Tax-saving investments, and what the deduction is actually worth

What a deduction is actually worth

A deduction reduces the income you are taxed on, not the tax you owe. So it returns the amount invested multiplied by your marginal rate — the rate on your topmost slice of income — and nothing at all on the default new regime, where it does not exist.

Same instrument, different people, different value. Write it as one line and the whole subject becomes tractable:

Cash back = amount invested × your marginal rate.

Two people put the identical sum into the identical instrument on the identical day. The cash each gets back differs in exactly the ratio of their marginal rates. Nothing about the instrument changed; nothing about their behaviour differed. The benefit was never a property of the product.

Three consequences follow, and they are the reason most tax-saving coverage is unhelpful:

• Someone whose income falls below the taxable threshold gets nothing. Their marginal rate is zero, so the deduction returns zero, however good the instrument may be on its own merits.
• The advice cannot be general. An article that says an instrument “saves you” a stated amount has silently assumed a marginal rate, and it is probably not yours.
• Under the default regime, the rate is irrelevant because the deduction is not available at all. That is the next section, and it comes before everything else.

Three distinct things get called “tax saving” and they behave differently. A deduction cuts taxable income, so it is worth your marginal rate. An exemption removes a receipt from the computation entirely. A rebate cuts the tax itself, rupee for rupee, and is therefore worth the same to everyone who qualifies. Only the first is what this article is about.

Two refinements to that one-line formula, both of them skipped in most coverage. The first: a deduction large enough to carry your income down past a slab boundary is not relieved at a single rate at all. The part above the boundary comes off at the higher rate and the part below it at the lower, so the r that matters is the blended rate over the slice actually removed, not the rate on your first rupee of income.

The second is sharper. Near a rebate threshold the arithmetic stops being marginal altogether: a deduction that brings total income under the threshold can extinguish the tax rather than shave a slice off it, so its value jumps rather than scaling. Both effects belong to the old regime, both turn on thresholds that move with each Finance Act, and both are worth checking against the current schedule rather than assumed.

Which regime you are in decides whether any of this exists

If you are on the new regime, none of this exists — and the new regime is the default under section 202 of the Income-tax Act 2025. Under it there is no deduction for the tax-saving basket at section 123 read with Schedule XV, none for health insurance premium at section 126, none for education loan interest at section 129, and no deduction for interest on a loan for a self-occupied house. A taxpayer on the default regime who invests the full ceiling receives, in tax terms, nothing.

The old regime remains available by election, and the two differ in their rate schedules as well as in what may be deducted. So the comparison is not “deductions versus no deductions”. It is: does the tax saved by the deductions you would genuinely claim exceed the extra tax the old regime's rate schedule charges you? Both halves have to be computed on your own numbers, and the answer changes as income and claims change.

The order of operations that follows from this is unusual, and it is the opposite of how most people do it in March:

1. Total the deductions you would actually claim — the ones already happening, not the ones you might make happen.
2. Compute tax under each regime on that basis.
3. Then decide whether any further tax-saving investment is worth making.

Doing it the other way round — investing first, choosing the regime afterwards — produces locked money that turns out to buy no deduction at all.

The basket is usually already partly full

Here is the mistake that costs the most, and it is arithmetic rather than judgement.

The main tax-saving deduction is one aggregate ceiling, currently ₹1.5 lakh under the old regime, shared across everything that qualifies. It is not a ceiling per instrument. And several of the qualifying items are things a salaried household is already doing without thinking of them as investments:

• the employee's own provident fund contribution, deducted from salary every month;
• premium on a life insurance policy already in force;
• children's school tuition fees;
• the principal portion of a home loan EMI — the principal only, which is the smaller part of an early-tenure instalment.

The named failure mode: the double-counted ceiling. Someone whose provident fund contribution, insurance premium and tuition fees together already reach the limit invests the full ceiling again in March. That second amount is locked for years and buys a deduction of zero, because the ceiling was consumed before the cheque was written. The investment may still be fine on its own merits — but it was bought for a benefit that was not there.

The remedy takes one sheet of paper: list what is already flowing into the ceiling for the year, subtract from the limit, and the remainder is the only amount where the tax question even arises. Everything above that is an ordinary investment decision and should be judged as one, alongside the goal it is meant to fund.

There is a second-order effect worth naming. If the provident fund contribution plus a tax-saving deposit fills the ceiling, then the entire tax-saving allocation is in debt — which is an asset-allocation decision, whether or not anyone made it as one. For money earmarked 20 years out, the tax rule has quietly set the mix.

