Chapter 2 · Indian Markets and Instruments

2.2 Mutual Funds by the Numbers: NAV, SIP, Costs and ETFs

A mutual fund is a shared portfolio priced once a day: how NAV and units work, what a SIP actually buys (the harmonic mean), how much a 1% expense ratio costs over 20 years, growth vs IDCW, exit loads, index funds and ETFs, tracking difference and tracking error, and STP/SWP calculations.

📍 The situation

Meera, 32, works in Hyderabad and wants to start a ₹10,000 monthly SIP. A relative who is a mutual-fund distributor suggests "Fund A — Flexi Cap, Regular Growth"; a colleague says she should buy the "Direct" plan of the same fund because it is "about 1% cheaper", and adds that "Fund B has a NAV of only ₹14, so it is cheaper than Fund A at ₹480". Her brother prefers a Nifty 50 index fund and wonders whether an ETF would be better, and why his index fund earned a little less than the index last year. Her father, who has just retired, wants to know whether he can withdraw ₹30,000 a month from ₹40 lakh, and whether the "IDCW" option gives him a regular income.

Every one of these questions has a precise numerical answer once we know how a mutual fund keeps its books. What is a NAV and how many units does ₹10,000 buy? What average price does a SIP pay? What does a 1% difference in expense ratio cost over 10, 20 or 30 years? How closely does an index fund follow its index, and how do we measure "closely"? How long does a monthly withdrawal last? The tools are the geometric series of §1.2, a classical inequality between means, the exponential function and a standard deviation.

⚠️ About the numbers on this page

All funds, NAVs, returns, expense ratios, loads and tax rates on this page are hypothetical or illustrative. An "assumed 12% return" is an input to a calculation, not a forecast — market-linked returns vary and can be negative, and past NAV performance does not predict future returns. Expense-ratio limits, exit loads, cut-off times, stamp duty, the riskometer, fund categories and the taxation of mutual funds are set by SEBI, the Government and each scheme's documents and change from time to time — check the scheme information document, AMFI and SEBI and the Income Tax Department for the current rules. Nothing here recommends any fund, plan or distributor.

NAV, Units, Purchases and Redemptions

A mutual fund pools money from many investors and invests it in a portfolio — shares, bonds, money-market instruments — managed by an asset management company (AMC) under SEBI regulation. Instead of owning shares directly, each investor owns units of the fund, and every unit is a claim on an equal slice of the portfolio. The price of one slice is the NAV.

📘 Definition 2.7 — Units and net asset value (NAV)

At the end of each business day a fund values everything it owns at market prices. Its net assets are $$\text{net assets}=\text{investments}+\text{cash}+\text{receivables}-\text{liabilities},$$ where liabilities include amounts payable and expenses accrued but not yet paid. If $U$ units are outstanding, the net asset value per unit is $$\text{NAV}=\frac{\text{net assets}}{U}.$$

Investors buy units from the fund, and sell them back to it (redemption), at the NAV of the applicable day. Which day's NAV applies depends on when the money or the request reaches the fund relative to a cut-off time set by SEBI's rules. Units are usually allotted to three decimal places.

📐 Theorem 2.6 (Formula) — Units, redemption proceeds, and why new money does not dilute
  1. Purchase. An investment of $X$ rupees at NAV $N$ buys $u=X(1-s)/N$ units, where $s$ is any stamp duty or transaction charge deducted (an illustrative $s=0.005\%$ below; $s=0$ if none).
  2. Redemption. Redeeming $u$ units at NAV $N$ pays $uN(1-\ell)$, where $\ell$ is the exit load rate applicable to those units (often zero after a stated holding period). Units are redeemed first in, first out (FIFO): the oldest units go first.
  3. No dilution. A purchase or redemption at the NAV does not change the NAV: if net assets are $A=NU$ and $X$ rupees come in, the new NAV is $\dfrac{A+X}{U+X/N}=N$.
  4. Return. An investor's return between two dates is $N_1/N_0-1$, whatever the NAV's level: a fund with NAV ₹14 and one with NAV ₹480 that both rise 10% give the same 10%.

Why no dilution. Since $A=NU$, $$\frac{A+X}{U+X/N}=\frac{NU+X}{U+X/N}=\frac{N(U+X/N)}{U+X/N}=N.$$ New investors pay exactly the value of what they get, so existing unit-holders gain or lose nothing when money flows in or out. (In practice large flows force the fund to trade, and those trading costs are shared by all — one reason for exit loads.) Part 4 is the reason that a "low NAV" is not "cheap": the NAV's level only reflects when the fund started and what it has earned since, and for the same money you simply get more units of smaller size. ∎

✏️ Example 1 — Computing a NAV and Meera's first units
At the close of a day, Fund A holds equity shares worth ₹1,182.40 crore, cash and money-market instruments worth ₹46.30 crore and dividends receivable of ₹1.85 crore. It owes ₹6.20 crore for shares bought but not yet paid for and has accrued expenses of ₹0.95 crore. There are 24.6 crore units. (a) Find the NAV. (b) Meera invests ₹50,000 at this NAV; how many units does she get without and with an illustrative 0.005% stamp duty? (c) Check that a ₹10 crore inflow at this NAV leaves the NAV unchanged.
(a) Net assets $=1{,}182.40+46.30+1.85-6.20-0.95=\text{₹}\,1{,}223.40$ crore. NAV $=1{,}223.40/24.6\approx\text{₹}\,49.7317$.
(b) Without charges: $50{,}000/49.7317\approx1{,}005.395$ units. With stamp duty, ₹2.50 is deducted and $49{,}997.50/49.7317\approx1{,}005.345$ units are allotted.
(c) ₹10 crore buys $10/49.7317\approx0.2011$ crore units; new NAV $=(1{,}223.40+10)/(24.6+0.2011)\approx\text{₹}\,49.7317$ — unchanged, as Theorem 2.6 says.
📘 Definition 2.8 — Growth and IDCW options; exit load
  • In the growth option the fund keeps all its income and gains invested; the NAV rises and falls, and the investor realises gains only by redeeming units.
  • In the IDCW option (income distribution cum capital withdrawal) the fund pays out amounts per unit from time to time, and the NAV falls by the amount distributed. The name is a reminder that the payout may come partly from the investor's own capital. With IDCW reinvestment, the payout buys new units at the ex-distribution NAV.
  • An exit load is a charge, as a percentage of the redemption value, on units redeemed within a stated period after purchase; it is credited back to the scheme.

