Icelandic personal-finance literacy: payslip and wage-rights math, the three-pillar pension system, verðtryggð mortgage modeling, séreignarsparnaður trade-offs, savings landscape, stacked staðgreiðsla tax math, and safety nets — with validated stdlib-only Python scripts.

GitHub stars
17

Install

/plugin install fjarmalalaesi@fjarmalalaesi

Run it inside Claude Code after adding the marketplace with /plugin marketplace add jokull/fjarmalalaesi.

Contents

Discovered from source

Skills 1

Fjármálalæsi — Icelandic Personal Finance

Iceland's personal-finance system is idiosyncratic: mortgage principals grow with CPI (verðtrygging), ~90% of workers are unionized with wage floors and benefit funds set by kjarasamningar, pensions come in three layers with very different tax and means-testing treatment, and tax on any extra income depends entirely on what it stacks on. Generic financial intuition — and generic calculators — get all of this wrong. This skill provides correct mechanics, validated modeling scripts, and the decision frameworks that actually matter.

Core mental models (read before modeling anything)

  1. Verðtryggð loans fight you with indexation. Each month: indexed_balance = balance × (1 + monthly_inflation), then the annuity payment is recalculated on the indexed balance over the remaining term. Early in a long loan, indexation typically exceeds scheduled principal repayment → the nominal balance rises for years ("höfuðstóllinn hækkar"). At ~4% inflation a fresh 40-year verðtryggt loan grows nominally for over a decade. This is not a bug in your model; it is the product.

  2. Never compare undiscounted nominal sums across decades. A "you save 47M over the loan's life!" figure mixes kronur from 2026 and 2049. Discount avoided-payment streams at the inflation assumption to get real, today's-money value. Rule of thumb from validated modeling: the passive value of a prepayment is nearly inflation-insensitive (higher inflation → more indexation avoided but a higher discount rate; they roughly cancel).

  3. Tax on extra income is a stacking question, never a flat rate. Staðgreiðsla is progressive per month. A séreignarsparnaður lump drawn on top of a top-bracket salary is taxed ~46%; the same pot spread over 15–25 retirement years stacks on pension income instead — anywhere from <20% (no other income, long drawdown) to ~38% (full career pension underneath). Always compute the stacked marginal rate with scripts/tax.py stacked.

  4. Passive vs snowball paydown is usually the deciding assumption. A lump-sum prepayment where the freed monthly payment is spent (passive) is worth far less than one where freed payments are recycled into extra principal every month (snowball) — in a validated real-household case, 28M vs 42M in today's kronur for the same 16M lump. The difference is behavioral discipline over ~16 years, not finance. Model both; present both; say which assumption the verdict rests on.

  5. Prioritize the highest-APR loan. Icelandic households often carry two mortgages (base loan + viðbótarlán at a higher rate). Lump sums and snowball extra go to the highest APR first. The scripts do this automatically.

  6. The payslip is verifiable arithmetic — verify it. Deduction order: lífeyrir 4% and séreign (pre-tax) → staðgreiðsla → stéttarfélagsgjald (post-tax). The 2% séreign employer match is a 100% instant return on the matched money — always ask whether it's being captured. The employer's true cost is ~120.8% of gross from scripts/payslip.py (pension + tryggingagjald + VIRK), ~122.4% including the union sjóðir (see unions-and-wages.md §2) — the "know your worth" number for negotiation and contractor comparisons.

  7. Kjarasamningar set floors for everyone (lög 55/1980 — member or not), so wage tables are floors, never benchmarks. Benchmark against Hagstofa's launarannsókn medians/percentiles (scripts/wages.py) using the right wage concept: regluleg laun for a plain monthly offer, heildarlaun for total-comp claims.

  8. Pension pillars have wildly different treatment. Frjáls séreign: TR-clawback-free, heritable, free from 60. Tilgreind séreign: counts against TR (since 2023), forced spread 62–67. Samtrygging: lifelong annuity + invisible disability insurance, but pays your estate nothing. Never say just "séreign" in a recommendation — qualify which.

  9. Kill expensive credit before saving. Overdraft/credit-card revolving runs ~15.5%, BNPL effectively 17–40%+; savings earn ~7% before tax. And the 300k/yr frítekjumark shelters interest on roughly the first ~4M kr per person — plan account ownership around it.

Scripts

All scripts are Python 3 stdlib-only, in scripts/. Each has --help and --json for machine-readable output. Run them; don't re-derive the math by hand — the mortgage formulas were validated against real Arion Banki loan data during development, and the payslip math verifies live against payday.is (payslip.py verify).

