Technical reference · Methodology

This is the complete technical reference for the simulator’s model: every stochastic component it uses, set down with its distribution, its formula, and the parameters behind it. It assumes you are comfortable with the notation below. For the same ground in plain language, see How it works; for what the model deliberately leaves out, see Assumptions & limits.

Notation

$N$ simulated paths (10,000 by default), $H$ years in the horizon, age $x$ , year index $t$ . Portfolio balance $B_t$ ; portfolio return $r_t^{\text{port}}$ , blended from equity, bond, and cash returns $r_t^{\text{eq}}, r_t^{\text{bond}}, r_t^{\text{cash}}$ with weights $w_{\text{eq}}, w_{\text{bond}}, w_{\text{cash}}$ ; annual management fee $\tau_{\text{mkt}}$ . Tax is a function of the year’s whole household position, written $T_t(g)$ for a gross withdrawal $g$ (§12); the single flat rate $\tau_{\text{port}}$ survives only as an override for a reader who prefers their own effective rate. Inflation rate $\pi_t$ (the inflation during year $t$ ), cumulative price index $\text{CPI}_t = \prod_{u<t}(1+\pi_u)$ , the price level at the start of year $t$ , so $\text{CPI}_0 = 1$ . All money is tracked in cash-of-the-day terms. Money spent or received during year $t$ is priced at $\text{CPI}_t$ , so the first year is in today’s money, and its today’s-money value divides by $\text{CPI}_t$ ; a balance at the end of year $t$ divides by $\text{CPI}_{t+1}$ .

1. How a simulation runs

The simulator follows one household across $N$ independent possible futures, year by year, from today to the last death in each future. Each path is its own roll of the dice: its own sequence of market returns, inflation, lifespans, and shocks. The numbers you see on the results page are summaries across all $N$ paths.

Every year runs through one cash-flow model: income and expenses meet and the portfolio funds any gap. Only when you let spending follow the markets does a withdrawal rule cap that draw and spending get cut when money is tight. What distinguishes the years before retirement is a coverage assumption: your expense inputs describe your retirement plan, not your working-age budget, so pre-retirement the model injects a synthetic coverage income that closes the gap in normal times. The household is assumed to live off its earnings; only your stated contributions and one-off lump sums (an inheritance, a severance payment) are added to the portfolio. Steady income like rent is assumed consumed rather than auto-invested, because whatever you do save from it is already part of your stated contribution.

In a covered year the balance update therefore reduces to simple compounding plus new savings:

B_{t + 1} = B_{t} (1 + r_{t}^{port}) + C_{t}

(1.1)

where $C_t$ is the year’s contribution: entered in today’s euros, grown with inflation and any real raise, and (only when job-loss risk is enabled) paused in years the person is out of work. From retirement onward the coverage assumption ends. The expense inputs become the authoritative spending plan, and every income stream (pensions, rent, a still-working partner’s salary) offsets the drawdown, with any genuine surplus invested.

What happens each year

Within a single year, the model resolves events in a fixed order, because later steps depend on earlier ones (a death changes expenses; inflation sets the price level before costs are computed):

Mortality: determine who, if anyone, dies this year.
First-death transitions (couples): when one partner dies, permanently adjust the survivor’s expenses and pensions.
Long-term care: test for the onset of a care episode, and age any episode already under way.
Job loss: before retirement, test for an employment shock (only when job-loss risk is enabled; it is off by default).
Inflation: advance the price level by last year’s inflation, so this year’s costs and income are priced as the year begins (the first year is in today’s money). This year’s inflation, which moves together with this year’s investment return, prices next year’s.
Investment return: draw the blended portfolio return.
Expenses: compute all seven spending buckets at this year’s prices.
Income: compute all income streams.
Withdrawal ceiling: only when spending follows the markets (a Premium switch, off by default), the withdrawal rule caps how much may be drawn. Otherwise the draw is whatever the plan needs.
Spending cuts: only under that switch: if income plus the allowed draw cannot cover expenses, cut spending in priority order down to its floors.
Balance update: settle the year: subtract the net withdrawal, apply the return, add contributions and any surplus.

The rest of this reference takes these components one at a time.

2. Investment returns

The portfolio holds three assets (equities, bonds, and cash) in weights that sum to one. Its return in any year is the weighted blend of the three asset returns, less a flat annual management fee:

r_{t}^{port} = w_{eq} r_{t}^{eq} + w_{bond} r_{t}^{bond} + w_{cash} r_{t}^{cash} - τ_{mkt}

(2.1)

Allocation: fixed or glidepath

You can hold a fixed allocation (by default 60% equities, 40% bonds, 0% cash) or follow a glidepath, a schedule that shifts the mix as you age. Under a glidepath the equity share is read off a line through your chosen age-and-weight points: interpolated between them and held flat beyond the ends. Whatever is not in equities is split between bonds and cash in a fixed ratio (by default 80% to bonds). For a couple, the allocation follows the first person’s age.

The distribution of returns and inflation

Each year’s equity, bond, cash, and inflation outcomes are drawn together from one joint distribution, so they move in a correlated way: equities and inflation lean against each other, cash and inflation lean together, and so on. The distribution is defined by each variable’s mean and volatility and by the correlations between them.

Two shapes are available. The default is a multivariate normal: each outcome is its mean plus a correlated random shock,

y = μ + ε, ε \sim N (0, Σ),

(2.2)

where $\mu$ collects the means and $\Sigma$ the variances and correlations. The alternative is a Student-t distribution, which keeps the same means and correlations but gives fatter tails: extreme good and bad years happen more often than the normal would predict. Its heaviness is set by a degrees-of-freedom parameter $\nu$ (lower $\nu$ = fatter tails); it is scaled so that the variances and correlations are preserved exactly.

Default market assumptions (annual)

Parameter	Default	Meaning
Equity return	8% ± 17%mean ± vol	Long-run equity mean and volatility
Bond return	4.5% ± 8%mean ± vol	Long-run bond mean and volatility
Cash return	3% ± 3%mean ± vol	Short-term/cash mean and volatility
Inflation	2.5% ± 3%mean ± vol	General CPI mean and volatility (default model)
Management fee	0.3%per year	Flat drag on the blended return
Default allocation	60 / 40 / 0eq / bond / cash %	Fixed allocation when no glidepath is set
Student-t tails	ν = 6	Degrees of freedom when fat tails are enabled (min 3)

3. Inflation: regime-switching

By default, inflation is one element of the joint draw above. An optional regime-switching model replaces it with something more realistic over long horizons: inflation alternates between a calm regime and a high regime, and persists in each.

A hidden two-state process governs which regime each year is in. From a calm year, inflation usually stays calm; from a high year, it usually stays high; occasionally it switches. The starting regime is drawn from the model’s long-run balance of the two.

P = [p_{N N} 1 - p_{H H} 1 - p_{N N} p_{H H}]

(3.1)

Here $p_{NN}$ is the chance a calm year is followed by another calm year (default 93%) and $p_{HH}$ the chance a high year stays high (default 80%). Within each regime, inflation is drawn around a regime-specific average (calm around 2.2%, high around 7%) with its own spread.

π_{t} ∣ calm \sim N (μ_{N}, σ_{N}^{2}), π_{t} ∣ high \sim N (μ_{H}, σ_{H}^{2})

(3.2)

Returns respond to inflation

The point of this model is that investment returns react to inflation. Once the year’s inflation is known, each asset’s return is drawn around its mean, tilted by how far inflation landed from its regime average:

r_{t} = μ_{r} + ρ \frac{σ _{r}}{σ _{π}} (π_{t} - μ_{π}) + σ_{r} 1 - ρ^{2} Z, Z \sim N (0, 1)

(3.3)

The sensitivity $\rho$ differs by asset, and for equities it is stronger in the high-inflation regime (a deeper negative link) than in the calm one. In this mode the assets are tied to one another only through their shared response to inflation, rather than through direct asset-to-asset correlations.

4. Mortality

Each person’s age at death is drawn once, at the start of their path, from a Gompertz distribution, the standard actuarial shape in which the chance of dying rises roughly exponentially with age. The distribution is pinned by two numbers: a mode $m$ (the most likely age at death) and a dispersion $b$ (how spread out it is). Drawing a uniform random number $U$ and inverting the survival curve gives the years remaining:

T_{death} = b ln (1 - ln (U) e^{- (x - m) / b}), U \sim Unif (0, 1)

(4.1)

with a hard cap at age 120.

The random number is drawn once, but the resulting age at death is not immutable: two in-simulation events can pull it earlier. Entering a care episode raises the death hazard for the episode’s duration (§6), and for couples the loss of a partner applies the widowhood adjustment below. Both work by re-solving the same survival curve with the elevated hazard. No new randomness enters, so comparisons across scenarios stay apples-to-apples.

The mode and dispersion are set per country and sex. They are approximate figures that roughly match each country’s life expectancy, not yet fitted to its official life tables. Self-reported health then shifts the mode: better-than-average health pushes the likely age at death later, worse health pulls it earlier.

Gompertz parameters (mode, dispersion) by country and sex

Parameter	Default	Meaning
Greece (male / female)	(84.6, 10.7) / (88.5, 9.9)years	Most likely age at death, and spread
United States (male / female)	(84.8, 9.5) / (88.2, 10.5)years	Most likely age at death, and spread
Health adjustment	excellent → poor	Shifts the mode earlier or later by self-reported health
Maximum age	120years	Hard cap on any drawn lifespan

The widowhood effect

For couples, losing a partner shortens the survivor’s own expected life. In the year the first death actually occurs, the model brings the surviving partner’s age at death forward by a fixed amount, a deliberately simple stand-in for the well-documented spike in mortality after bereavement. The shift is on the order of one to two years for typical dispersions, and the survivor is always left with at least one more year than the partner who died first. If a bereaved survivor later enters a care episode, the care-driven recalculation of their lifespan (§6) is done first and the widowhood shift is then re-applied on top. That layering is a documented approximation rather than a joint model of the two hazards.

5. Couples

Correlated lifespans

Partners’ lifespans are not independent: shared environment, lifestyle, and age tend to make them die closer together than chance alone would suggest. The model captures this with a Frank copula, which links the two random draws that set each partner’s death so that they are positively associated, while still letting each partner’s age at death follow its own Gompertz curve. A single parameter controls how tightly the two lifespans are coupled.

What changes at the first death

A couple’s path moves through four states: both alive, either partner alone as the survivor, or both gone. The first death is a turning point, and the model makes these permanent adjustments to the survivor’s finances:

Expenses step down, bucket by bucket, because one person costs less than two, though not proportionally and not for everything:

Survivor expense levels after the first death (share of the couple's level)

Parameter	Default	Meaning
Essential	60%of couple	Food, utilities, day-to-day basics
Discretionary	55%of couple	Travel, leisure, non-essentials
Healthcare	55%of couple	One person's care rather than two
Housing	100%of couple	Unchanged; see below
Long-term care	per person	Already individual; not scaled

Housing costs are left unchanged after a death. A mortgage payment, property taxes, and maintenance on the same home do not fall just because one resident is gone, so the model holds them at the couple’s level rather than discounting them.

Pensions are reduced. Each partner’s pensions are personal: when a partner dies, their pension streams continue to the survivor only at the stream’s survivor share, while the surviving partner’s own pensions are untouched. The share the input step prefills is each country’s statutory rule resolved to an effective share for a typical dual-pensioner couple: Greece’s 70% halves once the survivor draws their own pension, the Netherlands pays no inheritance but raises the survivor’s own pension to the single rate, Sweden and Norway pass on nothing at all to retirement-age survivors. The per-country table in §10 lists every rule with its source.

Two countries follow a different mechanism entirely. In the United States and Ireland the survivor does not keep their own pension plus a share of the deceased’s; the household keeps the larger of the two main pensions. The model implements this directly: at the first death the deceased’s main stream is reduced so that the household’s main-pension income equals the larger of the two. For an equal-earning couple that means the survivor inherits nothing on top, which a fixed share would badly overstate. This computation uses the two entered amounts, so it is exact while both main pensions follow the same uprating path, which both countries’ price-linked rules satisfy; a manually diverged indexation makes it an approximation. Spousal top-ups paid while both partners are alive are not modeled.

Accounts pass to the survivor. Each account belongs to one partner or to both. The survivor takes over the accounts of the partner who died, and from then on their access ages and required withdrawals follow the survivor’s age, as the tax on the survivor’s income does. Some rules change the money itself:

In the United States, Britain, Ireland, France, Spain, Portugal, Italy and Finland, the gain built up in the ordinary investments the partner owned is wiped for tax: the survivor’s cost becomes their value on the day of death, down as well as up. On a joint account that is the deceased’s half, or the whole account in the nine US community-property states (Arizona, California, Idaho, Louisiana, Nevada, New Mexico, Texas, Washington and Wisconsin), where shared property resets in full. Elsewhere the cost carries over to the survivor (Canada, Australia, Germany, Austria, Belgium, Sweden, Norway and Denmark), or gains on ordinary investments are not taxed at all.
In Britain, a pension left by a partner who died before 75 pays the survivor free of income tax. From 75 it is the survivor’s pension income, as before.
Some savings wrappers end at their holder’s death, and the law taxes them then or not at all. A French PEA closes and pays the social levies (18.6%) on its gain, and an assurance-vie pays the partner outside income and inheritance tax, with the social levies (17.2%) on its gain; a Finnish equity savings account and savings insurance pay out free of income tax. Each is paid into the survivor’s ordinary account at its value that day, which becomes its cost. A Finnish capitalisation contract carries on with its cost reset to that value. A UK or Irish investment bond written on the deceased’s life alone ends and its gain is taxed: 38% in Ireland, and in Britain the basic rate on the deceased’s last return, which leaves nothing more to pay on a UK bond and 20% on an offshore one. A bond the couple owns together is read as written on both lives and runs on to the second death. Any tax taken this way counts in the year’s bill as tax owed before any withdrawal.