The growth and IDCW options (and direct and regular plans) of a scheme hold the same portfolio; they differ only in what happens to the returns.

📐 Theorem 2.7 (Formula) — An IDCW payout is a withdrawal, not a gain

If a distribution of $\delta$ per unit is declared when the NAV is $N$, the NAV falls to $N-\delta$ and an investor with $u$ units receives $u\delta$: her wealth $u(N-\delta)+u\delta=uN$ is unchanged before tax. With reinvestment she receives $u\delta/(N-\delta)$ new units and her wealth is again $uN$. After tax, if distributions are taxed at a rate $\tau$ when paid (illustrative), she keeps $u\delta(1-\tau)$ — so the IDCW option is worse than the growth option by the tax paid early, unless the growth option's eventual tax on redemption is higher.

✏️ Example 2 — A redemption with an exit load (FIFO)
An investor bought 1,000 units of a fund at NAV ₹40 fourteen months ago and 500 units at ₹46 five months ago. The fund charges an illustrative exit load of 1% on units redeemed within 12 months. She redeems 1,200 units at NAV ₹50. Find the proceeds and the load. How could she have avoided the load?
FIFO The first 1,000 units redeemed are the oldest (held 14 months, no load); the remaining 200 come from the 5-month-old lot (load applies).
Proceeds $1{,}000\times50+200\times50\times0.99=50{,}000+9{,}900=\text{₹}\,59{,}900$; the load is ₹100.
Gain Cost of the units sold: $1{,}000\times40+200\times46=\text{₹}\,49{,}200$, so the gain before tax is $\text{₹}\,59{,}900-\text{₹}\,49{,}200=\text{₹}\,10{,}700$ (₹10,800 before the load).
Avoid Redeeming only 1,000 units now, and the other 200 after they complete 12 months, would cost no load. Tax on gains also depends on holding periods — check the current rules.
✏️ Example 3 — Growth or IDCW for a retired father?
A fund's NAV is ₹25.00 when it declares an IDCW of ₹2.00 per unit. Meera's father holds 4,000 units of the IDCW option; a friend holds 4,000 units of the growth option, whose NAV is also ₹25.00. (a) Find the ex-IDCW NAV, the payout and the father's total wealth. (b) With distributions taxed at an illustrative 30% when received, what does he keep? (c) Under IDCW reinvestment (ignore tax), how many units would he hold?
(a) Ex-NAV ₹23.00. Payout $4{,}000\times2=\text{₹}\,8{,}000$; units worth $4{,}000\times23=\text{₹}\,92{,}000$; total ₹1,00,000 — the same as the friend's $4{,}000\times25$.
(b) He keeps $8{,}000\times0.7=\text{₹}\,5{,}600$; his wealth is ₹97,600, against ₹1,00,000 for the growth holder, who pays tax only when (and if) she redeems.
(c) $8{,}000/23\approx347.826$ new units: $4{,}347.826\times23=\text{₹}\,1{,}00{,}000$.
Lesson IDCW is not "extra income" — it is part of his own money handed back, and the amount and timing are not guaranteed. For a regular income, a planned withdrawal from the growth option (an SWP, below) gives control over both.

SIP Units and Rupee-Cost Averaging

A SIP invests a fixed amount $A$ every month (§1.2 projects its value under an assumed return). Here we look inside: because the amount is fixed and the NAV varies, the SIP buys more units when the NAV is low and fewer when it is high. That simple fact has a precise mathematical consequence.

📐 Theorem 2.8 — The SIP's average cost is the harmonic mean of the NAVs

A SIP of $A$ rupees at NAVs $N_1,\dots,N_n\gt0$ buys $u=A\sum_{k=1}^n\dfrac1{N_k}$ units, at an average cost per unit $$\bar c=\frac{nA}{u}=\frac{n}{\sum_{k=1}^n 1/N_k}=H(N_1,\dots,N_n),$$ the harmonic mean of the NAVs. Always $$H\le\bar N=\frac{N_1+\dots+N_n}{n},$$ with equality if and only if all the NAVs are equal. For two NAVs $x,y$: $\bar N-H=\dfrac{(x-y)^2}{2(x+y)}$.

Proof. Month $k$ buys $A/N_k$ units, so the units and the average cost are as stated. By the Cauchy–Schwarz inequality applied to the vectors $(\sqrt{N_k})$ and $(1/\sqrt{N_k})$, $$n^2=\Big(\sum_k\sqrt{N_k}\cdot\tfrac{1}{\sqrt{N_k}}\Big)^2\le\Big(\sum_kN_k\Big)\Big(\sum_k\tfrac1{N_k}\Big),$$ which rearranges to $H\le\bar N$; equality in Cauchy–Schwarz holds exactly when the vectors are proportional, i.e. all $N_k$ equal. For two NAVs, $\bar N-H=\frac{x+y}2-\frac{2xy}{x+y}=\frac{(x+y)^2-4xy}{2(x+y)}=\frac{(x-y)^2}{2(x+y)}$. ∎

This is rupee-cost averaging: an investor who puts the same rupees in every month pays less per unit, on average, than one who buys the same number of units every month — and the more the NAV fluctuates, the bigger the gap. Note carefully what the theorem does not say: it does not say a SIP beats investing a lump sum, or that a SIP cannot lose money. If the NAV rises steadily, money invested earlier does better; if it falls and stays down, the SIP loses too.

