Home / CPCU / 520 / Assignment 4 / Part 2

Assignment 4 Part 2: Underwriting Decisions & Technology

How automation, tiered rating, and emerging tech are reshaping personal lines underwriting

Start Here: 5 Things You MUST Know

1

Straight-through processing (STP) means the application goes from submission to policy issuance with zero human underwriter involvement.

2

Tiered rating sorts applicants into risk buckets (preferred, standard, nonstandard) -- each tier has its own rate structure.

3

Credit-based insurance scores are used in most states but banned or restricted in California, Hawaii, Massachusetts, and Maryland.

4

Telematics/UBI uses actual driving data to price auto insurance -- discounts typically range from 5-30%.

5

Portfolio management is about managing ALL risks together -- geographic concentration, line mix, and loss ratio by segment.

Overview: Why Personal Lines Underwriting Is Different

Personal lines underwriting handles millions of individual policies -- homes, cars, renters. The sheer volume makes it impossible for human underwriters to touch every application. Instead, the industry has built automated systems that score, tier, and price applicants in seconds. This part covers how those systems work, what rating factors drive pricing, and the technologies transforming the field.

Exam Alert

Expect questions about what makes personal lines underwriting different from commercial lines (volume + automation), what STP means, and which states restrict credit-based insurance scores. Know the difference between rules-based systems and predictive models.

1. Straight-Through Processing (STP)

What Is STP?

Straight-through processing means an insurance application is submitted, automatically scored, priced, and issued as a policy -- all without a human underwriter ever looking at it. The system handles everything from start to finish.

How STP Works: The Flow

Application Submitted

Online or via agent

Data Prefill

MVR, CLUE, credit pulled automatically

System Scores Risk

Rules + predictive models

Tier Assigned

Preferred, Standard, etc.

Policy Issued

No human touch

When Does a Human Underwriter Step In?

Most applications sail through STP. A human underwriter only gets involved when the system flags something unusual:

  • -- Referrals: The risk score lands in a gray zone -- not clearly accept or reject
  • -- Exceptions: Unusual property type, exotic vehicle, prior cancellation history
  • -- Complex risks: High-value home, multiple claims in short period, mixed-use property
  • -- Conflicting data: Prefill data contradicts what the applicant reported

Real-World Scenario: STP in Action

The Setup: Sarah visits a direct-writer's website at 10 PM on a Sunday. She enters her name, address, DOB, and vehicle info for a new auto policy.

What Happens: The system instantly pulls her driving record (clean -- no violations in 5 years), CLUE report (no claims in 3 years), and credit-based insurance score (740). The rules engine checks: clean record? Yes. Good credit? Yes. No claims? Yes. Score = preferred tier.

The Result: Sarah gets a quote for $680/6 months, clicks "Buy Now," and her policy documents land in her email by 10:07 PM. No human underwriter was ever involved. This is STP.

2. Automated Underwriting: Rules vs. Predictive Models

Automated underwriting uses two main approaches. Most modern systems combine both.

Rules-Based Systems

Simple if/then logic that mirrors what a human underwriter would do -- just faster.

  • How it works: IF credit score > 700 AND no claims in 3 years AND no violations THEN auto-approve at preferred tier
  • Strengths: Transparent, easy to audit, regulators can see the logic
  • Weaknesses: Can't capture complex interactions between variables

Example: "If applicant is under 25 AND drives a sports car AND has a speeding ticket, decline." This is clear and simple but might miss that this particular 24-year-old completed a defensive driving course.

Predictive Models

Statistical models that analyze hundreds of variables simultaneously to produce a risk score.

  • How it works: Model weighs hundreds of data points -- age, credit, claim patterns, vehicle type, geography -- and produces a numeric score
  • Strengths: Finds hidden patterns, more accurate risk segmentation
  • Weaknesses: "Black box" -- harder to explain why someone was declined

Example: The model discovers that homeowners who also bundle auto insurance AND have a credit score above 720 AND live in suburban ZIP codes file 40% fewer claims. It automatically gives this group better pricing.

Real-Time Data Prefill

Instead of asking customers 50 questions, the system pulls data automatically from external databases:

MVR

Motor Vehicle Record -- driving history, violations, suspensions

CLUE

Claims history for auto AND property -- past 5-7 years

Credit Score

Credit-based insurance score from bureaus

Property Data

Year built, sq footage, roof type, construction

Vehicle Data

VIN decode -- safety ratings, theft rates, repair costs

Geo Data

Flood zone, wildfire risk, crime rate for the address

BEFORE: Manual Process

  • -- Agent fills out paper application
  • -- Mailed to home office
  • -- Underwriter orders reports manually
  • -- Reviews, decides, sets price
  • -- Policy mailed back to agent
  • Timeline: 3-7 business days

AFTER: STP with Automation

  • -- Customer enters basic info online
  • -- System prefills data instantly
  • -- Predictive model scores risk
  • -- Price calculated, policy issued
  • -- Documents emailed immediately
  • Timeline: 3-7 minutes

3. Tiered Rating Systems

Instead of setting a unique price for every single applicant (impossible at personal lines volume), insurers sort applicants into tiers -- groups of similar risk. Each tier has its own rate table. The better the tier, the lower the price.

