Best Funds for Retirement: Building and Spending a Corpus, by the Numbers

Retirement is the only financial goal you cannot borrow for, postpone honestly, or do over. It is also, mercifully, the goal where the mathematics is most on your side - if you respect one fact: retirement is two different investment problems wearing one name, and the fund that is best for one is frequently wrong for the other.

Problem one is building the corpus: decades long, survives volatility easily, rewards equity heavily. Problem two is spending the corpus: withdrawal-driven, punished brutally by early losses, rewards stability. This guide covers both with real numbers from our daily-refreshed dataset of 8,000+ Direct-Growth Indian mutual funds. Descriptive research, not advice; figures pre-tax.

First, the number: what does retirement actually cost?

Work backwards from monthly spending. The widely used planning shortcut says the corpus should be about 25 times annual expenses - the level at which a sustainable withdrawal rate has historically kept pace with inflation.

One adjustment most calculators soft-pedal: those are today's expenses. At 6% inflation, prices roughly double every 12 years - retire in 24 years and today's Rs 1 lakh lifestyle costs about Rs 4 lakh a month, pushing the required corpus toward Rs 12 crore in future rupees. The number is meant to be sobering; it is also entirely reachable with the arithmetic below, which is exactly why starting age dominates every other variable.

Phase 1: Building the corpus (age 25-55)

The engine: equity SIPs, held for decades

The median equity fund with a 10-year record in our data compounded at about 14.7% a year. Simulated SIPs through real NAV history show what quality funds turned that into: Rs 10,000 a month for a decade became Rs 30-39 lakh in the top tier (details and fund names in our long term SIP guide).

Stretch the horizon and the compounding becomes the whole story. Rs 25,000 a month at the historical median:

(Constant-rate illustration at the historical median - real paths swing hard around it and the median itself is not guaranteed.) Add a 10% annual step-up and the 25-year figure roughly doubles. The difference between starting at 30 and at 40 is not 40% - it is the difference between Rs 6.7 crore and Rs 1.5 crore. Time is the asset; the fund is only the vehicle.

Which funds for the accumulation decades

The long-horizon evidence points to a boring, powerful structure:

• Core (60-70% of instalments): a large & mid cap, flexi cap, or Nifty 50 index fund. The big Nifty 50 index funds have delivered about 12% over the last decade at 0.06-0.26% cost with identical -38% worst falls - predictable, cheap, no manager risk. Active core alternatives with stronger records live in our 5-year horizon ranking.
• Satellite (30-40%): a mid cap or small cap fund. Mid caps are the standout: 17.3% ten-year median with 92% of rolling years positive - the best consistency in equity. Small caps pay more (17.8% median) for a rougher ride (-32% median worst fall).
• If you use the 80C bucket anyway: ELSS. 13.9% ten-year median, and the 3-year lock per instalment is meaningless at a 20-year horizon. Current ranking.

What the accumulation phase does not need: capital "protection" products, insurance-linked investments, or more than 3-4 funds. Every rupee of drag - and Regular-plan commissions are pure drag, roughly 0.5-1% a year - compounds against the 25-year outcome. The Direct vs Regular math is worth ten minutes once.

What about NPS?

The National Pension System is the other serious accumulation vehicle - extra tax deduction, ultra-low cost, forced discipline, but locked till 60 with a mandatory annuity on 40% of the corpus. It is a genuine either-and-both question and we gave it its own full comparison: NPS vs mutual funds. Short version: the tax break is real, the lock cuts both ways, and the two work better as complements than rivals.

The glide path: the five years everyone skips

The most dangerous market crash of your life is the one just before or just after retirement day - maximum corpus, no salary to average with, withdrawals about to begin. A -35% year at age 59 does damage that a -35% year at 35 never could. This is sequence-of-returns risk, and the defence is mechanical:

From about five years out, move the corpus from equity toward hybrids and debt in planned annual tranches - not in one panicked switch, not based on market forecasts. A common destination mix at retirement day: roughly 40-50% in stability-first hybrids, 20-30% in high-quality debt, the remainder still in equity for the decades of inflation still ahead. Each switch is a taxable event, which is one more argument for spreading it over years.

