CAS Exam Errata

Study Guide & Review Videos

Werner/Modlin Ch. 6 (9/29/25) – On page 90, the LOGEST function should say “INDEX(LOGEST(Known Y Values, Known X Values),1)-1.”

Friedland Ch. 10 (9/29/25) – On page 354, under Step 1, it should say “Once we have the Used-Up Premium, we divided the total reported claims at the latest valuation by the used-up premium to arrive at the overall expected claim ratio.”

Werner/Modlin Ch. 15 (9/22/25) – For clarity, on page 250, the “Exp. Through LCF” bullet should say “Expected Loss Ratio X (LCF 1.0).”

Werner/Modlin Ch. 16 (9/22/25) – At the top of page 257, it should say “Insurers use step factors to reduce the premium for an immature claims-made policy…”

Werner/Modlin Ch. 14 (9/17/25) – At the bottom of page 222, the “product” operator in front of the hat(D_j) should be a “summation.”

Friedland Ch. 14 (9/5/25) – Under “Excess of Loss Arrangement,” the first sentence should say “Under an excess of loss arrangement, the insurer cedes…”

Werner/Modlin Ch. 12 (7/30/25) – At the top of page 189, the first line should say “Since the loss costs of a large related group are typically biased…”

Werner/Modlin Ch. 8 (7/29/25) – On page 120, the indicated average rate should be $660.05, leading to an indicate rate change of -11.99%.

Werner/Modlin Ch. 8 (7/29/25) – On page 117, the projected severity trend should be 2.5%, not 3%. This leads to the projected trend of 0.4% shown on page 118.

Werner/Modlin Ch. 7 (7/29/25) – On page 110, the general expense figures should be $5.83, $6.41, and $7.65, for CYs 2015-2017, respectively.

Werner/Modlin Ch. 9 (7/23/25) – On page 136, the variable expense ratio should be 30%.

Cookbook

Risk Classification: Loss Ratio Approach(8/12/25) – [PDF version] The “Cred-Wtd Indicated Relativity” values in the table for step 3 are incorrect. The excel version has the correct numbers. This is fixed on the course.

Practice Problems

Problem Set #8 Q2 (9/26/25) – On Q #2, the trend to date for fixed expenses should be 7/1/24, the average earned date, because the expenses are incurred throughout the policy period.

Problem Set #2 (8/2/25) – On Q #4, the final base rate change should be 0.6%, not -0.6%. On Q #7, the assumption that says “no development after 48 months” should be removed since a tail factor is provided in the problem.

Past CAS Problems

Spring 2018 #8b (10/6/25) – In the full Spring 2018 exam workbook, the premium trend for this problem should be 1/1/XX to 7/1/19.

Spring 2014 #1 (9/16/25) – In the solution to part b., it should say that “25% are in Group 4, written on the first day of 2012.”

Practice Exams

PE 3 Q#5 (9/23/25) – When calculating the ULAE ratios, we should be dividing the CY ULAE by the CY paid losses & ALAE. The solution has been update to subtract the prior diagonal to produce CY figures. The paid ULAE has also been revised to produce a reasonable ULAE factor.

Flashcards

Werner/Modlin Ch. 6 (9/29/25) – On flashcard 17 (page 34 of the standard PDF), the formula should say “INDEX(LOGEST(Known Y Values, Known X Values),1)-1.”

Werner/Modlin Ch. 15 (9/22/25) – On flashcard 15 (page 31 of the standard PDF), the “Exp. Through LCF” bullet should say “Expected Loss Ratio X (LCF 1.0).”

Werner/Modlin Ch. 16 (9/22/25) – On flashcard 5 (page 10 of the standard PDF), it should say “Step factors are used to reduce the premium for an immature claims-made policy…”

Werner/Modlin Ch. 14 (9/17/25) – On flashcard 5 (pages 9-10 of the standard PDF), the “product” operator in front of the hat(D_j) should be a “summation.”

Study Guide & Review Videos

Porter Ch. 4 (9/22/25) – On page 45, the final accreditation step should say “grants accreditation or suggests changes to the DOI…” The video has been revised as well.

CFR Ch. 19 (9/19/25) – On page 337, the locations of the initial and current values for the company development factor calculation are incorrect. The initial values are found along the diagonal of Schedule P, Part 2. The current values are found in column 10 of Schedule P, Part 2.

Various (8/15/25) – Fixed a few immaterial typos as we wrapped up the high-level summaries. Reposted a clean version of the full study guide.

CFR Ch. 19 (7/27/25) – On page 346, the loss-sensitive discount towards the bottom should say “0.30(0.18)(567,428) + 0.15(0)(567,428) = 30,641.”