Everything after the second death, such as inheritance tax or what heirs pay on the accounts they receive, is outside the plan: the money left at the end is shown before any of it.

6. Long-term care

Long-term care is modeled as its own spending bucket with its own onset, separate from ordinary healthcare. In any year, a person who is alive and not already in care can enter a care episode. The annual chance of onset rises with age along a Gompertz curve anchored at 65 and set slightly higher for women than men:

h_{ltc} (x) = a_{ltc} \cdot exp (β_{ltc} (x - 65)), p_{onset} (x) = 1 - e^{- h_{ltc} (x)}

(6.1)

When an episode begins, its length is drawn from a lognormal distribution (a short-skewed shape: most episodes are brief, a few run for years), rounded up to whole years. The rounding leans long: at the default settings it adds about half a year to the average episode, and the care cost and the raised death hazard below both run over the rounded length. While the episode is active, the long-term-care bucket carries that year’s care cost; when it ends, the bucket returns to zero. If the person dies mid-episode, both the episode and its cost end with them.

A care episode also raises the person’s mortality for its duration. Concretely: at onset, the person’s death hazard is multiplied by ×4 for the episode’s length, then returns to the normal age curve if they recover. Their age at death is re-derived under that elevated-hazard window from the same random draw that set it originally (a closed-form recalculation on the Gompertz curve of §4, not a re-roll). Someone drawn to die at 90 who enters care at 80 will typically die years earlier; someone who recovers from a short episode keeps most of their remaining life. Repeat episodes compound correctly. This matters for the headline number: without it, care paths would pay care costs over a full unshortened lifetime, systematically overstating cumulative care expense (the model’s scariest cost) and understating the chance of the money lasting for care-exposed plans.

By default the full market cost of care applies. If the family-care option is enabled, each episode is additionally tagged (with a country-specific probability) as one where family provides care, in which case a much lower cost applies for that episode instead.

What care costs, and who pays it

The cost of care is the single most country-dependent number in the model, and not because care is cheaper in some places. It is because the share of the bill that reaches the household differs enormously. In the United States a resident with savings pays the full private rate. In Denmark the care itself is free by law and what a resident pays is essentially rent and food. Those are not small differences of degree: they are the difference between a risk that can end a plan and one that barely registers.

Every figure below is what the household itself pays in a year, in today’s money and the country’s own currency, for a household that has retirement savings. That qualification does real work. Where help is means-tested, having savings is what disqualifies you from it, so the modelled household pays the full rate. Where care is funded for everyone against a capped fee, savings change nothing and the capped fee is what you pay.

What a year of residential care costs you, by country

Country	Per year	Who pays	Where the figure comes from
Norway	NOK 285,918	Care is funded by the municipality, and you pay a share of your income for it: 75% up to a threshold and 85% above it, covering room, meals and medicines. The figure is that charge at an assumed income after tax of NOK 362,100 a year in today's money, the median for people 67 and over. The simulation charges it as a flat amount for as long as care lasts and does not recompute it from your own income.	Lovdata / Helse- og omsorgsdepartementet (Norway)sourcechecked 2026-08-05
Denmark	DKK 188,400	The state funds the care. You pay a capped fee plus your room and meals.	Borne-, Aeldre- og Boligministeriet (Denmark)sourcechecked 2026-08-05
Sweden	SEK 153,564	The state funds the care. You pay a capped fee plus your room and meals.	Socialstyrelsen (Sweden)sourcechecked 2026-08-05
United States	$129,575	State help is means-tested, and a household with retirement savings pays the full price. The simulation charges that price for as long as care lasts and never switches to state help as your savings run down.	Genworth Financial / CareScoutsourcechecked 2026-08-05
New Zealand	NZ$84,990	State help is means-tested, and a household with retirement savings pays the full price. The simulation charges that price for as long as care lasts and never switches to state help as your savings run down.	Ministry of Health, New Zealandsourcechecked 2026-08-05
Switzerland	CHF 80,946	Costs are split between the public system and you, by type of cost.	Bundesamt fuer Gesundheit (BAG)sourcechecked 2026-08-05
Australia	A$71,639	Costs are split between the public system and you, by type of cost.	Australian Government Department of Health, Disability and Ageingsourcechecked 2026-08-05
United Kingdom	£67,496	State help is means-tested, and a household with retirement savings pays the full price. The simulation charges that price for as long as care lasts and never switches to state help as your savings run down.	Department of Health and Social Care (GOV.UK)sourcechecked 2026-08-05
Ireland	€51,050	This is not the price of a place. Under the Fair Deal scheme you pay 80% of your income plus 7.5% a year of your assets, and the state pays the rest; the figure is that contribution for an assumed single retiree with €25,000 a year of income, a €350,000 home and €100,000 of other savings, in today's money. The simulation charges it as a flat amount for as long as care lasts: it does not recompute it from your own income and savings, or drop the home from the sum after three years as the scheme does.	Health Service Executive (HSE), Irelandsourcechecked 2026-08-05
Germany	€40,368	Public care insurance pays a fixed amount by care level. You pay the rest.	Verband der Ersatzkassen e. V. (vdek)sourcechecked 2026-08-05
Netherlands	€36,742	The state funds the care. You pay a capped fee plus your room and meals.	CAK (Centraal Administratie Kantoor)sourcechecked 2026-08-05
Austria	€34,776	The state funds the care home, and you pay 80% of your pension and hand over the care allowance (Pflegegeld) it pays you, keeping a small personal allowance. The figure is that contribution at an assumed pension of €2,586 a month before tax in today's money, with the allowance counted in although the simulation does not pay it to you. It is charged as a flat amount for as long as care lasts and is not recomputed from your own pension.	oesterreich.gv.at (Republic of Austria, Bundeskanzleramt)sourcechecked 2026-08-05
Spain	€27,961	State help is means-tested, and a household with retirement savings pays the full price. The simulation charges that price for as long as care lasts and never switches to state help as your savings run down.	European Commission (Social Protection Committee / DG EMPL)sourcechecked 2026-08-05
Belgium	€27,200	Costs are split between the public system and you, by type of cost.	Departement Zorg, Flemish Government (Belgium)sourcechecked 2026-08-05
France	€25,968	Costs are split between the public system and you, by type of cost.	Caisse nationale de solidarite pour l'autonomie (CNSA)sourcechecked 2026-08-05
Canada	CA$25,550	The state funds the care. You pay a capped fee plus your room and meals.	Government of Ontario, Ministry of Long-Term Caresourcechecked 2026-08-05
Finland	€25,400	Care is publicly funded, and you pay rent plus a fee of up to 85% of your net income, keeping at least €195 a month. The figure is that outlay at an assumed net income of €2,366 a month in today's money. The simulation charges it as a flat amount for as long as care lasts and does not recompute it from your own income.	Pirkanmaan hyvinvointialue (Finland)sourcechecked 2026-08-05
Portugal	€22,224	State help is means-tested, and a household with retirement savings pays the full price. The simulation charges that price for as long as care lasts and never switches to state help as your savings run down.	OneValue, Government of Portugalsourcechecked 2026-08-05
Italy	€19,164	Costs are split between the public system and you, by type of cost.	Altroconsumo (Euroconsumers)sourcechecked 2026-08-05
Greece	Not publishedwe ask you instead	State help is means-tested, and a household with retirement savings pays the full price. The simulation charges that price for as long as care lasts and never switches to state help as your savings run down.	No statistics office, regulator or ministry publishes one.

Amounts are what the household pays out of its own pocket in a year, in today's money and the country's own currency, for a household that has retirement savings. That last part matters: where help is means-tested, savings disqualify you from it, and where care is funded for everyone, the charge is capped or set as a share of income. The simulation charges each figure as a flat amount for as long as care lasts; it never switches to state help as savings run down, and where the charge follows income it is worked out at an assumed income, not yours.

The same figures, one country at a time and alongside everything else the model assumes there, are on your country’s page.

Why the model does not bill you the whole figure

Each of those charges buys a room and meals as well as care. Your plan is already paying for a roof and for groceries in its housing and essential spending, and the model keeps paying for both throughout a care episode. Billing the published charge on top would charge you twice for the same bed and the same dinner.

So the model bills the published figure minus what moving into care replaces, worked out from your own budget rather than a national average:

If you own your home, it subtracts your food and utilities only. You still owe the mortgage, the property tax and the upkeep on a house you are no longer living in, which is a real cost that people entering care genuinely carry.
If you rent, it subtracts your food and utilities and your rent, because the lease ends.
In a couple where one partner stays home, the housing is not touched. Only the partner in care stops eating at home.
Where family provides the care, nothing is subtracted at all. That care happens in the person’s own home, where they keep buying their own food and paying their own rent.

Where a country publishes a breakdown of its charge, that breakdown caps the subtraction, so the model can never credit back more living cost than the charge actually contains. Six countries publish one:

Published share of the care charge that is room, board and utilities

Parameter	Default	Meaning
Denmark	100%	Of the published charge
France	91.53%	Of the published charge
Sweden	79.88%	Of the published charge
Switzerland	66.53%	Of the published charge
Germany	47.24%	Of the published charge
Australia	45.32%	Accommodation is charged separately here, so rent is never subtracted

Australia is the exception that has to be handled by name: its published charge covers care and everyday living but not accommodation, which is billed separately. Subtracting an Australian household’s rent from it would discount the same thing twice.

Long-term care assumptions

Parameter	Default	Meaning
Annual cost of care	€129,575 (US) / €25,968 (FR)today's money, per year	What the household pays while in care; see the table above for all twenty
Food and utilities share	38.1% (US) / 66.33% (IT)of essential spending	The part of your essentials a care home replaces
Family-care cost	€3,000today's money, per year	Applied when family provides care (if enabled)
Family-care chance	66.4% (GR) / 20% (US)	Probability an episode is family-provided, when enabled
Mortality during care	×4	Multiplier on the death hazard while in care

Care costs rise each year with medical inflation, which runs above general inflation (see §8).

How much of this we can stand behind

Ten of the twenty care costs come from an official statistics office, regulator or ministry publishing the figure directly. The rest are pieced together from published rules, and each one links to what it was built from. Greece is the one country where no figure exists at all: no statistics office, regulator, ministry, World Bank or European Commission source publishes what a care home charges, so the model asks you rather than inventing one.

Two further limits worth knowing. Ireland, Austria, the Netherlands, Norway and Finland charge for care as a percentage of your own income or assets rather than a fixed price, so a single annual figure can only stand for a typical retiree there. And the chance of an episode being family-provided is an upper bound everywhere: published statistics count everyone not receiving formal care, which mixes family care together with privately purchased care.

The full reference list

The source column in the table above carries one link per country, the strongest publication for the headline charge. The figures behind a country rest on more than that one page, so here is everything each country’s care model was built from, from the nursing-home fee itself down to the household budget survey that sets how much of your food spending a care home replaces.

