The Hidden Cost of "Regular" Mutual Fund Plans in India
Most Indians who invest in mutual funds are quietly sitting in the costlier version of the exact same fund — and losing lakhs to crores over a lifetime for nothing in return. This project measures that cost from official data, fund by fund.
Every open-ended Indian mutual fund is sold in two forms of the identical portfolio:
| Regular plan | Direct plan | |
|---|---|---|
| Portfolio | Same stocks, same manager | Same stocks, same manager |
| Distributor commission | Yes — skimmed from the fund every year | No |
| How you usually get it | Default via banks, agents, many apps | You have to choose it |
SEBI made Direct plans mandatory in January 2013, yet most retail investors have never heard of them and stay in the Regular version by default. Because the two plans hold the same portfolio, the gap between their returns is a clean, assumption-free measurement of the commission you're paying — and almost nobody cross-checks it. This project does.
- Across 1,084 cleanly matched funds (50 fund houses, 2013–2026), the Direct plan beat the Regular plan in 99% of funds.
- The typical fund skims ~0.8 percentage points/year from Regular investors overall, and ~1.4 pp/year for equity funds — the funds people hold longest.
- The penalty is a flat toll, uncorrelated with how well the fund performs (it's a fee, not a performance share).
- On a modest ₹10,000/month SIP in a typical equity fund, the Regular plan costs roughly ₹27 lakh over 20 years (18% of the corpus) and ~₹1.9 crore over 30 years (28% of the corpus) — versus the identical Direct plan.
- The gap is statistically ironclad (paired Wilcoxon p ≈ 10⁻¹⁷⁶) and structural: fund category alone explains ~59% of its variation, concentrated in equity.
2. Equity investors pay the steepest hidden fee
Switching from a Regular plan to the Direct plan of the same fund is one of the only financial decisions that is pure upside: identical portfolio, lower cost, no downside. For a default-Regular investor with no advice attached, the commission is simply leakage.
- Source: AMFI daily NAV history, accessed via the free public mfapi.in API (a mirror of AMFI's official files).
- Universe: all 37,631 listed schemes → filtered to 1,084 open-ended funds with a matched Direct-Growth and Regular-Growth plan and ≥2 years of overlapping history.
- Reproducibility:
src/download.pyfetches and caches everything. The small scheme list and the processed dataset are committed; the bulk raw NAV cache (~300 MB) is regenerated by the download script (seedata/README.md).
# 1. Environment (Python 3.12 recommended)
python3.12 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
# 2. Download + cache the raw data (~5 min; skips anything already cached)
python -m src.download
# 3. Run the analysis, in order
jupyter notebook # then run 01 -> 02 -> 03 -> 04
# or headless:
for nb in notebooks/0*.ipynb; do
jupyter nbconvert --to notebook --execute --inplace "$nb"
donehidden-mutual-fund-fees/
├── data/
│ ├── raw/ scheme_list.json (committed) + nav/ cache (regenerated)
│ └── processed/ fund_returns.csv <- the analysis-ready dataset
├── notebooks/
│ ├── 01_eda.ipynb profile the raw data, surface the messiness & questions
│ ├── 02_cleaning.ipynb parse names, pair plans, compute drag — every decision documented
│ ├── 03_analysis.ipynb the four headline questions, with charts + interpretation
│ └── 04_modeling.ipynb paired hypothesis test + regression + honest limitations
├── src/ reusable logic, imported by the notebooks
│ ├── download.py fetch & cache AMFI NAV data
│ ├── cleaning.py name parsing, plan pairing, return/drag computation
│ ├── plotting.py shared chart styling
│ └── paths.py canonical project paths
├── reports/figures/ exported charts (embedded above)
├── requirements.txt
└── README.md
- Pair each fund's Direct-Growth plan with its Regular-Growth twin via a normalised name key (exactly one candidate per side — correctness over coverage).
- Align both plans to their common date window (Direct plans start in 2013).
- Measure each plan's CAGR over that identical window;
drag = direct − regular. - Aggregate robustly (median headline), translate into rupee SIP cost, and confirm with a paired test + regression.
Full reasoning, including every cleaning decision, lives in the notebooks.
Python 3.12 · pandas · numpy · matplotlib · seaborn · scipy · statsmodels ·
scikit-learn · Jupyter · requests.
- Survivorship bias (the effect is likely understated): only funds alive today with both plans are pairable; merged/closed funds — often poor performers — are absent.
- Tracking noise over short windows; mitigated with a ≥2-year floor and by reporting the median rather than the mean.
- No AUM / exact expense ratio in the free feed. Next step: join AMFI's TER disclosures to model the commission directly and add fund-size effects.
- Distributor advice has value for some investors — the commission is its price. The finding targets the large set of investors who hold Regular plans by default, with no advice attached.
- Possible extensions: a public "what is your Regular plan costing you?" calculator; per-category SIP cost tables; tracking how the gap evolves as SEBI tightens expense-ratio rules.
Returns are historical (2013–2026) and not a guarantee of the future. The mechanism — Direct = Regular minus commission — is structural and forward-looking; the exact rupee figures depend on future returns.


