the formulas. I have some fundamental understanding of it and don’t plan to dig into its theory.

The only things I’m memorizing from that part of Brosius are Z = VHM/(VHM + EPV), VHM = d^2(sigma_y^2), EPV = sigma_d^2 (sigma_y^2+E[y]^2), and L(x) = Z(x/d) + (1-Z)E[y]

I don’t understand the part in Brosius (in the Linear Approximation section) where if x is greater than expected, having Cov(X,Y) < Var(X) would cause a reserve decrease. To me, the formula will still show L(x) > E(Y), which is increasing ultimate losses from the initial estimate, if x > E(X). Intuitively the statement makes sense, but not based on the formula.

The entire formula is L(x) = [x - E(X)] * [Cov(X,Y)/Var(X)] + E(Y)

If x>E[X] & Cov(X,Y)<Var(X), then your revised reserve L(x)-x will be less than the expected reserve E[Y]-E[X]. See the screenshot for clarification. The Difference will be < 0 in this case.

Got it, thanks very much. I think my sticking point was not comparing to the initial, expected reserve. I was just thinking about how the current reserve changes, as x changes.

So saying it in words for the Cov(X,Y) < Var(X), if x is greater than E(X), L(X) will also increase from E(Y) but not as much as the increase in x from E(X). Hence the reserve goes down from what we expected it to be, at the outset.

Actually, not sure how I was missing it before, but realized that if you take L(x) - x, which is the reserve definition, you get a function where the coefficient for x is Cov(X,Y)/Var(X) - 1, which can be both positive and negative.

I think I was just not looking at the reserve function before, and instead just ultimate losses.

Are you guys bothering to study the marginal allocation formulas for each risk measure in Brehm 2.2?

Nope, the only marginal allocation I know is the Myers-Read one and that’s just because I’m taking 9 too. Brehm chapter 2 is definitely my weakest paper on 7 though.

I think the section where Mango is showing how each method would compute a marginal approach is not very testable. After that it seems to have a lot of crossover with other exams so far.

Just finished 2016. It seemed a bit harder than I expected, definitely harder than 2011-2015.