Every publication behind the care figures, by country

Australia9 publications

Accommodation payments and contributions for residential aged careAustralian Government Department of Health, Disability and Ageing · 2026
Basic daily feeAustralian Government Department of Health, Disability and Ageing · 2026
Means tested fees for residential aged careAustralian Government Department of Health, Disability and Ageing · 2026
Refundable accommodation deposit and contribution (RAD and RAC) retentionAustralian Government Department of Health, Disability and Ageing · 2026
Schedule of fees and charges for residential and home care, from 20 March 2026Australian Government Department of Health, Disability and Ageing · 2026
Consumer Price Index, Australia - index numbers, weighted average of eight capital cities (dataflow ABS,CPI,1.1.0)Australian Bureau of Statistics · 2025
Health expenditure Australia 2023-24Australian Institute of Health and Welfare · 2025
Disability, Ageing and Carers, Australia: Summary of Findings, 2022Australian Bureau of Statistics · 2022
Household Expenditure Survey, Australia: Summary of Results, 2015-16, data cube 10 'Age of reference person' (65300do010_201516), table 10.3AAustralian Bureau of Statistics · 2016

Austria14 publications

Konsumerhebung 2024/25: Monatliche Verbrauchsausgaben nach Alter der Referenzperson (Tab_3a, Tab_1b)Statistik Austria · 2026
Kosten fuer Alten- und Pflegeheimeoesterreich.gv.at (Republic of Austria, Bundeskanzleramt) · 2026
Pflegegeld: Voraussetzungen, Hoehe und Pflegestufen 2026Pensionsversicherungsanstalt (PV) / Sozialministerium · 2026
Soziale Dienste fuer pflegebeduerftige MenschenBundesministerium fuer Soziales, Gesundheit, Pflege und Konsumentenschutz (Sozialministerium), Austria · 2026
Betreuungs- und PflegediensteSTATISTIK AUSTRIA · 2025
Gender-Statistiken: PensionenSTATISTIK AUSTRIA · 2025
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Informationen zur Berechnung des Kostenbeitrags fuer Wohnen und PflegeFonds Soziales Wien (FSW) · 2024
Visualisierung des oesterreichischen Pflegesystems: Herausforderungen und PerspektivenOesterreichisches Institut fuer Wirtschaftsforschung (WIFO), Famira-Muehlberger and Weingaertner · 2024
Verbrauchsausgaben 2019/20, Sozialstatistische Ergebnisse der Konsumerhebung, Kapitel 1 Tabelle 4Statistik Austria · 2021
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Expenditure for selected health care functions by health care financing schemes (hlth_sha11_hchf)Eurostat / OECD-Eurostat-WHO System of Health Accounts · 2026
HICP - annual data (average index and rate of change) (prc_hicp_aind)Eurostat · 2026
Eurostat HBS 2020: hbs_exp_t135 (mean consumption expenditure by age of reference person), hbs_str_t225 / hbs_str_t211 (structure of consumption, CP06), hbs_car_t314 (household characteristics by age)Eurostat · 2020

Belgium11 publications

Depenses de sante a charge des menages (SHA) / Quelle part des depenses de sante en Belgique est a la charge des menages?SPF Securite sociale (Federal Public Service Social Security), Belgium · 2026
Iriscare report on daily prices in Brussels facilities for older peopleIriscare (Brussels bicommunity social protection body) · 2025
Meting dagprijzen in woonzorgcentra (measurement of 1 May 2025)Departement Zorg, Flemish Government (Belgium) · 2025
Zorgbudget voor zwaar zorgbehoevendenDepartement Zorg, Flemish Government (Belgium) · 2025
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Performantie van het Belgische gezondheidssysteem, rapport 2024: Zorg voor ouderenKCE / Sciensano / RIZIV (gezondbelgie.be) · 2024
Wat is in de dagprijs van een woonzorgcentrum inbegrepen? (implementing the Ministerial Decree of 7 June 2023)Departement Zorg, Flemish Government (Belgium) · 2023
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
HICP annual average indices, dataset prc_hicp_aind (unit INX_A_AVG, 2015=100), COICOP CP00 and CP06Eurostat · 2026
Health care expenditure by financing scheme, dataset hlth_sha11_hf (unit EUR_HAB)Eurostat · 2026
Eurostat HBS 2020: hbs_exp_t135, hbs_str_t225 / hbs_str_t211 (CP06), hbs_car_t314, BelgiumEurostat · 2020

Canada8 publications

Paying for long-term careGovernment of Ontario, Ministry of Long-Term Care · 2026
Survey of Household Spending 2023, Public Use Microdata File (catalogue 62M0004X)Statistics Canada · 2026
Consumer Price Index, annual average, not seasonally adjusted (table 18-10-0005-01)Statistics Canada · 2025
Long-term care home level-of-care per diem funding summaryGovernment of Ontario, Ministry of Long-Term Care · 2025
User Guide for the Survey of Household Spending, 2023 (62F0026M2025001), Appendix G Table G1Statistics Canada · 2025
Household spending by age of reference person (table 11-10-0227-01), Survey of Household Spending 2023Statistics Canada · 2023
Table 11-10-0227-01: Household spending by age of reference person (Survey of Household Spending)Statistics Canada · 2023
Care receivers in Canada, 2018 (The Daily, 22 January 2020)Statistics Canada · 2020

Denmark16 publications

Hvad koster det at bo i plejebolig (2026 takster)Furesoe Kommune (Denmark) · 2026
Satser pa aeldreomradet for 2026Borne-, Aeldre- og Boligministeriet (Denmark) · 2026
Bekendtgorelse nr. 1719 af 18. december 2025 om betaling for generelle tilbud og for tilbud om personlig og praktisk hjaelp m.v. efter 79, 83, 83 b og 84 i lov om social serviceSocial- og Boligministeriet / Retsinformation (Denmark) · 2025
Huslejestatistik 2025 (rent as at 1 January 2025)Landsbyggefonden (Denmark) · 2025
Sundhedsloven 138Danske Love (rendering of lovbekendtgorelse om sundhedsloven), Denmark · 2025
Forbrugsundersogelsen: FU18 (forbrug efter alder, COICOP 06) og FU11 (personer pr. husstand), 2015-2024Danmarks Statistik · 2024
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Register of Medicinal Product Statistics (medstat.dk), 2024 ATC-level sales with patient shares by single year of ageSundhedsdatastyrelsen · 2024
Statbank table FU11, Household budget survey: background information by group of households (persons per household, head aged 60-74)Danmarks Statistik · 2024
Statbank table FU18, Consumption by group of consumption, age, price unit and time (Forbrugsundersogelsen)Danmarks Statistik · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Harmonised index of consumer prices, annual average index, by COICOP (prc_hicp_aind, unit INX_A_AVG)Eurostat · 2025
Persons in households with at least one person that reported a need for long-term care, by care status and household composition (ilc_ats22), EU-SILCEurostat · 2024
2021 Long-Term Care Report: Trends, challenges and opportunities in an ageing society, Volume IEuropean Commission (DG EMPL) and the Social Protection Committee · 2021
Self-reported use of home care services by people with severe difficulty in personal care activities, by sex and age (hlth_ehis_am7sa), EHIS wave 3Eurostat · 2019
Betaling i plejebolig og aeldreboligAeldre Sagen (Denmark) · 2026

Finland11 publications

Asiakasmaksut - Keski-Uudenmaan hyvinvointialue (2026)Keski-Uudenmaan hyvinvointialue / Keusote (Finland) · 2026
Sosiaali- ja terveydenhuollon asiakasmaksut 2026 (client fee guide implementing asiakasmaksulaki 734/1992)Pirkanmaan hyvinvointialue (Finland) · 2026
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Sairaanhoitokorvausten saajat ja maksetut korvaukset, kunnittain 2024 (avoindata.fi open data)Kela · 2024
Kulutustutkimus: kotitalouksien kulutusmenot kotitalouden viitehenkilon ian mukaan (StatFin ktutk 14q4, 14q9; waves 2012, 2016, 2022)Statistics Finland · 2022
StatFin tables ktutk 14q4 (Household consumption expenditure by age of reference person) and 14q9 (Background information by age of reference person), Household Budget SurveyStatistics Finland (Tilastokeskus) · 2022
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Harmonised index of consumer prices, annual average index, by COICOP (prc_hicp_aind, unit INX_A_AVG)Eurostat · 2025
Persons in households with at least one person that reported a need for long-term care, by care status and household composition (ilc_ats22), EU-SILCEurostat · 2024
Self-reported use of home care services by people with severe difficulty in personal care activities, by sex and age (hlth_ehis_am7sa), EHIS wave 3Eurostat · 2019
Ikaantyneiden ymparivuorokautisen hoivan kustannuslaskenta 2023Hyvinvointiala HALI ry (Finland) · 2023

France10 publications

Prix des EHPAD en 2024: un ecart important entre les chambres habilitees a l'aide sociale et les autresCaisse nationale de solidarite pour l'autonomie (CNSA) · 2026
Reperes statistiques no 27: En 2024, un ecart moyen de 1 000 euros par mois entre les chambres habilitees a l'aide sociale en EHPAD et les autresCaisse nationale de solidarite pour l'autonomie (CNSA) · 2026
Comprendre les prix, tarifs et prestations affiches dans l'annuaire des EHPADPortail national d'information pour l'autonomie des personnes agees (pour-les-personnes-agees.gouv.fr), France · 2025
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
L'aide sociale aux personnes agees ou handicapees, Edition 2024, fiche 04: La perte d'autonomie et les aidants des personnes ageesDREES, France · 2024
La complementaire sante, edition 2024, Fiche 11: primes des contrats individuels par age (enquete OC 2021, data file)DREES · 2024
La complementaire sante, edition 2024, Fiches 14-15: reste a charge apres AMO et AMC par age (Ines-Omar 2019, data files)DREES · 2024
La complementaire sante: acteurs, beneficiaires, garanties, Edition 2024 (updated 2025)DREES (Direction de la recherche, des etudes, de l'evaluation et des statistiques), France · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
HICP annual average indices, dataset prc_hicp_aind (unit INX_A_AVG, 2015=100), COICOP CP00 and CP06Eurostat · 2026

Germany12 publications

Eigenbeteiligungen steigen erneut: Pflegeheimbewohner und -bewohnerinnen zahlen im ersten Jahr durchschnittlich 3.364 Euro im Monat aus eigener TascheVerband der Ersatzkassen e. V. (vdek) · 2026
FAQ Eigenanteile in der vollstationaeren PflegeVerband der Ersatzkassen e. V. (vdek) · 2026
Sozialgesetzbuch (SGB) Elftes Buch, Paragraf 82: Finanzierung der PflegeeinrichtungenBundesministerium der Justiz / Bundesamt fuer Justiz, gesetze-im-internet.de · 2026
5,7 Millionen Pflegebeduerftige zum Jahresende 2023 (Pflegestatistik)Statistisches Bundesamt (Destatis) · 2024
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Statistischer Bericht, Laufende Wirtschaftsrechnungen, Einkommen, Einnahmen und Ausgaben privater Haushalte 2022 (sheets 63121-13, 63121-01)Statistisches Bundesamt (Destatis) · 2022
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Einkommens- und Verbrauchsstichprobe 2018, Konsumausgaben privater Haushalte, Fachserie 15 Heft 5, Tabelle 1.4 (Konsumausgaben nach dem Alter der Haupteinkommenspersonen)Statistisches Bundesamt (Destatis) · 2018
Preisvergleich Betreuungs- und Aufenthaltstaxen von Schweizer Alters- und PflegeheimenPreisueberwachung (PUE), Eidgenoessisches Departement fuer Wirtschaft, Bildung und Forschung · 2018
Expenditure for selected health care functions by health care financing schemes (hlth_sha11_hchf)Eurostat / OECD-Eurostat-WHO System of Health Accounts · 2026
HICP - annual data (average index and rate of change) (prc_hicp_aind)Eurostat · 2026
Eurostat HBS 2020: hbs_exp_t135, hbs_str_t225 / hbs_str_t211 (CP06), hbs_car_t314, GermanyEurostat · 2020

Greece10 publications

Ereuna Oikogeneiakon Proypologismon (Household Budget Survey) 2024, press release with Tables 1, 5, 6 and 8ELSTAT (Hellenic Statistical Authority) · 2025
Household Budget Survey 2024, table T1.10: purchases and receipts in kind by age of household head (SFA05)ELSTAT · 2025
Monades Frontidas Ilikiomenon (Elderly Care Units) - programme pageMinistry of Social Cohesion and Family, Greece · 2025
Prosopikos Voithos gia Atoma me Anapiria (personal assistant programme)Ministry of Social Cohesion and Family, Greece · 2025
Household Budget Survey 2023, table T1.10 (SFA05)ELSTAT · 2024
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
HICP - annual data (average index and rate of change), dataset prc_hicp_aindEurostat · 2026
2021 Long-Term Care Report: Trends, challenges and opportunities in an ageing society, Volume II (country profiles)European Commission (Social Protection Committee / DG EMPL) · 2021
Greece: Options for meeting the increasing demand for elderly care needsWorld Bank · 2020

Ireland12 publications

Health Insurance in Ireland: Market Report 2025Health Insurance Authority · 2026
Fair Deal Scheme: financial assessmentHealth Service Executive (HSE), Ireland · 2025
Fair Deal scheme: about the scheme, and Financial assessment: how much you pay towards careHealth Service Executive (HSE), Ireland · 2025
Health Insurance in Ireland: Market Report 2024Health Insurance Authority (Ireland) · 2025
Household Budget Survey 2022-2023, Results and AnalysisCentral Statistics Office (Ireland) · 2024
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Special Report 110: Nursing Homes Support Scheme (Fair Deal)Office of the Comptroller and Auditor General, Ireland · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
TILDA research offers fresh insight into how Ireland's system of caring affects older adults (Wave 5 Key Findings Report)The Irish Longitudinal Study on Ageing / Trinity College Dublin · 2020
HICP annual average indices, dataset prc_hicp_aind (unit INX_A_AVG, 2015=100), COICOP CP00 and CP06Eurostat · 2026
Nursing Homes Ireland Pre-Budget Submission 2026Nursing Homes Ireland (citing National Treatment Purchase Fund data) · 2025
Over 65s pay 43% more for health insurance - report (secondary confirmation of HIA figures: EUR 2,367 over-65 vs EUR 1,648 under-65, market average EUR 1,740)RTE News · 2024

Italy9 publications

Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Spese per consumi delle famiglie - COICOP 2018 (dataflow IT1,31_740), year 2024ISTAT, Italy · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
DPCM 12 gennaio 2017 (nuovi Livelli Essenziali di Assistenza), art. 30 - Assistenza sociosanitaria residenziale e semiresidenziale alle persone non autosufficientiPresidenza del Consiglio dei Ministri, Italy (text mirrored at medicoeleggi.com; official text in Gazzetta Ufficiale 18/03/2017, 17A02015) · 2017
HICP - annual data (average index and rate of change), dataset prc_hicp_aindEurostat · 2026
2021 Long-Term Care Report: Trends, challenges and opportunities in an ageing society, Volume II (country profiles)European Commission (Social Protection Committee / DG EMPL) · 2021
Inchiesta RSA: rette alte, attese lunghe e soddisfazione ferma a 6,8Altroconsumo (Euroconsumers) · 2026
Ma la pensione basta per permettersi una badante? (5o Rapporto annuale sul lavoro domestico)Osservatorio Nazionale DOMINA sul Lavoro Domestico, Italy · 2024
Invalidita civile: importi delle provvidenze economiche 2026 e limiti di reddito (reproducing INPS Circolare n. 153 of 19 December 2025)AIPD (Associazione Italiana Persone Down), reproducing INPS · 2026