Preferred Tier

Best risks, lowest rates

  • -- Clean driving record (0 violations)
  • -- High credit score (700+)
  • -- No claims in 3+ years
  • -- New or well-maintained home
  • -- Continuous prior insurance

Example: Maria, age 42, credit score 780, no accidents ever, drives a Honda CR-V, owns her home. She qualifies for the best tier and pays $520/6 months for auto.

Standard Tier

Average risks, standard rates

  • -- Minor violation (1 speeding ticket)
  • -- Average credit score (620-700)
  • -- Maybe 1 small claim
  • -- Older home, acceptable condition
  • -- Some lapse in prior coverage

Example: Dave, age 35, credit score 660, one speeding ticket 2 years ago, drives a Ford F-150. Standard tier -- pays $780/6 months for auto.

Nonstandard / High-Risk Tier

Worse-than-average, highest rates

  • -- Multiple violations or at-fault accidents
  • -- Poor credit (below 600)
  • -- Multiple claims in recent years
  • -- Lapse in coverage
  • -- DUI/DWI on record

Example: Kevin, age 23, credit score 540, DUI last year, two at-fault accidents. Nonstandard tier -- pays $2,400/6 months for auto (if he can find coverage at all).

Fine-Grained Tiers

Many insurers go well beyond three tiers. Some companies use 5 to 10+ tiers for even finer segmentation. For example: Super Preferred, Preferred, Preferred Standard, Standard, Standard Non-Standard, Non-Standard. More tiers = more accurate pricing = competitive advantage for the best risks while still writing borderline ones.

4. Rating Factors: Personal Auto

These are the variables that determine your auto insurance premium. Each one moves your price up or down.

Category Factor Impact on Premium
Driver Age Young (16-25) and elderly drivers pay more
Marital status Married drivers statistically file fewer claims
Driving record Violations and accidents = higher premium
Years licensed More experience = lower risk
Gender (where permitted) Some states ban gender as a rating factor
Vehicle Year, make, model Expensive/sporty cars cost more to insure
Safety features Airbags, ABS, backup cameras = discounts
Anti-theft devices Alarms, GPS tracking = lower comp premium
Vehicle use / mileage Pleasure < commute < business. More miles = more risk
Other Territory (ZIP code) Urban = higher theft/accidents. Rural = lower
Credit-based insurance score Strong predictor of claims. Used in most states
Prior insurance Continuous coverage = positive factor. Lapse = red flag

Credit-Based Insurance Scores: State Restrictions

Credit-based insurance scores are one of the strongest predictors of claims frequency, but some states restrict or ban their use:

California

Banned for auto

Hawaii

Restricted

Massachusetts

Restricted

Maryland

Restricted

Real-World Scenario: Same Driver, Different Factors

The Setup: James, 30, clean driving record, credit score 750, drives a 2023 Toyota Camry for his 10-mile commute. He lives in suburban New Jersey.

What Happens: His brother Mike, also 30 with the same credit and clean record, drives the same car -- but lives in Newark (higher-crime ZIP code) and uses his car for DoorDash deliveries (business use, 25,000 miles/year).

The Result: James pays $640/6 months. Mike pays $1,050/6 months. Same car, same age, same credit -- but territory and vehicle use alone create a 64% premium difference.

5. Rating Factors: Homeowners

Homeowners rating is more complex than auto because the property itself introduces dozens of variables.

Property Factors

  • -- Construction type (frame, masonry, etc.)
  • -- Age of home
  • -- Square footage
  • -- Roof type and age
  • -- Heating / electrical / plumbing systems

Protection Factors

  • -- Fire protection class (1-10 scale)
  • -- Distance to fire station
  • -- Smoke detectors / sprinklers
  • -- Security system
  • -- Smart home devices

Location Factors

  • -- ZIP code
  • -- Wildfire zone
  • -- Hail zone
  • -- Coastal proximity (hurricane)
  • -- Flood zone designation

Policyholder Factors

  • -- Credit-based insurance score
  • -- CLUE claims history
  • -- Age of insured
  • -- Insurance score

Coverage Factors

  • -- Deductible amount chosen
  • -- Coverage limits selected
  • -- Endorsements added
  • -- Bundling with auto

Real-World Scenario: Roof Age Matters

The Setup: Two identical 2,000 sq ft colonial homes on the same street in suburban Ohio. Same credit score, same owner age, same coverage limits.

