Guide · Costs and taxes · Updated 2026-09-18

The exit load is paid to the other unitholders, and its calendar does not agree with the tax one

Ask an investor what an exit load is and the answer is almost always a penalty the fund house charges for leaving too soon. Half of that is right. It is charged for leaving soon. It is not the fund house's money, has not been since 2012, and in the financial year now running there is no longer even an indirect allowance compensating the manager for handling it. Treat the load as something the industry extracts and you optimise around the only date anybody advertises, which is not the date that costs more.

The load is not the manager's money

Before October 2012 a portion of an exit load could be kept outside the scheme, in a separate account funding distribution and marketing expenses. That is precisely the arrangement most people still describe when they explain what a load is.

The SEBI (Mutual Funds) (Second Amendment) Regulations 2012 inserted Regulation 51A, which requires that any exit load charged be credited to the scheme. Not a portion of it. The load stops being revenue and becomes an inflow to net assets, and whoever holds units at that point owns a share of it.

One deduction survives the journey. Goods and Services Tax at 18 percent applies to the load, and what reaches the scheme is the load net of that tax. On an illustrative load of 1,000 rupees treated as inclusive of tax, roughly 153 rupees goes to the exchequer and roughly 847 rupees to the scheme. The manager's share of both figures is nil.

Because the 2012 change took real revenue away from asset management companies, SEBI allowed schemes that levy a load to charge an additional expense of up to 0.20 percent of daily net assets, reduced to 0.05 percent in February 2018 once the regulator concluded the original size could not be justified. The SEBI (Mutual Funds) Regulations 2026, notified on 14 January 2026 and in force from 1 April 2026, removed the remaining 5 basis points on the reasoning that the allowance had always been transitory, folding a partial offset into the first two expense ratio slabs instead, and cut the maximum permissible load from 5 percent to 3 percent.

What the leaver is actually paying for

A redemption is an instruction to raise cash, and raising cash costs something the continuing unitholders pay. The scheme must either hold idle cash against the possibility, which drags on the portfolio it was mandated to run, or sell securities to fund the payment. Selling incurs brokerage, exchange charges, Securities Transaction Tax and stamp duty, and on anything short of the most liquid names it moves the price against the seller. Those costs land inside the scheme, on everybody still in it, including investors who did nothing that day.

The load pushes that cost back to the person who caused it. Read that way it is not a penalty but a correction for an externality, and crediting it to the scheme is the only design under which the correction works. Paying it to the manager would take money from the leaver without ever compensating the people who were diluted.

It also explains why the load is short dated. An unexpected redemption imposes a one time dealing cost, not a recurring one, and the window is a rough proxy for how long the manager treats the money as genuinely invested. A scheme holding less liquid instruments needs a longer or larger deterrent than one holding treasury bills, which is why the categories differ so widely.

The structure, and the day it counts from

The usual shape is a single percentage applying if units are redeemed within a stated number of days of allotment, falling to nil thereafter. Tiered loads step down instead, for instance one rate inside twelve months and a lower one between twelve and twenty four. Graded loads are the most granular form, and SEBI mandated one for liquid schemes by a circular dated 20 September 2019, effective from 20 October 2019, expressly to move very short parking into overnight schemes.

Four features of the structure are load bearing and routinely missed.

It is charged on redemption value, not on gain. The base is units multiplied by the applicable net asset value, so a holding that has fallen still pays and the load comes out of capital. The tax clock behaves in the opposite way, since it only ever operates on a gain.

The clock starts at allotment, per tranche. Not when the instruction was given, not when the money left the bank, not when the systematic plan was registered. Every instalment is its own purchase with its own window.

It cannot be imposed or increased retrospectively. Any introduction or enhancement applies only to prospective investments, and units already allotted keep the terms in force when they were allotted, which is why one folio can hold tranches on different load terms.

It cannot discriminate by size. No distinction may be made between unitholders on the basis of the amount subscribed, and no load is charged on bonus units or on units allotted against reinvestment of income distribution.

Two calendars, two authors, no coordination

Here is the part that costs money. Both are described casually as how long you have to stay in. They are not the same rule, not written by the same body, and not expressed in the same unit.

One is a commercial term the asset manager chooses within the regulatory cap. The other is a statutory threshold in section 2(42A) of the Income-tax Act 1961, which asks whether the units were held for more than the stated period rather than for it. They differ on every dimension that decides an outcome.

Nothing in either framework requires the two to line up, and nothing in the redemption process shows both at once. The scheme information document states the load. Section 2(42A) states the threshold. The redemption screen states neither.