✏️ Example 4 — Six months of Meera's SIP
Meera invests ₹10,000 at the start of each month; the NAVs on the six dates are ₹50, ₹45, ₹40, ₹42, ₹48 and ₹55. (a) Find the units bought each month and in total. (b) Find her average cost per unit and compare it with the average NAV. (c) What is her investment worth at the last NAV? (d) Compare with buying 100 units each month, and with investing ₹60,000 at once in the first month.
(a) $10{,}000/N_k$: 200.000, 222.222, 250.000, 238.095, 208.333, 181.818 — total 1,300.469 units. The most units were bought in the month of the lowest NAV.
(b) Average cost $60{,}000/1{,}300.469\approx\text{₹}\,46.14$ — the harmonic mean of the NAVs. The average NAV is $(50+45+40+42+48+55)/6\approx\text{₹}\,46.67$.
(c) $1{,}300.469\times55\approx\text{₹}\,71{,}526$: a gain of about 19.2% on ₹60,000 invested over six months (the money was invested for different lengths of time; XIRR, §1.3, gives the annualised rate).
(d) 100 units a month costs $100\times(50+45+40+42+48+55)=\text{₹}\,28{,}000$ for 600 units — an average of ₹46.67, higher than the SIP's ₹46.14. A ₹60,000 lump sum in month 1 buys 1,200 units, worth ₹66,000 at ₹55: fewer units than the SIP here, because the NAV dipped after month 1. Had the NAV risen every month, the lump sum would have won.
Interactive 1 — Rupee-Cost Averaging on a Simulated Fund

A ₹10,000 SIP runs for 60 months on a simulated NAV path (not real data). The blue line is the NAV; the amber bars are the units bought each month (taller when the NAV is low). The green line is your average cost per unit (harmonic mean) and the purple dashed line the average NAV (arithmetic mean). Choose the market's shape and its volatility, and press Play.

18%
month—
invested—
units held—
average cost (HM)—
average NAV (AM)—
SIP value now—
same total, all in month 1—

Costs: Expense Ratio, Direct vs Regular Plans

A fund pays for its management, operations and distribution out of the investors' money. Nobody sends you a bill: the costs are taken out of the fund's assets a little every day, so they show up only as a slightly lower NAV. Small percentages, applied every day for decades, compound into large sums.

📘 Definition 2.9 — Total expense ratio; direct and regular plans

The total expense ratio (TER) $c$ is the annual percentage of net assets that a scheme charges for its expenses. It is accrued daily — roughly $c/365$ of net assets each calendar day — and the published NAV is always after expenses. SEBI sets maximum TERs, which depend on the type and size of the scheme; each fund discloses its actual TER.

Every scheme has a direct plan, bought straight from the AMC, and a regular plan, bought through a distributor. Both hold the same portfolio; the regular plan's TER is higher because it includes the distributor's commission, so the two plans have different NAVs. Whether a distributor's service is worth its cost is a personal decision; the mathematics below measures the cost.

📐 Theorem 2.9 (Formula) — Expense-ratio drag

Model the fund day by day: on day $d$ the portfolio earns a gross return $g_d$ and the day's expense $c/365$ of assets is deducted, so $\text{NAV}_d=\text{NAV}_{d-1}(1+g_d)(1-c/365)$. Then over $T$ years ($365T$ days) $$\frac{\text{NAV}_T}{\text{NAV}_0}=\prod_d(1+g_d)\cdot\Big(1-\frac{c}{365}\Big)^{365T}\approx G_T\,e^{-cT},$$ where $G_T$ is the gross growth factor of the portfolio. Consequently:

  1. Two plans of the same portfolio with TERs $c_1\lt c_2$ end in the ratio $$\frac{V_2}{V_1}\approx e^{-(c_2-c_1)T},$$ whatever the market does — the path cancels out.
  2. With an assumed gross annual return $g$, the net annual return is $(1+g)e^{-c}-1\approx g-c$.
  3. For a SIP, the instalment invested $t_k$ years before the end is reduced by the factor $e^{-\Delta c\,t_k}$; the whole corpus's ratio is a weighted average of these factors, so it lies between $e^{-\Delta c\,T}$ and $1$.

Why. Multiplying the daily factors gives the product formula. Since $(1-c/365)^{365}$ is within a millionth or so of $e^{-c}$ for realistic $c$ (this is the limit of §1.1's continuous compounding, run in reverse), the cost factor is $e^{-cT}$. The gross factor $G_T$ is the same for both plans, so it cancels in the ratio. For a SIP, each instalment is a separate "lump sum" held for its own time $t_k\le T$. ∎ The key lesson: a cost difference compounds like a negative return. A 1 percentage point difference costs about $1-e^{-0.01T}$ of a lump sum's final value: 9.5% after 10 years, 18.1% after 20 and 25.9% after 30.

✏️ Example 5 — ₹10 lakh for 20 years: direct vs regular
₹10,00,000 is invested for 20 years in the direct plan (TER 0.6%) or the regular plan (TER 1.6%) of the same fund. Assume, for illustration only, a gross portfolio return of 12% a year. Find each plan's net annual return and value after 20 years, the rupee gap, and check the ratio against Theorem 2.9.
Net returns Direct: $1.12\,e^{-0.006}-1\approx11.33\%$; regular: $1.12\,e^{-0.016}-1\approx10.22\%$ (close to the shortcut $12-0.6$ and $12-1.6$).
Values Direct: $10{,}00{,}000\times1.11330^{20}\approx\text{₹}\,85{,}55{,}494$. Regular: $10{,}00{,}000\times1.10222^{20}\approx\text{₹}\,70{,}04{,}646$.
Gap About ₹15.51 lakh — more than one and a half times the original investment. The ratio is $70{,}04{,}646/85{,}55{,}494\approx0.8187=e^{-0.2}$, exactly as predicted; it would be the same if the market returned 5% or 20%.
Interpret The extra 1% a year is the price of the distributor's service. It may be worth paying for good advice; the point is to know the price — here, about 18% of the final corpus.
📈 Market case — Meera's ₹10,000 SIP: direct vs regular over 10, 20 and 30 years

Meera compares the two plans of the same flexi-cap fund for her ₹10,000 monthly SIP, invested at the start of each month. For illustration she assumes a gross portfolio return of 12% a year and TERs of 0.6% (direct) and 1.6% (regular). Net annual returns are 11.33% and 10.22% (Example 5); she converts each to a monthly rate $i=(1+r)^{1/12}-1$ and uses the SIP formula of §1.2, $FV=A\frac{(1+i)^n-1}{i}(1+i)$.

