# Errata for 2018 Exam 7 Products

## Errata for Exam 7 Cookbook

**Brehm - Efficient Frontier (4/21/18)**: In the discussion, the risk and return should be switched to match the solution. The risk is the Loss at the 0.5% level. The return is the probability of making plan.

**Verrall - Fully Stochastic BF Model (2/27/18)**: On page 116, the value in sell AY 2, Period 4 should match the value at the end of the last step, 42.4. The inconsistency was due to rounding.

**Brosius - Least Squares (2/20/18): ** The triangle of losses should be labeled "Cumulative Paid Losses ($000)" instead of reported losses to be consistent with the solution which solves for unpaid losses.

**Verrall - Bayesian Model for the BF Method (2/11/18)**: After the first diagonal of incremental losses, the chain ladder method portion should be calculated using the prior incremental loss from the credibility-weighted method, *not* the prior chain ladder incremental.

I fixed the recipe as well as RF Verrall - 2 to correct this. Make sure to download the fixed recipe and problem (links below).

Verrall - Bayesian Model for the BF Method (Fixed)

## Errata for Practice Problems

**RF Shapland - 6 (4/27/18)**: In the solution for part b, the incremental loss should labeled *q*(4,2) *not* *q*(4,3).

**RF Siewert - 8 (4/21/18)**: The problem should include an assumption that the claim count is assumed to have no development after 24 months.

**RF Verrall - 3 (4/20/18)**: This problem assumes "there is no further development after 36 months." It should be **48 months**.

**RF Meyers - 1 (4/3/18): **The graphs in the solution didn't make it through to the PDF when I combined all the underlying files and exported it as a PDF. I fixed the graphs in the solution and you can download the fixed problem below.

**RF Brosius - 4 (4/12/18)**: The prompts in parts a and b ask about the unpaid claims estimate. Since reported loss information is given, the question should be asking about IBNR instead.

**RF Goldfarb - 11 (3/25/18)**: The assumption in part a should say: "Assuming a constant dividend payout rate and constant growth rate in perpetuity." This is the assumption on pg. 34 of Goldfarb.

**RF Shapland - 6 (3/21/18)**: The solution for part b about stratified sampling should say: "For each simulated incremental loss, only sample residuals from the same * group*."

**RF Verrall - 2 (2/11/18)**: This problem is fixed to correct the chain ladder calculation in the Bayesian Model for the BF method.

**RF Clark - 3**: The solution in Part b should state "... overestimates losses for CY 2014 and underestimates losses for CY **2015**..."

**RF Teng - 3 (1/27/18)**: The wording in the discussion should state: "The reason the enhanced method produces a lower premium asset is that the ultimate reported loss for 2015 and 2016 ($12,640) is * higher* than the

*a priori*expected ultimate loss ..."

## Errata for High Level Summaries

**Patrik summary**: The subtitle of the Patrik summary should state "Reinsurance Loss Reserving." Also, the referenced recipes are incorrect and should reference the "Stanard-Bühlmann (Cape Cod) Method" and "IBNR Monitoring."