CFR Ch. 19 (7/24/25) – On page 332, the total “amount held” in the bottom table should be 13,195,000.

SSAP 65 (7/15/25) – The UEPR test example in the guide and video have been slightly reworded to make it clear that we are considering a “representative policy” from a cohort of policies written throughout the year. In addition, the first column of the UEPR table in the example should say “Year” rather than “Policy Year.” Finally, negatives were added to the exponents for the discount factor calculation.

CFR Ch. 9 (7/15/25) – On page 193, the change in surplus notes should be color-coded as a “other surplus change” rather than “additional capital contribution.” In addition, in the calculation, the gross paid-in and contributed surplus should just be the 7,000 from the current year, not 7,000 – 5,000.

CFR Ch. 14 (7/15/25) – On page 259, a note was added below the formula for the stressed net recoverable stating that it has a “floor of zero.” In other words, it can’t be negative. The credit risk video was revised as well.

Freihaut & Vendetti (7/15/25) – On page 624, the final sentence of the first paragraph should say “…risk contracts without risk limited features (such as loss ratio caps).”

Cookbook

Practice Problems

COPLFR (9/1/25) – In Q #1, we need to comment on the reasonableness of the gross reserves as well. In this case, they are reasonable. An updated version was uploaded to the course.

SSAP 5R (8/7/25) – In Essay Problem #5, the heading of the final column in the table should say “Chance that the Outcome Will be Unfavorable.”

CFR Ch. 25 (7/30/25) – The cost of capital (i.e., required rate of return) was revised to 7.5% so that R – i = 6% as required under Solvency II. The problem has been revised and reposted.

Webel (7/24/25) – The header in the solution should say “150M” instead of “150B.” Problem has been reposted.

Cedar (7/18/25) – Q #6 was missing a data column. The problem was revised and reposted.

SSAP 65 (7/15/25) – Q #1 was revised to make it clear that the assumption of uniform writings applies to a cohort of policies.

CFR Ch. 9 (7/15/25) – In Q #1, the change in surplus notes should be coded as an “other surplus change.” In addition, the gross paid-in and contributed surplus should not be computed as a change. Instead, the $6,000 from the current year should be included directly as an “additional capital contribution.” The problem has been revised and reposted.

High-Level Summaries

Porter Ch. 4 (9/22/25) – On page 15, the final accreditation step should say “grants accreditation or suggests changes to the DOI…”

CFR Ch. 19 (9/19/25) – On page 100, the locations of the initial and current values for the company development factor calculation are incorrect. The initial values are found along the diagonal of Schedule P, Part 2. The current values are found in column 10 of Schedule P, Part 2.

CFR Ch. 9 (9/6/25) – The first two rows (unrealized capital gains & unrealized foreign exchange capital gains) of the table on page 1 should say increases surplus.

Flashcards

Porter Ch. 4 (9/22/25) – On flashcard 14 (page 28 of the standard PDF), the final step of the accreditation process should say “grants accreditation or suggests changes to the DOI…”

CFR Ch. 19 (9/19/25) – In flashcards 32 and 33 (pages 63-66 of the standard PDF), the locations of the initial and current values for the company development factor calculation are incorrect. The initial values are found along the diagonal of Schedule P, Part 2. The current values are found in column 10 of Schedule P, Part 2.

Study Guide & Review Videos

Mach Chain-Ladder (10/2/25) – On page 134, in the allocation example, the guide implies that we cannot sum the individual AY CI limits to get the total reserve CI limit due to the correlation between the AYs. But even if the AY reserve estimates were independent, the CI limits cannot just be summed. The key thing provided by independent reserve estimates is that we can sum the squared standard errors of the individual AY reserve estimates to get the total reserve squared standard error (see Problem Bank Q10). The wording has been revised at the bottom of the page.

Verrall (9/15/25) – On page 358, the advantage of the normal approximation could be clearer. It’s better to say “the advantage of this model is that it can handle negative development (i.e., a negative column sum of incremental values) because the variance is no longer tied to lambda_j.

Mack Chain-Ladder (8/30/25) – On page 137, the second bullet should say “third chain-ladder” assumption, not second.

Shapland (8/16/25) – On page 179, F(d) should say “the factor applied to c(w,d) to estimate c(w,d+1). On page 180, the formula for F(d) should be sum(c(w,d+1))/sum(c(w,d)), where the limits of the sum are w = 1 to n-d.

Taylor & McGuire (8/4/25) – On page 339, the formulas for the scaled and unscaled deviances have extra parentheses after the “ln’s”. Those can be removed.

Venter Factors (7/24/25) – On page 154, there is a typo at the top of the page. The second bullet should show “0.073/0.007 = 10.136.”

