Guide · Costs and taxes · Updated 2026-09-19

The expense ratio is taken daily before the NAV is published, and the figure you compare stopped meaning what it used to mean

An investor asked what a scheme costs will name a percentage. Asked what it is a percentage of, most name the gain, as though the charge bites in a good year and disappears in a bad one. It does neither. It is a share of the entire balance, taken every day, on money contributed and on money already earned, whether the scheme rose or fell.

That misreading is applied to the largest number in a portfolio for the longest period available, which is why it outranks every other cost error open to an Indian investor.

A share of the balance, not a share of the return

A transaction cost is charged on the amount transacted. A performance fee, where one exists, is charged on the gain. An expense ratio is charged on neither. It is charged on net assets: the whole accumulated balance, contributions and accrued growth together.

So the charge scales with success. A holding that has quadrupled pays four times what it paid at the start, for the same mandate and usually the same work. Nor does it pause in a year the scheme falls: the holder is charged on the reduced balance, smaller in rupees and identical in rate.

The second consequence makes the arithmetic later on this page unforgiving. What the fee takes is not just the fee. It is the fee plus everything it would have earned had it stayed. Money removed in year three cannot compound through years four to thirty, and that foregone compounding, not the headline percentage, is what turns one point into a quarter of a terminal balance.

Charged daily, and gone before the NAV is published

At the end of each dealing day the scheme's portfolio is valued. Expenses for that day are accrued against it, at the annual rate divided by the number of days in the year. Only then is the net asset value struck and published. The figure an investor looks up is already net of everything the scheme charged that day.

None of this is visible to the unit holder. No units are redeemed, nothing is debited, and the statement carries units and a value with no cost line on it. The charge exists only as an amount the published figure is lower by, and nobody sees a number never shown. That is not concealment: accruing into the price is the only workable way to charge a pooled vehicle whose holders enter and leave daily, because it makes whoever held units on a day bear that day's expense in proportion. The design is correct and the behavioural consequence still real. The same person who compares two one time charges to the last rupee will carry a recurring one for twenty years without once computing what it came to.

One practical point follows. An annual rate is not a bill falling due once a year, so leaving early avoids nothing. A month of holding costs roughly one twelfth of it. Cost accrues from the first day, in proportion, exactly as growth does.

The slabs are marginal, so the permitted rate is a blend

The regulator sets a ceiling that falls in steps as daily net assets rise, on the reasoning that the cost of running a scheme is largely fixed. Research, compliance, accounting, custody and management do not double when assets double, so a flat percentage would hand the entire scale benefit to the manager.

The step that catches people is that these slabs are marginal, exactly as income tax slabs are. A scheme with Rs 6,000 crore of daily net assets does not charge the rate attached to its band on all Rs 6,000 crore. It charges the top rate on the first Rs 500 crore, the next rate on the next tranche, and so on up through its size.

Run the arithmetic on that Rs 6,000 crore scheme and the blended ceiling is about 1.57 percent, not 1.40 percent. The tranches permit Rs 10.50 crore, Rs 4.75 crore, Rs 20 crore, Rs 45 crore and Rs 14 crore, which is Rs 94.25 crore a year against Rs 6,000 crore of assets.

Two things fall out. The blended ceiling falls far more slowly than the staircase, because every earlier tranche keeps charging its own higher rate for as long as the scheme exists. And the lines never converge: a very large scheme carries its first Rs 500 crore in the blended figure forever.

A ceiling is also only a ceiling. Crossing a slab boundary gives the manager room it may or may not use, and the published rate is the only evidence of which. On that Rs 6,000 crore scheme, Rs 94.25 crore a year is about Rs 25.8 lakh accrued every single day.

The figure changed on 1 April 2026, and so did what it contains

This is the part that dates most of what is currently published. The SEBI (Mutual Funds) Regulations 2026 were notified in January 2026 and came into force on 1 April 2026, replacing a framework in place since 1996. The expense provisions were not trimmed. They were restructured.

Under the old framework the total expense ratio was one capped number swallowing almost everything the scheme spent, including Goods and Services Tax on the management fee and a long tail of statutory levies. Under the new one, what an investor may be charged is a short list: a base expense ratio, brokerage, transaction costs, statutory levies and exit load where it applies. Only the first is inside the slab table above. The rest sit outside it, on actuals or under their own caps, disclosed separately.

The practical instruction is blunt. A base expense ratio quoted after 1 April 2026 and a total expense ratio quoted before it are not the same measurement, and the newer number is the narrower one. Setting them side by side and concluding costs collapsed is a category error, the same family as comparing a figure before tax with one after. What genuinely fell is the ceiling on the manager's own take, cut by ten basis points on most slabs and fifteen on some, plus a sharp cut to the brokerage a scheme may pass on.

That creates a new disclosure to look for. The industry body now publishes the base expense ratio and the brokerage and transaction costs separately for every scheme. Anyone wanting the old all in number has to add them back together, and anyone comparing two schemes has to match component against component.

The extra charge for smaller cities is no longer charged to you

Almost every live page still says a scheme may charge up to thirty extra basis points where enough inflows come from beyond the largest thirty cities. As a description of today, that is wrong twice over.