yearsinvested (₹)direct (₹)regular (₹)gap (₹)gap %
1012,00,00021,61,88120,38,4321,23,4495.7%
2024,00,00084,85,33674,33,40610,51,93012.4%
3036,00,0002,69,81,3012,17,11,90252,69,39919.5%
  1. The gap grows faster than time. Doubling the horizon from 10 to 20 years multiplies the rupee gap by about 8.5; by 30 years it exceeds ₹52 lakh — more than everything she will have invested.
  2. A SIP loses less, in percentage terms, than a lump sum. The SIP gaps (5.7%, 12.4%, 19.5%) lie between 0 and the lump-sum gaps $1-e^{-0.01T}$ (9.5%, 18.1%, 25.9%), as Theorem 2.9(3) says: recent instalments have had little time to suffer the drag.
  3. The assumption matters less than it seems. If the gross return is lower, both values shrink, but the lump-sum ratio is still $e^{-0.01T}$; the cost is a fixed fraction of whatever the market gives.
  4. What she decides is not arithmetic. If she needs help choosing funds and staying invested through bad years, a distributor may be worth the cost; if she is confident doing it herself, the direct plan keeps the difference. Switching an existing holding from regular to direct is a redemption, which may attract exit load and tax (see Practice corner E4).

Interactive 1 shows why the SIP's average cost is low in volatile markets; this table shows why that advantage is small compared with a cost that is charged every single day.

Index Funds, ETFs and Tracking

An index fund does not try to beat the market: it holds the stocks of an index in index weights (§2.1) and aims to earn the index's total return minus as little as possible. That makes it the cleanest test of how costs and practical frictions show up in returns.

📘 Definition 2.10 — Index fund, ETF and iNAV
  • An index fund is a mutual fund that replicates an index; it is bought from and sold to the AMC at the day's NAV, like any other fund.
  • An exchange-traded fund (ETF) is also an index-replicating fund, but its units are listed and traded on the stock exchange during market hours through a broker and a demat account, at a market price set in an order book (§2.1). Large investors and designated market makers can create or redeem ETF units with the AMC in big blocks (a "creation unit") by exchanging the underlying basket, which keeps the market price close to the value of the holdings.
  • The indicative NAV (iNAV) is an estimate of the ETF's NAV per unit published during the trading day. The ETF trades at a premium if its price $P$ exceeds the iNAV and at a discount if below: premium $=(P-\text{iNAV})/\text{iNAV}$.
📘 Definition 2.11 — Tracking difference and tracking error

Let $R_F$ and $R_I$ be the returns of a fund and of its benchmark's total return index over a period, and $d_t=r_{F,t}-r_{I,t}$ the difference of their daily returns.

  • The tracking difference is $\text{TD}=R_F-R_I$: by how much the fund lagged (or led) the index — usually negative, roughly the costs.
  • The tracking error is the annualised standard deviation of the daily differences: $\text{TE}=s_d\sqrt{252}$, where $s_d$ is the sample standard deviation of the $d_t$ and 252 is the conventional number of trading days in a year.

TD measures the average shortfall; TE measures how erratically the fund follows the index. A fund can have a large TD and a tiny TE (a steady, expensive fund) or the reverse. Indian index funds and ETFs disclose both.

📐 Theorem 2.10 (Formula) — Where tracking difference comes from

Suppose a fund keeps a fraction $w$ of its assets in cash earning $R_c$ and the rest exactly in the index, pays TER $c$ and incurs trading costs $\kappa$ over a year. Then, to first order, $$R_F\approx(1-w)R_I+wR_c-c-\kappa,$$ so $$\text{TD}\approx-c-\kappa-w\,(R_I-R_c).$$ The term $w(R_I-R_c)$ is the cash drag: it costs return in rising markets and helps in falling ones. The tracking error comes from the parts that vary day to day — cash, imperfect replication, the timing of dividends and of trades — not from the TER, which is charged smoothly.

Why. A portfolio's return is the weighted average of its parts' returns (the same fact as Theorem 2.2), and the expenses and costs are subtracted. A constant daily deduction shifts every $d_t$ by the same amount; it changes the mean of the differences (TD) but not their standard deviation (TE). ∎

✏️ Example 6 — Ten days of an index fund
Daily returns (%) of a total return index and of an index fund tracking it:
day12345678910
index0.620−0.3501.1000.050−0.8000.4800.270−0.1200.950−0.400
fund0.612−0.3441.0930.047−0.8060.4780.262−0.1210.946−0.408
Find the daily differences, the tracking difference over the ten days, and the annualised tracking error. What would the TD be over a year if the average daily difference persisted?
Differences $d_t$ (%): −0.008, +0.006, −0.007, −0.003, −0.006, −0.002, −0.008, −0.001, −0.004, −0.008. Mean $\bar d=-0.0041\%$ a day.
TD Compounding the daily returns: index $+1.797\%$, fund $+1.756\%$, so TD $\approx-0.042$ percentage points over ten days (close to $\sum d_t=-0.041$).
TE Sample standard deviation $s_d\approx0.00441\%$; $\ \text{TE}=0.00441\%\times\sqrt{252}\approx0.070\%$ a year.
A year $-0.0041\%\times252\approx-1.03\%$ a year — far more than an index fund's TER would usually explain, so one would look for cash drag or trading costs. But ten days is much too short a sample: in practice TD and TE are computed from at least a year of daily data.
✏️ Example 7 — Buying an ETF at a premium
A Nifty 50 ETF's units trade at ₹245.80 (best ask) while its iNAV is ₹244.90. (a) Find the premium. (b) An investor buys ₹1,00,000 worth (whole units only). How many units does he get, and how much more than the iNAV value does he pay? (c) Why does the premium usually stay small for a large, liquid ETF, and when can it be large?
(a) $(245.80-244.90)/244.90\approx0.367\%$.
(b) $\lfloor100{,}000/245.80\rfloor=406$ units for $406\times245.80=\text{₹}\,99{,}794.80$. Their iNAV value is $406\times244.90=\text{₹}\,99{,}429.40$: he pays ₹365.40 more — more than a year's TER at 0.3% on this amount — before brokerage.
(c) If the price is well above the value of the basket, market makers can create new units by delivering the basket and sell them on the exchange, pushing the price down; at a discount they do the reverse. When creation is slow or costly, the ETF is thinly traded, or the underlying market is closed or volatile, premiums and discounts can widen. Using limit orders and checking the iNAV protect the buyer.
Interactive 2 — Tracking an Index: Where Tracking Difference and Error Come From

One simulated year (252 trading days) of an index and an index fund. The amber line is the fund's cumulative return minus the index's (the tracking difference so far); the purple dashed line is what the TER alone would cost. Raise the cash holding to see cash drag follow the market; add replication noise to see the tracking error grow while the average shortfall hardly changes. Press Play to run the year.

