Chapter 34 Quiz

Twenty questions to check your grasp of InsurTech and the digital transformation of insurance: fifteen multiple choice and five short answer. Answers are in the collapsed key at the bottom — try the whole set before you open it.

Multiple choice

1. The single most important distinction the chapter draws between InsurTech players is:

  • A. Whether they use machine learning or rule-based systems
  • B. Whether they are based in the United States or abroad
  • C. Whose balance sheet carries the insurance loss
  • D. Whether they are publicly listed or privately held

2. A digital MGA differs from a full-stack InsurTech carrier in that the digital MGA:

  • A. Holds its own insurance license and policyholder surplus
  • B. Underwrites on a backing carrier's paper under delegated authority and does not own the loss in the same existential way
  • C. Cannot use technology to quote business
  • D. Is prohibited from earning commissions

3. Embedded insurance is best described as:

  • A. Insurance sold inside the purchase of the product or service it protects, at the moment of that purchase
  • B. Insurance that embeds a reinsurer's capital into the policy
  • C. A policy with an embedded deductible
  • D. Insurance sold only through independent agents

4. The chapter says three things converged in the 2010s to enable InsurTech. Which one turned out to be the unsound premise?

  • A. The flood of data from phones, telematics, and satellites
  • B. The genuine ugliness of the customer's buying experience
  • C. That cheap capital would last and underwriting losses could be funded until scale fixed them
  • D. That software could improve insurance intake

5. "Combined ratio = loss ratio + expense ratio." The defining mistake of many full-stack InsurTechs was to assume that:

  • A. A low loss ratio would deliver a low combined ratio
  • B. A low expense ratio would deliver a low combined ratio, when the loss ratio is the larger, harder term
  • C. The combined ratio did not matter at all
  • D. Investment income would always exceed underwriting losses

6. Basis risk in a parametric policy is:

  • A. The risk that the reinsurer defaults
  • B. The gap between the parametric payout (on a trigger) and the policyholder's actual loss
  • C. The risk that interest rates rise
  • D. The commission the MGA charges

7. A parametric weather policy must be carefully designed around which Chapter 4 doctrine so that it is insurance and not a wager?

  • A. Subrogation
  • B. Utmost good faith
  • C. Insurable interest
  • D. Indemnity by replacement cost

8. API distribution moves the underwriting decision so that it is:

  • A. Made by a human agent at the point of sale every time
  • B. Encoded in advance as rules the system runs unattended, thousands of times
  • C. Eliminated entirely, because no underwriting is needed
  • D. Performed only by the reinsurer

9. In a quote-in-seconds API flow, the chapter calls which element "the single most important piece of underwriting in the whole flow"?

  • A. The customer's email address field
  • B. The speed of the rating engine
  • C. The referral rule that stops the auto-quote and sends an unusual risk to a human
  • D. The color scheme of the checkout page

10. Which InsurTech businesses does the chapter identify as having aged best?

  • A. Full-stack carriers that owned their own loss ratio
  • B. Digital MGAs, embedded platforms, and software enablers
  • C. Companies that refused to use any third-party data
  • D. Peer-to-peer pools with no backing carrier

11. The chapter's "master lesson" of the InsurTech cycle is:

  • A. Technology never works in insurance
  • B. Growth is easy and underwriting is hard; the loss ratio cannot be outrun
  • C. Expense ratios are the only thing that matters
  • D. Customer love is a reliable hedge against losses

12. When several InsurTechs pivoted from full-stack carriers to MGAs, enablers, or heavy reinsurance buyers, the chapter reads this as:

  • A. Proof that InsurTech was entirely hype
  • B. The market correctly relocating the insurance risk to balance sheets built to hold it
  • C. Regulatory failure
  • D. A sign that underwriting no longer matters

13. Embedded insurance changes the underwriting problem mainly by turning it into a form of:

  • A. Individual medical underwriting
  • B. Class underwriting, where the program (eligibility, terms, price) is set once for a whole population
  • C. Surety analysis
  • D. Reinsurance treaty negotiation