Phase 2: Spending the corpus (60 onwards)

Now the problem inverts, and so does the fund profile. The categories that dominate accumulation - small cap, mid cap - are exactly wrong for withdrawals: their -32% falls meet your monthly redemptions and destroy units at the bottom. The data now favours the funds that never have a catastrophic year:

The striking fact repeats from our SWP research: the median conservative hybrid and equity savings fund has never posted a negative rolling 12-month return in our data - the worst year was still positive. Standout individual records at the moment include a conservative hybrid compounding 10% over five years with a worst-ever fall of just -2.3%. Named funds, the two-bucket withdrawal structure, and the exit-load traps are all detailed in the companion piece: best funds for SWP.

The withdrawal rate is the lever that decides everything: at 6% a year (Rs 50,000 monthly per Rs 1 crore) the corpus has historically kept growing through withdrawals in these categories; at 10% it is being consumed by arithmetic. And the single strongest safety feature is behavioural - the willingness to trim the withdrawal 10-15% after a bad year.

The complete lifecycle in one table

The retirement mistakes the data flags hardest

1. Starting at 40 what should have started at 28. Every table above says the same thing: the last decade of compounding is where most of the money appears. Delay is the one error that cannot be fixed by fund selection.
2. Being 100% in equity at 59. The corpus is largest exactly when a bad sequence hurts most. The glide path is boring; so are parachutes.
3. Being 0% in equity at 65. Retirement can last 30 years, and at 6% inflation prices double twice in that span. A corpus entirely in 6.5% debt is shrinking in real terms; the balanced advantage sleeve exists to fight that.
4. Products sold as "retirement plans". Insurance-investment hybrids and high-commission pension products routinely trail the plain SIP-then-SWP structure above by percentage points a year. Complexity in this space is usually a fee in costume.
5. Ignoring the behaviour gap. Across funds we track, investors' actual money-weighted returns routinely trail the funds' own returns by 2-4 points - the cost of buying high, stopping SIPs in crashes and switching at the wrong times. The plan above only works if it is left alone to work.

A worked case: 32 years old, Rs 1 lakh monthly income

Frameworks convince; examples convert. An illustration at the historical medians (pre-tax, unguaranteed):

The target. Current spending Rs 55,000 a month. At 6% inflation over 28 years to age 60, that lifestyle costs about Rs 2.8 lakh a month - roughly Rs 34 lakh a year, implying a corpus target near Rs 8.5 crore at 25x. The number looks impossible; the plan below reaches it with a savings rate that starts at 20% of income.

The engine. Rs 20,000 a month into the two-fund SIP structure (core large & mid cap + mid cap satellite), with a 10% annual step-up tracking salary growth. At the equity median of 14.7%: the corpus passes Rs 1 crore around year 12, Rs 3 crore around year 18, and lands in the Rs 8-10 crore zone by year 28 - the step-up doing fully half the work. (A flat Rs 20,000 without the step-up lands near Rs 4.5 crore: the difference between comfortable and constrained retirement is the annual 10% instruction.)

The layers. EPF continues untouched in the background - its assured-return corpus becomes part of the debt allocation at retirement. Rs 50,000 a year into NPS captures the 80CCD(1B) deduction (about Rs 15,600 of tax saved annually at the 30% slab). Term life and health insurance sit outside the investment stack entirely - protection bought as protection, never bundled.

The glide. From 55, one-fifth of the equity corpus moves each year into the conservative hybrid and balanced advantage funds from the table above. At 60: roughly half the corpus in stability-first vehicles, a Rs 20+ lakh liquid-fund runway, an SWP starting near 6% - and the remaining equity sleeve still compounding against the three decades of inflation ahead.

Every number scales linearly - a Rs 50,000 income runs the same plan at half the amounts. What does not scale is the start date: begin the identical plan at 42 instead of 32 and the terminal corpus divides by roughly three.

Where EPF and PPF fit - and where they cannot

Most Indian retirement conversations start with EPF and PPF, so the honest placement:

EPF is the automatic backbone - assured returns (historically 8-8.5%), employer matching, brutal-to-withdraw. Treat it as the permanent debt allocation of the retirement plan: it is why the SIP engine above can afford to be fully equity. Counting EPF as "already saving for retirement" and skipping equity entirely is the classic middle-class error: at 8.5% pre-tax, EPF alone rarely outruns lifestyle inflation to a 25x corpus.