What Happens: Home A has a 3-year-old architectural shingle roof. Home B has a 22-year-old roof that's past its expected lifespan. The insurer's aerial imagery confirms Home B has visible wear.

The Result: Home A: $1,200/year. Home B: $1,750/year -- and the insurer requires a roof inspection before renewal. If the roof isn't replaced, they may non-renew the policy. That old roof is a ticking time bomb for hail and wind claims.

6. Technology Transforming Personal Lines

Telematics & Usage-Based Insurance (UBI)

Instead of guessing how someone drives based on age and ZIP code, telematics measures actual driving behavior using a plug-in device or smartphone app.

Progressive Snapshot

Plug-in OBD device or app

State Farm Drive Safe & Save

Smartphone app-based

Allstate Drivewise

Smartphone app-based

What Telematics Tracks:

Hard Braking

Rapid Acceleration

Speed

Time of Day

Total Mileage

Phone Use

WITHOUT Telematics

Jake, 22, male, drives a Mustang GT. Based on demographics alone, the system assumes high risk. Premium: $2,100/6 months.

WITH Telematics

Jake enrolls in Snapshot. After 90 days, data shows: no hard braking, drives only 6,000 mi/year, rarely drives after 11 PM. Actual behavior = low risk. Premium drops to $1,470/6 months (30% discount).

Key number: Telematics discounts typically range from 5% to 30% based on driving behavior.

Smart Home Integration

Connected home devices reduce claims frequency, so insurers offer premium discounts for using them.

Water Shut-Off Devices

Flo by Moen, Phyn -- detect leaks and automatically shut off water. Prevents $10,000+ water damage claims.

Security Systems

Ring, SimpliSafe, ADT -- reduce burglary claims. Some insurers partner directly with these companies.

Smart Smoke Detectors

Nest Protect -- alerts you AND the fire department instantly. Early detection = smaller fire losses.

Key number: Smart home device discounts typically range from 5% to 15% on homeowners premiums.

Aerial / Satellite Imagery

Insurers use aerial photos and satellite images to assess properties without sending a physical inspector. The system can identify:

Roof Condition

Missing shingles, wear patterns

Property Size

Accurate square footage

Pool Presence

Undisclosed liability risk

Tree Overhang

Risk of falling branches on roof

Example: An applicant says their roof is in "good condition." The insurer's aerial imagery system flags that 15% of shingles are missing and there's significant moss growth. The underwriter orders an inspection before binding coverage -- this would have been missed without the technology.

Parametric Insurance

Pays out automatically when a triggering event occurs -- no claims adjuster needed.

Example: A parametric earthquake policy pays $10,000 automatically if an earthquake measuring 6.0+ hits within 50 miles of the insured's home. No need to prove damage -- the seismometer reading IS the trigger. Also emerging for hurricanes (wind speed trigger) and flight delays (departure time data).

Embedded Insurance

Insurance sold at the point of sale of another product -- seamlessly bundled in.

Example: You buy a Tesla -- Tesla offers its own auto insurance at checkout, priced using the car's own telematics data. Or you book a flight on Expedia and get offered travel insurance for $12 with one click. Or you buy an iPhone and Apple offers AppleCare+ insurance at purchase. The insurance is embedded in the buying experience.

7. Portfolio Management

The Big Picture View

Underwriters don't just evaluate one application at a time -- they manage the entire portfolio. A policy that looks good individually might create a concentration problem when combined with thousands of other similar policies.

Portfolio Concern What It Means Real-World Example
Geographic Concentration Too many policies in one area = massive catastrophe exposure An insurer has 40% of its homeowners book in coastal Florida. One Category 4 hurricane could bankrupt them.
Line of Business Mix Balance between auto, home, umbrella to diversify risk A company that writes only homeowners in tornado-prone states has no diversification. Adding auto helps balance volatility.
Loss Ratio by Segment Which tiers, territories, or products are profitable vs. bleeding money? Preferred tier auto: 55% loss ratio (great). Nonstandard auto in urban NJ: 92% loss ratio (barely breaking even). Need to raise rates or tighten underwriting.
Rate Adequacy Are current rates sufficient given inflation, loss trends, and claim severity? Auto repair costs jumped 15% due to supply chain issues. If rates aren't adjusted, combined ratio exceeds 100%.
Reinsurance Catastrophe reinsurance protects the portfolio from massive events The insurer buys a cat reinsurance treaty: "If total hurricane losses exceed $50M, the reinsurer covers 80% of losses above that." This protects the personal lines portfolio.

