Prefer to understand the assumptions first? Start with the basics below, then return to the calculator with more context.
There is no universal retirement corpus. A useful estimate starts with the spending your portfolio must support at retirement, subtracts reliable non-portfolio income, then models how long withdrawals may last under inflation, investment returns, taxes, fees and uncertain market sequences. A single salary-multiple or “25× expenses” shortcut can be a useful checkpoint, but it is not a complete retirement plan.
Start with spending, then model how long the money may need to last.
There is no single retirement corpus that is right for everyone. Two people with the same salary can need very different amounts because their spending, retirement age, housing costs, healthcare needs, pensions, family commitments and desired safety margin may be different.
A practical retirement estimate usually has four stages: Estimate retirement spending in today's money, grow those costs for inflation until retirement, model the years after retirement, and then compare the required corpus with the assets and income you expect to have.
Why salary is not the best starting point
Salary is useful for understanding how much you may be able to save, but retirement is funded by expenses. If you earn a high income but live on a modest portion of it, a salary multiple can overstate your needs. If you spend most of your income, support family members or expect large healthcare and housing costs, the same salary multiple can understate them.
For planning purposes, separate expenses into categories. Housing may fall after a mortgage ends. Commuting costs may disappear. Travel may rise. Healthcare may rise later in retirement. Some costs continue for life while others are temporary.
Inflation can change the number dramatically
Suppose a household spends 100 units per month today. At a constant 5% annual inflation assumption, the same basket would cost about 163 units in 10 years, 265 units in 20 years and 432 units in 30 years. That is not a forecast; it is a scenario showing why today's spending and future spending should not be mixed.
Retirement length matters as much as retirement age
Retiring at 55 rather than 65 has two effects at the same time: There are fewer years to build the portfolio and more years in which the portfolio may need to fund spending. Planning to age 90 rather than 80 also changes the required amount. A retirement calculator should therefore ask both when retirement begins and how long the plan should run.
What about the 25x rule?
A common shortcut is to multiply annual retirement spending by 25, which is mathematically the same as dividing spending by a 4% first-year withdrawal rate. It can be a useful sense check, but it is not a personalized retirement plan. Taxes, inflation, investment returns, retirement length, sequence of returns, pensions and one-time goals can all move the answer materially.
Existing savings should reduce the gap, not disappear from the calculation
What you have already saved has time to grow before retirement. A good model projects existing savings separately from new contributions, then shows the remaining gap. That is more informative than calculating a target corpus in isolation.
Use a range, not one magical number
Try more than one inflation rate, return assumption and retirement age. If a plan only works under the most optimistic inputs, that is useful information. If it remains reasonably funded across a range of assumptions, that is also useful information.
Build the target from retirement cash flow, not from salary alone
Salary matters mainly because it helps describe your current lifestyle and saving capacity. Retirement funding is driven by what you expect to spend, which expenses disappear or begin, and what income will arrive from pensions, annuities, government benefits, rent or part-time work. Two people with identical salaries can require very different portfolios if their retirement spending and guaranteed income differ.
Retirement spending − reliable non-portfolio income = amount the portfolio must fundInflate today’s spending to the retirement date
If your estimate begins in today’s money, move it forward consistently. A $6,000 monthly lifestyle today becomes about $19,200 per month after 20 years at 6% inflation. That is not a forecast of one country’s inflation; it simply shows why mixing today’s expenses with a future nominal corpus can understate the target.
Current monthly spending × (1 + inflation)^yearsThen model the retirement phase as a stream of withdrawals
A corpus is not just annual spending multiplied by the number of retirement years because the remaining portfolio may keep earning returns while withdrawals rise with inflation. Under a simplified deterministic model, the present value of a growing withdrawal stream can estimate the starting portfolio. Real life is less tidy because returns and inflation vary year by year.