PPF offers tax-free assured compounding with a 15-year lock - excellent as a conservative sleeve, capped at Rs 1.5 lakh a year, and sharing the crowded 80C bucket with ELSS and insurance. In the framework above it substitutes for part of the debt glide-path destination, not for the equity engine.

What neither can do is the 14-17% compounding work that turns Rs 68 lakh of stepped-up contributions into several crores. The assured-return instruments are the floor of the plan; the equity SIP is the escalator. Portfolios that confuse the two jobs - all floor, no escalator - arrive at 60 with safety and insufficiency simultaneously.

The double squeeze: healthcare and longevity

Two variables deserve explicit space in any retirement model, because both run worse than general inflation:

Healthcare costs in India have compounded meaningfully faster than CPI - medical inflation estimates routinely run 10-14%. The defence is structural: a substantial health insurance cover maintained without gaps from well before retirement (fresh cover after 60 is expensive and exclusion-ridden), plus a dedicated medical buffer - several years of premiums and deductibles in a liquid fund, outside the SWP arithmetic entirely.

Longevity cuts the other way: a healthy 60-year-old today plans for 25-30 years of withdrawals. That is the quiet argument for keeping 25-40% equity through retirement (the balanced advantage sleeve in the spending table) and for the 6%-not-10% withdrawal discipline. A corpus that merely preserves its nominal value halves in purchasing power by 72 at 6% inflation. The retirement plan that survives is the one still growing at 75.

FAQ

Which single mutual fund is best for retirement?
No single fund does both jobs well. The closest one-fund compromise for later life is a balanced advantage fund (10.2% median, -16% median worst fall), and for early accumulation a flexi or large & mid cap fund - but the two-phase structure above beats any single-fund answer.

Is Rs 1 crore enough to retire in India?
At a 6% withdrawal it generates Rs 50,000 a month pre-tax - the 25x table converts that: enough for Rs 50,000 of monthly expenses at the moment of retirement, before inflation. The honest answer depends on age, inflation ahead, and other income; the framework above lets you compute rather than guess.

Mutual funds or NPS or both for retirement?
Usually both, doing different jobs - the full comparison walks through tax, lock-in, cost and flexibility line by line.

When should I start moving out of equity?
The data-based answer: begin about five years before withdrawals start, in annual tranches. Earlier wastes compounding years; later gambles the largest corpus you will ever have on the market's next mood.

Every category median and fund record in this article comes from the terminal's daily-recomputed dataset - and every fund mentioned has a full page there: rolling windows, drawdowns in rupees, SIP and SWP simulations. Free account. The retirement you fund at 30 costs about a third of the one you fund at 40; the data is simply the reminder.

How much should health insurance cover be at retirement?
More than feels comfortable: with medical inflation running 10-14%, a Rs 10 lakh family floater at 45 is a Rs 40+ lakh equivalent need by 65. The investment plan above assumes the medical risk is insured separately - one uninsured hospitalisation can consume years of SWP withdrawals, which is why the cover and the buffer sit outside the corpus arithmetic entirely.

Should I use my retirement corpus to pay off my home loan first?
The arithmetic comparison is loan rate vs conservative portfolio return - a 8.5-9% loan against 8.5-10% hybrid medians is close to a wash, and entering retirement debt-free has a sequence-risk value the spreadsheet understates: it lowers the mandatory monthly withdrawal, which lowers the rate, which is the single strongest lever in the whole plan. Most retirements are safer with the loan gone.

What about rental income in the retirement plan?
Treat it as a parallel SWP with its own risks (vacancy, maintenance, illiquidity) and count only a haircut version - say 70-80% of gross rent - in the income arithmetic. Its real virtue is inflation linkage; its real cost is concentration. The framework above simply runs on a smaller required corpus when reliable rent covers part of the monthly need.

Is it too late to start at 50?
Late is not never: a 50-year-old with ten working years still has one full compounding decade, and the data's Rs 30-39 lakh per Rs 10,000 SIP outcome applies to exactly that window. What changes is honesty about the arithmetic - a shorter runway means a higher savings rate does the work that time cannot: at Rs 50,000 a month with a step-up, the ten-year corpus lands near Rs 1.3-1.9 crore at historical medians. The glide path compresses too: de-risking starts almost immediately after accumulation peaks. Late plans succeed on aggression of contribution, never on aggression of allocation.