Clark (7/15/25) – In the Misc. Topics video, there was a typo in the discounted reserves example. The expected losses should have been 3,541. The video has been corrected.

Friedland (7/7/25) – On page 78, there is a typo. The exchange rates on the latest diagonal for AYs 2015-2019 should say 1.12270.

Past CAS Problems

Verrall 2017 #11 (8/19/25) – A few of the alpha/beta symbols didn’t show up for some. We fixed the file to correct this.

Practice Exams

PE #1 Q21 (9/16/25) – Options 4 and 5 of part b. were unclear and appeared to be referring to the same thing. These were revised.

Flashcards

Anki Deck (7/26/25) – The Marshall cards were not displaying correctly in the Anki deck. A new version was posted on the course.

Formula Summary

Brosius – Least Squares Method (9/22/25) – The inputs to the formulas for b and a in excel should be switched to:
b = SLOPE( Ultimate Loss Ratio, Undeveloped Loss Ratio )
a = INTERCEPT( Ultimate Loss Ratio, Undeveloped Loss Ratio )

Study Guide & Review Videos

Bahnemann Ch. 5 (9/27/25) – A few notational issues. On page 365, when referring to X^2_{a,a+l}, it should say X^2_{3k,5k} = X^2_{3k}; 5k. On page 371, it should say X_{5k,4k}. All calculations are correct as shown, just a labeling issue here.

GLM Ch. 2 (9/2/25) – The bullet points around CIs are misleading. For instance, for the second bullet, it should say “If the hypothesized coefficient lies outside of the confidence interval, then we reject the null hypothesis and we conclude that the hypothesized coefficient is not within the reasonable range of estimates for the coefficient.” A similar correction is needed for the first bullet. For simplicity, we just removed the portions about the reasonable range. The key point is “inside CI, fail to reject” and “outside CI, reject.”

NCCI Circular (8/25/25) – Typo on page 349. In step 9, r_G should say 1.74, not 1.72.

Holmes & Casotto App. C (8/15/25) – On page 223, the original and rebased factors are incorrect for Level C (they also incorrect in the textbook). They should be 1.649 and 1.351, respectively.

Holmes & Casotto Ch. 3 (8/12/25) – On page 171, the final line should say “the stepwise approach ensures control variables…”

Bahnemann Ch. 5 (7/24/25) – On page 294, under the first bullet, the denominator in the second to last line should say “1 – 0.8594.”

Practice Problems

ISO Q5b (10/6/25) – The improved training should already be reflected in 1/3 of the experience period (there is a 1 year lag for the experience period).

Fisher Retro Q16 (9/30/25) – The wording around the two policy options was misleading. Assume that both options are large deductible policies. For Option 1, the per-occ. ded. is $100k and the aggregate ded. is $720k. For Option 2, the aggregate ded. is $1.2M.

Fisher Retro Q18 (9/30/25) – For clarity, on part f., the problem prompt should say “…this table produces a higher expected total loss cost.”

NCCI Circular – CAS Fall 2018 #14 – In part a, the formula for expected retro premium should be shown as:
E[R] = (B + c(E – I))*T

The calculation is correct.

GLM 5-7 Q20 (9/11/25) – In part b, the answer should state that Model 2 is selected (not 1), because it is the larger model.

Fisher Retro 2019 #16 (9/7/25) – The % and # of risks above r=3 were corrected.

Bahnemann 26 (9/7/25) – The formulas for excess layer loss and inflation effects have been corrected.

Bahnemann 9c (8/29/25) – In part c, CV should simplify to 1/sqrt(p*E[X]).

GLM 5-7 Q18 (8/25/25) – In the solution to Q18, the fifth bullet point was not copied correctly.

ISO (8/22/25) – In Q5 part d, the sample solution should be proposing a debit, not a credit. This has been fixed.

GLM 5-7 Q19 (8/19/25) – In Q19, the formula for AIC was missing parenthesis and has been fixed. Model 2 is now selected as the better model. In Q20, selections in the legend for Model A and Model C were switched.

Fisher Experience (8/18/25) – In Q9, the statements about primary and excess credibility were switched. This has been fixed.

Fisher Experience (8/16/25) – In Q10, the mod should use losses capped at the MSL. This makes Plan 3 the correct answer.

Holmes & Casotto (8/7/25) – Corrected the Beta subscripts for Q13, 14, 15, and 18.

Holmes & Casotto – Q13 (8/3/25) – In Q13 part e., the linear predictors for Model 2 and Model 3 should include the complement coefficients. The original percent difference calculation was also incorrect.