Why a withdrawal-rate shortcut gives a different answer
The familiar “25× annual spending” shortcut corresponds to a 4% first-year withdrawal. It was popularized from historical withdrawal research, but the original research had specific asset-allocation, historical-period and inflation-adjustment assumptions. Forward-looking research also produces different starting rates depending on time horizon, asset mix and spending flexibility. Treat the multiplier as a lens, not a law.
| Starting withdrawal rate | Approximate spending multiple | Interpretation before taxes/fees |
|---|---|---|
| 5.0% | 20× | Higher starting spending; less margin for long horizons or poor early returns |
| 4.0% | 25× | Common historical-rule shorthand; not a universal guarantee |
| 3.5% | 28.6× | Larger initial portfolio for the same spending |
| 3.0% | 33.3× | More conservative initial spending relative to portfolio |
Retirement income sources reduce the portfolio burden
If an inflation-linked pension covers essential spending, the investment portfolio is funding a smaller and potentially more flexible liability. If income is fixed in nominal terms, its real purchasing power declines. Model each income source by start date, duration and inflation behavior instead of subtracting one lifetime number.
The target is a range because several risks interact
- Sequence risk: Poor returns early in retirement can permanently damage a withdrawal plan even if long-run average returns later recover.
- Longevity risk: The portfolio may need to fund more years than an average life-expectancy estimate suggests.
- Inflation risk: Persistent price increases raise the currency amount needed to preserve purchasing power.
- Spending shocks: Healthcare, care needs, family support, home repairs and taxes can differ from the base budget.
- Behavioral risk: Abandoning the plan after a market shock or overspending after a strong market can change outcomes materially.
A stronger corpus process uses three layers
- Base case: Expected spending, income, longevity and a moderate return assumption.
- Stress cases: Lower returns, higher inflation, an early bear market, longer life and one-off expenses.
- Decision rules: What you would change if the stress case occurs—delay a discretionary purchase, reduce flexible spending, work longer, rebalance, or use other income.
What to review every year before retirement
- Retirement date and planning horizon.
- Current essential versus discretionary spending.
- Expected pension/benefit income and start dates.
- Assets genuinely available for retirement after excluding emergency funds and other goals.
- Investment allocation, fees and tax structure.
- Whether the corpus still works under less favorable assumptions.
Do not mistake a deterministic corpus for a probability of success
A present-value formula assumes one return and one inflation rate every year. It is excellent for understanding the relationship among spending, return, inflation and time, but it cannot show how many bad return sequences the portfolio survives. Simulation can add that dimension, yet Monte Carlo output is also model-dependent: The result changes with return assumptions, correlations, inflation process and the definition of “success.”
Define success before choosing the corpus
Does success mean never reaching zero? Preserving the starting real principal? Leaving a legacy? Maintaining full discretionary spending? Each objective implies a different required margin. A plan should not advertise “90% success” without stating what the 10% failure and the terminal-value distribution actually mean.
Use mortality and planning horizon separately
Planning to age 95 or 100 is a risk-management choice, not a prediction that the individual will live exactly that long. For couples, model the last survivor. If lifetime income covers essential expenses, the investment portfolio may be allowed more flexibility in later years than a plan where every expense depends on the portfolio.
Frequently asked questions
Is 25 times annual expenses enough to retire?
It is a useful shorthand for a 4% starting-withdrawal concept, but it ignores individual longevity, taxes, non-portfolio income, spending flexibility, asset allocation and current market conditions. Model those separately.
Should I use current expenses or expected retirement expenses?
Use a retirement budget. Start with current spending, remove expenses likely to end, add new retirement costs, and then express everything consistently in today’s money or future money.
Should my house be counted in the retirement corpus?
Only to the extent you have a realistic plan to convert housing wealth into retirement cash flow. A home that you intend to live in is an asset, but it does not automatically pay monthly expenses.
Why does the result change so much with inflation?
Retirement can span decades. A small annual inflation difference compounds repeatedly, raising both the first-year retirement budget and later withdrawals.
Sources and how this guide was prepared
This Carrowmont guide was drafted with AI assistance, then structured and checked against the cited external sources and Carrowmont calculation conventions. Worked examples are illustrative, use simplified assumptions, and are not forecasts or individualized recommendations.