14. Which statement best captures the chapter's view of the underwriter's future in an InsurTech world?

  • A. The underwriter is fully replaced by algorithms
  • B. The routine work is automated, while the complex-judgment work and the work of governing the automated systems become more valuable
  • C. Nothing about the job changes
  • D. Underwriters should refuse to use any InsurTech tools

15. "The system did it automatically" offered to a regulator questioning an API-delivered rate is:

  • A. A complete legal defense
  • B. Not a defense — the rates must match the filing and the encoded logic is examinable for prohibited factors and proxies
  • C. Irrelevant, because API rates are unregulated
  • D. Acceptable only for commercial lines

Short answer

16. In two or three sentences, explain the structural moral hazard in the digital-MGA relationship and name two defenses a backing carrier uses against it.

17. Why does a fast-growing delegated MGA book "look like a triumph for two or three years and then reveal itself as a disaster"? Use the timing of revenue versus losses.

18. Give one way usage-based auto insurance genuinely attacks adverse selection, and one limit the chapter insists on.

19. Could a digital MGA's quote-in-seconds platform appropriately bind Harbor Steel? State the correct outcome and the reason, in two sentences.

20. State, in your own words, why the public InsurTech stumbles are "the same event at the scale of a whole business" as the model-override you studied in Chapter 32.


Answer key (try the questions first) **1.** C — whose balance sheet carries the loss separates a carrier (insurance risk) from an MGA/distributor (distribution risk). **2.** B — the digital MGA underwrites on a carrier's paper under delegated authority; the carrier and its reinsurers own the loss. **3.** A — cover sold inside the purchase it protects, at the moment of purchase. **4.** C — the data and the bad customer experience were sound premises; the bet that cheap capital would last long enough to fund years of underwriting losses was not. **5.** B — they optimized the expense ratio and assumed it would deliver the combined ratio, but the loss ratio is the larger and harder term for most lines. **6.** B — the gap between the trigger-based payout and the actual loss. **7.** C — insurable interest, the doctrine that keeps insurance from being a wager. **8.** B — the underwriting is encoded in advance and the system runs it unattended at volume. **9.** C — the referral rule, the safety valve that stops the auto-quote for unusual risks. **10.** B — digital MGAs, embedded platforms, and enablers (the capital-light, distribution-or-software plays). **11.** B — growth is easy, underwriting is hard, and the loss ratio cannot be outrun. **12.** B — the market relocating risk to where it belongs (and technology to where it adds value). **13.** B — class underwriting at scale; the program is priced once for the whole embedded population. **14.** B — routine work automates; complex judgment and system-governance work become more valuable. **15.** B — not a defense; rates must match the filing and the encoded logic is examinable for prohibited factors and proxies (Chapters 4 and 35). **16.** The MGA earns fees/commissions on *volume* while the carrier bears the *losses* — a misalignment of incentive (a structural moral hazard). Defenses include tight delegated-authority limits, a referral grid for out-of-appetite risks, a hard audit of the MGA's underwriting, and a contract clause to claw back the pen if the loss ratio drifts (any two). **17.** The MGA's revenue (fees/commissions) arrives *now*, with the premium, while the losses arrive *later*, with the claims; so a growing book books healthy revenue and immature, low-looking losses for a couple of years, and the true loss ratio only emerges as the claims develop and the growth slows. **18.** It replaces a *proxy* for driving risk (age, zip code) with a measurement of *how the person actually drives*, which is harder to game and lets a safe-but-risky-looking driver prove it. The limit: it is a *data* relationship with data-quality, privacy, and consent problems (and a fairness question deferred to Chapter 35), and the device or the scored month may not represent true behavior. **19.** No — the platform's intake could speed the *quote*, but a good referral rule should stop the auto-bind and send the account to a human, because the two fires, the aging roof and sprinklers, the named-windstorm exposure, the pending products claim, and the multi-line structure are exactly the complexity that requires judgment. The correct outcome is *refer to an underwriter*. **20.** In both cases the data and the algorithms were genuinely useful but genuinely insufficient: a model scored one risk and an underwriter overrode it on judgment; a whole business model bet that data could price and select risk well enough to win, and the industry's accumulated loss experience overruled it. The algorithm is neither enemy nor savior — its limits are the same at one risk and at one company.