The Three-Bucket Strategy for Indian Early Retirees

Somebody retired in December 2007 with a corpus they had checked three times. Fourteen months later the Sensex was down about 60% from its peak and they were still redeeming units every month to pay for groceries. Their arithmetic had been fine. What it never covered was the order the returns would turn up in.

That is the problem the bucket strategy exists to solve, and it has a name - sequence risk. Two retirements with identical average returns over thirty years can end very differently depending on whether the bad years came first or last, and you do not get to choose.

Why not just do SWP from one balanced fund?

You can, and plenty of people do. A 60:40 fund with a systematic withdrawal plan is a perfectly defensible retirement. The fund house rebalances internally, you get one statement, and there is nothing to maintain.

What you give up is control over which asset gets sold in a bad month. When you redeem units from a balanced fund in March 2020, you sell equity and debt in proportion. You had no choice in the matter. The bucket structure is essentially a manual version of the same portfolio, where you keep the right to decide what gets liquidated and when.

It is more work. Two or three folios, a refill decision once a year, and the temptation to fiddle. If you know you will not do the maintenance, the balanced fund is the better answer for you and this guide is optional reading.

The three buckets

Bucket 1 is not an emergency fund. Keep the emergency fund separate and on top - the safety bucket is your planned spending for the next few years, and if you raid it for a car repair your runway gets shorter without anything on your statement telling you so.

Sizing bucket 1

Every article you will read says “two to three years of expenses” and moves on. The number is roughly right and the reasoning is usually missing, so here it is.

The bucket has to fund N years of spending. Each year's withdrawal is bigger than the last because of inflation. Meanwhile the money still sitting in the bucket keeps earning the safe rate. So the correct size is the present value of a rising withdrawal stream, discounted at your post-tax safe return:

Withdrawals happen at the start of each year, because that is how spending works. Year one's money comes out on day one and earns nothing.

Now the useful bit. When your post-tax return exactly equals inflation, k is 1 and the formula collapses to annual expense × N. Which is the rule of thumb. So the two-to-three-years rule turns out to be the exact answer to one specific case, the case where your safe money keeps pace with prices, and whether it does depends almost entirely on tax.

The spread is about ₹1.3 lakh on a ₹48 lakh bucket, which is small in percentage terms. But look at which direction it moves. In the 30% slab, where most people reading this actually sit, the number lands above the rule of thumb. Your safe money is shrinking in real terms at 4.55% against 6% inflation, and the bucket has to be bigger to absorb that.

Which is also the argument for arbitrage funds over bank FDs for this slice. Same practical risk for a 3-4 year horizon, taxed at 12.5% LTCG once held past a year instead of at your slab. On the numbers above that is ₹79,000 of corpus you get to leave in equity instead.

How many years should it cover?

Three to five. The reasoning is recovery time, not a formula. Indian equity drawdowns have historically taken somewhere between two and four years to get back to the previous peak - the 2008 crash took about three years in Sensex terms, and 2020 took under one. A bucket sized to cover that window means you can leave bucket 3 alone through the worst of it.

Going bigger is expensive and it feels prudent, which is a bad combination. Every rupee sitting at 4.55% post-tax is a rupee not compounding at 12%. Over a thirty-year retirement that drag is not small. Ten years of expenses in FDs is insuring against something that has not happened in the history of the Indian market.

If you are retiring at 42 rather than 58, lean towards five. You have more years of sequence risk ahead and no EPF or pension arriving to bail you out mid-way.

Sizing buckets 2 and 3

Bucket 2 is the same arithmetic, one layer out. Eight years of expenses in short-duration debt, corporate bond funds and a little hybrid. Its job is to still be there when bucket 1 has run out, so it needs to cover several consecutive bad years on its own without you ever going near equity.

Add buckets 1 and 2 together and you get twelve years of expenses in non-equity. On a ₹4 crore corpus with ₹12 lakh annual spending, that is roughly ₹1.4 crore, so about 36% of the portfolio. Which is not a coincidence - it lands close to a conventional 70:30 equity-debt split. The bucket framing just tells you why 30, and gives you a rule for what to sell.

Bucket 3 is whatever is left. There is no separate calculation for it. If the remainder after buckets 1 and 2 is too small to grow at the rate your plan assumes, then you are looking at the wrong problem. Your corpus is too small and you are not ready yet.

Refilling, which is the hard part

Sizing the buckets on day one takes an afternoon. Keeping them sized for thirty years is the actual discipline, and it is where most people drift.

Pick a date. Once a year, same week, ideally not in a month when the market has just done something dramatic. Then three rules, in order:

• Spend bucket 1 down to nothing before you touch bucket 2. There is no annual top-up between them. Bucket 2 earns more than bucket 1 does, so every year you leave it invested is a year it is working, and letting bucket 1 empty out first costs you nothing you can spend. Together they are really one pool of safe money with the cheaper half spent first.
• When that pool drops below about four years of spending, go to equity and refill it back to twelve. Not on a calendar, not because markets look good, but because the safe money has run down to the point where another bad stretch would reach equity anyway.
• Unless the sale would cost you more than half of bucket 3. In that case sell one year of spending, no more, and come back to it next year. This is the rule that stops a crash turning into a liquidation - if equity has fallen far enough that a full refill would take half of what is left, you take the minimum and give it time.

The reason the second and third rules are separate is that they run on different clocks. The refill trigger is about your own runway and it fires every eight to ten years on a 4-and-8 split. The half rule is about price, and it only ever fires after equity has already fallen a long way. Most years neither one does anything and equity is left completely alone, which is the entire point.

The failure mode to watch for is selling equity during a crash because the safe pool looks uncomfortably thin. That is the exact behaviour the structure was built to prevent, done with extra steps. The pool is supposed to look thin after three bad years, which is what it was filled up for, and the correct response is to leave it alone.

Where the money actually comes out

Bucket 1 is where your monthly spending lands, and there are two reasonable ways to run it. A sweep-in FD that credits your savings account, or an SWP from an arbitrage or liquid fund into the same account on the 1st. The second is usually better on tax and both are fine.

The tax treatment differs by bucket, and it is worth setting up with that in mind rather than fixing it later:

The debt fund line is the one that surprises people who set their plan up before 2023, because indexation is gone. A debt fund and an FD are now taxed the same way, which removes most of the reason to prefer the fund for bucket 2 and leaves you choosing on liquidity and convenience instead. For a fuller treatment of the withdrawal order across EPF, PPF and NPS as well, see the post on withdrawal sequencing.

What this does not fix

The bucket strategy protects you from having to sell equity at a bad price. It does not protect you from a corpus that was too small, and it does not create returns. If your plan needs 11% real and the market hands you 6% for two decades, no arrangement of the same money survives that.

It also has a cost that people who advocate for it tend to skip past. Holding 30% in low-return instruments drags your overall return down compared to a 100% equity portfolio that you never panic-sell from. Over thirty years that difference is large. The bucket strategy is insurance, and like all insurance you are paying a premium for the scenario where you need it. If you are certain you would hold through a 60% drawdown without touching anything, you would end up wealthier without the buckets. Most people are not certain, and most people who think they are have not lived through one while unemployed.

There is also a live debate about whether buckets do anything mathematically at all, or whether they are the same portfolio as a static allocation with rebalancing, dressed up in a story. I think the critics are largely right on the maths and largely wrong on what matters, because the story is doing real work here. A structure you will actually follow in March 2020 beats a marginally better one you abandon in March 2020.

TL;DR