Netherlands11 publications

Bewonersbijdrage (Wlz)Zorginstituut Nederland · 2026
De Wlz in 2026 / Wat kost een verpleeghuis in 2026CAK, reported via Nationale Zorggids · 2026
Hoe berekent het CAK de hoge eigen bijdrage?CAK (Centraal Administratie Kantoor) · 2026
Nominale premie zorgverzekeringStaat van Volksgezondheid en Zorg (VWS / RIVM), on Vektis and NZa data · 2026
Welke 55-plussers ontvangen mantelzorg en thuiszorg?Centraal Bureau voor de Statistiek (CBS), Statistische Trends · 2026
Eigen risico en eigen bijdrage van personen Zvw, Wmo en/of Wlz (StatLine/MLZ table 40094NED, 2015-2023)Centraal Bureau voor de Statistiek · 2024
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Expenditure for selected health care functions by health care financing schemes (hlth_sha11_hchf)Eurostat / OECD-Eurostat-WHO System of Health Accounts · 2026
HICP - annual data (average index and rate of change) (prc_hicp_aind)Eurostat · 2026
Factsheet 15 jaar Zorgverzekeringswet (eigen risico by age charts)Vektis · 2020

New Zealand7 publications

Aged residential care fundingHealth New Zealand / Te Whatu Ora · 2026
Consumers price index: June 2026 quarter - index numbers (CPIQ.SE906 Health, CPIQ.SE9A All groups)Stats NZ · 2026
Maximum Contribution Applying in Each Region From 1 July 2026Ministry of Health, New Zealand · 2026
Residential Care SubsidyWork and Income, Ministry of Social Development, New Zealand · 2026
Aotearoa Data Explorer SDMX dataflow list (HES_HES_001 to HES_HES_007)Stats NZ · 2025
Household expenditure statistics: Year ended June 2023 - data tables, table 1Stats NZ · 2023
New Zealand Retirement Expenditure Guidelines, Annual Update 2025New Zealand Financial Education and Research Centre (Fin-Ed Centre), Massey University · 2025

Norway12 publications

Egenandel for opphold pa institusjon (2026 rates)Bergen kommune (Norway) · 2026
Egenandeler for helsetjenester / egenandelstakHelfo (Helsedirektoratet), Norway · 2026
Forskrift om egenandel for kommunale helse- og omsorgstjenester (FOR-2011-12-16-1349), kapittel 1, 3-4, as amended by FOR-2025-12-17-2619 in force 1 January 2026Lovdata / Helse- og omsorgsdepartementet (Norway) · 2026
Statbank table 12293 (KOSTRA), Omsorgstjenester - supplerende nokkeltallStatistisk sentralbyra (SSB), Norway · 2025
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Statbank table 06946, Inntekt for husholdninger, etter husholdningstype (median)Statistisk sentralbyra (SSB), Norway · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Harmonised index of consumer prices, annual average index, by COICOP (prc_hicp_aind, unit INX_A_AVG)Eurostat · 2025
Persons in households with at least one person that reported a need for long-term care, by care status and household composition (ilc_ats22), EU-SILCEurostat · 2024
Health care expenditure by financing scheme (hlth_sha11_hf) and Expenditure for selected health care functions by financing scheme (hlth_sha11_hchf), SHA 2011Eurostat · 2023
Eurostat HBS 2020: hbs_exp_t135, hbs_str_t225 / hbs_str_t211 (CP06), hbs_car_t314, NorwayEurostat · 2020
Self-reported use of home care services by people with severe difficulty in personal care activities, by sex and age (hlth_ehis_am7sa), EHIS wave 3Eurostat · 2019

Portugal11 publications

Person in a care home for elderly people (custo OneValue)OneValue, Government of Portugal · 2026
Carta Social - Rede de Servicos e Equipamentos, Resultados Globais 2023 (nota tecnica)GEP / Ministerio do Trabalho, Solidariedade e Seguranca Social, Portugal · 2024
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Inquerito as Despesas das Familias 2022/2023, Quadro 10 'Despesa total anual media por divisao da COICOP e composicao familiar'Instituto Nacional de Estatistica (INE), Portugal · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Portaria n.º 67/2012, de 21 de marco - Estruturas Residenciais para Pessoas Idosas, arts. 1, 7 and 8Diario da Republica, 1.a serie, n.º 58, Portugal · 2012
HICP - annual data (average index and rate of change), dataset prc_hicp_aindEurostat · 2026
2021 Long-Term Care Report: Trends, challenges and opportunities in an ageing society, Volume II (country profiles)European Commission (Social Protection Committee / DG EMPL) · 2021
Lares de idosos: reformas nao chegam para pagar servicoDECO PROteste (Euroconsumers), Portugal · 2026
Preco das camas em lares de idosos aumenta 10% em 2026Jornal Economico, reporting the Lares Online 2026 price study · 2026
Evolucao das despesas diretas em saude em PortugalNova SBE, Catedra em Economia da Saude / Iniciativa para a Equidade Social · 2025

Spain9 publications

Encuesta de Presupuestos Familiares, ficheros de usuario (microdata), waves 2024 and 2025Instituto Nacional de Estadistica (INE) · 2026
IMSERSO press release on SAAD beneficiaries and the mix of benefits (prestacion economica para cuidados en el entorno familiar 31%, telecare 27%, home help 17%)IMSERSO / Ministerio de Derechos Sociales, Consumo y Agenda 2030, Spain · 2026
Encuesta de Presupuestos Familiares 2025, table 'Gasto segun tipo de hogar' (id 73807) and 'Gasto segun sexo y edad del sustentador principal' (id 73801)Instituto Nacional de Estadistica (INE), Spain · 2025
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Ley 39/2006, de 14 de diciembre, de Promocion de la Autonomia Personal y Atencion a las personas en situacion de dependencia (consolidated text), arts. 17, 18, 25, 33Boletin Oficial del Estado (BOE), Spain · 2006
HICP - annual data (average index and rate of change), dataset prc_hicp_aindEurostat · 2026
2021 Long-Term Care Report: Trends, challenges and opportunities in an ageing society, Volume II (country profiles)European Commission (Social Protection Committee / DG EMPL) · 2021
Informe de precios del 2025 de las residencias de mayores de EspanaInforesidencias.com · 2025

Sweden11 publications

Avgifter inom aldre- och funktionshindersomsorg (regelverk, 2026 amounts)Socialstyrelsen (Sweden) · 2026
Hogkostnadsskyddet januari 2026 and Patientavgifter i halso- och sjukvarden i alla regioner 2026E-halsomyndigheten and Sveriges Kommuner och Regioner (SKR) · 2026
Kommunernas avgifter enligt socialtjanstlagen - Kartlaggning och analys (art.nr 2025-1-9412)Socialstyrelsen (Sweden) · 2025
Household budget survey (hbs) ESMS reference metadataEurostat · 2024
Kolada KPI N23009, Kostnad sarskilt boende aldreomsorg, kr/brukare, national (Riket)Radet for framjande av kommunala analyser (RKA) / SKR, source data SCB Rakenskapssammandraget and Socialstyrelsen individstatistik · 2024
Structure of consumption expenditure by age of the reference person and COICOP consumption purpose (hbs_str_t225), unit PM (per 1000)Eurostat · 2020
Harmonised index of consumer prices, annual average index, by COICOP (prc_hicp_aind, unit INX_A_AVG)Eurostat · 2025
Persons in households with at least one person that reported a need for long-term care, by care status and household composition (ilc_ats22), EU-SILCEurostat · 2024
Health care expenditure by financing scheme (hlth_sha11_hf) and Expenditure for selected health care functions by financing scheme (hlth_sha11_hchf), SHA 2011Eurostat · 2023
Self-reported use of home care services by people with severe difficulty in personal care activities, by sex and age (hlth_ehis_am7sa), EHIS wave 3Eurostat · 2019
Eurostat HBS 2015: hbs_exp_t135, hbs_str_t225 / hbs_str_t211 (CP06), hbs_car_t314, SwedenEurostat · 2015

Switzerland12 publications

Krankenversicherung: PflegeleistungenBundesamt fuer Gesundheit (BAG) · 2026
Krankenkassenpraemien erhoehen sich durchschnittlich um 4,4 Prozent - Kostendaempfung bleibt DaueraufgabeBundesamt fuer Gesundheit (BAG) · 2025
Statistik der obligatorischen Krankenversicherung 2024, Tabelle T 2.11 (Kostenbeteiligung nach Altersklasse und Geschlecht)Bundesamt fuer Gesundheit (BAG) · 2025
Haushaltsbudgeterhebung 2020-2021: Gesundheitsausgaben der Haushalte nach Alter (T20.02.01.02.11) und Haushaltsgroesse (T20.02.01.00.11)Bundesamt fuer Statistik (BFS) · 2023
Statistik der obligatorischen Krankenversicherung 2021, Tabelle T 2.11Bundesamt fuer Gesundheit (BAG) · 2023
Haushaltsbudgeterhebung 2015-2017: Kommentierte Ergebnisse und TabellenBundesamt fuer Statistik (BFS) · 2022
Haushaltsbudgeterhebung, table T20.02.01.00.11: Haushaltseinkommen und -ausgaben nach Altersklasse der Referenzperson, wave 2020-2021 (je-d-20.02.01.00.11.xlsx)Bundesamt fuer Statistik (BFS) · 2021
Haushaltsbudgeterhebung, table T20.02.01.02.11: Detaillierte Haushaltsausgaben nach Altersklasse der Referenzperson, wave 2020-2021 (je-d-20.02.01.02.11.xlsx)Bundesamt fuer Statistik (BFS) · 2021
Preisvergleich Betreuungs- und Aufenthaltstaxen von Schweizer Alters- und PflegeheimenPreisueberwachung (PUE), Eidgenoessisches Departement fuer Wirtschaft, Bildung und Forschung · 2018
Expenditure for selected health care functions by health care financing schemes (hlth_sha11_hchf)Eurostat / OECD-Eurostat-WHO System of Health Accounts · 2026
HICP - annual data (average index and rate of change) (prc_hicp_aind)Eurostat · 2026
Faktenblatt: Pflegefinanzierung in der Schweiz (Daten 2024)CURAVIVA / ARTISET, on BFS Statistik der sozialmedizinischen Institutionen (SOMED) 2024 definitive Standardtabellen · 2025

United Kingdom16 publications

CPI INDEX 06: HEALTH 2015=100 (series D7BZ) and CPI INDEX 00: ALL ITEMS 2015=100 (series D7BT)Office for National Statistics · 2026
Family spending in the UK, Workbook 2: expenditure by income, Table A15 (HRP aged 65 and under 75)Office for National Statistics · 2026
Increased funding for care homes providing nursing (NHS-funded Nursing Care standard rate)Department of Health and Social Care (GOV.UK) · 2026
Family Spending in the UK, Workbook 1, table A1: Detailed expenditure with full-method standard errors, financial year ending 2025Office for National Statistics · 2025
Family Spending in the UK, Workbook 2: Expenditure by income, tables A15 (HRP aged 65 to 74) and A16 (HRP aged 75 or over), financial year ending 2025Office for National Statistics · 2025
Family Spending in the UK, Workbook 5: Housing, table 2.4 Housing expenditure by age of household reference person, financial year ending 2025Office for National Statistics · 2025
Free Personal and Nursing Care: methodology and background information 2024-25, Appendix 1 FPNC payment amountsScottish Government (gov.scot) · 2025
Social care: charging for care and support 2025 to 2026 (local authority circular)Department of Health and Social Care (GOV.UK) · 2025
Introducing a cap on care costs (research briefing CBP-9315) and Social care: announcements delaying the introduction of funding reforms (CBP-7265)House of Commons Library · 2024
The future of care needs: a whole systems approachJoseph Rowntree Foundation · 2024
Family Spending in the UK, Workbook 1, table A11: Detailed household expenditure by age of household reference person, financial year ending 2022Office for National Statistics · 2022
Care homes market study: summary of final reportCompetition and Markets Authority (GOV.UK) · 2017
Care Homes for Older People UK Market Report, 36th editionLaingBuisson · 2026
Paying for care at home: cost of home care serviceshomecare.co.uk / Homecare Association · 2026
Care home fees and costs: how much do you pay?carehome.co.uk · 2025
New data reveals stark regional differences in care home costs for self-fundersCare Home Magazine · 2025