0.20%2%
trading day—
index (TRI) return—
fund return—
tracking difference—
tracking error (annualised)—
of which TER so far—

Fund Categories, the Riskometer, STP and SWP

Categories. SEBI groups open-ended schemes into broad families with defined mandates, so that funds with the same label invest in a comparable way. Equity schemes include large-cap, mid-cap, small-cap, flexi-cap and multi-cap funds (defined by the market-cap ranks of §2.1), ELSS (tax-saving) funds and sectoral or thematic funds. Debt schemes are defined mainly by the maturity or quality of what they hold — overnight, liquid, short- and medium-duration, corporate bond, gilt and credit-risk funds (their interest-rate sensitivity is studied in §2.3). Hybrid schemes mix equity and debt in stated proportions (conservative, aggressive, balanced-advantage, multi-asset, arbitrage). Index funds, ETFs and funds of funds form another group. The exact category definitions are in SEBI's circulars and change occasionally.

The riskometer. Every scheme displays a riskometer, a dial that currently has six levels from "Low" to "Very High", based on the risk of the scheme's portfolio (for example its equity share, volatility, credit quality and interest-rate sensitivity), and it is reviewed regularly. It ranks risk, not return: "Very High" does not mean high returns are likely, and "Low" does not mean the NAV cannot fall. It describes the product, not whether the product suits a particular investor.

📘 Definition 2.12 — STP and SWP
  • A systematic transfer plan (STP) redeems a fixed amount $A$ every month from one scheme (often a liquid fund holding a lump sum) and invests it in another (often an equity fund): it turns a lump sum into a SIP. Month $k$ redeems $A/N^{\text{src}}_k$ units of the source and buys $A/N^{\text{dst}}_k$ units of the destination.
  • A systematic withdrawal plan (SWP) redeems a fixed amount $W$ every month from a scheme to provide a regular income; month $k$ redeems $W/N_k$ units.
📐 Theorem 2.11 (Formula) — How long an SWP lasts

A corpus $C$ earns an assumed constant monthly return $i\gt0$, and $W$ is withdrawn at the end of every month. The balance after $k$ withdrawals is $$B_k=C(1+i)^k-W\,\frac{(1+i)^k-1}{i}.$$ If $W\le iC$ the balance never falls (the withdrawal is at most the monthly return). If $W\gt iC$ the corpus is exhausted after $$n=\frac{-\log\left(1-iC/W\right)}{\log(1+i)}\ \text{months.}$$ Because $W$ is fixed in rupees, an SWP redeems more units when the NAV is low — rupee-cost averaging in reverse.

Why. An SWP is a loan in reverse: the investor "lends" $C$ to the fund and receives "EMIs" of $W$. The balance formula is the outstanding-balance formula of §1.2, and setting $B_n=0$ gives $(1+i)^n=\frac{W}{W-iC}$, which is the stated $n$ (it needs $W\gt iC$). If $W\le iC$, each month's return $iB\ge iC\ge W$ covers the withdrawal. ∎ Real NAVs do not grow at a constant rate: a fall early in the withdrawal period forces more units to be sold at low prices, which shortens the life of the corpus even if the average return is unchanged. This sequence-of-returns risk is simulated in §8.1.

✏️ Example 8 — Can Meera's father withdraw ₹30,000 a month?
Her father has ₹40,00,000 and wants an SWP of ₹30,000 a month. Assume, for illustration, a steady return of 7% a year (effective). (a) Find the monthly rate and the largest withdrawal that never touches the capital. (b) How long does ₹30,000 a month last, and what is left after 5 and 10 years? (c) If the NAV is ₹32 in one month and ₹25 in another, how many units does the SWP redeem in each?
(a) $i=1.07^{1/12}-1\approx0.5654\%$ a month; $iC\approx\text{₹}\,22{,}617$ a month can be withdrawn indefinitely (in rupees — its purchasing power still falls with inflation, §1.1).
(b) $W=30{,}000\gt iC$, so $$n=\frac{-\log(1-22{,}617/30{,}000)}{\log1.005654}\approx248.7$$ months, about 20.7 years. Balance after 60 months: $B_{60}\approx\text{₹}\,34.74$ lakh; after 120 months $\approx\text{₹}\,27.37$ lakh.
(c) $30{,}000/32=937.5$ units and $30{,}000/25=1{,}200$ units: the low-NAV month sells 28% more units, permanently.
Interpret About 21 years may or may not be enough for a 60-year-old; the answer is very sensitive to the assumed return and to the order in which good and bad years arrive. The arithmetic gives the starting point; planning (§8.1) has to test many scenarios.
⚠️ Common mistakes and myths
  1. "A ₹14 NAV fund is cheaper than a ₹480 NAV fund." Only percentage changes matter (Theorem 2.6); a low NAV just means more, smaller units.
  2. "Past NAV performance shows what the fund will earn." Past returns depend on the period chosen (point-to-point bias, §1.3) and on markets that will not repeat; they are not a forecast.
  3. "IDCW is extra income on top of my investment." It is paid out of the fund's assets, and the NAV falls by the same amount (Theorem 2.7); it may also be taxed earlier.
  4. "SIPs cannot lose money" / "SIPs always beat lump sums." Rupee-cost averaging only guarantees an average cost at or below the average NAV (Theorem 2.8). In a steadily rising market a lump sum does better; in a market that falls and stays down, a SIP loses.
  5. "1% a year is too small to matter." Over 20 years it removes about 18% of a lump sum's final value (Theorem 2.9) — lakhs of rupees on a typical SIP.
  6. "Direct plans are a different, riskier fund." The portfolio is identical; only the expense ratio, and hence the NAV, differ.
  7. "An index fund earns exactly the index." It earns the total return index minus costs, cash drag and frictions; compare it with the TRI, not the price index (§2.1).
  8. "Tracking error is how much the fund lagged." That is the tracking difference; tracking error is the variability of the daily gap.
  9. "An ETF always trades at its NAV." It trades at a market price that can be at a premium or discount to the iNAV, plus a bid–ask spread (Example 7).
  10. "A Low riskometer means no risk." It ranks risk; even low-risk categories can see their NAVs fall.