ScriptPurpose
verdtrygging.py scheduleFull amortization of one verðtryggt loan (annuity/jafngreiðslur or equal-principal/jafnar afborganir), with optional lump sum or fixed extra monthly principal
verdtrygging.py lumpsumReal (PV) value of a one-time prepayment, passive assumption
verdtrygging.py snowballMulti-loan: lump applied by APR priority + freed payments recycled into principal; PV and payoff acceleration
lanareiknivel.py projectCalendar-dated projection of one or more verðtryggð loans: real CPI history + a bracketed inflation path (per calendar year), three scenarios, rate steps, extra-principal plans, séreign-to-principal with the program cap, yearly tables, milestones (peak balance, 50% paid, payoff). The engine behind lanareiknivel.solberg.club, verified 1:1
lanareiknivel.py equityProperty value vs debt over time: eigið fé, LTV vs market and vs fasteignamat, and real-terms equity erosion under each scenario
lanareiknivel.py linkURL that opens the same inputs in the interactive web UI
namslan.py old/newNámslán: lump-sum payoff vs let it run. Old LÍN loans (income-linked, 0.4% real, 5–15% prepayment discount, cancelled at death) and Menntasjóður H-lán (4% cap, no discount): real IRR of prepaying, cash-flow effect, ranked against the mortgage and taxed savings
portfolio.pyHistory + projection of the whole mortgage portfolio: exact balance reconstruction from the bank's payment export (Arion base-index kronur convention, verified to the krona against Staða láns), what the séreign and manual innborganir actually did, the variable-rate history implied by the interest column, property value from HMS vísitala íbúðaverðs anchors, equity/LTV monthly from purchase to payoff in nominal and real terms
tax.py monthlyStaðgreiðsla on a month's income (2026 brackets, persónuafsláttur)
tax.py stackedEffective marginal rate of extra income stacked on existing income — the tool for séreign withdrawal modeling
paydown_vs_invest.pyFull comparison: draw down a pot now (taxed, stacked) and pay down mortgages (passive AND snowball) vs leaving it invested until retirement (growth, stacked drawdown tax, discounted to today)
cpi.pyLive CPI from Hagstofa's PX-Web API — latest gives the current VNV til verðtryggingar value and YoY inflation; use the indexation series (M+2 lag already applied) for loan models
payslip.py grossFull launaseðill from gross salary: lífeyrir, séreign, staðgreiðsla, stéttarfélagsgjald → take-home, PLUS the employer side (mótframlag, tryggingagjald → total cost, the "know your worth" number). Verified field-for-field against the payday.is calculator
payslip.py total-costInverse: employer total cost → gross → payslip (for contractor-vs-employee and salary-negotiation framing)
payslip.py verifyDiff the local math against the live payday.is calculator across a salary list — run after each January bracket update
wages.py groups/percentiles/occupationLive salary benchmarks from Hagstofa's launarannsókn (medians by occupation group, P10–P90 distributions, fine-grained ÍSTARF95 lookup). Data lags ~5 months — uprate by the Jan 1 contractual raise

Typical session for "should I use savings/pension to pay down my mortgage?":

# 1. What's the tax hit if drawn now? (stacks on current salary!)
python3 scripts/tax.py stacked --withdrawal <pot> --other-income <monthly_salary>

# 2. Full comparison across both paydown styles and the invest alternative
python3 scripts/paydown_vs_invest.py \
    --loans <bal:apr:months>,<bal:apr:months> \
    --pot <gross_pot> --tax-rate-now <from step 1> \
    --inflation 4.3 --fund-return 7.0 --horizon-years <years_to_retirement> \
    --retirement-other-income <expected_pension_per_month> --drawdown-years 15

# 3. Sensitivity: rerun step 2 varying horizon (5/10/15/20), fund return
#    (4.5–8.5%), and retirement income (0/300k/600k). Present as a table.

Interview the user for: loan balances/APRs/remaining months (from their bank's netbanki), gross monthly salary, years to retirement, expected pension income (TR + lífeyrissjóður — Lífeyrisgáttin or the fund's estimator), and current inflation forecast (see below).

Two modeling notes: (1) óverðtryggt loans are the same engine with --inflation 0 and the nominal rate — use that for verðtryggt-vs-óverðtryggt payment-path comparisons; (2) TR skerðing is not inside paydown_vs_invest.py — frjáls séreign is exempt, but for any other retirement income stream apply tax.py tr-clawback on top (45% above the frítekjumark; see tax-and-benefits.md §9).

"What will my verðtryggt loan / equity look like?" (lanareiknivel.py)

Pick the engine by the question: verdtrygging.py values a decision (prepay or not, in today's kronur); lanareiknivel.py projects the trajectory (balance, payments, equity, LTV by calendar year, under an explicit inflation path). Model details: references/verdtrygging.md §9.

