Actuarial Analyst

Actuarial Analyst

When to Use

• Build or refresh loss development triangles and support IBNR / ultimate loss estimates
• Prepare pricing support: experience summaries, trend, credibility-weighted indications
• Validate, reconcile, and document actuarial data feeds and model inputs/outputs
• Produce reporting packs: exhibits, bridges, roll-forwards, and memo-ready tables
• Tie statutory or management reporting schedules to actuarial workpapers
• Apply approved assumptions in models under actuary direction (not set enterprise policy)
• Run reasonability checks and peer-review prep before actuary sign-off
• Support pension or group benefit valuation data prep and schedule tie-outs (analyst depth)

When NOT to Use

• Appointed actuary opinions, statement of actuarial opinion, or final sign-off → actuary
• Engagement scoping, SOW, client decks, M&A consulting framing → actuarial-consulting
• Enterprise assumption governance, assumption papers, and change control → assumption-setting
• ALM strategy, duration matching, and investment policy → asset-liability-management
• P&C coverage legal wording, claims handling, or underwriting authority → property-casualty-insurance
• Life/health product design and reserving sign-off → life-health-insurance (line context); technical judgment → actuary
• Pension plan design, funding policy, or trustee advice → pension-retirement-funds (strategy); analyst prep stays here
• Chart design, dashboards, and visualization craft only → data-visualization
• Source-system ETL, deduplication pipelines without actuarial judgment → data-scrubbing
• Capital markets quant research and backtesting → quantitative-researcher

Related skills

Core Workflows

1. Task intake and scope (analyst)

Before opening triangles or models:

1. Assignment type — Reserving, pricing support, reporting tie-out, model run, data prep 2. Basis — Statutory, GAAP, IFRS 17, management; valuation date and cutoff 3. LOB and segment — Direct vs assumed; reinsurance netting rules 4. Deliverable — Exhibit list, workpaper IDs, deadline, reviewer (actuary name if known) 5. Source of truth — Prior memo, approved assumptions version, model build ID 6. Out of scope — Flag sign-off, legal interpretation, or assumption changes needing actuary / assumption-setting

See `references/actuarial_analyst_scope.md`.

2. Reserving and loss development

1. Pull paid, incurred, case, and count triangles by origin and development age 2. Run diagnostics (factor stability, calendar-year check, paid/incurred divergence) 3. Apply selected methods (chain ladder, BF, Cape Cod) per actuary instruction—document picks and alternatives 4. Build IBNR and ultimate tables; reconcile to prior evaluation 5. Prepare PYD bridge and reserve roll-forward exhibits 6. Escalate tail, cat, or large-loss treatment questions to actuary

See `references/reserving_and_loss_development.md`.

3. Pricing and rate indication support

1. Aggregate experience by segment (exposure, frequency, severity, pure premium) 2. Compute trend and on-level adjustments per documented methodology 3. Apply credibility weights as specified; show gross vs credibility-weighted indications 4. Summarize indicated rate change vs current; tie to loss ratio targets if provided 5. Document data cuts, outliers, and one-time events in a support schedule 6. Outline GLM or multivariate specs for actuary or modeler—do not present as final model without review

See `references/pricing_and_rate_indication.md`.

4. Data validation and model I/O

1. Reconcile premium, exposure, and claim feeds to control totals 2. Validate triangle dimensions, accident/valuation dates, and duplicate keys 3. Compare model inputs to source exhibits; log exceptions with $ impact where possible 4. Review model outputs for reasonability vs prior run and plan metrics 5. Archive run logs, parameter files, and data cut manifests for audit trail 6. Route assumption changes to actuary; do not alter approved assumption tables without approval

See `references/data_validation_and_model_io.md`.

5. Reporting and workpapers

1. Build exhibit index aligned to filing or management pack structure 2. Produce bridges (reserve change, premium earned, claim counts) with footnoted definitions 3. Tie external schedules (statutory blanks, management reports) to internal workpapers 4. Draft memo support tables (facts and exhibits)—not full executive narrative unless asked 5. Version files: YYYYMMDD_basis_LOB_vNN with change log tab or README row

See `references/reporting_and_workpapers.md`.

6. Quality review and documentation

1. Complete analyst checklist before actuary review (below) 2. Ensure formulas trace and no hard-coded plugs without comment 3. Cross-foot subtotals and tie to GL or bordereaux where required 4. List open items with owner and target date 5. Package review folder: inputs, outputs, diagnostics, reconciliation, change log

See `references/quality_review_and_documentation.md`.