Exercises

All funds and numbers are hypothetical and all rates illustrative. Units to three decimals, NAVs to four, rupees to the nearest rupee unless stated.

I
NAV, units, loads and IDCW.
  • 1 A fund holds investments worth ₹845.60 crore, cash ₹12.40 crore and accrued income ₹0.90 crore; its liabilities are ₹3.10 crore and 31.2 crore units are outstanding. Find the NAV.
  • 2 How many units does ₹25,000 buy at a NAV of ₹27.45, ignoring charges?
  • 3 Repeat part 2 with an illustrative stamp duty of 0.005% deducted from the amount.
  • 4 In part 1, ₹50 crore of new money comes in at the NAV. Show that the NAV is unchanged.
  • 5 You hold 500 units bought 400 days ago and 300 units bought 100 days ago; the exit load is an illustrative 1% within 365 days. Redeeming 650 units at NAV ₹36, find the proceeds and the load (FIFO).
  • 6 An IDCW of ₹1.50 per unit is declared when the NAV is ₹18.20. For 2,000 units, find the ex-NAV, the payout, and the number of units held under IDCW reinvestment (ignore tax).
  • 7 A friend says Fund X (NAV ₹12) "has more room to grow" than Fund Y (NAV ₹480). Explain why this is wrong.
1 Net assets $845.60+12.40+0.90-3.10=\text{₹}\,855.80$ crore; NAV $=855.80/31.2\approx\text{₹}\,27.4295$.
2 $25{,}000/27.45\approx910.747$ units.
3 $24{,}998.75/27.45\approx910.701$ units (₹1.25 duty).
4 New units $50/27.4295\approx1.8229$ crore; NAV $=(855.80+50)/(31.2+1.8229)\approx\text{₹}\,27.4295$ — unchanged (Theorem 2.6).
5 FIFO: 500 old units free of load, 150 new units with 1% load: $500\times36+150\times36\times0.99=18{,}000+5{,}346=\text{₹}\,23{,}346$; load ₹54.
6 Ex-NAV ₹16.70; payout ₹3,000; reinvestment buys $3{,}000/16.70\approx179.641$ units, so 2,179.641 units worth $\text{₹}\,36{,}400=2{,}000\times18.20$.
7 Returns are percentage changes in NAV; a 10% rise is a 10% rise whether NAV is ₹12 or ₹480. The NAV level reflects the fund's age and past gains, not its potential.
II
SIP units and rupee-cost averaging.
  • 1 A ₹5,000 SIP buys at NAVs ₹20, ₹25, ₹16, ₹20. Find the total units and the average cost, and compare with the average NAV.
  • 2 For two NAVs ₹20 and ₹30, compute the harmonic and arithmetic means and verify $\bar N-H=(x-y)^2/(2(x+y))$.
  • 3 When is a SIP's average cost exactly equal to the average NAV? Give the reason.
  • 4 In part 1 the NAV is ₹22 after the fourth instalment. Find the value and the percentage gain on the amount invested.
  • 5 NAVs rise steadily: ₹10, 11, 12, 13, 14, 15 in six months. Compare a ₹6,000 SIP with ₹36,000 invested in month 1, both valued at ₹15.
  • 6 "Rupee-cost averaging guarantees a profit." Disprove this with a simple NAV sequence.
1 Units $250+200+312.5+250=1{,}012.5$; average cost $20{,}000/1{,}012.5\approx\text{₹}\,19.75$, below the average NAV ₹20.25.
2 $H=2/(1/20+1/30)=24$, $\bar N=25$; $(20-30)^2/(2\times50)=1=25-24$. ✓
3 Only when all the NAVs are equal: equality in the Cauchy–Schwarz (AM–HM) inequality needs proportional vectors (Theorem 2.8).
4 $1{,}012.5\times22=\text{₹}\,22{,}275$: a gain of 11.4% on ₹20,000.
5 SIP units $6{,}000(1/10+\dots+1/15)\approx2{,}935.564$, worth $\approx\text{₹}\,44{,}033$; lump sum $3{,}600$ units worth ₹54,000. In a steadily rising market, earlier money wins.
6 NAVs 50, 45, 40, 35, 30, valued at 30: every instalment was bought above ₹30, so the SIP shows a loss (average cost $5/(1/50+1/45+1/40+1/35+1/30)\approx\text{₹}\,38.72$). The average cost is lower than the average NAV, but not lower than the final NAV.
III
Expense ratios and plans.
  • 1 A ₹500 crore scheme has a TER of 1.2%. Roughly how much is accrued as expenses each calendar day?
  • 2 ₹5,00,000 is invested for 15 years at an assumed gross 11%, in plans with TERs 0.5% and 1.5%. Find both values and their ratio.
  • 3 Explain why the ratio in part 2 would be the same if the market had fallen over the 15 years.
  • 4 An ₹8,000 monthly SIP (start of month) runs for 25 years at an assumed gross 12%, TERs 0.7% and 1.7%. Find both corpuses and the percentage gap.
  • 5 After how many years does a 1 percentage point TER difference cost a lump sum 25% of its final value?
  • 6 Compare an exit load of 1% with a TER difference of 1%: which costs more over 10 years, and why?
1 $500\times0.012/365\approx0.0164$ crore, about ₹1.64 lakh a day.
2 $5{,}00{,}000\,(1.11e^{-0.005})^{15}\approx\text{₹}\,22{,}19{,}436$ and $5{,}00{,}000\,(1.11e^{-0.015})^{15}\approx\text{₹}\,19{,}10{,}286$; ratio $\approx0.8607=e^{-0.15}$.
3 Both plans hold the same portfolio, so the gross factor $G_T$ — whatever it is — cancels: the ratio is $e^{-0.01\times15}$ in any market (Theorem 2.9).
4 Monthly net rates from $(1.12e^{-c})^{1/12}-1$; corpuses $\approx\text{₹}\,1{,}20{,}34{,}868$ and $\approx\text{₹}\,1{,}01{,}20{,}128$; gap about ₹19.1 lakh, 15.9% — between 0 and the lump-sum gap $1-e^{-0.25}\approx22.1\%$.
5 $e^{-0.01T}=0.75$ gives $T=\log(4/3)/0.01\approx28.8$ years.
6 The exit load is charged once (and only on early redemptions): at most 1% of the value. The TER difference is charged every day: over 10 years it costs about $1-e^{-0.1}\approx9.5\%$ of a lump sum's value.
IV
Index funds, ETFs, STP and SWP.
  • 1 In a year an index fund returned 9.62% and its benchmark TRI 10.05%; its TER is 0.30%. Find the TD and say what might explain the rest of it.
  • 2 The daily differences (fund minus index, in %) on six days are +0.01, −0.02, 0.00, −0.01, +0.02, −0.01. Find their sample standard deviation and the annualised tracking error.
  • 3 A fund keeps 3% in cash earning 6% while the index returns 15%. Estimate the cash drag. What happens if the index returns −10%?
  • 4 An ETF trades at ₹112.40 with iNAV ₹112.10. Find the premium and roughly how much extra a ₹50,000 purchase pays.
  • 5 ₹6,00,000 is placed in a liquid fund at NAV ₹3,000 and moved by an STP of ₹1,00,000 a month for six months into an equity fund. The liquid NAVs on the transfer dates are ₹3,000, 3,015, 3,030.1, 3,045.2, 3,060.4, 3,075.7 and the equity NAVs ₹80, 76, 79, 83, 81, 85. Find the liquid units redeemed, the equity units bought and the average cost of the equity units.
  • 6 An SWP withdraws ₹20,000 a month from ₹30,00,000 at an assumed 6% a year (effective). Is it sustainable for ever? If not, how long does it last?
  • 7 What does a fund's riskometer tell an investor, and what does it not tell her?
1 TD $=9.62-10.05=-0.43$ percentage points. The TER explains 0.30; the other 0.13 comes from cash drag in a rising year, trading costs at rebalancing and inflows, and the timing of dividends (Theorem 2.10).
2 Mean $-0.00167\%$; sample standard deviation $\approx0.0147\%$; TE $\approx0.0147\%\times\sqrt{252}\approx0.23\%$ a year.
3 Drag $\approx0.03\times(15-6)=0.27$ percentage points. If the index returns −10%, the term is $0.03\times(-10-6)=-0.48$, so cash adds about 0.48 points: the fund falls less than the index.
4 $(112.40-112.10)/112.10\approx0.268\%$; on ₹50,000 about $50{,}000\times0.30/112.40\approx\text{₹}\,133$ extra.
5 Liquid units redeemed $100{,}000/N$: 33.333, 33.167, 33.002, 32.839, 32.676, 32.513 — total 197.530 of the 200 units (the 2.470 left are the liquid fund's earnings, worth about ₹7,600). Equity units: 1,250.000, 1,315.789, 1,265.823, 1,204.819, 1,234.568, 1,176.471 — total 7,447.470; average cost $6{,}00{,}000/7{,}447.470\approx\text{₹}\,80.56$.
6 $i=1.06^{1/12}-1\approx0.4868\%$, $iC\approx\text{₹}\,14{,}603\lt20{,}000$: not sustainable. $n=-\log(1-14{,}603/20{,}000)/\log(1.004868)\approx269.7$ months, about 22.5 years.
7 It ranks the risk of the scheme's portfolio on a scale (currently six levels, Low to Very High). It does not predict returns, does not say the NAV cannot fall, and does not say whether the fund suits her goals and time horizon.
🎯 Practice corner NISM-styleCase studyInterview