Gather inputs yourself; do not build parsers. Everything needed is on the loan-overview page of any netbanki (Arion, Landsbankinn, Íslandsbanki, lífeyrissjóðir), on a greiðsluseðill, or in the bank's payment-history export — read the screenshot, PDF or xlsx directly and map the fields:

InputIcelandic label to look forNotes
balanceEftirstöðvar / Staða láns / Uppreiknaðar eftirstöðvarThe indexed balance as of the statement — not the original lánsfjárhæð
aprVextir / VaxtakjörThe loan's real rate on verðtryggt loans (e.g. 4.49%); note whether fastir (fixed until when → rate_changes) or breytilegir
remaining_monthsGjalddagar eftir / Lokagjalddagi / LánstímiConvert a date to months from start_month
methodGreiðslumáti: jafngreiðslur (annuity) / jafnar afborganir (equal principal)Annuity is the default on Icelandic mortgages
start_monthStatement dateThe month the balance belongs to
pension_monthlySéreignarsparnaður → inn á höfuðstólOnly if enrolled; the cap is per person per year (pension_cap)
extraThe user's planAmount/month from a year for N years; ask, don't assume
property.value / fasteignamatFasteignamat (skatturinn.is / fasteignaskra), market estimateLTV rules use fasteignamat; equity uses market value — get both
inflation_pathPeningamál Tafla 4Forecast years, then 2.5% terminal (see Inflation assumptions)

Write a loans.json (schema in lanareiknivel.py --help), then:

python3 scripts/lanareiknivel.py project --loans-file loans.json          # scenarios, yearly table, milestones
python3 scripts/lanareiknivel.py equity  --loans-file loans.json \
    --property-value <market> --fasteignamat <mat> --property-growth 4.5
python3 scripts/lanareiknivel.py link    --loans-file loans.json          # interactive version for the user

When the user has their bank's payment history (Arion netbanki "Lán → Greiðslur", the LoanPayments xlsx export, or screenshots of it), use portfolio.py instead of guessing balances: read the rows yourself, write them into history.json as exported (schema in portfolio.py --help), and run it. It reconstructs every month's balance exactly, verifies against "Staða láns", and prints what the séreign and manual innborganir actually achieved, the variable-rate history implied by the interest column, and the equity/LTV path from purchase to payoff. Read references/verdtrygging.md §10 first: the export's amounts are base-index kronur and its verðbætur columns are not the balance indexation — an agent that sums columns will be off by millions. For the property side set property.index_series to the HMS sub-index matching the home (fjolbyli_hbs, serbyli_hbs, fjolbyli_landsbyggd, serbyli_landsbyggd, hbs, landsbyggd, all) and the script fetches HMS's monthly vísitala íbúðaverðs itself; index_anchors is the offline fallback.

python3 scripts/portfolio.py --history history.json          # report; --json for the monthly series to chart

Rerun equity with --property-growth 0 and with growth = the inflation path: the gap between those two is the real equity story.

"Should I pay off my námslán in one go?"

First establish the loan class from Mitt Lán (letter: V/S/R/G = old LÍN loan, H = Menntasjóður). Then:

# Old loan: the 5–15% discount vs the 0.4% real rate, ranked against the mortgage
python3 scripts/namslan.py old --class G --balance <eftirstöðvar> --income <tekjustofn> \
    --age <age> --prepay all --mortgage-real-rate <mortgage APR> --savings-nominal 7
# H-lán: prepaying earns the loan rate (4% real today), no discount
python3 scripts/namslan.py new --balance <eftirstöðvar> --months <remaining> --prepay <amount> \
    --mortgage-real-rate <mortgage APR>

Present the ranking the script prints (mortgage, H-lán, prepay-IRR, taxed savings, keep-old-loan) and the cash-flow line: partial prepayment on an income-linked loan changes no monthly payment. Add the non-return reasons from namslan.md §5 (greiðslumat, death) when they apply. Use --json when you are going to chart or tabulate the output yourself. When the user asks "should I prepay", hand the same loans to verdtrygging.py / paydown_vs_invest.py — do not read a verdict off nominal totals.

Inflation assumptions

For current inflation and the live indexation series, run python3 scripts/cpi.py latest. For forward assumptions use Seðlabanki's Peningamál forecast (published Feb/May/Aug/Nov), Tafla 4 (annual averages) — not stale third-party numbers. The Hagstofa þjóðhagsspá API can lag the latest Peningamál by weeks or months; cross-check against the PM PDF at sedlabanki.is. For long horizons beyond the forecast window, 2.5% (the inflation target) is the standard terminal assumption; run sensitivity at ±1.5pp regardless.

