Start Here: 5 Things You MUST Know
Five types of rate regulation from most to least restrictive: Prior Approval, File-and-Use, Use-and-File, Flex Rating, No Filing (Open Competition)
Credibility formula: New Rate = Z x (Own Data) + (1 - Z) x (Industry Data), where Z ranges from 0 to 1
Large losses must be capped (truncated) so one giant claim does not distort the rate for everyone
GLMs are the standard modern rating tool; GBMs/ML are more powerful but face "black box" regulatory pushback
Proxy discrimination is the top regulatory concern -- using allowed factors (credit, ZIP) that correlate with prohibited factors (race, income)
1 Five Types of Rate Regulation
State regulators control how insurance rates are approved. The five regulatory approaches range from extremely strict (Prior Approval) to completely hands-off (Open Competition). This is one of the most heavily tested topics on the exam.
Restrictiveness Spectrum
Prior Approval
Most Restrictive
File-and-Use
Moderate
Use-and-File
Less Restrictive
Flex Rating
Flexible
No Filing
Least Restrictive
a) Prior Approval
Insurer must file rates AND get explicit approval from the state regulator BEFORE using them. Most restrictive type.
Advantage
Strong consumer protection -- regulator reviews every change
Disadvantage
Slow -- can take months. Rates may be outdated by approval date
Real-World Scenario:
The Setup: Insurer files a 12% auto rate increase in January.
What Happens: State DOI reviews for 6 months. Approved in July.
The Result: By July, loss trends have worsened, and the 12% is already insufficient. The insurer must file again.
Common in: Personal lines (auto, homeowners) in many states.
b) File-and-Use
Insurer files rates with the regulator and can use them immediately, but the regulator can disapprove them later.
Key risk: The insurer takes the chance that rates may be ordered withdrawn after they are already in use.
Real-World Scenario:
The Setup: Insurer files new homeowners rates on March 1 and starts using them immediately.
What Happens: In May, the regulator reviews and says "the 15% increase is too high."
The Result: Regulator orders a rollback to 10%. Insurer may need to issue refunds for the overcharge from March to May.
c) Use-and-File
Insurer uses rates FIRST, then files them with the regulator within a specified period (typically 15-30 days).
Real-World Scenario:
The Setup: Sudden spike in auto theft hits the metro area.
What Happens: Insurer raises comprehensive rates immediately to reflect the new risk.
The Result: Insurer files the rate change with the state within 30 days. Fast response keeps the insurer solvent.
d) Flex Rating (Band Rating)
Insurer can change rates within a pre-approved range (band) without filing. Changes outside the band require a full filing.
Example: State allows +/- 7% band
+5%
No filing needed
+12%
Full filing required
Combines speed (within band) with oversight (outside band).
e) No Filing Required (Open Competition)
No rate filing needed. Market competition is trusted to keep rates fair. Least restrictive approach.
Real-World Scenario:
The Setup: A large manufacturer needs commercial property insurance.
What Happens: The buyer is sophisticated, shops multiple insurers, and negotiates terms.
The Result: Competition among insurers keeps pricing fair. No regulator involvement needed.
Common in: Commercial lines in many states. Illinois is a notable example.
| Type | When Insurer Files | When Insurer Can Use | Regulator Power | Best For |
|---|---|---|---|---|
| Prior Approval | Before using rates | Only after approval | Must approve first | Personal lines, consumer protection |
| File-and-Use | At same time as using | Immediately on filing | Can disapprove later | Moderate regulation states |
| Use-and-File | Within 15-30 days after use | Immediately, before filing | Can disapprove after review | Responsive markets |
| Flex Rating | Only if outside band | Immediately if within band | Sets the band limits | Balancing speed + oversight |
| No Filing | Never | Anytime | Market competition only | Commercial lines, sophisticated buyers |
Memory Trick: "Please File Usually For Nothing"
Prior Approval -- File-and-Use -- Use-and-File -- Flex Rating -- No Filing. That is the order from most to least restrictive. The first letters spell PFUFN -- "Please File, Usually For Nothing."
2 The Rate Filing Process
Rate Filing Flow
Insurer or Advisory Org
Prepares filing
SERFF Submission
Electronic filing
State DOI Review
Actuaries evaluate
Approved or Objected
Hearing if disputed
Who Files?
The insurer itself, or an advisory organization on its behalf (ISO for property/liability, NCCI for workers' comp). Advisory orgs collect industry-wide data and file reference rates that individual insurers can adopt or modify.
What Is Included?
Supporting loss data, actuarial analysis, justification for the rate change, expense projections, and trend factors. The filing must demonstrate the rate is adequate, not excessive, and not unfairly discriminatory.
How Is It Filed?
SERFF (System for Electronic Rate and Form Filing) -- the electronic platform used by most states. Standardizes the filing process and speeds up review times.
What If the Regulator Objects?