Cookbook

Fisher – Lee Diagrams and Expected Retro Premium (9/17/25) – In the discussion section of the pdf,the bullet point for the third property should show the savings increasing (unbounded) as r increases.

Holmes & Casotto – Penalized Regression (9/9/25) – In the pdf, part c step 4 should include 0.5 in the formula for parameter set 1 after lambda (3). This is fixed.

GLM – Model Validation: Lorenz Curve (8/25/25) – The first couple numbers in the “Area between the curves” table were incorrect in the PDF. We updated these to match the Excel version.

Penalized Regression (8/15/25) – In step 2: Parameter set 3 only has 3 parameters. Because beta_1 = 0, it isn’t a parameter used in the model. This is fixed.

Lasso Credibility Rating Factors Table (8/14/25) – A few of the boxed formulas weren’t quite right. These are fixed.

High-Level Summaries

Fisher 2/Retro Rating (8/30/25) – Under “Other Loss-Sensitive Rating Plans,” it says “Deductible Plan” by mistake. It should say “Dividend Plan.”

Formula Summary

Lasso Credibility Rating Factors Table (8/14/25) – A few of the boxed formulas weren’t quite right. These are fixed.

Flashcards

GLM Ch. 2 (9/2/25) – On Flashcard 10 (pages 19-20 of the standard PDF), the bullet points around CIs are misleading. For instance, for the second bullet, it should say “If the hypothesized coefficient lies outside of the confidence interval, then we reject the null hypothesis and we conclude that the hypothesized coefficient is not within the reasonable range of estimates for the coefficient.” A similar correction is needed for the first bullet. For simplicity, we just removed the portions about the reasonable range. The key point is “inside CI, fail to reject” and “outside CI, reject.”

Bahnemann Ch. 6 (8/28/25) – Flashcard 18 (pages 35-36 of the standard PDF) should say “Explain why it’s NOT appropriate to use risk-loaded ILDFs to price an excess layer.”

Study Guide & Review Videos

Clark (1/5/25) – On page 20, the third bullet should say “R/IV = 57%” instead of “(R+L)/IV = 57%.”

Mildenhall Ch. 4 (2/4/25) – On page 108, it should say “If p = 1 – 1/n,” rather than “Since F(x) = p = 1 – 1/n.” The “p” here incorporates frequency and severity, not just severity. The video will be updated as well.

Mildenhall Ch. 4 (2/12/25) – In the Ch. 4 video, when the n-year occurrence PML is defined, it still says “at least 1/n.” This will be revised to “at most 1/n” to align with the study guide (this is incorrect in the textbook).

Brehm Ch. 2 (2/26/25) – On page 286, in the second to last bullet, economic margin is incorrectly described as the return above the cost of capital. It should just refer to economic margin as the “profit margin.”

Panning (3/3/25) – On page 263, under the “Advantage,” it should say “seek to reduce the duration of the firm’s total economic value by…”

Mildenhall Ch. 5 (3/17/25) – On page 127, the bid-ask spread should be defined as rho(X) – (–rho(-X)). Then, in the example below the definition, the final spread should be rho(H) – (-rho(-H)) = rho(H) – (-rho(C-1)) = rho(H) + rho(C) – 1 = 0.316 + 0.95 – 1 = 0.266 > 0. The definition of the bid-ask spread will be corrected in the video as well.

Brehm Ch. 2 (3/24/25) – On page 282, the marginal decomposition of the exponential moment has a typo. The “r'(X_j)” should be replaced with E[X_j * e^(cY/E[Y])]. This will be corrected in the video as well.

Cookbook

Mildenhall Pt 2 – Classical PCPs (2/9/25) – The PDF version shows a typo in the semivariance formula. The squared should be outside the parenthesis instead of inside squaring E(X):

E( (X-E(X))^2 where X>= E(X) )

Practice Problems

RF Bernegger – 2  (11/25/24) – The data in the table didn’t exactly sum to a 100% probability. We fixed this.

High-Level Summaries

Mildenhall 14-15 (4/7/25) –
The unit premium formula using risk adjusted expected proportion of recoveries has a typo. At the end of the formula should be g(S) but it just has S. It has been fixed.

Clark (12/20/24) – On page 3 of the Clark summary, the ELF formula has a ‘U’ in the numerator instead of an ‘L’. It has been fixed.

Flashcards

Mildenhall Ch. 8 (1/22/25) – On page 24 of the standard PDF, the fifth reason should say “…technically insolvent…”

Clark (12/16/24) – On page 48 of the standard PDF, there was an “U” in the subscript of the ELF that should not have been there. It has been removed.

Panning (3/3/25) – On page 26 of the standard PDF, under the “Advantage,” it should say “seek to reduce the duration of the firm’s total economic value by…”