Reference documents

Load these only when the question requires depth in that area:

  • references/verdtrygging.md — legal basis (lög 38/2001), index mechanics, vísitala publication schedule, verðtryggt vs óverðtryggt trade-offs, 40-year-loan restrictions, current rate environment, Hagstofa CPI API
  • references/sereignarsparnadur.md — third-pillar pension rules, the séreign-to-mortgage tax-free redirect program (reinstated June 2026; applications from Oct 1 2026, retroactive to Jan 1 2026; 120-month cumulative cap), fyrsta íbúð program, withdrawal taxation, TR interactions
  • references/mortgage-market.md — lender landscape (banks, lífeyrissjóðir, HMS hlutdeildarlán), current rates, Seðlabanki LTV/DSTI rules, refinancing costs and mechanics, fasteignamat, prepayment mechanics
  • references/tax-and-benefits.md — staðgreiðsla brackets, álagning cycle, vaxtabætur status, rental income (incl. Airbnb/heimagisting), capital gains on housing, fasteignagjöld, gifts/arv toward down payments, TR skerðingar
  • references/unions-and-wages.md — the union system (what membership buys, sjúkrasjóðir, starfsmenntasjóðir), exact payslip mechanics and employer-side costs, employee rights (veikindaréttur, uppsagnarfrestur, yfirvinna, jafnaðarkaup pitfalls, gerviverktaka), salary benchmarking sources, negotiation practicals incl. skattmat hlunnindi values
  • references/pension-system.md — the three pillars, age-priced accrual, tilgreind vs frjáls séreign (TR treatment differs!), fund/provider comparison incl. fees, Lífeyrisgáttin, abroad/gaps/divorce/death rules
  • references/financial-institutions.md — banks vs indó/Auður, deposit insurance (TVF), savings rates and the frítekjumark real-return math, government bonds, funds and the cheap index options, investing abroad, crypto tax, expensive-credit traps and complaint routes
  • references/namslan.md — námslán payoff decisions: the two regimes (old LÍN loans at 0.4% real with income-linked repayment, a 5–15% prepayment discount and cancellation at death; H-lán at the 4%/9% caps with no discount), the IRR framework, non-return reasons (greiðslumat, death), inputs to collect. Load for any "should I pay off my student loan" or verðtryggt-vs-óverðtryggt H-lán question
  • references/safety-nets-and-life-events.md — fæðingarorlof math and pitfalls, atvinnuleysisbætur, the illness income ladder, Menntasjóður, barnabætur/meðlag/childcare costs, framfærsluviðmið (what greiðslumat uses), debt trouble (UMS, greiðsluaðlögun, Creditinfo), renting rights
  • references/tax-law-navigation.md — which statute governs what and how to verify any tax number: the lög 90/2003 article map, the annual fjárlög/bandormur cycle, yskn.is/bindandi álit escalation, the 3-step verification workflow and its traps. Load when citing or verifying tax rules rather than just values
  • references/icelandic-data.md — where to get reliable numbers: APIs and official sources for CPI, forecasts, policy rates, lender rates, fasteignamat, tax figures, and legislation, with each source's lag traps. Load this before fetching any external data or citing current values
  • references/practical-moves.md — verified, quantified money moves (FX-cheap cards, the séreign match, unclaimed union grants, frítekjumark account-splitting, fee-free prepayment windows, time-sensitive deadlines). Comes with presentation rules: always relay the number, the condition, and the verify-by source, and compute the value for the user's own situation — never present as generic "hacks"
  • references/glossary.md — Icelandic ⇄ English term glossary with usage gotchas (gjalddagi vs eindagi, afborgun vs innborgun, ráðstöfun vs úttekt). Load whenever writing Icelandic or translating financial terms

Guardrails

  • Always run sensitivity on inflation, fund return, and horizon before giving a verdict. Single-scenario answers on 20-year questions mislead.
  • State the behavioral assumption (passive vs snowball) any paydown verdict rests on.
  • Bracket table staleness: tax.py embeds 2026 values; verify against skatturinn.is each January (the script's --year flag fails loudly), then run payslip.py verify as a regression check.
  • Legal availability ≠ modeling: séreignarsparnaður cannot normally be drawn before retirement age outside the housing-redirect programs. Flag hypotheticals as hypothetical.
  • Official pages lag new laws — a Skatturinn info page is never proof a scheme doesn't exist (see the verification workflow in tax-law-navigation.md).
  • Námslán: the discount is the question on old loans, the rate is the question on H-lán. Never say "pay off debt" generically — an old G/R-lán at 0.4% real is the cheapest money in Iceland and is cancelled at death; run namslan.py and rank against the mortgage.
  • Point users to their union's kjaramál desk for payslip disputes and wage claims — it's free and it's what dues pay for.
  • This is decision-support modeling, not financial advice; recommend users verify big moves with their bank, union, or a licensed advisor.

Sponsors

Details

Marketplace
jokull/fjarmalalaesi
Category
Finance
License
MIT
Keywords
iceland, fjármálalæsi, verðtrygging, mortgage, húsnæðislán, séreignarsparnaður, launaseðill, pension, kjarasamningar