The insurer can request a formal hearing to defend its rate. State DOI actuaries test for adequacy, excessiveness, and unfair discrimination. The burden of proof depends on the state.
3 Credibility
Credibility is a statistical measure of how reliable an insurer's own loss data is. The more data you have, the more you can trust it. Think of it as a "confidence dial" from 0 (no trust) to 1 (full trust).
The Credibility Formula
New Rate = Z x (Own Experience) + (1 - Z) x (Industry Data)
Z = 1
Full credibility. Use own data entirely.
Z = 0.5
50/50 blend of own + industry.
Z = 0
No credibility. Use industry data entirely.
What Determines Z?
Volume
More policies = higher Z
Homogeneity
Similar risks = higher Z
Time
Longer experience period = higher Z
Worked Example: Two Insurers, Same State
Big National Insurer
Policies: 200,000 auto policies in the state
Z = 0.95 (nearly full credibility)
Own experience says: Rates need +8%
Industry data says: +6%
Result: 0.95 x 8% + 0.05 x 6% = 7.6% + 0.3% = +7.9%
Almost entirely driven by own data.
Small Regional Insurer
Policies: 500 auto policies in the state
Z = 0.15 (low credibility)
Own experience says: Rates need +8%
Industry data says: +6%
Result: 0.15 x 8% + 0.85 x 6% = 1.2% + 5.1% = +6.3%
Heavily reliant on ISO industry data.
4 Large Loss Limitations
A single massive loss can distort the entire dataset. Actuaries must cap or truncate large losses to prevent one outlier from unfairly spiking rates for everyone.
Real-World Scenario: The Mansion Fire
The Setup: Typical homeowners claims in a state average $20,000-$50,000.
What Happens: One mansion burns down -- $2 million claim. If included at full value, it spikes the average claim for all 50,000 homeowners policies.
The Result: The actuary truncates (caps) that claim at $250,000 for ratemaking purposes. The excess above $250K is analyzed separately in an excess layer.
Truncation
Cap each loss at a threshold (e.g., $250K). Any amount above the cap is excluded from the base rate calculation. Simple and widely used.
Separate Excess Layer Analysis
The amounts above the cap are pooled and analyzed separately, often using industry-wide data because individual insurer data is too sparse at high loss levels.
5 Modern Actuarial Techniques
The actuarial profession has moved far beyond simple averages and manual calculations. Modern tools use advanced statistics, machine learning, and real-time data to price risk more accurately than ever before.
Generalized Linear Models (GLMs)
The standard modern rating tool. Analyzes multiple rating factors simultaneously and detects interactions between variables. Replaced the old one-variable-at-a-time approach.
Example: GLM discovers that credit score, driving record, and vehicle type interact to predict risk better than looking at each variable in isolation.
GBMs & Machine Learning
Even more powerful than GLMs -- can find complex, non-linear patterns in data. But the "black box" problem makes them harder to explain to regulators.
Usage: More common for internal pricing sophistication scores, less common in filed rates because regulators demand explainability.
Telematics-Based Rating (UBI)
Usage-Based Insurance rates drivers on actual behavior: speed, braking, acceleration, time of day, and miles driven.
Programs: Progressive Snapshot, State Farm Drive Safe & Save, Root Insurance. Data collected via phone app or OBD-II plug-in device.
Climate Modeling
Catastrophe models from AIR, RMS, and CoreLogic estimate hurricane, earthquake, wildfire, and flood risk. Climate change means historical data alone is no longer enough.
Example: Traditional model shows Florida hurricane risk based on last 100 years. Climate-adjusted model shows 20-30% higher risk due to warming oceans.
6 Regulatory Concerns with Modern Techniques
More powerful tools create more powerful concerns. Regulators are increasingly focused on fairness, transparency, and privacy in algorithmic rating.
Proxy Discrimination
When a permitted rating factor (credit score, ZIP code) correlates with a prohibited factor (race, income). The algorithm is technically legal but produces discriminatory outcomes.
Example: Using ZIP code as a rating factor may effectively charge more to minority neighborhoods, even though race is never explicitly used. The ZIP code acts as a proxy for race.
Algorithmic Bias
ML models trained on biased historical data will perpetuate and even amplify that bias. If past underwriting unfairly penalized certain groups, the algorithm learns to do the same thing -- automatically and at scale.
Explainability
Regulators increasingly demand that insurers can explain WHY a rate is what it is. "The algorithm said so" is not an acceptable answer. This is why GLMs (explainable) are preferred over GBMs (black box) for filed rates.
Data Privacy
GDPR (Europe), state privacy laws (CCPA in California), and emerging frameworks limit what data can be collected and used in rating. Telematics data is a particular flashpoint.
Colorado SB 21-169: Leading the Way
Colorado now requires insurers to test their algorithms for unfair discrimination before using them. Other states are watching closely and considering similar legislation. This is a landmark piece of insurance regulation for the AI age.
