Who sells mutual funds: methods

Code: pipeline/tipsheet/compute/mf_distribution.py, pipeline/tipsheet/publish/mf_distribution.py, step mf_distribution (pipeline/tipsheet/steps/insights_mf_distribution.py). Tests: pipeline/tests/test_mf_distribution.py. Page: /funds/distribution/.

What we measure

• Channel split, monthly from March 2014: the share of mutual fund average AUM held in direct plans, sold through distributors in the fund house’s own sponsor group (“associate distributors”) and sold through every other distributor. It is broken down by fund group, investor type and city group.
• Direct share by SEBI category, quarterly from January–March 2013: direct-plan average AUM over the category’s average AUM.
• The distribution bill, quarterly from April–June 2018: what regular-plan investors pay each year, in rupees, over what direct-plan investors in the same schemes pay. It is shown by fund group, by SEBI category and by financial year.

ETFs have a single plan and trade on exchanges, so they are left out of every direct share unless a figure says otherwise. Counting them raises the direct share by about four points (July 2026: 48.8% with ETFs, 44.4% without).

Sources

1. AMFI’s category-wise average AUM report (Data bank amfi_aum_catwise, API scheme-catwise-data), monthly from March 2014. It follows the format in SEBI’s circular CIR/IMD/DF/05/2014 of 24 March 2014. Every cell has five keys:
   • Category: AMFI’s twelve broad rows (liquid/money market, gilt, FMP, other debt, ELSS, other equity, balanced, gold ETF, other ETFs, overseas FoFs, and two empty legacy rows). AMFI still uses this pre-2018 list in this report. Domestic fund-of-funds sit outside the grand total, as AMFI keeps them out to avoid counting money twice.
   • Channel: TDP direct plans, TAD through associate distributors, TNAD through non-associate distributors.
   • City group: T15 top 15 cities, B15 the rest. The report keeps these labels even though SEBI’s incentives moved to T30/B30 in 2018.
   • Sponsor bucket: I and II (see below).
   • Investor type: sub-columns 1 to 5.
2. AMFI’s scheme-wise average AUM (amfi_aum_schemewise), quarterly, one row per plan.
3. AMFI’s daily TER (amfi_ter_daily_fy2018_fy2020, amfi_ter_daily_fy2021_fy2023, amfi_ter_daily), from 6 April 2018, regular and direct plan side by side.

Reading the category-wise report’s codes

AMFI’s API returns the codes without labels, so two readings were checked before use:

• Investor types 1–5 are retail, corporates, banks/FIs, FIIs/FPIs and HNIs, in that order. For June 2026 the five sums (₹22.6, 31.5, 1.5, 0.06 and 28.5 lakh crore) line up with AMFI’s investor-class report for the same quarter end (retail 23.7, corporates 29.3, banks/FIs 1.0, FIIs 0.06, HNIs 30.2). The quarter-end figures differ a little from a monthly average, but no other ordering fits.
• Bucket I is money from the AMC’s sponsor and its associates; II is everyone else. The 2014 circular requires that split. Bucket I never holds any retail money in any month, sits mostly with corporates and banks, and is 0.9–2.4% of AUM. That is what sponsor-group investments in their own schemes look like.

AMFI defines retail and HNI investors by the size of their investment, not their wealth.

Calculations

• Checks on every run: each row’s cells must sum to its printed total, and the rows to the grand total, within ₹5 crore. Every category label must be known.
• Direct share by category: a plan whose name contains “direct” is a direct plan; every other plan of a non-ETF scheme is a regular plan. AMFI’s API returns some quarters twice, under two financial-year ids, with identical AUM. The step keeps one row per plan and quarter and fails if the two copies ever disagree.
• Categories: AMFI prints each scheme under its current SEBI category, in two naming schemes, plus legacy labels (“Growth”, “Income”, “Balanced”) for schemes closed before 2018. All are mapped to one list (sw_category). Sectoral and thematic funds are one category, and so are value and contra. Debt is split into liquid, overnight, money market and the rest.
• Cross-check: the scheme-wise quarterly direct share (ETFs excluded) is compared with the category-wise report’s first month of the same quarter. The two are separate AMFI reports. Since 2014 they never differ by more than 2.2 percentage points (median 0.2), and AUM agrees within about 5%. The step fails if the share gap ever exceeds 3 points.