Analyst checklist (before actuary review)

Analyst QA:
- [ ] Valuation date, basis, and segment definitions match memo / prior period
- [ ] Triangles tie to source data; paid + case = incurred (where applicable)
- [ ] Development factors documented with $ impact of alternative picks (if material)
- [ ] Reserve roll-forward reconciles opening + movement = closing
- [ ] Pricing experience A/E and exposure bases foot to bordereaux
- [ ] Credibility and trend applied per approved assumption memo
- [ ] Model run ID and assumption version recorded
- [ ] Exhibits numbered; cross-references in memo outline updated
- [ ] Open questions listed for actuary (not buried in cells)

Key metrics (analyst must define each time)

Tool patterns (agnostic)

Prefer reproducible steps actuaries can re-run; avoid opaque manual cell edits without audit trail.

Line-of-business notes (analyst)

Deliverable standards (analyst)

State limitations (data gaps, one-time events, immature years). Do not present analyst work as opinion, legal advice, or filed regulatory submission without qualified actuary and counsel review.

Assignment type matrix

When tasks span reserving and pricing on the same block, separate workpapers and flag independence concerns to the actuary.

Ethics and reliance (analyst)

• Label work as draft or analyst support until actuary review
• Cite data limitations, restatements, and one-time events in open items
• Do not change approved assumptions locally—request update through actuary and assumption-setting
• Distinguish technical indication from implemented rate or posted reserve
• Refer legal or filing questions to counsel; refer sign-off to actuary

When to load references

• Scope, boundaries, ethics → references/actuarial_analyst_scope.md
• Reserving and triangles → references/reserving_and_loss_development.md
• Pricing and indication → references/pricing_and_rate_indication.md
• Data and model I/O → references/data_validation_and_model_io.md
• Reporting and workpapers → references/reporting_and_workpapers.md
• QA and documentation → references/quality_review_and_documentation.md

Actuarial analyst scope

Table of contents

1. Role boundary 2. Relationship to actuary and consulting 3. Typical assignments 4. Lines of business (analyst lens) 5. Professional standards (analyst) 6. Escalation triggers

Role boundary

The actuarial analyst executes technical preparation, calculation support, reconciliation, and exhibit production under actuarial direction. The analyst does not own:

Relationship to actuary and consulting

Client / management question
        ↓
actuarial-consulting (engagement framing) — optional
        ↓
actuary (methods, assumptions, sign-off)
        ↓
actuarial-analyst (data, exhibits, reconciliations, runs)

Analysts implement methods the actuary selects; they propose diagnostics and alternatives in a support memo, not as final professional judgment.

Typical assignments

Lines of business (analyst lens)

Coordinate line context with property-casualty-insurance, life-health-insurance, or pension-retirement-funds when the user needs product or plan mechanics beyond exhibit production.

Professional standards (analyst)

Align workpapers so a qualified actuary can evaluate:

• Traceability — Source → transformation → exhibit cell
• Definitions — Every metric footnoted (basis, cutoff, net/gross)
• Versioning — Data cut date, model build, assumption memo reference
• Transparency — Hard-coded adjustments explained; no hidden plugs
• Confidentiality — Redact policyholder fields per firm policy

Do not misrepresent analyst tables as signed actuarial opinion.

Escalation triggers

Escalate to actuary (and document in open-items log) when:

• Triangle diagnostics imply method change (e.g., shift from chain ladder to BF)
• Tail factor or cat treatment is material and undocumented
• Credibility or trend differs materially from approved assumptions
• Model output diverges from plan or prior quarter beyond tolerance
• Regulatory or legal interpretation is requested
• Assumption change is needed → also involve assumption-setting for governance

Data validation and model I/O

Table of contents

1. Validation principles 2. Source reconciliation 3. Triangle and bordereaux checks 4. Assumption and parameter control 5. Model run workflow 6. Output reasonability 7. Handoffs to data-scrubbing

Validation principles

Actuarial conclusions are only as good as inputs. Analyst validation focuses on:

Log issues in a validation register: ID, severity, $ impact, owner, status.

Source reconciliation

Typical control totals:

Σ policy earned premium  =  GL earned premium (± known adjustments)
Σ claim payments         =  claims warehouse paid (± timing)
Σ case reserves          =  claims system outstanding

Steps:

1. Import control report from finance or data team 2. Compare at company, LOB, and segment level 3. Investigate differences > materiality threshold (from actuary) 4. Document timing (cutoff, ETL lag) vs error

Use data-scrubbing for systematic source fixes; analyst documents actuarial impact of corrections.