7 The Actuary's Expanding Role Across the Value Chain
Traditional Role
- --Ratemaking (setting premiums)
- --Reserving (setting aside funds for claims)
Just two functions -- relatively narrow scope.
Modern Role (2024-2026)
- +Pricing strategy and product development
- +Underwriting strategy optimization
- +Claims analytics: fraud detection, severity estimation
- +Catastrophe modeling and climate risk
- +Capital allocation and surplus modeling (links to CPCU 540)
- +Enterprise Risk Management (ERM)
- +Predictive analytics across the entire value chain
Exam Alert!
The exam may ask about the actuary's role beyond traditional ratemaking. Know that predictive analytics in claims (fraud detection, severity estimation, reserve optimization) is a major growth area. Capital modeling ties directly to CPCU 540 content.
Cheat Sheet
Print this page for quick referencePrior Approval
File first, wait for approval, THEN use. Most restrictive.
File-and-Use
File and use immediately. Regulator can disapprove later.
Use-and-File
Use first, file within 15-30 days.
Flex Rating
Free within the band. File only if exceeding band limits.
No Filing / Open Competition
No filing. Market forces regulate. Commercial lines.
Credibility Formula
Z x Own + (1-Z) x Industry. Z = 0 to 1.
Large Loss Limitation
Truncate big losses at a cap. Analyze excess separately.
GLMs vs GBMs
GLMs = standard, explainable. GBMs = powerful, black box.
SERFF
System for Electronic Rate and Form Filing. Most states use it.
Proxy Discrimination
Allowed factor (ZIP) correlates with prohibited factor (race).
Telematics / UBI
Rate on actual driving behavior. Progressive Snapshot, Root.
Colorado SB 21-169
Requires testing algorithms for unfair discrimination.
Exam Trap Alerts
1. File-and-Use vs. Use-and-File -- Do NOT Mix These Up
File-and-Use: File and use at the same time. Use-and-File: Use FIRST, file LATER. The exam loves switching these two. Remember: the first word tells you what comes first. "File-and-Use" = filing comes first. "Use-and-File" = using comes first.
2. Flex Rating Is NOT No Filing
Flex rating still requires filing for changes OUTSIDE the band. Only "No Filing / Open Competition" has zero filing requirements. If the exam describes a +/- percentage band, that is flex rating, not open competition.
3. Credibility Z = 0 Means Industry Data, NOT Zero Rate
When Z = 0, the formula becomes: New Rate = 0 x Own + 1 x Industry = Industry data entirely. It does NOT mean the insurer charges nothing. Z = 0 just means the insurer's own data is too thin to rely on.
4. GLMs Are the Standard -- Not GBMs
If the exam asks what is the "standard modern rating tool" or "most widely used in filed rates," the answer is GLMs. GBMs are more powerful but not the standard for filed rates because they lack explainability.
5. Proxy Discrimination vs. Unfair Discrimination
Proxy discrimination uses a permitted factor that correlates with a prohibited one. The factor itself (e.g., credit score) is legal. The issue is the disparate impact. Direct unfair discrimination explicitly uses prohibited factors. Proxy discrimination is the sneakier, harder-to-detect problem.
6. Advisory Orgs File REFERENCE Rates, Not Final Rates
ISO and NCCI file reference (advisory) rates. Individual insurers can adopt, modify, or deviate from these rates. Advisory orgs do NOT set the final rate that consumers pay. The insurer always makes the final pricing decision.
7. Large Loss Truncation Does NOT Ignore the Loss
Truncation caps the loss at a threshold for base rate calculation, but the excess amount is still analyzed in a separate excess layer. The loss is not thrown away -- it is just handled differently to avoid distorting the average.
8. Prior Approval's Biggest Flaw Is TIMING, Not Fairness
Prior approval provides strong consumer protection. Its weakness is that rates can become outdated before they are approved. The exam may frame this as a trade-off question: protection vs. responsiveness.
Quick Reference Summary
Prior Approval
Most restrictive. File, wait, then use. Common in personal lines.
File-and-Use
File and use simultaneously. Regulator reviews after the fact.
Use-and-File
Use first, file within 15-30 days. Fastest response to market.
Flex Rating
Free within band (e.g., +/-7%). File only outside the band.
Open Competition
No filing. Market forces regulate. For sophisticated commercial buyers.
Credibility (Z)
0 to 1 scale. More data = higher Z = more weight on own experience.
Large Loss Truncation
Cap outlier claims. Analyze excess separately. Prevents rate distortion.
GLMs
Standard modern tool. Multi-variable. Transparent. Regulator-approved.
GBMs / Machine Learning
More powerful but black box. Used internally, hard to file.
Telematics / UBI
Rates based on actual driving. Privacy vs. accuracy trade-off.
Proxy Discrimination
Legal factor correlates with illegal factor. Top regulatory concern.
Actuary's Modern Role
Beyond ratemaking: claims, UW, cat modeling, ERM, capital allocation.