Compare the basket on lock-in and liquidity, not on returns

Ranking these instruments by return is the standard treatment and it is close to meaningless, because they are not the same kind of thing. An equity fund's return is market-determined and unknown; an administered-rate instrument's is set by the government and revised periodically; a deposit's is contractual. Lining up three numbers of different kinds in one column invites a comparison that cannot be made.

What genuinely distinguishes them is the shape of the lock-in, what the money is doing economically, and what can end the lock-in early. Everything in the table below qualifies for the deduction under the old regime only; on the default new regime the column that groups them does not exist, and each row has to stand on the other three columns alone.

Read down the second column rather than the third. Four of the eight rows are debt of one sort or another, one is equity, one is an insurance contract and one is not an investment at all. The tax rule has grouped together instruments that have almost nothing in common except the deduction, which is why comparing them on returns produces nonsense and comparing them on liquidity produces something usable.

One row deserves a note. A life insurance premium belongs in this table because it qualifies, not because it competes. Term cover and an investment are separate purchases that a bundled policy combines, and the arithmetic of separating them is worth running before the deduction is allowed to decide it.

The deduction is a discount on the purchase price, not a return

This is the reframe that makes the basket comparable, and it is arithmetic anyone can redo.

Call your marginal rate r, including any surcharge and cess that applies to you. You invest an amount A. Taxable income falls by A, so tax falls by A×r. You now hold an asset worth A, for which you parted with only A×(1 − r) of your own money.

So the deduction is not a return on the investment. It is an instant uplift on the capital actually committed, and its size is:

r ÷ (1 − r)

Note that this is larger than r itself, which is the pleasant surprise. Now the unpleasant one. That uplift happens once, and the lock-in lasts years. Spread across a lock-in of n years, the equivalent annual rate is

(1 ÷ (1 − r))^1/n − 1

— which shrinks as n grows. The same deduction, at the same marginal rate, is worth materially more per year on a short lock-in than on a long one. Put your own r and each instrument's n into that expression and you have a single number that makes the basket comparable for the first time, because it converts a one-off cash benefit into the annual yield equivalent that the illiquidity cost you.

Worked out, the spread is wide enough to change decisions:

The rates in the first column are round numbers chosen to show the shape of the arithmetic, not slab rates — substitute your own r and redo it. What the table shows is that the deduction and the lock-in are the same transaction seen from two ends. At 30%, a 3-year lock-in converts the deduction into something worth 12.6% a year; the identical deduction on a 15-year lock-in is worth 2.4% a year, which most administered-rate instruments would cover several times over on their own. The long lock-ins are not being paid for by the tax benefit. They are being paid for by whatever the instrument itself does, and the deduction is a rounding adjustment on top.

Three honest caveats, because the expression flatters the deduction if you skip them. It assumes you are on the old regime — on the default new regime r is zero for this purpose and the whole calculation collapses. It assumes the amount sits within the unused part of the ceiling — above that, r is effectively zero for the excess. And it ignores tax at the other end: an instrument taxed on withdrawal gives part of the uplift back. That is the next section.

The trade-off, stated plainly: the discount is real, it is one-off, and it is paid for with years of illiquidity. Whether that is a good trade depends on whether you would have locked the money anyway. If you would — because the goal is genuinely distant — the discount is close to free. If you would not, you have sold liquidity to buy a single year's tax benefit — and the money you might need at short notice is the worst possible source for it, because illiquidity is not a side effect of these instruments, it is the feature the deduction is buying.

Three points where tax can land, and only one of them is the deduction

A tax-saving instrument meets the tax system three times, not once. Most coverage discusses the first meeting and skips the other two, which is how a product with a generous entry benefit and a taxed exit gets sold as though the entry benefit were the whole story.

The middle row is the one that compounds, and it is the mechanism behind every deferral argument in investing. Money taxed as it accrues is money removed from the base that generates next year's growth; money taxed only at exit keeps working until then. Over one year the difference is trivial. Over a 15-year lock-in it is not, for the same reason that time does the heavy lifting in any compounding argument. The direction of that effect never changes, whatever the Finance Act does to the rates.

Instruments differ at every stage, and several combinations exist — deductible going in and taxed coming out, deductible going in and exempt throughout, not deductible but exempt at exit. Check all three stages for each instrument you are considering, from the current Act rather than from any article, because this is exactly where secondary coverage is stale most often.

One stage is settled enough to state, because ELSS is an ordinary equity-oriented fund and the equity rules are unambiguous. Gains on redemption above ₹1.25 lakh in a year are taxed at 12.5% under section 198 of the Income-tax Act 2025, the provision that was section 112A. And there is a small structural neatness here: the long-term holding period for equity units is 12 months, while the lock-in is three years, so an ELSS redemption is always long-term by construction. The short-term rate at section 196 — formerly section 111A — cannot apply to it. That is a genuine, if modest, structural advantage of the lock-in rather than a cost of it.