Triangle and bordereaux checks

Triangle tie-out:

Σ incremental cells for origin AY = cumulative at latest lag
Σ all origins at latest valuation = bordereaux total incurred

Assumption and parameter control

If model requires assumption upload:

1. Export from approved repository 2. Check row counts and key totals vs memo 3. Store file in run folder with timestamp

Model run workflow

1. Pre-run checklist (data, assumptions, version)
2. Execute model (script, vendor tool, spreadsheet)
3. Capture log (start/end, errors, warnings)
4. Export outputs to standard template
5. Post-run reasonability vs prior
6. Package for actuary review

Document:

Output reasonability

Compare current run to prior quarter and plan (tolerances from actuary):

Flag statistical noise vs data breaks (system migration, coding change).

Handoffs to data-scrubbing

Route to data-scrubbing when:

• Repeated ETL failures require pipeline fix
• Master data (class codes, territory) needs normalization at source
• Duplicate policy records need systematic dedupe

Analyst remains accountable for actuarial reconciliation after scrubbing completes—re-run controls.

Pricing and rate indication (analyst)

Table of contents

1. Analyst role in pricing 2. Experience aggregation 3. Trend and on-level 4. Credibility 5. Indication summary 6. GLM and multivariate (spec level) 7. Constraints and implementation 8. Analyst exhibits

Analyst role in pricing

Analysts prepare technical support for rate indications and renewals. The actuary or pricing actuary owns:

• Segment definition and homogeneity
• Method selection and final indication
• Regulatory filing narrative (with compliance/counsel as needed)

Analyst deliverables: clean experience, documented adjustments, indication math reproducible from source data.

Experience aggregation

Steps:

1. Filter experience period per memo (e.g., 3 accident years, 12 months earned) 2. Segment per rating variables (class, territory, limit band) 3. Remove outliers only per written rules; log removed records 4. Compute pure premium = losses ÷ exposure (or frequency × severity)

Produce A/E vs expected if expected table provided in assumption memo.

Trend and on-level

Document:

• Formula and indices (CPI, industry benchmark, internal index)
• From-date / to-date for trend factors
• Sensitivity: ±1 point trend on indicated change (if material)

Do not invent trend without approved assumption or actuary instruction.

Credibility

Apply weights per memo (examples—actuary selects formula):

Show columns:

• Observed pure premium or loss ratio
• Complement (manual, industry, or prior approved)
• Credibility-weighted indication

indicated = Z × observed + (1 - Z) × complement

Indication summary

Standard summary table:

Separate gross indication from implemented rate if underwriting constraints provided.

GLM and multivariate (spec level)

When actuary requests GLM support, document specification only unless trained to run production models:

Route model fitting and selection to actuary or advanced analytics (quantitative-researcher for non-standard ML).

Constraints and implementation

Capture non-technical constraints in a separate tab (do not bury in formula):

• Competitive floor/ceiling
• Regulatory maximum increase
• Filings already submitted
• Portfolio mix targets

Analyst does not override technical indication without actuary approval—show both columns.

Analyst exhibits

Minimum pricing support pack:

1. Data reconciliation (premium, exposure, losses) 2. Segment experience tables 3. Trend and on-level support 4. Credibility calculation 5. Indication summary 6. Sensitivities (if requested) 7. Open questions

Cross-check combined ratio impact if expense and profit loads provided in separate tab.

Quality review and documentation

Table of contents

1. Analyst QA layers 2. Formula and spreadsheet controls 3. Peer review prep 4. Documentation standards 5. Open items and escalation 6. Archive and retention 7. Common defects

Analyst QA layers

Complete self-QA using the checklist in SKILL.md every time.

Formula and spreadsheet controls

Run balance checks:

• Triangles sum to control totals
• Subsegments sum to total
• Roll-forward closes

Peer review prep

Package for reviewer:

review_folder/
  00_README.md          # Purpose, basis, valuation date
  01_index.xlsx         # Or index tab
  02_inputs/            # Frozen data
  03_workpapers/        # Main file(s)
  04_outputs/           # PDF exhibits if needed
  05_open_items.md      # Questions and resolutions

In 05_open_items.md use:

Reviewer comments: use cell comments or review tool; resolve with initials and date.

Documentation standards

Every exhibit should answer:

1. What is this table? 2. Source data and cutoff? 3. Definition of each column? 4. Change from prior period? 5. Limitations (thin data, one-time events)?

Assumption references:

Assumptions per: [Memo ID], version [X], effective [date]
Model: [Name], build [Y], run [timestamp]

Open items and escalation

Escalate to assumption-setting when enterprise assumption change is requested, not just application error.