United States20 publications

Medicare Supplement Rate Guide, rates as of January 1, 2026Maryland Insurance Administration · 2026
Medigap Rate Shopper Tool (Plan G, all carriers, every age 65 to 95-and-over)Missouri Department of Commerce and Insurance · 2026
QHP Landscape PY2026 Individual Medical, public use file (dataset 6fe7fb77-7291-4104-952f-7c7e2c5d0c45)CMS / CCIIO via data.healthcare.gov · 2026
2026 Medicare Parts A & B Premiums and DeductiblesCenters for Medicare & Medicaid Services · 2025
Annual Release of Part D National Average Monthly Bid Amount and Other Part C & D Bid Information for CY2026Centers for Medicare & Medicaid Services · 2025
Consumer Price Index for All Urban Consumers: Medical care (CUUR0000SAM) and All items (CUUR0000SA0), annual averagesUS Bureau of Labor Statistics · 2025
Medicare Program; Medicare Part B Monthly Actuarial Rates, Premium Rates, and Annual Deductible Beginning January 1, 2026Federal Register (CMS-8089-N) · 2025
Consumer Expenditure Surveys 2024, consumer units with reference person 65-74 (series CXUHEALTHLB0408M, CXUHLTHINSRLB0408M, CXU980040LB0408M, CXU980060LB0408M)US Bureau of Labor Statistics · 2024
Consumer Expenditure Surveys, 2024 annual data (series CXUHEALTHLB0408M, CXUHLTHINSRLB0408M, CXU980040LB0408M, CXU980060LB0408M)US Bureau of Labor Statistics · 2024
Consumer Expenditure Surveys, Table 1300: Age of reference person - annual expenditure means, shares, standard errors and relative standard errors, 2024U.S. Bureau of Labor Statistics · 2024
Consumer Expenditure Surveys, Table 2500: All consumer units - detailed annual expenditure means, 2024U.S. Bureau of Labor Statistics · 2024
National Health Expenditure Accounts: Health Expenditures by Age and Sex, data year 2020CMS Office of the Actuary · 2023
Long-Term Services and Supports for Older Americans: Risks and Financing, 2022 Research BriefUS Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation (ASPE), with the Urban Institute · 2022
Medicare Supplement Premium Comparison Guide 2022-23 (Northern/Central and Southern Illinois editions)Illinois SHIP, Illinois Department on Aging · 2022
2017 Medicare Supplement Comparison GuideLouisiana Department of Insurance, SHIIP · 2017
Insurance Standards Bulletin Series: Guidance Regarding Age Curves and State Reporting, Appendix I (federal default standard age curve)CMS / CCIIO · 2016
CareScout Releases 2025 Cost of Care Survey ResultsGenworth Financial / CareScout · 2025
Senior care rates balloon in every state: CareScout report / 10 Cheapest States for Long-Term Care: 2025McKnight's Senior Living; ThinkAdvisor · 2026
Marketplace Average Benchmark Premiums (state health facts indicator)KFF · 2026
A Snapshot of Sources of Coverage Among Medicare BeneficiariesKFF · 2025

The table above links one source per country: the strongest publication for the headline charge. Each country's figures rest on more than that, including what the charge covers, the room and board share the model credits back, the chance that family provides the care, and health costs outside care. Open a country to see everything it was built from.

7. Job loss

Job-loss risk is opt-in and off by default. When it is off, it is fully off: no unemployment shocks are simulated and contributions are never scaled by employment state. When you switch it on, an after-tax employment income is required for every person still working. Unemployment is modeled as the loss of that income’s expense coverage: the year’s drain on the portfolio is your expenses minus whatever household income remains, capped at the lost salary. The unemployed year runs through the same withdrawal and spending-cut machinery as any tight retirement year.

With the toggle on, an earner faces an annual risk of losing their job before retirement. The risk rises with age, because older workers are displaced more often and re-hired more slowly:

λ_{js} (x) = ⎩ ⎨ ⎧ 2% 3.5% 6% x < 55 55 \leq x < 60 x \geq 60

(7.1)

When a shock hits, it is either permanent (about a third of the time), meaning employment and contributions stop for good in an early forced retirement, or temporary. A temporary spell lasts a short, randomly drawn number of years, during which employment income stops entirely, after which the person returns to work.

Re-employment carries a lasting penalty: wage scarring. Each temporary spell permanently reduces earning power by a fixed fraction, and repeated spells compound (two spells leave roughly three-quarters of the original wage). Permanent losses end earnings outright.

One caveat to know about: the drain uses your retirement-plan expense estimate as the working-age burn rate, because the model has no separate “spending while working” input. It is a documented approximation, not a bug.

Job-loss assumptions

Parameter	Default	Meaning
Enabled	off	Opt-in; off means no shocks and no contribution scaling
Permanent share	35%of shocks	Fraction of shocks that end employment for good
Temporary spell length	~0.75years (avg)	Average unemployment duration, rounded up to ≥1 year
Wage scar	15%per spell	Permanent earnings reduction on re-employment, compounding

8. Expenses

Spending is built from seven independent buckets, each computed at that year’s prices and then summed. Three inflation series drive them: general inflation, medical inflation (general plus a premium, default +1.8%), and home inflation (general plus a premium, default +0.5%).

Essential: food, utilities, and basics. Flat in real terms: it simply tracks general inflation.

Housing: either rent, which tracks general inflation, or ownership, which sums the mortgage payment (fixed in cash terms, ending when the mortgage is paid off), maintenance and property tax (each a fraction of the home’s value, which itself grows with home inflation), and home insurance (rising about 1% per year above general inflation).

Healthcare: a base cost at age 65 that grows with medical inflation and moves with age along a calibrated cost-by-age curve. The input step asks for this per person and counts it once per partner for couples; the figure it prefills is each country’s out-of-pocket healthcare spend per person at 65, on one defined basis: costs the household pays directly, including insurance premiums it pays itself, excluding anything already withheld from a pension before it lands, and excluding long-term care, which has its own bucket (§6). Published statistics rarely arrive on that basis, so each figure is normalized and the method recorded: figures published per household are divided by the official household size of the surveyed group, and figures published as all-ages averages are raised by the official age profile of health spending. This is why the prefill spans a wide range across countries: where premiums are withheld from pensions at source the direct figure is small, and where households pay premiums themselves it is large.

The curve itself is measured per country where a source exists. Fifteen countries carry their own, taken from the country’s household budget survey, its insurers’ administrative statistics or its regulators’ published rate tables, and the measured shapes disagree with the old one-curve-for-everyone assumption in a consistent direction: where the state caps patient fees or cover is priced the same at every age, a retired household’s own late-life spending rises gently into the early 80s and then levels off, and in Austria it measurably falls back after 80. Each curve’s source, arithmetic and open questions live in that country’s data file, and the publications behind it appear in the country’s reference list on the country pages. The five countries where no official source splits late-life spending by age (Norway, Sweden, Italy, Belgium and New Zealand) share a fallback curve calibrated so its late-life rise matches the US health accounts with long-term care removed; the evidence from their measurable neighbours suggests even that overstates them slightly, and the routes tried are recorded so the next attempt does not repeat them. In every country with its own curve, a couple’s healthcare is billed at each partner’s own age.

The United States has its own curve, and it falls at 65 and climbs again after. A retiree there buys their own cover until Medicare begins, at a premium the law prices by age, so the years just before 65 cost several times the years just after: the US curve carries that step exactly, billed at each partner’s own age for couples. After 65 the curve keeps rising, about a third by 95, because the top-up cover most American retirees buy alongside Medicare is priced by attained age in most policies; the slope is taken from four state insurance regulators’ published rate tables. The premium the pre-65 half prices is the full sticker price, and the government help with it is handled where the law puts it, in the tax bill: while household income stays under a cap the plan receives a credit against the benchmark plan’s premium, priced against every withdrawal decision, and one dollar of income above the cap removes the entire year’s help. The accounts step asks what the benchmark plan costs where you live, prefilled with the national average for your ages; the tax page records what the credit model simplifies.

What a retired household spends

The essential and discretionary prefills come from each country’s best retiree budget source: a published retirement budget standard where one exists, otherwise the national household budget survey’s retired-household tables. Every figure is decomposed into the app’s buckets, excluding housing and healthcare which are collected separately, and each country’s data file records exactly which survey categories went where.

Retired-household budget prefills by country: essential plus discretionary spending, single and couple, and healthcare per person at 65

Country	Single, essential + discretionary	Couple, essential + discretionary	Healthcare at 65, per person	Where the figures come from
Australia	A$2,552 + A$993/mo	A$3,452 + A$1,496/mo	A$2,365/yrsourcechecked 2026-08-07	ASFA Retirement Standard, Comfortable lifestyle, homeowners aged 65-84, March quarter 2026sourcechecked 2026-08-07
Austria	€1,406 + €423/mo	€2,109 + €635/mo (derived)	€1,419/yrsourcechecked 2026-08-07	Statistik Austria Konsumerhebung 2024/25, Einpersonenhaushalt 60 Jahre und aelter (monthly Verbrauchsausgaben, detailed results published 2026-03-04)sourcechecked 2026-08-07
Belgium	€1,377 + €435/mo	€2,066 + €653/mo (derived)	€1,795/yrsourcechecked 2026-08-07	Statbel Huishoudbudgetonderzoek (HBS) 2024, TAB04_BE group 'Alleen ouderen' (households consisting only of persons 65+), published 2 Sep 2025sourcechecked 2026-08-07
Canada	CA$1,290 + CA$530/mo	CA$1,936 + CA$795/mo (derived)	CA$1,804/yrsourcechecked 2026-08-07	Statistics Canada, Survey of Household Spending 2023: table 11-10-0227-01 (age of reference person 65+) decomposed by household size using table 11-10-0224-01 (household type)sourcechecked 2026-08-07
Denmark	DKK 9,750 + DKK 2,763/mo	DKK 18,469 + DKK 6,479/mo	DKK 8,271/yrsourcechecked 2026-08-07	Danmarks Statistik Forbrugsundersogelsen, StatBank FU03 2022, enlig 60 ar og over uden born / 2 voksne hovedperson 60+ uden born (annual DKK per household incl VAT, current prices)sourcechecked 2026-08-07
Finland	€612 + €48/mo	€1,077 + €97/mo	€1,100/yrsourcechecked 2026-08-07	Kohtuullisen minimin viitebudjetit 2025 (University of Helsinki / Kuluttajatutkimuskeskus reference budgets for a decent minimum, published 19.11.2025), single 65+ (man/woman averaged) and couple 65+, rest-of-Finland variantsourcechecked 2026-08-07
France	€1,035 + €335/mo	€1,552 + €502/mo (derived)	€1,691/yrsourcechecked 2026-08-07	Enquete Budget de famille 2017 (Insee Resultats 'Les depenses des menages en 2017'), table TF102: depenses annuelles moyennes par menage, personne de reference 65-74 ans (all household sizes)sourcechecked 2026-08-07
Germany	€935 + €319/mo	€1,403 + €479/mo (derived)	€523/yrsourcechecked 2026-08-07	EVS 2023 (Einkommens- und Verbrauchsstichprobe), Konsumausgaben privater Haushalte, soziale Stellung der Haupteinkommensperson: 'Nichterwerbstaetige: im Ruhestand' (retiree households, all household sizes)sourcechecked 2026-08-07
Greece	€597 + €101/mo	€896 + €151/mo (derived)	€1,096/yrsourcechecked 2026-08-07	ELSTAT Household Budget Survey 2024, one-person household aged 65+ (monthly purchases per household, COICOP 2018, English-edition tables T1.05/T1.10)sourcechecked 2026-08-07
Ireland	€976 + €261/mo	€1,293 + €391/mo	€2,367/yrsourcechecked 2026-08-07	MESL 2026 (Minimum Essential Standard of Living), Older Adult living alone / Older Couple, URBAN, Vincentian MESL Research Centre at SVPsourcechecked 2026-08-07
Italy	€896 + €110/mo	€1,504 + €231/mo	€1,234/yrsourcechecked 2026-08-07	Istat Spese per consumi delle famiglie 2024, persona sola 65 anni e piu / coppia senza figli con p.r. 65 anni e piu (mean monthly EUR per household, COICOP 2018, published 2025-10-07)sourcechecked 2026-08-07
Netherlands	€955 + €287/mo	€1,604 + €470/mo	€2,514/yrsourcechecked 2026-08-07	CBS Budgetonderzoek 2020 (StatLine 83678NED amounts x 83679NED detailed shares), household types 'Alleenstaande man/vrouw vanaf AOW-leeftijd' and 'Paar zonder kind, vanaf AOW-leeftijd'sourcechecked 2026-08-07
New Zealand	NZ$1,809 + NZ$666/mo	NZ$4,968 + NZ$1,225/mo	NZ$1,409/yrsourcechecked 2026-08-07	NZ Retirement Expenditure Guidelines 2025 (Massey University NZ Fin-Ed Centre), Choices budget, Metropolitan, one-person and two-person householdssourcechecked 2026-08-07
Norway	NOK 13,524 + NOK 1,080/mo	NOK 20,983 + NOK 2,160/mo	NOK 18,653/yrsourcechecked 2026-08-07	SIFO Referansebudsjett 2026 (OsloMet, prices Feb 2026, 70+ rows), plus SSB Forbruksundersokelsen 2022 (StatBank 14227 single 65+, 14152 couple) for utilities, housing and healthcaresourcechecked 2026-08-07
Portugal	€511 + €120/mo	€861 + €187/mo	€757/yrsourcechecked 2026-08-07	INE Portugal IDEF 2022/2023 (Inquerito as Despesas das Familias), agregado '1 adulto idoso' 65+ / '2 ou mais adultos pelo menos 1 idoso' (annual EUR per household, COICOP 2018)sourcechecked 2026-08-07
Spain	€911 + €177/mo	€1,367 + €266/mo (derived)	€1,009/yrsourcechecked 2026-08-07	INE Encuesta de Presupuestos Familiares (EPF) 2025, hogar de una sola persona de 65 o mas anos (COICOP 2018/EPF, annual EUR per household, published 2026-06-25)sourcechecked 2026-08-07
Sweden	SEK 6,545 + SEK 1,200/mo	SEK 11,140 + SEK 1,810/mo	SEK 12,918/yrsourcechecked 2026-08-07	Konsumentverket Hushallskostnader 2026 reference budget (71+ rows), plus SCB Boendeutgifter 2024 for housing and SCB HUT 2021 for healthcaresourcechecked 2026-08-07
Switzerland	CHF 1,440 + CHF 613/mo	CHF 2,729 + CHF 1,134/mo	CHF 7,290/yrsourcechecked 2026-08-07	BFS HABE (Haushaltsbudgeterhebung) 2020-2021 pool, Einpersonenhaushalte ab 65 Jahren / Paare ab 65 Jahren ohne Kinder (mean CHF per household per month)sourcechecked 2026-08-07
United Kingdom	£1,470 + £984/mo	£1,921 + £1,475/mo	£367/yrsourcechecked 2026-08-07	PLSA/CRSP Retirement Living Standards, Moderate tier, outside London (2026 update, 'RLS 2025 uprated' budgets)sourcechecked 2026-08-07
United States	$1,489 + $611/mo	$2,686 + $1,283/mo	$5,511/yrsourcechecked 2026-08-07	BLS Consumer Expenditure Surveys, cross-tabulated Table 3600 (consumer units of one person by age of reference person) and Table 3620 (consumer units of two people by age), '65 years and older' column, 2023-2024 two-year averagesourcechecked 2026-08-07

