An investor will put Rs 25,000 a month into the index for ten years. Compare an ETF with a low fee but brokerage, a spread and a demat account against an index fund with a higher fee. Which costs less, and when does the answer flip?

The ETF ends about Rs 27,375 ahead after ten years, roughly 0.5% more, on these assumptions. The index fund's 0.20% fee costs about Rs 59,340 of growth; the ETF's 0.05% fee plus Rs 20 an order, half the spread on every trade and demat charges cost about Rs 31,965. In year one the index fund is actually ahead, and below about Rs 7,500 a month it stays ahead for all ten years.

One cost is a percentage of what you hold; the other is a toll on each purchase. A gym that charges a monthly fee and a gym that charges per visit cost the same for someone, and which is cheaper depends on how you use it. The expense ratio grows with the corpus, while brokerage is a fixed Rs 20 whatever the size of the order, so small tickets early on favour the index fund and a large corpus later favours the ETF. The spread sits in between: you pay about half of it, 0.05%, each time you buy at the ask price and again when you finally sell at the bid.

Measure both against the same money earning 12% with no costs at all, which grows to about Rs 56,00,897. The index fund ends at about Rs 55,41,557, a cost of Rs 59,340; the ETF ends at about Rs 55,68,932, a cost of Rs 31,965. Of the ETF's cost, Rs 14,897 is its fee; the rest is 121 brokerage charges of Rs 20 (Rs 2,420), half-spreads on each purchase and on the final sale (Rs 4,285), ten years of demat charges (Rs 4,000), and the growth all of that money would have earned had it stayed invested.

Run the same ten years at smaller amounts. Below about Rs 7,500 a month the fixed Rs 20 and the demat charge are a large share of each order, and the index fund wins over the whole ten years. Beyond cost, three practical points belong in the answer. An ETF needs him to place an order every month, and a missed month costs more than any fee difference; many brokers offer automated ETF purchases, which removes that. An ETF can trade slightly above or below its underlying value on a thin day, a cost this model ignores. And the index fund's SIP never lets him buy at a bad intraday price by mistake.

State the limit: the 12% is a planning assumption, not a forecast, and the gap scales with it. If returns were lower, both costs shrink in rupees and the gap narrows. The structure of the answer holds either way: a percentage fee is cheap on a small corpus and expensive on a large one; a per-order toll is the reverse.