The deduction you would rather not use

Health insurance premium sits in a separate deduction at section 126 — formerly section 80D — with its own ceiling — ₹25,000, raised to ₹50,000 where the insured is a senior citizen — applied separately to your own family and to your parents. It does not draw on the section 123 ceiling, so it is genuinely additional — but like the rest of the basket it exists only under the old regime, and is worth nothing on the default one.

It is worth separating from the investment instruments for a reason that has nothing to do with tax. The other items in the basket are purchases you hope to benefit from. This one is a purchase you hope never to claim on. Its value is the transfer of a risk you cannot absorb, and the deduction lowers the cost of that transfer without changing what it is.

Which means the sizing question is not a tax question. The cover you need is set by what a hospital admission would cost you, not by what the limit permits you to deduct, and those two numbers have no reason to agree. Buying to the limit when you need more cover is a failure dressed as optimisation; buying more than you need because the deduction is available is spending money to save a fraction of it. The mechanics of choosing the cover are in the health insurance guide, and the tax treatment should be the last input, not the first.

Four ways this goes wrong

Each of these is specific enough to recognise yourself in.

1. Deciding in March. The deduction is claimed for the year in which the payment is made, which concentrates the entire decision into the last few weeks of the financial year — the point at which the least comparison and the most selling happens. The same investment made in April buys the identical deduction, with 11 more months of thought behind it and, for a market-linked instrument, 11 more months of compounding. It also leaves time to discover that you are on the default new regime and that there was no deduction to buy.
2. Buying insurance as an investment because it qualifies. A deduction lowers the cost of a product. It does not improve the product. A policy with a mediocre internal return is still a policy with a mediocre internal return after the deduction, and the deduction is received once while the return runs for decades. Compute the rate the cashflows actually imply before the tax benefit is allowed into the argument.
3. Locking money that is not spare. Most of this basket cannot be accessed at all before its term ends. A household with no liquid buffer that puts its buffer into a locked instrument for a deduction has converted a solvable problem into an unsolvable one, and meets the next emergency on a credit card.
4. Treating the deduction as the return. The clearest symptom is someone who can quote the amount they saved in tax but not what the instrument earned. The first is a one-off; the second is what the money does for the whole lock-in and beyond, and what it does after inflation is the only version that matters for a long horizon.

None of the four is exotic. All four come from the same root: treating the tax rule as the decision rather than as one input into it.

Working it out for yourself

The whole subject reduces to five lines, in this order. None of them requires anyone's opinion.

1. Which regime applies to you once both are computed on your own figures. If it is the default new regime, the deduction arithmetic stops here and every instrument below has to justify itself as an investment alone.
2. What is already flowing into the ceiling — provident fund contribution, existing premium, tuition fees, home loan principal. Subtract from ₹1.5 lakh.
3. Your marginal rate, r, including surcharge and cess. The remaining headroom multiplied by r is the largest cash benefit available to you this year, and it is usually smaller than expected.
4. For each candidate, the annual yield equivalent of that benefit: (1 ÷ (1 − r))^1/n − 1, where n is that instrument's lock-in.
5. Whether you would hold the instrument without the deduction. If the honest answer is no, the deduction is buying you a position you do not want at a discount.

FNOTrader is not a SEBI-registered investment adviser and this is not tax advice. What is described here is how the arithmetic works, so that the decision can be made on your own numbers rather than on a general claim about what an instrument “saves”.

Testing the market-linked half on real data

The administered-rate instruments in the basket declare their own terms, so there is nothing to test. The market-linked ones do not, and for those the question — what did holding this through a full lock-in actually feel like? — is answerable from history rather than arguable.

FNOTrader's Mutual Funds app runs on the full AMFI NAV history, around 34 million rows, and simulates a lumpsum or a monthly contribution into any scheme over any period, reporting the internal rate of return for irregular cashflows — XIRR — alongside invested versus value and maximum drawdown. For a locked instrument the relevant output is the rolling-return distribution over windows the length of the lock-in: because the money genuinely cannot be withdrawn, the worst such window is not a hypothetical, it is the outcome an unlucky start date would have delivered.

Past returns describe what happened over a stated period and are not a guide to what any scheme will return next. The reason to look at the worst window rather than the average is precisely that it is the case the average conceals.

Common questions

Continue reading

More in Tax Planning · App: Mutual Funds · Definitions: glossary · Free tools: calculators · All: every article