Essential and discretionary exclude housing and healthcare, which the app collects separately. Couple budgets marked derived apply a documented equivalence factor of about one and a half to the single budget; the rest are published couple rows. The standards are not one concept: some countries publish comfortable-retirement standards, others actual-spending survey averages, and each row names its own.

For the country-shaped view of these prefills, with their published basis spelled out, see your country’s page.

The free tier collects one number instead of buckets, and all of it is treated as essential spending: it tracks general inflation, is never cut, and follows no age curve. Until v0.32 it was split into the four buckets by each country’s budget shares, which gave part of the number a medical-inflation path, part an age taper and part to the cut order below, none of which the person who typed it could see. A plan that wants those paths sets the buckets itself.

Discretionary: travel and leisure. Tracks general inflation. When the plan says extras shrink with age (a checkbox, off by default), it also tapers along an S-shaped curve, settling toward a floor as activity declines in later years:

φ (x) = f + \frac{1 - f}{1 + exp ( k ( x - a _{mid} ) )}

(8.1)

with a floor $f$ (default 5%), a midpoint age $a_\text{mid}$ (default 80), and a steepness $k$ . The taper is part of the plan, not a cut: it lowers the planned line on the results page, and it never counts against how much of the plan was paid.

Long-term care: the care cost from §6, active only during a care episode.

One-off: planned, dated events (a new roof, a wedding), entered in today’s or fixed euros and applied in their year.

Family support: recurring help to family over a chosen age range, tracking inflation with an optional real growth rate.

Home equity as a last resort

For homeowners, the plan can treat the home as a final backstop. It is on by default, with a toggle on the Housing step. If a year arrives where income plus the entire remaining portfolio, after the tax that liquidating it would trigger, cannot fund the survival floor, the home is sold instead of the plan failing: the sale realises that year’s home value less a haircut (default 10%, covering transaction costs and the urgency discount), any remaining mortgage is settled out of the proceeds (its balance estimated as the present value of the remaining payments at an assumed 5% rate), and the rest joins the portfolio. From the following year the household rents: the rent equivalent defaults to the home’s value times a per-country gross rent yield (5% US, 4% GR), rising with general inflation, unless you set an explicit monthly rent instead. The plan counts as failed only if the floor is still unfundable after the sale.

The model is deliberately simple, and one-way. There is no buying back in, no downsizing or partial sale, no reverse mortgage; the home’s value follows its deterministic inflation drift, and the post-sale rent derives from the home’s value today rather than its (per-path) value at sale. The sale itself is not taxed. The proceeds land in your ordinary investment account as money already taxed, so spending them is not taxed again; only what they earn from then on is. Because the household sells only at the brink of insolvency, real households would typically act earlier; the fallback is a floor on how bad things get, not a strategy.

On the results page the sale shows up three ways: a sentence under the headline (“in X% of futures the home is sold, median age Y”), the median sale age in the list under the portfolio chart (listed, not drawn, because it is a median over only the futures that sold), and a “plus home equity” note beside what’s left at the end. That figure itself stays portfolio-only. An unsold home is shown as context, never added to it.

Selected expense assumptions

Parameter	Default	Meaning
Medical inflation premium	+1.8%per year	Medical inflation above general inflation
Home inflation premium	+0.5%per year	Home-value inflation above general inflation
Maintenance	1%of home / yr	Annual upkeep as a share of home value
Property tax	0.3% (GR) / 1.2% (US)of home / yr	Annual property tax as a share of home value
Home-sale haircut	10%of sale price	Transaction costs + urgency discount when the fallback sells the home
Post-sale rent yield	5% (US) / 4% (GR)of home / yr	Gross yield deriving the rent equivalent after a fallback sale
Discretionary taper floor	5%of base	Lower limit the optional age taper settles toward
Essentials floor	80%of the year's essentials	Least of essentials kept when spending follows the markets
Extras floor	5%of the year's extras	Least of discretionary kept when spending follows the markets

9. Spending cuts under stress

Spending is cut only when the plan lets it follow the markets, a Premium switch that is off by default. With it off, nothing is cuttable: the household spends its whole plan every year, and a future fails the year that plan can no longer be paid.

With it on, when income plus the most the withdrawal rule allows cannot cover the year’s expenses, the household trims spending. Not all spending can be cut, though, and the order matters. The model cuts in priority, taking as much as it can from each before moving on:

Discretionary, down to its floor (by default 5% of that year’s discretionary).
Family support, down to zero.
One-off expenses, deferred or cancelled.
Essential, down to its floor (by default, no more than a 20% trim).

Both floors are the user’s to set on the spending step, which shows, as they type, the least the household could live on in a bad year. Before v0.32 this machinery ran on every plan, under a 4% fixed-amount ceiling nobody chose, so a plan asking for more than the ceiling was cut every year and still reported as a success.

Housing, healthcare, and long-term care cannot be cut: these are obligations. Whatever shortfall remains after exhausting the cuttable buckets is a forced shortfall: spending the household needed but could not fund. Because the floors above define the minimum the household can survive on, a forced shortfall arises only when a path has genuinely run out of money. It is the signature of a ruined year. For homeowners with the home-equity fallback on (§8), one more thing happens before that verdict: the home is sold, and ruin is declared only if the floor is still unfundable with the proceeds in hand.

10. Income

Each income stream carries its own basis: you say whether the amount you entered is before tax or after tax, and the model routes it accordingly (§12). “Real growth” always means growth above general inflation. The model tracks six kinds of income.

Employment: salary, growing with inflation plus a real raise, gated by whether the person is working and scaled by any wage scarring, and ending at retirement:

Y_{emp} (t) = w_{0} \cdot CPI_{t} \cdot (1 + g_{w})^{t} \cdot Emp (t) \cdot s

(10.1)

Pensions: up to four streams (each partner’s main and supplementary). Each begins at its own start age and is indexed one of five ways: by full inflation, by inflation plus or minus a fixed margin, by inflation capped at a ceiling, at a fixed annual rate, or by a custom schedule. Indexation begins the year after the pension starts; before the start age the entered amount is uprated with full inflation so it keeps its purchasing power.

The margin mode exists because most countries’ statutes are not plain price indexation: wage-linked systems and guarantee rules have historically run above prices, and some capped or conditional rules have run below them. The input step’s default, “as your country’s pensions do”, writes the researched rule for the selected country; the margin for each country is calibrated against its realized uprating history over the last fifteen to twenty-five years, and the evidence window is recorded in the data file. The default start age is likewise the country’s fully phased-in state pension age, not a global constant.

State pension rules by country: the default start age, how payments rise, and what a surviving partner keeps

Country	Pension age	How payments rise	What the survivor keeps
Australia	67	With inflation, plus 0.8 points a yearsourcechecked 2026-08-07	33% of the deceased's pensionsourcechecked 2026-08-07
Austria	65	With inflationsourcechecked 2026-08-07	40% of the deceased's pensionsourcechecked 2026-08-07
Belgium	67	With inflationsourcechecked 2026-08-07	10% of the deceased's pensionsourcechecked 2026-08-07
Canada	65	With inflationsourcechecked 2026-10-25	32% of the deceased's pensionsourcechecked 2026-10-25
Denmark	65 to 75	With inflation, plus 0.5 points a yearsourcechecked 2026-08-07	35% of the deceased's pensionsourcechecked 2026-08-07
Finland	67	With inflation, plus 0.2 points a yearsourcechecked 2026-08-07	25% of the deceased's pensionsourcechecked 2026-08-07
France	64	With inflation, minus 0.4 points a yearsourcechecked 2026-08-07	54% of the deceased's pensionsourcechecked 2026-08-07
Germany	67	With inflation, plus 0.2 points a yearsourcechecked 2026-08-07	55% of the deceased's pensionsourcechecked 2026-08-07
Greece	67	With inflation, minus 0.5 points a yearsourcechecked 2026-08-07	35% of the deceased's pensionsourcechecked 2026-08-07
Ireland	66	With inflationsourcechecked 2026-08-07	Keeps the larger of the two pensionssourcechecked 2026-08-07
Italy	68	With inflationsourcechecked 2026-08-07	60% of the deceased's pensionsourcechecked 2026-08-07
Netherlands	65 to 70	With inflation, plus 0.5 points a yearsourcechecked 2026-08-07	46% of the deceased's pensionsourcechecked 2026-08-07
New Zealand	65	With inflation, plus 1.0 points a yearsourcechecked 2026-08-07	30% of the deceased's pensionsourcechecked 2026-08-07
Norway	68	With inflation, plus 0.3 points a yearsourcechecked 2026-08-07	Nothing passes onsourcechecked 2026-08-07
Portugal	68	With inflationsourcechecked 2026-08-07	60% of the deceased's pensionsourcechecked 2026-08-07
Spain	67	With inflationsourcechecked 2026-08-07	52% of the deceased's pensionsourcechecked 2026-08-07
Sweden	68	With inflation, minus 0.4 points a yearsourcechecked 2026-08-07	Nothing passes onsourcechecked 2026-08-07
Switzerland	65	With inflation, plus 0.3 points a yearsourcechecked 2026-08-07	30% of the deceased's pensionsourcechecked 2026-08-07
United Kingdom	66 to 68	With inflation, plus 1.0 points a yearsourcechecked 2026-08-07	Nothing passes onsourcechecked 2026-08-07
United States	67	With inflationsourcechecked 2026-08-07	Keeps the larger of the two pensionssourcechecked 2026-08-07

These are the values the input step prefills and the simulation executes. Rise labels compress each statute into the nearest rule the engine can run, calibrated against the country's realized uprating history. A survivor share is the statute's own share where the law pays a fixed one, and otherwise the effective share for a typical couple who both draw a pension. Either way the simulation applies it from the first death, without the income tests or time limits some countries attach; each country's page says what that leaves out.

Each of these rules is also explained in words, with its statute and sources, on your country’s page.

Supplementary pensions get none of these defaults: workplace and private schemes are contracts, so they start on plain inflation indexation with a neutral 50% survivor share, and your plan statement is the source that matters.

Rental: net rental income that grows with inflation plus real growth and is scaled by a constant occupancy rate; an expected maintenance cost (a fraction of a month’s rent per year, from the chance and size of repairs) subtracts from it. In extreme settings the net can go negative, and it is treated as a cash flow rather than floored at zero.

Lump sums: dated one-off receipts (an inheritance, a property sale), in today’s or fixed euros.

Other: any recurring income over a chosen age range, with optional real growth.

Selected income assumptions

Parameter	Default	Meaning
Real wage growth	+1%per year	Salary growth above inflation
Pension indexation cap	2%per year	Ceiling when capped-inflation indexation is chosen
Rental occupancy	90%	Constant occupancy rate
Maintenance cost	5% chance × 3 monthsper year	Expected repair cost, deducted every year in months of rent

11. Withdrawal strategies

When spending follows the markets, a withdrawal rule sets a ceiling on how much may be taken from the portfolio each year. It does not generate spending. Your expense plan drives what the household actually spends; the rule only caps how much of it the portfolio will fund before spending cuts kick in, and never raises spending above the plan. With the switch off, the default, there is no ceiling at all and the draw is whatever the plan needs. The model offers three ceilings; all are then subject to a survival override (below).