E1. NISM-style A scheme's assets are worth ₹520 crore and its liabilities ₹8 crore; 16 crore units are outstanding. Its NAV is (A) ₹32.50 (B) ₹32.00 (C) ₹33.00 (D) ₹31.50.

Compute $(520-8)/16=512/16=32.00$. (A) ignores the liabilities; (C) adds them; (D) subtracts them twice.
Answer (B).

E2. NISM-style An investor's monthly SIP of ₹6,000 buys units at NAVs of ₹30, ₹20 and ₹24. Her average cost per unit is (A) ₹24.67 (B) ₹24.00 (C) ₹24.33 (D) ₹25.00.

Compute Units $200+300+250=750$; cost $18{,}000/750=\text{₹}\,24.00$, the harmonic mean. (A) is the arithmetic mean of the NAVs, which is what buying equal units would cost.
Answer (B).

E3. NISM-style The tracking error of an index fund is best described as (A) the difference between the fund's return and the index's return over a year (B) the annualised standard deviation of the differences between the fund's and the index's daily returns (C) the fund's total expense ratio (D) the fund's beta with respect to the index.

Reason (A) is the tracking difference. The TER (C) contributes to the tracking difference but, being smooth, hardly to the tracking error. Beta (D, §3.2) measures sensitivity, not closeness.
Answer (B).

E4. Case study An investor holds ₹15,00,000 in the regular plan (TER 1.8%) of a fund, bought more than a year ago, with an unrealised gain of ₹5,00,000. Switching to the direct plan (TER 0.8%) is a redemption plus a purchase. Assume no exit load and an illustrative tax of 12.5% on the gain (ignore any exemption). (a) How much is invested in the direct plan after the switch? (b) Ignoring the tax the regular plan would pay later, after how many years does the direct plan catch up? (c) Why does part (b) overstate the break-even time?

(a) Tax $0.125\times5{,}00{,}000=\text{₹}\,62{,}500$, so ₹14,37,500 goes into the direct plan.
(b) By Theorem 2.9 the direct plan grows faster by the factor $e^{0.01}$ a year, whatever the market: $14.375\,e^{0.01T}=15$ gives $$T=\frac{\log(15/14.375)}{0.01}\approx4.26\text{ years}.$$
(c) Staying in the regular plan does not avoid the tax — it postpones it, and the gain taxed later will be larger. Comparing after-tax values at the end, the switch pays off even sooner. (Tax rules and exemptions change; check the current ones before acting.)