The additional thirty basis points was introduced to push distribution into smaller towns, originally against the top fifteen cities and later the top thirty. It was kept in abeyance from February 2023, after the regulator wrote to the industry body setting out inconsistencies in how it was applied and citing transaction splitting and churning as evidence of misuse. It has not been chargeable to schemes since.

What replaced it differs in kind, not just in size. The current framework pays a flat incentive on the first investment of a genuinely new individual investor from beyond the top thirty cities, and of a new woman investor from anywhere: one percent of that application, capped in low four figures, with a minimum holding period. Crucially the asset manager pays it out of the allocation it already sets aside for investor education, so it never touches the scheme or the expense ratio. It was deferred from 1 February 2026 to 1 March 2026.

The current position, then, is that the city an investor lives in no longer changes the ratio their scheme charges.

One percentage point, thirty years, and no assumption about returns

Now the arithmetic. The usual demonstration assumes a return, projects two corpuses and shows the gap, producing a large number hostage to the assumption and close to a forecast. There is a better route, and it is both more rigorous and more honest.

Take two units of the same portfolio, identical except that one is charged one percentage point more each year. Whatever the portfolio does, both experience it, and the only difference is the fee applied to the balance each period. The ratio of the costlier unit's value to the cheaper one's is therefore the product of the annual fee differences alone. The market's contribution cancels out entirely.

That cancellation is the whole insight. The share of terminal wealth a fee difference takes depends on only two things: the size of the difference and the number of years. Not on what the portfolio earned, and not on the path. It is the same in a decade that rose and a decade that fell.

Read the third row. One percentage point, routinely waved away as a rounding error, gives up close to a quarter of the final balance over thirty years. Not a quarter of the growth. A quarter of what would have been there.

These numbers exceed intuition for the reason compounding works in the other direction. In year one the charge takes a small amount. In year two it takes a small amount, and the year one amount is no longer there to grow. By year twenty five the missing balance is mostly not the fees paid; it is what those fees would have become. Stretching to thirty years adds nearly four more points on that row.

A stream of instalments carries less drag than a single sum

The table above describes a single amount held for the whole period. Most Indian household investing is not shaped that way. It is a monthly instalment, so each contribution has its own horizon, and applying a thirty year figure to a thirty year instalment plan overstates the damage. The instalment paid in month one is exposed for the full term and carries the full drag. The one paid in the final month is exposed for weeks and carries almost none. Everything in between sits along that curve.

Where the average lands is the one place in this article where the answer genuinely depends on the return. Terminal value weights each instalment by how much it grew, so a higher path shifts weight toward the earliest instalments, precisely those carrying the longest exposure and the most drag. The stream's drag is therefore a function of the path, the independence the single sum case has and the stream does not.

So treat the single sum table as a firm upper bound for an instalment stream of the same length. A long running plan gives up meaningfully less than a quarter of its terminal balance to one percentage point, and meaningfully more than nothing. Anyone needing the exact figure must compute it on their own schedule, and should distrust any page offering one without stating the return it assumed.

What the ratio does not capture

The cost of trading inside the portfolio. When a scheme buys and sells it pays brokerage and taxes and moves the price against itself. Brokerage and statutory levies are now disclosed separately and are at least visible. Market impact, the price movement caused by the scheme's own order, is disclosed nowhere and borne entirely by unit holders. A high turnover portfolio can cost holders far more than the base ratio suggests, and nothing published makes that gap legible.

The difference between plans on the same portfolio. A direct plan and a regular plan hold identical securities managed identically. The gap between their ratios is the distribution commission, inside the base expense ratio and paid from scheme assets on one, not at all on the other. Run through the table above, that gap is among the largest controllable numbers in a long horizon portfolio.

Exit load. Charged on redemption, not accrued daily, and outside the ratio entirely. It now goes back to the scheme rather than the manager, which changes who benefits but not who pays.

Everything about how the money is managed. The ratio is silent on mandate drift, on concentration, and on whether two schemes held together contain the same positions. Holding the same companies twice through two schemes is a commoner and costlier problem than twenty basis points of ratio.

A low rate on a poorly run scheme is not a bargain

Everything above argues that cost compounds and that a percentage point is not small. None of it argues the cheapest number is automatically right, and reading it that way inverts the point. The ratio measures what is charged. It says nothing about what is delivered, whether the mandate is coherent, whether the scheme does what its documents say it does, or whether it still will in five years. A low ratio on an incoherent mandate can lose more in one poor allocation decision than the cost advantage recovers in a decade, and on an impatient portfolio it spends the difference on trading costs that never reach the headline figure.

The correct weight is this. Cost is the only input known in advance with certainty; everything else, the mandate included, is judgement under uncertainty. Take it seriously because so little else is certain, and refuse to let it be the only thing you look at because it is not the largest. That is three habits. Know the blended ratio you actually pay, not the slab your scheme's size falls in. Read the base ratio and the separately disclosed brokerage together, because after 1 April 2026 the base ratio is the narrower number. And put any gap through the horizon table before deciding it is too small to act on, because that table is the only part of this subject that does not depend on anybody's opinion.