Real-World Scenario: Portfolio Concentration Problem

The Setup: Sunshine Mutual writes personal lines in Florida. Their agents are excellent -- they've grown the homeowners book to 85,000 policies. But 60,000 of those policies are within 20 miles of the Gulf Coast.

What Happens: Hurricane season arrives. A Category 3 storm makes landfall near Tampa. 12,000 of those 60,000 coastal homes file claims. Average claim: $45,000.

The Result: Total losses: $540 million from one event. Without adequate catastrophe reinsurance, Sunshine Mutual faces insolvency. This is why portfolio management -- monitoring geographic concentration and buying reinsurance -- is just as important as underwriting individual risks correctly.

Cheat Sheet

Print this page for quick reference

STP & Automation

  • -- STP = application to policy, zero human touch
  • -- Rules-based = if/then logic (transparent)
  • -- Predictive models = statistical scoring (black box)
  • -- Data prefill: MVR, CLUE, credit, property, VIN, geo
  • -- Human referral only for exceptions/gray zones

Tiered Rating

  • -- Preferred = best risk, lowest rate
  • -- Standard = average risk, standard rate
  • -- Nonstandard = worst risk, highest rate
  • -- Some insurers use 5-10+ tiers
  • -- Each tier = its own rate structure

Auto Rating Factors

  • -- Driver: age, marital, record, years licensed
  • -- Vehicle: make/model, safety, anti-theft, use, mileage
  • -- Territory: ZIP-code based
  • -- Credit score: banned in CA (auto), restricted in HI, MA, MD
  • -- Prior insurance: continuous = good

Homeowners Rating Factors

  • -- Property: construction, age, roof, systems
  • -- Protection: fire class, alarms, sprinklers
  • -- Location: wildfire, hail, coastal, flood zone
  • -- Policyholder: credit, CLUE, age
  • -- Coverage: deductible, limits, endorsements

Key Technology

  • -- Telematics/UBI: 5-30% discount
  • -- Smart home devices: 5-15% discount
  • -- Aerial imagery: roof, pools, tree overhang
  • -- Parametric: auto-payout on trigger event
  • -- Embedded: insurance at point of sale

Portfolio Management

  • -- Geographic concentration = cat risk
  • -- Line mix diversifies volatility
  • -- Monitor loss ratio by segment
  • -- Rate adequacy vs. loss trends
  • -- Cat reinsurance protects portfolio

Exam Trap Alerts

1. STP Does NOT Mean No Underwriting

STP means no human underwriter is involved. The underwriting still happens -- it's just done by the automated system. The rules, the scoring, the tier assignment -- that's all underwriting. Don't confuse "automated" with "no underwriting."

2. Credit Score vs. Insurance Score

A credit-based insurance score is NOT the same as a regular credit score (FICO). Insurance scores weigh factors differently -- they're optimized to predict claims likelihood, not creditworthiness. Same data, different model, different purpose.

3. California Bans Credit for Auto, Not Homeowners

California bans credit-based insurance scores for auto insurance. The exam may try to trick you into thinking it's banned for all lines. Other states (HI, MA, MD) have various restrictions but not complete bans across all lines.

4. Tiered Rating Is Not Cherry-Picking

Tiered rating sorts applicants into risk groups and prices accordingly -- every tier gets a rate. Cherry-picking means only writing preferred risks and declining everyone else. An insurer using tiered rating writes all tiers, just at different prices.

5. Parametric Insurance Has No Claims Process

With parametric insurance, the payout is triggered by a measurable event (earthquake magnitude, wind speed), NOT by proof of actual loss. This means you could receive a payout even if you had minimal damage -- or get nothing if the event didn't meet the threshold even though you had damage.

6. Portfolio Management vs. Individual Underwriting

An individually profitable policy can still be bad for the portfolio. If you already have too many coastal homeowners, writing one more excellent coastal risk still increases your catastrophe concentration. Portfolio thinking overrides individual-risk analysis.

Quick Reference Summary

Straight-Through Processing

Application to policy issuance with no human underwriter involvement

Rules-Based Systems

If/then logic -- transparent and auditable but limited in complexity

Predictive Models

Statistical scoring using hundreds of variables -- powerful but less transparent

Tiered Rating

Preferred (best risk/lowest rate) to Nonstandard (worst risk/highest rate)

Telematics / UBI

Actual driving behavior data -- 5-30% discounts for safe drivers

Smart Home Discounts

Connected devices (water sensors, security) -- 5-15% premium reduction

Aerial Imagery

Satellite/aerial photos to assess roof condition, pools, tree risk remotely

Parametric Insurance

Auto-pays on triggering event (earthquake magnitude, wind speed) -- no claims process

Portfolio Management

Geographic concentration, line mix, loss ratios, rate adequacy, reinsurance