E5. Case study A retiree has ₹60,00,000 and needs ₹40,000 a month through an SWP. Assume a steady 8% a year (effective). (a) Is the SWP sustainable for ever? (b) How long does it last? (c) Suppose instead that markets fall 20% just after he starts, so the corpus becomes ₹48,00,000, and then returns 8% a year. How long does it last now?

(a) $i=1.08^{1/12}-1\approx0.6434\%$, $iC\approx\text{₹}\,38{,}604\lt40{,}000$: not quite.
(b) $n=-\log(1-38{,}604/40{,}000)/\log(1.006434)\approx523$ months, about 43.6 years — effectively a lifetime, under this steady assumption.
(c) $iC=\text{₹}\,30{,}883$, and $n=-\log(1-30{,}883/40{,}000)/\log(1.006434)\approx231$ months, about 19.2 years. One bad start more than halves the life of the plan: sequence-of-returns risk (§8.1).

E6. Interview You invest equal rupee amounts on two days at NAVs $x$ and $y$; a friend buys equal numbers of units on the same two days. Who has the lower average cost, by exactly how much, and what does this say about volatility?

Means Your cost is the harmonic mean $H=2xy/(x+y)$; your friend's is the arithmetic mean $(x+y)/2$.
Gap $\frac{x+y}2-\frac{2xy}{x+y}=\frac{(x-y)^2}{2(x+y)}\ge0$, so you pay less unless $x=y$. With NAVs 40 and 60 the gap is $400/200=\text{₹}\,2$ per unit (48 vs 50).
Volatility The gap grows with the square of the price difference: rupee-cost averaging is worth more when prices swing more — a second-order (variance) effect, like the volatility drag of §3.1. It is an advantage relative to buying equal units, not a promise of profit.
🧑‍🏫 For the teacher

Suggested plan — 2 lectures × 60 min.

  1. NAV, units and SIPs. Open with the 📍 situation and ask: "Which is cheaper, a ₹14 NAV or a ₹480 NAV?" NAV and units (Definition 2.7, Theorem 2.6, Example 1); FIFO and exit loads (Example 2); growth vs IDCW (Theorem 2.7, Example 3). Then the SIP's harmonic mean (Theorem 2.8) with Example 4 on the board, and project Interactive 1: first "fall, then recovery" (the SIP shines), then "steady rise" (the lump sum wins).
  2. Costs and tracking. Expense drag (Theorem 2.9) with Example 5 and the 📈 Market case table — ask the class to guess the 30-year gap before revealing it. Index funds, ETFs and tracking (Definitions 2.10–2.11, Theorem 2.10, Examples 6–7); project Interactive 2 and change one slider at a time (TER moves the purple line, noise makes the amber line wobble, cash drag follows the market). End with SWP (Theorem 2.11, Example 8) and Practice corner E5.

Class activity. Give each pair a printed factsheet of a (hypothetical) fund with its direct and regular TERs. Each pair computes the 20-year cost of the TER difference on a ₹5,000 SIP under two different return assumptions and shows that the lump-sum ratio does not depend on the assumption.

Misconceptions to watch for. Low NAV = cheap; IDCW = extra income; SIP = guaranteed profit; "1% doesn't matter"; tracking error confused with tracking difference; comparing an index fund with the price index.

Suggested homework. I(1, 4, 5, 6), II(1, 2, 5, 6), III(2, 3, 4, 5), IV(1, 2, 3, 6); Practice corner E2, E4, E6.

🛠️ Do it yourselfSpreadsheet mini-project · about 60 minutes

Project. Download the NAV history of the direct and regular plans of one index fund, and of its benchmark's total return index, and measure the cost of the regular plan, the fund's tracking difference and its tracking error — then replay a SIP through the history.

Why it matters. Factsheets quote these numbers; computing them yourself shows exactly what they mean and how sensitive they are to the period chosen.

What to do.

  1. From the AMFI website (historical NAVs) download two years of daily NAVs of both plans of an index fund; from the index provider's website download the same dates of its TRI. Put dates in A, direct NAV in B, regular NAV in C, TRI in D, aligned by date.
  2. Daily returns: E2 =B3/B2-1, F2 =C3/C2-1, G2 =D3/D2-1; differences H2 =E2-G2, I2 =F2-G2.
  3. Over the last 252 rows: TD of each plan =PRODUCT(1+E2:E253)-PRODUCT(1+G2:G253) (array formula) and TE =STDEV.S(H2:H253)*SQRT(252). Compare the TD gap between plans with their TER difference.
  4. Check Theorem 2.9: =(C_last/C_first)/(B_last/B_first) against =EXP(-(TER_reg-TER_dir)*years).
  5. SIP replay: pick the first trading day of each month, put ₹5,000 in column K, units =K2/B2 in L; average cost =SUM(K:K)/SUM(L:L) and the harmonic mean =COUNT(L:L)/SUMPRODUCT(1/B_dates); compare with =AVERAGE(B_dates).

Python version: d=fund.pct_change()-tri.pct_change(); TD=(1+fund.pct_change()).prod()-(1+tri.pct_change()).prod(); TE=d.std()*252**0.5; units=(5000/nav_monthly).sum().

Hand in: a one-page note with each plan's TD and TE, the measured cost ratio vs $e^{-\Delta c\,T}$, and your SIP's average cost vs the average NAV.

Going further. Do the same for an ETF on the same index, using its exchange closing price instead of its NAV: how much of its tracking error comes from premiums and discounts?

Instructor note: the measured ratio of the two plans' growth should match $e^{-\Delta c\,T}$ closely if the TERs were constant; differences usually mean a TER change during the period (check the AMC's TER disclosures). Students often forget to align dates across files and to use the TRI rather than the price index — the second mistake typically makes the TD look about one percentage point better than it is.

🧪 Self-check quiz

0/5 correct

Q1. Fund P has NAV ₹20 and fund Q has NAV ₹200; both hold similar portfolios. Which statement is right?

Q2. The average cost per unit of a SIP of a fixed rupee amount is

Q3. Two plans of the same fund differ in TER by 1 percentage point. After 25 years a lump sum in the costlier plan is worth about

Q4. A fund declares an IDCW of ₹2 per unit when its NAV is ₹30. Immediately afterwards

Q5. Index fund A has tracking difference −0.20% and tracking error 0.05%; index fund B has tracking difference −0.20% and tracking error 0.60%. Which statement is right?