The distribution bill

For each regular plan in each quarter:

bill = quarterly average AUM × (regular-plan TER − direct-plan TER) ÷ 4

• The gap is averaged over the quarter’s days. Up to March 2026, AMFI’s printed total TER is base TER plus additional expenses plus GST. From April 2026 it also includes brokerage and transaction costs. Those are the same for both plans, but their regular-minus-direct difference is subtracted anyway, so the gap means the same thing throughout.
• GST: the gap includes the GST charged on it (18% throughout the record), and the ex-GST figure divides by 1.18.
• What it is: the extra cost borne by regular-plan investors. Fund houses pay distributors’ trail commission out of it, but can pay more or less than the gap. So it is not a measure of what any distributor received.
• Matching TERs to AUM: TER files print scheme names as they were at the time, while the AUM file prints current names. A rename is detected when a TER name stops and another from the same fund house (same first word) starts within seven days with identical regular and direct TERs. Ties are broken by SEBI category, otherwise left unlinked. That gives 917 renames, listed in .cache/derived/qa_mfd_ter_renames.csv for review. A regular plan is matched when its plan-free base name equals the base name of any name in a TER rename chain, and base names claimed by two chains are dropped.
• Coverage: matched regular-plan AUM rises from 56% in 2018 to 90% in 2026, and is published for every quarter and year. Unmatched AUM takes the AUM-weighted gap of matched plans in the same category and quarter. The matched-only figure is published alongside (bill_matched_q_cr).
• Outside check: AMFI’s disclosure of commission paid to its 3,158 largest distributors totalled ₹21,106 crore for FY25. Press estimates of the whole industry’s commission are about ₹27,000 crore (Business Standard, 31 August 2025; Business Today, August 2026). Our FY25 estimate is ₹23,936 crore excluding GST and ₹28,244 crore including it.

What it cost in returns (from NAVs)

compute/mf_realised_cost.py, bundles funds/distribution/realised_yearly and realised_trailing.

• Pairs: every scheme with exactly one direct and one regular growth plan in NAV history, matched by fund house and plan-free name. Legacy retail, institutional, super and premium plans are left out. That gives 2,033 pairs, closed and merged schemes included for the years they existed. ETFs and domestic fund-of-funds are left out.
• Gap: (1 + direct return) / (1 + regular return) − 1 over the same window, from month-end NAVs. A simple difference of returns would grow with the year’s return (a 1% fee costs about 1.3 points in a +30% year); the relative gap does not. Calendar years run December to December. Trailing windows end at the latest complete month and are annualised.
• Weights: calendar years use each regular plan’s December-quarter average AUM; trailing windows use the latest quarter’s. Medians are shown alongside.
• Cleaning: a pair-year whose gap exceeds 5 points either way is not one portfolio (a mislabelled plan or a segregated portfolio) and is dropped. That’s 6 of 9,839, listed in .cache/derived/qa_mfr_dropped.csv.
• Agreement with the TER gap: for equity schemes the NAV-based gap is within 0.05 points of the TER-based gap in every year from 2018 (largest: 0.04 in 2020; the 2018 TER figure covers April–December). The two measures share no input except scheme-wise AUM weights.
• ₹10 lakh over ten years: each pair’s growth of ₹10 lakh in both plans. The page shows the medians of each and the median of the pair-by-pair shortfall, which need not equal the difference of the two medians.

Advised or do-it-yourself

AMFI’s bifurcation of direct-plan average AUM (Data bank amfi_direct_plan_aaum, monthly from January 2024) splits it into money placed through SEBI-registered investment advisers (RIA), portfolio managers (PMS) and the rest, which AMFI calls do-it-yourself. The step fails if the three parts differ from the total by more than 1%. AMFI’s direct total in this report equals the category-wise report’s direct AUM including ETFs, so ETF purchases count as do-it-yourself.

Source breaks (not adjusted, marked on the charts)

• November 2016: about ₹8,000 crore of retail direct-plan equity money (a third of it) disappears from the category-wise report in one month, with no offsetting rise elsewhere. In the same month about ₹14,000 crore of corporate ETF money moves from “direct” to “non-associate distributor”. Both look like one fund house reclassifying. They lower the retail, individual and B15 direct shares from that month on.
• June–July 2017: associate-distributor AUM falls by about a fifth across debt, equity, hybrid and liquid, moving to non-associate distributors, and returns in August 2017. This looks like two months of misreporting.

Limits

• The category-wise report’s equity row (“Others”) includes index funds after 2018. The scheme-wise table separates them.
• The TER record starts in April 2018, so the bill starts in the April–June 2018 quarter. An estimate before that would need the gap inferred from NAV differences between plans; we have not done it.
• The scheme-wise categories are current categories applied to past quarters. A scheme that changed category in 2018 appears under its new one throughout.