Where the two calendars sit, category by category

The gap is not uniform. In one direction the load clears long before the tax threshold, which is the common and expensive case. In the other the tax clock does not exist at all, so waiting buys nothing.

The section 50AA rows deserve more than a glance. Inside that section the gain is deemed short term however long it was held and taxed at slab rates, so an investor who waits three years for long term treatment gets nothing for the wait. The definition narrowed from assessment year 2026-27 to schemes investing more than 65 percent of proceeds in debt and money market instruments, moving gold and several international fund of fund structures out of it and back onto an ordinary holding period. Which side of that line a scheme sits on is a portfolio fact rather than a label, and it can change.

First in, first out, and the redemption with two answers

A folio is not a single holding. It is a stack of tranches, each with its own allotment date, cost, load terms and position on the tax clock. A partial redemption has to decide which of them it consumes, and the answer is the oldest first.

For units held in dematerialised form the ordering is statutory. Section 45(2A) of the Income-tax Act 1961 applies first in, first out to dematerialised securities for the date of transfer and the period of holding, and CBDT Circular 768 dated 24 June 1998 directs that it be applied account wise, so units in two accounts do not pool. In statement of account form the registrar runs the same ordering at folio level, for both the load computation and the capital gains statement.

Work it through. A monthly plan of 10,000 rupees runs into an equity oriented scheme carrying a load of 1 percent inside 365 days of allotment, and eighteen instalments have been allotted. At the redemption date instalments 1 to 8 are past both 365 days and 12 months, instalments 9 to 18 are inside both, and the net asset value is an illustrative 138.00.

Request B produces two loads and two tax characters inside one transaction the investor described only as a rupee amount, and the split is invisible at the point of instruction. Nothing on a redemption form tells you which tranche the number you typed will reach.

One further detail is worth confirming in the scheme information document rather than assuming. Where a redemption is requested as an amount, many schemes redeem units to the value requested and then deduct the load from the proceeds, so less reaches the bank than was asked for. Others compute the units so the requested amount arrives net.

Four ways the two calendars disagree, worked

The first row is the ordinary case and the most expensive, because the mistake feels like diligence: the investor checked a rule, satisfied it and acted, and the rule they checked was the cheaper of the two. The fourth row catches careful people instead. A 365 day window and a 12 month threshold look identical and are not, because one counts days and the other counts calendar months and demands strictly more than twelve of them. In a leap year the counts separate by a day in one direction, and a clause drafted as within rather than up to and including separates them by a day in the other.

What the load does not do, and where that bites

A switch is not a move, it is an exit and an entry. Switching between two schemes of the same fund redeems the first and buys the second, so the load of the scheme being left applies, the gain on the units leaving is a taxable transfer, and both clocks restart from zero in the scheme being entered. The same holds for a move between the growth and income distribution options of one scheme, and from a regular plan to a direct plan, which surprises people who thought they were only lowering their expense ratio.

A systematic transfer or withdrawal plan is a series of redemptions. Each instalment is tested against both calendars on its own date and draws on the folio by first in, first out. A withdrawal plan running out of a folio built by a systematic investment plan is two first in, first out queues working against each other, and the tranche sold in any given month is rarely the one the investor pictures.

The cap is not the practice. Cutting the maximum permissible load from 5 percent to 3 percent matters for schemes holding less liquid instruments, but the retail equity norm has long been 1 percent or lower. A page presenting the cap as the typical figure is quoting a ceiling, not a rate anybody pays.

The GST treatment has been contested. That the load reaches the scheme net of Goods and Services Tax is settled in practice, but the indirect tax position on loads has been the subject of departmental proceedings against fund houses. That is open on the collection side, it does not change the crediting principle, and it is a reason to read the current position rather than a page written a few years ago.

What the two dates are actually for

The load window is a transaction cost defence for the people who stayed, calibrated to what unwinding costs in that portfolio. The tax holding period is fiscal policy, calibrated to the behaviour the government of the day wants to encourage. They were never negotiated with each other because they were never answering the same question: one asks who should bear a dealing cost, the other how long is long enough to deserve a concessional rate. No mechanism exists through which the answers would converge, so expecting them to is the root of the error rather than a fair assumption that fails.

Which leaves a short discipline. Before any redemption read the load clause in the scheme information document, including whether the day count is inclusive; the allotment dates of the tranches the redemption will reach, since first in, first out decides that and nothing on the form shows it; and the holding period threshold for the category, which is where the larger cost usually sits. All three are knowable in advance. None appears on the screen where the decision gets made.