Archive and retention

At actuary sign-off (or project end):

1. Freeze final workpaper and inputs manifest 2. Store in firm document system per retention policy 3. Record distribution list if sent externally 4. Delete scratch copies from shared drives or mark superseded

Do not archive PII in unsecured locations—follow firm privacy policy.

Common defects

Use data-visualization only after numbers are correct—charts do not fix logic errors.

Reporting and workpapers

Table of contents

1. Workpaper structure 2. Exhibit numbering and index 3. Roll-forwards and bridges 4. Statutory and management tie-outs 5. Memo support vs narrative 6. Version control 7. Pension and group benefit schedules

Workpaper structure

Recommended workbook tabs (adapt to firm standards):

Keep logic on calc tabs; exhibits link or paste values—avoid fragile cross-book links for filings.

Exhibit numbering and index

Example index columns:

Align numbering with actuarial memo outline. If memo not yet drafted, use placeholder titles actuary can rename.

Roll-forwards and bridges

Reserve roll-forward (calendar period):

Premium earned bridge (if in scope):

written premium
± change in unearned
= earned premium

Footnote reinsurance (gross vs net) on every bridge.

Statutory and management tie-outs

Build a mapping table:

Materiality: actuary sets threshold; analyst lists all differences with suggested classification (timing, reclass, error).

Do not force tie with unexplained plugs—use open items.

Memo support vs narrative

Use neutral labels in exhibits ("Selected factor" not "Correct factor").

Version control

For collaborative tools, export PDF or xlsx archive at sign-off milestone.

Pension and group benefit schedules

Analyst support for pension-retirement-funds / actuary valuations:

Do not certify funding adequacy or ERISA conclusions—prepare schedules only.

Reserving and loss development (analyst)

Table of contents

1. Data preparation 2. Triangle construction 3. Development factors 4. Ultimates and IBNR 5. Diagnostics checklist 6. Bridges and roll-forwards 7. Large losses and catastrophes 8. Analyst exhibits

Data preparation

Before building triangles:

1. Confirm valuation date and accident period definition (AY, policy year, report year) 2. Align paid, incurred (paid + case), and counts from same cutoff 3. Map reinsurance (gross, ceded, net) consistently across periods 4. Flag large losses per threshold in actuarial memo 5. Reconcile triangle totals to bordereaux or claims warehouse control report

Triangle construction

Standard layout:

• Rows — Origin periods (oldest at top or bottom—match prior workpaper)
• Columns — Development age (12, 24, … months or annual)
• Cells — Cumulative paid or incurred; incremental optional tab

Build separate tabs for:

• Paid losses
• Incurred losses
• Reported claim counts (if BF or count-severity support)
• Optional: incremental triangle for diagnostic charts

Development factors

Methods (execute per actuary direction—see actuary for judgment):

• Chain ladder — Cumulative age-to-age to ultimate
• Bornhuetter-Ferguson — Expected LR × % reported
• Cape Cod — Iterative blend when specified

Do not finalize method or tail without actuary sign-off.

Ultimates and IBNR

ultimate = reported_to_date + IBNR   (incurred basis, typical)
IBNR = ultimate - reported_incurred

Produce tables:

Compare implied ultimate loss ratio to pricing plan and prior memo; flag outliers in open-items list.

Diagnostics checklist

Run and retain screenshots or tabs for:

• [ ] Factor stability — Last N ages vs historical
• [ ] Calendar-year implied ultimates vs accident-year
• [ ] Paid vs incurred factor divergence
• [ ] Count × severity vs incurred ultimates (if applicable)
• [ ] Oldest origin immaturity — BF vs CL spread
• [ ] Outlier AYs — cat, large loss, one-time reserve releases

Document alternative factor sets with dollar impact on total IBNR.

Bridges and roll-forwards

Reserve roll-forward (analyst exhibit):

opening_reserve
+ incurred_losses (calendar period)
- paid_losses
+/- prior_year_development (if shown separately)
= closing_reserve

PYD bridge (change in prior AY ultimates):

prior_ultimate_at_old_valuation
- prior_ultimate_at_new_valuation
= PYD (sign convention per firm standard)

Tie every line to GL or statutory schedule when tie-out is in scope.

Large losses and catastrophes

Never delete cat years without actuary approval—flag distortion in diagnostics.

Analyst exhibits

Minimum reserving pack:

1. Data reconciliation to source 2. Triangles (paid, incurred, counts) 3. Factor development and selections 4. Ultimates and IBNR summary 5. Diagnostic charts or tables 6. PYD and roll-forward 7. Open questions for actuary

File naming example: 20241231_stat_P&C_CL_v03.xlsx