Constant-real (the “4% rule”)

The classic Bengen rule. The first year’s withdrawal is a fixed percentage (default 4%) of the balance at retirement, and that amount thereafter only keeps pace with inflation. It does not rise or fall with the portfolio:

W_{t} = w_{0} \cdot B_{retirement} \cdot CPI_{t} / CPI_{retirement}

(11.1)

The first retired year therefore withdraws exactly $w_0 \cdot B_{\text{retirement}}$ .

Constant-percentage

A fixed percentage of the current balance each year. Spending rises and falls directly with the portfolio.

Variable percentage withdrawal (VPW)

Each year’s withdrawal is the share of the current balance that an annuity over the remaining horizon would pay out, given an assumed real return:

W_{t} = \frac{r}{1 - ( 1 + r ) ^{- n}} \cdot B_{t}, n = max (5, 100 - x)

(11.2)

Here 100 is the age the payout is planned to run to. It is a setting of its own, not the lifespan cap of §4: a horizon at the cap (120) would spread the balance over decades almost nobody lives, and hold the ceiling far below what the rule intends. The assumed return $r$ is a blend of equity and bond real-return assumptions at the current allocation, so it drifts as a glidepath de-risks. The withdrawal share is capped from age 85 on. VPW’s ceiling grows with the balance and pulls back when it shrinks, and rises naturally with age as the horizon shortens. Because every rule here is a ceiling on planned spending, a rising VPW ceiling never spends more than the plan: it only cuts less. The first-year rate plays no part in it.

The survival override

Whatever the rule proposes, the household must still cover its hard floor, the irreducible minimum left after every allowed spending cut: floored essentials, all housing, healthcare, any care costs, any tax on the balance, and the floored share of that year’s discretionary.

H_{t} = θ_{ess} E_{ess} + E_{house} + E_{hc} + E_{ltc} + E_{tax} + f_{cut} E_{disc}

(11.3)

With the switch off, nothing is cuttable and the hard floor is the year’s whole planned spending.

If meeting this floor requires more than the rule allows, the model overrides the rule upward and takes what is needed. The actual withdrawal is therefore the larger of “what the rule advises” and “what survival demands,” grossed up for the tax the withdrawal itself causes (§12) and capped at the balance. A path is ruined in the first year its income plus its entire remaining portfolio cannot meet the hard floor.

One subtlety: the ceiling is stated after tax, so it caps what lands in your pocket, not what leaves the portfolio. The gross draw that clears a given ceiling is therefore the solution of $g - [T_t(g) - T_t(0)] = W_t$ , which §12 solves exactly rather than dividing by one minus a rate.

Withdrawal assumptions

Parameter	Default	Meaning
Spending follows the markets	off	Whether any ceiling and cuts apply at all
Initial rate	4%	Starting withdrawal rate for constant-real / constant-%
VPW assumed return	5% eq / 1.9% bondabove inflation	Return assumptions above inflation in the annuity factor

12. Tax

Tax is computed inside the year loop, on that year’s actual position, against the rules of the country in the plan. It is not a rate applied to the draw. Twenty countries are modelled; the tax page sets out country by country what each one covers and what it does not.

The year’s position

Each simulated year builds one household position: every income stream at its stated basis (before or after tax, per stream), the portfolio account by account with each account’s type, the cost basis, this year’s asset mix and this year’s returns. From it a country rule produces two quantities:

$C_t^{\text{bal}}$ , a charge on the balance, independent of every decision the household makes that year. Wealth taxes and assumed-yield charges live here. Thirteen of the twenty countries have at least one.
$T_t(g)$ , the income tax at a gross withdrawal of $g$ , which is the whole household bill at that draw and not the increment over drawing nothing.

The two are billed differently, and the split follows from the order the year loop has to run in. A spending bucket must be known before the withdrawal is solved, which $C_t^{\text{bal}}$ is and $T_t(g)$ is not. So the balance charge becomes a genuine eighth bucket and the income tax is netted out of income, leaving the whole gap, ceiling, cuts and ruin chain in the after-tax terms it has always used. That is arithmetically the same as billing both, and it keeps the solve in one place.

The tax bucket is exempt from the spending-cut cascade (§9): you cannot cut your tax bill, and letting the cascade shave it would understate ruin. The figure reported on the results page is neither of the two above on its own but the whole household bill at the draw actually taken, which is the number you would recognise from a tax return.

The gross-up, solved exactly

The engine needs the inverse: the draw that leaves a required amount in hand. Two properties make it cheap. Marginal rates are always below 100%, so $g - T_t(g)$ is increasing; and a tax scale is piecewise-polynomial, so the function is made of smooth segments.

find g such that g - [T_{t} (g) - T_{t} (0)] = n

(12.1)

Each country returns its own analytic $T_t'(g)$ alongside $T_t(g)$ , so a Newton step costs one evaluation rather than the two a numerical derivative needs. The slope between segment boundaries is not monotone (a realistic United States position has sixteen corners inside a 300,000 range), so plain Newton can alternate across a corner instead of converging. The solve is therefore bracketed, taking the Newton step when it stays inside the bracket and bisecting when it does not. Both ends of the bracket are free: the low end is a zero draw, already evaluated for the year, and the high end follows from the floor on the slope. Three iterations plus a final linear correction land exactly on a piecewise-linear segment.

Accounts, cost basis and the asset mix

Your savings sit in accounts, and the kind of account changes the bill by more than any rate does. Each country offers its own kinds (a 401(k) and a traditional IRA, a Danish ratepension, Australian super in each of its two phases) and each kind carries its own law: how withdrawals are taxed, any tax charged inside it while it is held, the earliest age it pays out, and any schedule it must pay out on. Every kind falls into one of three broad groups for the charts and for the withdrawal orders:

Account groups

Parameter	Default	Meaning
Tax-free	invisible	Nothing while held, nothing on withdrawal, and absent from every threshold. That last part is most of its value.
Tax-deferred	ordinary income	Nothing while held; the whole withdrawal is income in the year it is taken, unless the country charges it at a set rate instead.
Taxable	as it arises, then on the gain	Distributions or an assumed yield each year, then the profit portion on sale. The only kind of account where the asset mix matters.

Each country’s accounts page, linked from the tax page, describes every type it offers in plain words, with its sources and its disclosed simplifications. What follows is the machinery they share, and the country rules that need more precision than those pages give.

Withdrawals come out in the order the plan chooses: a bit of each account in proportion to what it holds, or one group emptied before the next. Whatever the order, two things come first and two come last. First, whatever the law makes an account pay this year (a required minimum, a payout schedule’s instalment, an account that pays out whole in its year) leaves that account. Last, an account reachable only at a charge (the US 10% before 59½, cashing in a Danish ratepension or a Dutch lijfrente early) is drawn only once everything else is spent.

Cost basis is tracked rather than assumed. Your ordinary investment account keeps one per-path figure, seeded from the starting portfolio, increased by contributions and by income taxed as it arises, and reduced pro rata on withdrawal. Crediting taxed income to the basis is not a refinement. The balance update compounds a total return, so distributions are implicitly reinvested; taxing them without crediting them would tax the same money again as a gain on sale.

An account whose law separates what you paid in from what it earned keeps its own figure, and its own order. A Norwegian share savings account (aksjesparekonto) gives back what you paid in first, untaxed, and taxes only what comes out beyond it, less a yearly untaxed slice that builds up while it goes unused. The part of a Danish kapitalpension paid in without a tax deduction comes out first, free of its 40% charge. A Dutch lijfrente into which premiums went that were never deducted pays those premiums back untaxed before any instalment is taxed as income. Each of these asks what you paid in. Skip the question and the plan takes the usual case: a share savings account with no gain in it yet, a kapitalpension paid in entirely from deducted pay.

Some accounts are taxed every year on what they hold instead of on what comes out. A Swedish investment savings account (ISK), endowment insurance (kapitalförsäkring) and pan-European pension product pay 30% on an assumed return of 3.55% of their value, above the first 300,000 kronor a person, which the three share in that order. Sweden averages the value at the start of each quarter; the plan takes the value at the start of the year plus a quarter of the year’s deposits (half, for the insurance), so in a year you draw one down it charges slightly more than the law does. A Danish share savings account (aktiesparekonto) pays 17% of each year’s change in value from inside the account, and a fall is carried forward against the next rise. Wealth taxes are charged on what you hold at the start of the year, where Norway, Spain and Switzerland count the end of it: each year’s charge is the previous year’s figure, and over a retirement the total is the same before inflation.

In a couple, each account can be marked as yours, your partner’s or shared, except the ones the law opens in one person’s name only (pensions, IRAs, ISAs and their like), which are always one of you. A shared one you saved before is split between you by the share you set. Where the law gives each person their own allowance, like the Swedish 300,000, or taxes each partner on their own income, the plan counts each account against its owner, and splits a shared one between you, half each unless you set another split.

Within an ordinary account, interest, dividends and gains are charged at the rates their country actually applies to each, using the plan’s asset weights.

What you hold inside that account can change the bill by more than any rate does, so where a country taxes two instruments differently, the Detailed view asks. Greece charges nothing on a European fund and 15% on a bank deposit. The Netherlands puts a savings account and a money-market fund in different statutory categories and charges them almost five times apart. Every menu is prefilled with the instrument that country’s own tax literature says people actually hold, so leaving it alone gives you the same answer as before the question existed.

Most countries are asked nothing, and that is a finding rather than a gap. A Dutch retiree can hold an Irish fund, an American one, a Dutch one or individual shares and the charge on their shares and bonds is identical, because it falls on what you hold and not on what it pays you. The United States is the same once both sides are counted: tax-exempt municipal bonds do cut the bill, but the market prices that exemption into a lower yield, and this model has no way to charge you the lower yield. Showing you one half of that trade would be worse than showing you neither.

When money can come out

A pension account usually cannot pay out before an age the law sets, and sometimes only once you have retired: a UK pension from 55, or 57 for payments from April 2028; Australian super from 60 and retired, or from 65; a Swedish occupational pension from 55; a Norwegian one from 62; an Irish PRSA from 60, or 50 once retired; a Swiss pillar 3a or vested benefits account from five years before the reference age, which the plan takes as the age your main pension starts, or 65. Before that age the money is simply out of reach. It still grows, and nothing is charged on it, but the plan cannot spend it.

Some accounts pay out on a schedule instead of on demand. A Danish ratepension pays the balance divided by the years left of the term you chose. A Norwegian occupational pension does the same, over at least ten years and to at least 77. A Dutch lijfrente pays fixed instalments: started before the AOW age, over at least 20 years plus the years before it; from the AOW age, over at least five years, unless that would pay more than 27,192 a year in today’s money, when the plan lengthens the term to what the law allows, up to 20 years less the years past the AOW age. A German pension depot (Altersvorsorgedepot, from 2027) starts paying between 65 and 70, or earlier alongside your state pension: up to 30% as a lump in the first year, then each year at least 80% of what is left spread over the years to 85, fixed every three years. A Swiss pillar 3a or vested benefits account pays out whole in the year you choose, and by your reference age unless you are still working, five years later at the latest. Everything a Swiss household takes out of these accounts and as pension capital in one year is added up and taxed on its own scale, apart from income: a fifth of the federal income tax scale plus your canton’s own rate, measured where the canton’s capital city is. So two accounts paid out in different years cost less than one paid out at once. The plan takes exactly what the schedule pays, no more and no less, and reinvests what you do not need in your ordinary account.

Small pensions are let out faster. A Norwegian pension is paid over fewer years when that keeps each payment at about 20% of the base amount (G), about 27,300 kroner in 2026, as either the provider or the saver may ask. A Swedish pension insurance worth at most one price base amount, 59,200 kronor in 2026, may be cashed in at any age, and so may a Dutch lijfrente account worth at most 5,513 euro, without the 20% charge. These amounts are in today’s money and move with prices.

Some pensions pay part of themselves tax free. A quarter of a UK pension pot comes out tax free, up to £268,275 over a lifetime for each person, by one of two routes you choose for each pot. The usual one, and what the plan assumes unless you say otherwise, takes the tax-free quarter as a lump when you start drawing: it moves into your ordinary investment account, and every later payment from the rest of the pot is taxed as income. In the Financial Conduct Authority’s figures for 2024/25, more than nine in ten pots of £30,000 or more that stayed invested when first drawn from went this way. The other route leaves the pot invested and takes a quarter of each withdrawal tax free until the £268,275 is used. Both routes share the one allowance.

In the United States, a health savings account (HSA) pays the year’s medical and care costs tax free, and we take those costs from it first. Anything more is taxed as income, plus 20% before 65. A deferred annuity bought with money already taxed pays out its earnings first, taxed as income and counted for the 3.8% tax on investment income, plus 10% before 59½. Cash value life insurance pays back what you put in first, untaxed. A 457(b) government plan has no 10% early charge at any age, and a SIMPLE IRA’s early charge is 25% in the first two years after you join. A Roth IRA pays back what you put in and what you converted first, tax free at any age; growth taken out before 59½, or before the account is five years old, is taxed as income, plus 10% before 59½. A Roth 401(k) pays each withdrawal before then partly from contributions and partly from growth. If you do not say what share of a Roth is your own contributions, we count all of it. Money in a traditional IRA you paid tax on already comes back tax free in the share it makes of all your traditional IRAs together, whichever one you take from or convert, so converting a fresh after-tax IRA while another holds pre-tax money is mostly taxed. A schedule of equal yearly payments (a SEPP) takes money from an IRA or 401(k) before 59½ without the 10%: we pay exactly its amount each year, worked out by the method you pick, until five years have passed or you reach 59½, whichever is later. In the year you turn 59 we charge the 10% on half that year’s early withdrawals, since we do not know your birthday.

