Chapter 8 Exercises

Work these with the chapter's central discipline: a decision is only as good as the information behind it, so ask of every source what can this actually prove, what can't it, and am I allowed to use it? Items marked with a dagger (†) have worked solutions in Appendix: Answers to Selected Exercises; the rest are for discussion or self-test. Section references like (§8.3) point you back to the relevant part of the chapter.

A. Recall and definitions

  1. † Define the application (submission) and name the three jobs it does — describe the exposure, ask the underwriting questions, and create the representations. Why is the third one a legal matter, not just an informational one? (§8.1)
  2. State, in one sentence each, what a loss run is and what it is for. Why do you read it before you fully believe the application? (§8.2)
  3. Define third-party data and explain in one sentence why its independence is exactly what makes it valuable. (§8.3)
  4. † Distinguish the three major third-party products — the MVR, CLUE, and the credit-based insurance score — by stating, for each, what it measures. (§8.3)
  5. What does an inspection report do that the application cannot? Name the three functions from the chapter. (§8.4)
  6. Name three specific signals an underwriter reads from a company's financial statements, and say why financial distress is a loss signal and not merely a business concern. (§8.5)
  7. State the three core obligations the FCRA imposes when an insurer uses a consumer report to underwrite personal insurance. (§8.6)

B. Reading the loss run

  1. † A five-year commercial property loss run shows: Year 1 — a \$15K and a \$22K water-damage claim; Year 2 — a \$9K, \$31K, and \$12K claim (slip-and-fall, water, theft); Year 3 — clean; Year 4 — a \$280K fire; Year 5 — clean. (a) Characterize this history as primarily a frequency story, a severity story, or both. (b) What is the single most useful question to ask about the Year 4 fire? (c) Does the clean Year 5 reassure you, and what would you want to confirm before relying on it? (§8.2)
  2. Explain why two property fires both coded "fire" on a loss run tell you far less than two fires coded "electrical" and "hot work." What does the difference in cause imply about the building? (§8.2)
  3. † A loss run arrives valued fourteen months ago and shows a \$150K reserve on an open workers'-comp claim. Why is the valuation date a problem, and what specifically would you ask for before you price the account? (§8.2)
  4. A broker sends a one-page "loss summary" letter stating the account has been "essentially loss-free" for five years, in place of the carrier-produced loss runs. Why is this not an acceptable substitute, and what do you do? (§8.2)
  5. Harbor Steel has one pending products-liability claim with nothing yet paid. A colleague says "no payment, no problem — treat it as zero for now." Explain why that is wrong and how an open claim should be weighed. (§8.2; The Underwriting File)

C. Third-party data and its limits

  1. † Explain precisely what the industry does and does not claim about the credit-based insurance score. Why is "poor people cause more losses" a misstatement of the claim, and what is the accurate statement? (§8.3)
  2. A homeowner applicant has two CLUE claims in five years. Before treating this as a high-risk signal, what three things about each claim do you need to read — and why does a raw count of claims punish the unlucky alongside the genuinely high-risk? (§8.3)
  3. An MVR comes back clean. Give two different reasons a clean MVR does not prove the driver is low-risk. (§8.3)
  4. † A data vendor offers you a new alternative-data product that, it says, "predicts loss with strong lift." List three questions you would ask before using it in underwriting decisions — at least one about accuracy/provenance and at least one about legality. (§8.3, §8.6)

D. Underwrite this submission

  1. † Order the information. A new commercial submission arrives: a mid-size regional distributor seeking property, GL, and commercial auto on a 9-vehicle fleet, ~\$28M revenue, two warehouses. The broker has attached the ACORD application and three years of property loss runs only. Write the information order you send back: list every additional source you would require before quoting, and for each, one line on why. (§8.1–§8.5)
  2. For the distributor in #17, the application lists revenue at \$28M but the GL section is rated on \$19M of "sales." State why this discrepancy matters, what you do about it, and which source would settle it. (§8.5)
  3. A small, low-hazard risk (a two-person accounting office seeking a BOP) and a \$20M fabrication plant both cross your desk the same morning. Explain why you might write the first on the documents alone but would never write the second without an inspection. What principle governs how much verification a risk gets? (§8.4)
  4. † Build the risk picture. You receive an application for a 40,000 sq ft plastics manufacturer. The application is glowing: new ownership, a new safety committee, "state-of-the-art" equipment. It mentions no prior losses and provides no loss runs ("available on request"). List four specific things you would go verify and the source for each — and name the one thing the absence of in this submission you find most concerning. (§8.1, §8.2, §8.7)