An Irish pension pot pays up to a quarter of itself as a lump sum when you take the benefits, and the rest moves into an ARF you draw from. Lump sums over your lifetime are tax free up to €200,000, charged 20% up to €500,000, and taxed as income above that. Money drawn from an ARF before 66, or until the State Pension is awarded, also pays social insurance (PRSI): 4.2%, rising to 4.35% from October 2026, at least €650 a year, and nothing if that income is under €5,000. An Irish life assurance bond pays back what you put in, and its profit is taxed at 38% when it comes out and every eighth year as if it had.

Belgian pension savings pay one levy and are then free of tax: 8% of the account at 60, or 10 years after you opened it if that was at 55 or later, taken from the account itself; long-term savings pay 10% the same way. After it, nothing the account earns or pays out is taxed. For a pension savings fund the law levies what you paid in grown at 4.75% a year rather than the fund’s value; we do not know what you paid in, so we use the value. Cashed in before 60, either is taxed at 33%.

A UK investment bond pays back what you put in first, untaxed. What comes out beyond that is profit, taxed as savings income on top of all your other income, with 20% counted as already paid on a UK bond and nothing on an offshore one. The law also lets you take 5% a year of what you paid in without tax for now, and spreads a large profit over the years you held the bond when it is taxed. Both change when tax falls more than how much, and the plan leaves them out, so in a year you cash in a big bond it can show too much tax.

Your choices about an account (when it starts paying, over how many years, how its tax-free part comes out, whether it may be cashed in early) are made once, at the start, with what you know then. The plan does not change them partway through: a Danish ratepension’s term stays the one you chose, and an account is never converted into another kind along the way.

Where the law lets money out early against a charge, the plan uses that route only if you allow it, and the box is ticked unless you untick it. A Danish ratepension cashed in off its terms pays a 60% charge instead of income tax, and a Dutch lijfrente surrendered early pays 20% on top of income tax. The US 10% before 59½ is always available.

When a year cannot be paid because the money is out of reach, that future counts as failing for that year, even though the money pays out later. It is not wiped out: the accounts out of reach keep growing, and the plan carries on once they open. The results say how often this happens, from what age and for how long.

Some pensions pay their tax from inside the account rather than out of your spending: Denmark’s 15.3% on the yearly return, Sweden’s yield tax, Australia’s 15% on super that is still being built, Italy’s 20% on a pension fund. The plan takes these from the account they fall on, so they show in your tax but not in your spending.

Roth conversions (United States)

A premium United States plan can carry a conversion schedule: a yearly amount, in today’s money, moved from a tax-deferred account into a tax-free one across a chosen age window. The move executes at the top of each simulated year, before that year’s tax context is built, and it is capped at what the tax-deferred account holds above the year’s required minimum, because the law makes the required withdrawal come out first. The moved amount is ordinary income in the year it moves and nothing else: it is not spendable cash, so the solver funds its tax with a correspondingly larger withdrawal, under whatever account order the plan uses. Everything downstream reprices on its own, because it all reads the same income: the taxable share of Social Security, the stacking of investment income, the deduction taper for 65 and over, the pre-65 health-cover help and its cap, the Medicare surcharge two years later, and every future required minimum, which shrinks because the account it is computed on shrank.

The five-year rule is charged. Money converted while under the access age skips the 10% early charge on the way in, and each year’s conversion then carries its own five-tax-year clock; taking converted money back out before its clock has run pays the 10% after all. The engine tracks each vintage separately, lets money out oldest first, and treats the Roth balance the plan started with, plus later contributions, as never clocked. All of these amounts are tracked in the money of the year they moved, deliberately: growth inside the Roth never shortens a clock. Two edges are approximate and both are disclosed where they bite: earnings drawn past every clocked and unclocked dollar before the access age owe only the 10% here, where the law would also income-tax them, and a tax-deferred account’s after-tax contributions count only where you enter them as its basis (your Form 8606 basis), so where they are left out the conversion’s tax is overstated.

Two more deliberate positions. The schedule converts on every surviving, solvent path, crash years included: a real household might pause, but a schedule that quietly paused in bad years would make the comparison against not converting dishonest. And the product ships no optimiser and no fill-to-bracket mode: the cost of the next converted dollar is not monotone, so the results page draws the cost curve with its steps named and compares whole schedules on identical futures instead of asserting a stopping point.

Thresholds over time

Whether a bracket or an allowance keeps pace with inflation is researched per parameter and per year, not assumed per country: United States brackets are indexed while the thresholds that decide how much of Social Security is taxable have been frozen since 1983. Where a freeze has a statutory end date the projection uses it, so the United Kingdom’s personal allowance stands still through 2030/31 and then tracks prices, because the annual uprating duty is suspended for named years rather than repealed. The default follows the research. Two overrides let you stress it either way, assuming every threshold keeps pace or that none of them moves.

Tax settings

Parameter	Default	Meaning
Mode	country	"country" uses the modelled rules; "flat" falls back to a single rate on the draw; "off" treats every amount as already after tax
Threshold indexation	researched	"researched" follows each parameter's own history, including a statutory end date on a freeze; "full" and "frozen" are the two stress tests
Withdrawal order	proportional	"proportional" draws a bit of each account in proportion to what it holds; "taxable_first" drains the ordinary account first and the tax-free pot last; "deferred_first" drains the tax-deferred account first. Enforced in the shared composition layer, so every modelled country honours every order
Charged accounts last		An account the law only pays out early against a charge (the US 10% before 59 and a half) is drawn after every account that is not, so the plan never pays the charge while untaxed money sits elsewhere
Flat-mode rate	0%	The override rate, used only when mode is "flat". Applies to the whole draw, not the gain
Assumed dividend yield	1.6%	The income share of the equity return, needed because dividends and gains carry different rates in most countries

13. Sensitivity analysis

Sensitivity analysis answers: of the things you could change, which ones move your outcome the most? It works by varying your inputs (starting portfolio, contributions, retirement age, and so on) within plausible bands around your actual plan, and measuring how much each one drives the variation in an outcome such as success probability.

The method is Sobol’ variance decomposition. Inputs are varied together using a low-discrepancy sampling design that covers the space efficiently, and the model reports two numbers per input. The first-order index is how much of the outcome’s variation that input explains on its own; the total-effect index adds its variation through interactions with other inputs:

S_{1, i} = \frac{variation explained by input i alone}{variation your inputs cause}, S_{T, i} = \frac{... including interactions}{variation your inputs cause}

(13.1)

A key technique makes this reliable: every input combination is evaluated against the same set of random market and lifespan draws, so the differences in outcome reflect the input change, not luck. Without it, the sensitivity signal would be drowned in Monte Carlo noise.

Some of the outcome’s variation is still luck: which markets and lifespans a sample happened to draw, which no input of yours can change. We measure that part and leave it out, so both numbers are shares of the variation your inputs cause, and an input that changes nothing scores exactly zero.

Three tiers vary more inputs at higher tiers, and the one your plan gets is chosen for you: the smallest set that still covers everything your plan actually uses. Nothing is asked of you here, and there is nothing to trade off. Inputs your plan leaves empty, such as a mortgage you do not have, are skipped rather than ranked, so they cost you no waiting and produce no empty bar.

Sensitivity tiers

Parameter	Default	Meaning
Low	7 inputs	Portfolio, contributions, retirement age, expenses, pension, care cost, equity weight
Medium	10 inputs	Adds withdrawal rate (live only when spending follows the markets under a rate-reading rule), where your savings sit, healthcare cost
High	13 inputs	Adds supplementary pension, wage growth, mortgage

14. Failure-mode attribution

When a plan fails, this analysis asks which risk caused it. It works by replaying each failed path with one risk at a time switched off, holding everything else identical, and checking whether the path now survives. A path that survives once a given risk is removed was, in part, sunk by that risk.

Five risks are tested, each neutralized in a natural way:

Long-term care: no care episode ever begins: the costs disappear, but so does care’s shortening of life (§6), so the “no care” path also lives longer. The channel measures the honest net effect.
Job loss: no employment shock occurs.
Longevity: nobody lives unusually long (lifespans capped at their typical value).
Market returns: returns replaced by their average.
Inflation: inflation held at its average.

Counting how often each risk rescues a failed path gives its share of the blame. Two further figures fall out. Failures that two or more of the single removals each rescue on their own are over-determined: there is more than one way out, which is why the shares can add up to more than 100%. Failures that no single removal rescues are structural: no one risk is to blame, so the plan falls short either on several fronts at once or because it is under-funded. The model never runs a path with every risk removed together, so it does not separate those two.

15. Counterfactuals

A counterfactual answers “what if you changed one thing?” and shows the effect on your plan. The model runs each change against the same random futures as your baseline, so the difference is attributable to the change alone, not to luck.

The changes are chosen for you, adaptively. The model takes the inputs that the sensitivity analysis found matter most, keeps those that map to an action you can actually take, and skips changes that wouldn’t apply (suggesting “save more” when you’re already retired, for instance). Where that leaves fewer than three, or there is no ranking, saving more and retiring later fill in. There are five levers:

Save more: 500 more a month, or double the current contribution if that is less.
Retire later: 2 years later for each partner still working, capped at 75.
More in shares: the equity weight raised by 15 percentage points, bonds taking the difference. This is a move toward more market risk. It is skipped on a glidepath, whose equity weight follows age instead.
Family-rate care: every care episode priced at the family-care cost instead of the full one.
Keep the home: the last-resort sale (§8) switched off. It joins the list ahead of the ranking, but only when the plan actually sells the home in some futures.

Up to 4 are shown, ordered by how much they move your success probability.

For each, two differences from your baseline are reported:

Δ success probability, Δ median terminal wealth

(15.1)

The changes themselves are evaluated on the full model, so couples and every withdrawal rule are handled correctly.

How you take your money out. A second set of counterfactuals leaves the plan as it is and changes one of the household’s own drawdown decisions. Each runs several options and reports the best (highest success probability, then the highest median terminal wealth), listing the others:

the withdrawal order: each other named order, and, where a tax-deferred account opens after you retire, ordinary money first until it opens and tax-deferred money first from then;
the conversions you planned, against the same plan with none;
each account setting that is a decision rather than a fact, at a few of its legal values: a payout’s start age (the earliest, five years later than chosen, and the latest up to 75), its term (the legal ends and the middle), a tax-free lump’s route, share or age, and whether the plan may break a pot early at its charge.

An option the account’s own rules refuse (one that needs an answer the plan does not give) is dropped rather than guessed. Facts the household reports, such as a scheme’s start year or the interest a fund credits, are never varied. Up to 6 of these cards are shown, after the changes to the plan. A free run shows at most one card in all, rotating between the changes to the plan and these. Every option named here is a decision, which a free plan runs on its default: the proportional order, no conversions, and each account setting’s cited default where it has one (a setting with none is asked on every plan).

16. Sustainable spending

This answers the planning question in reverse: how much could you spend each month and still hit your target success probability (90% by default)? Rather than reporting the success of one spending level, the model searches for the spending level that lands on the target.

It does so by bisection, repeatedly testing a spending level and then narrowing: if that level is still safe, try higher; if it fails, try lower; converge on the boundary. The search is bracketed first (does your current spending already pass, and is the plan fundable even with essentials, discretionary and healthcare all at zero?), then halved a handful of times until the answer is pinned to within about €25/month. If even those three at zero cannot reach the target, the bills that remain (housing, care, tax, and any one-off or family costs) already exceed what your income and portfolio can support, and the model says so rather than returning a figure.

17. How we check the engine against history

Historical calculators such as cFIREsim and FI Calc replay your plan through every retirement start year since 1871, using Robert Shiller’s US market data. We run the same check on our engine in our automated tests, every time the code changes.

The arithmetic, checked to the cent. We feed the engine the actual historical returns and inflation, one simulated future per start year, on a plain plan: a fixed spending amount rising with prices, a stock and bond mix rebalanced every year, Social Security, no tax, no fees, and a fixed 30 or 40 year horizon. We run cFIREsim’s own open-source code (cFIREsim-open, Apache 2.0) on the same plans. For every start year, both fail in the same year, and the money left at the end agrees to within cFIREsim’s rounding to the cent. A 4% withdrawal for 30 years, 75% in stocks, lasts in 112 of 116 start years in both.

Our random futures, compared with history. Our futures are not a replay of history, so their success rates differ from a historical calculator’s, and we measure by how much and why:

The order of the years (long bad decades, the recoveries after them) moves a historical success rate by up to about 7 points against the same years in random order. History in order is kinder at low withdrawal rates and harsher at high ones.
How we generate a year adds nothing measurable. Given history’s own averages, spreads and correlations, our futures land within one point of shuffled history in every case we test.
What we assume about the future is the whole of the remaining gap. By default we expect stocks to return 8% a year before inflation, with 2.5% inflation; from 1871 to 2014 US stocks returned about 10.3% with about 2.2% inflation. So for the same plan our chance of success is lower than a historical calculator’s: for 4% over 30 years, about 78% against about 95%. You can set your own return assumptions on the money step.