E. Find the red flag

  1. For each, identify the red flag and the source that would confirm or resolve it: (a) the prior-losses field on the application is left blank; (b) the loss runs cover years 2–5 of a five-year request; (c) the financial statements are provided for two of three requested years; (d) the submission for a welding shop never mentions hot-work permits. (§8.1, §8.2, §8.5, §8.7)
  2. † An applicant's submission is unusually polished and reassuring — every answer smooth, every concern pre-addressed, the new safety program described in detail. A newcomer reads this as a clean risk. Explain why an experienced underwriter slows down, and what specifically they would look to confirm. (§8.7)
  3. A news search on a manufacturing applicant turns up a fire two years ago that does not appear on the loss runs the broker provided. What are the possible explanations, and why does this stop the quote until resolved? (§8.5, §8.2)

F. Compliance and ethics

  1. † You decline a personal-auto applicant in part because of a CLUE report and in part because of a credit-based insurance score. Walk through exactly what the FCRA requires you to do, what the adverse-action notice must contain, and the consumer's rights it must disclose. (§8.6)
  2. An underwriter says: "We charged this applicant a higher rate, we didn't decline them, so no adverse-action notice is needed." Is this correct? Explain. (§8.6)
  3. † Ethics dilemma. A rating factor your company uses in a particular state is statistically predictive of loss, but you suspect it functions as a proxy for a protected class in that market. Your manager says, "It's filed, it's legal here, and it predicts — use it." Lay out the genuine tension between actuarial validity and fairness, what "proxy discrimination" means, and what a responsible underwriter does rather than either using everything that predicts or refusing all contested data. (§8.6; preview of Ch. 35)
  4. A colleague pulls a CLUE report on a neighbor who has not applied for a policy, "just to check something." Which FCRA principle does this violate, and why does permissible purpose exist? (§8.6)

G. Write the memo

  1. † Draft a short (150–250 word) information-request memo to a broker for a commercial account of your choice. Specify exactly what you need (loss runs — how many years, which lines, how valued; inspection; financials; MVRs; SOV; any supplements), state the gaps you have identified, and make clear that you cannot quote until the file is complete — without sounding adversarial. (§8.1, §8.7, The Underwriting File)
  2. Write a brief (under 150 words) plain-language explanation, suitable for a non-underwriter colleague in sales, of why the company orders independent loss runs and a CLUE report instead of trusting the applicant's own loss summary. Use the idea of adverse selection without jargon. (§8.2, §8.3)

H. The Underwriting File

  1. † Reproduce, from memory and in your own words, the information order for Harbor Steel: the six categories of information you would request and, for each, the single most important thing it will tell you. (The Underwriting File)
  2. The Underwriting-File checkpoint ends with the disposition "information ordered; gaps flagged" and explicitly decides nothing. Explain why it would be a mistake to grade or price Harbor Steel at this point, and which specific pieces of still-missing information make any decision premature. (The Underwriting File)
  3. † Of everything you ordered on Harbor Steel, identify the two sources most likely to change your eventual view of the two fires — and explain what each could reveal that the loss-run dollars alone cannot. (§8.2, §8.4, The Underwriting File)
  4. List the gaps the chapter flags in the Harbor Steel file (the things still missing even after the order goes out), and for each, name which later chapter is most likely to close it. (The Underwriting File; §8.2, §8.4, §8.5)

I. Synthesis and connection

  1. The chapter argues that every information-gathering technique is, at root, a way of managing adverse selection. Choose any three sources (application, loss run, MVR, CLUE, credit score, inspection, financials, public records) and explain, for each, the specific information imbalance it corrects. (§8.1– §8.7; adverse selection from §1.4)
  2. † "The data tells you what is there; the underwriter notices what is not." Explain what this means using a concrete example of a missing field, and tie it to the theme that technology augments but does not replace the underwriter. Why can a model not, today, do the noticing? (§8.7; theme 5)
  3. Connect this chapter forward: the information you gather here becomes the input to the predictive model in Chapter 32, which will score Harbor Steel. Why does the quality and completeness of your information order now directly affect whether the model's score later is worth trusting? (§8.1–§8.5; preview of Ch. 32)