The right track

Index funds track the broad market, and thus make for a good investment in a rising market—but watch out for the tracking error.
Narendra Nathan
6 Jul 2001

SEASONED INVESTORS know that the best time to buy is when everyone else is selling. Which should make it a good time to invest in equities now. That, however, has its own associated difficulties—stock-picking is not easy.

There is, however, a special class of mutual funds—index funds—that are designed to track the market's every move. That makes them an attractive investment in a market recovering from a bear run.

No fund-manager risk. Mutual funds have historically offered lower returns than the broader market. That is because fund managers, in their attempts to beat the market, often go overboard and deviate from their mandates. LIC’s Dhanasamrudhi Fund, for instance, has a portfolio that is heavily skewed in favour of just four stocks: ITC (37 per cent), Hind. Lever (27 per cent), Bharat Petroleum (18 per cent) and Sterlite (17 per cent). Others are heavily concentrated on just one sector. For instance, ING Growth Fund is biased in favour of ICE (Information, communication and entertainment) stocks, which constitute 92 per cent of its portfolio. Not surprisingly, the fund’s NAV has fallen 67 per cent as compared to the broad market’s fall of 28 per cent.

Managers of index funds, on the other hand, have a simple mandate—to closely track a chosen index. That precludes any attempts to beat the market, and therefore the low returns that most mutual funds are prone to. It also makes for a well-diversified and balanced portfolio. The table below details the four index funds available in India.

Beware of tracking error. Theoretically, returns from index funds should not deviate at all from the indices they are based on. In practice, however, there is a slight variance, known as the tracking error.

For an illustration, take the returns from a fund based on the Sensex. Rs 100 invested in a Sensex index fund (on 1 April, when the index was launched) should now be worth Rs 3,457. This works out to a reasonable long-term return of 17.15 per cent per annum. A tracking error of 1 per cent would mean that returns from this fund would vary from the Sensex by 1 percentage point on an yearly basis. If that 1 percentage point tracking error is on the negative side (that is, the fund’s returns have lagged the Sensex by 1 percentage point year-on-year), the index fund would have grown at 16.15 per cent since its launch. The Rs 100 investment would now be worth Rs 2,867—a gap of Rs 590. So, the lower the tracking error, the better the index fund.

Index funds in India typically have higher tracking errors than the ones in developed markets. For instance, the tracking error of UTI Nifty Index Fund exceeds 5 per cent—that’s more than 10 times the average figure for index funds in developed markets.

Causes of tracking errorExpenses. Expenses are the biggest cause of tracking error. Fund’s expenses include asset management fees, agents’ commissions, brokerage on buying and selling of shares and stationery expenses. Index funds in developed markets have very low expenses, thus reducing their tracking errors. By contrast, the best index fund in India, IDBI Principal Index Fund, has a tracking error of 0.74 per cent. The figure is even higher for the two index funds managed by UTI, as it charges expenses at the highest level allowed by Sebi. Explains Nilesh Shah, the fund manager, "We have to spend a lot of time and money to educate investors—what an index fund is, what can be expected from them…"

A closer look reveals that the major constituent of the expenses is the asset management fee—all the four available funds charge expenses at the highest permitted levels. (Two of these funds charge 1.25 per cent, the maximum allowed for funds of a size below Rs 100 crore.)

That seems unreasonable, as there is no active trading or stock selection involved—after all an index fund merely tracks the index. Rajat Jain, chief investment officer, IDBI Principal, counters this: "Even though there is no active trading, we still have to manage the funds and we have one person dedicated for this fund." Nilesh Shah, for his part explains: "We have forced to charge the maximum possible AMC fee because of the small corpus. Once the corpus improves, the AMC fee will come down."

Lack of competition is probably the biggest reason index funds charge the maximum allowed fund management fees. As more fund houses enter the fray, these fees should fall.

Dividends. Most index committees do not factor in the dividends declared by companies while computing the index. For instance, the dividend yield of Sensex and Nifty stocks is around 1.5 per cent, but this is not considered in the index calculations. An index fund’s NAV would rise when it receives dividend, and this would introduce a positive tracking error (which is a good thing, as opposed to a negative tracking error). The argument is that dividends represent a very small fraction of a fund’s corpus, and so are not worth the trouble.

The index committee at the NSE (National Stock Exchange) is now working on the Total Return Nifty (inclusive of the dividend), which would be used as the benchmark to compute the tracking error. Such a reworked index is not available for the UTI Master Index Fund, the only fund tracking the index.

Trading expenses. Tracking error is the difference between a fund’s NAV and the index’s closing price on any given day. Both the NAV and the index are calculated at the end of each trading day, based on the closing prices of stocks. However, funds have to do their buying and selling throughout the day, and their prices may vary from the closing prices.

The NSE’s post-closing session enables funds to buy and sell at the closing price, thus eliminating one constituent of the tracking error.

Market lots. Assume that an index fund receives an inflow of Rs 5 lakh in a day. This can’t be parked on the same proportion on the same day, as the exact division will leads to investment in shares in fractions, which is not possible. The allotment has to be rounded off to the nearest whole number, which introduces another element of tracking error. This problem used to be even more severe before the introduction of dematerialisation, when stocks had to be traded in market lots of 50 or 100 shares.

Index futures have helped funds reduce this element of tracking error. Explains Rajat Jain, chief investment officer, IDBI Principal, "What we can do now is park these small (residual) amounts in index futures and convert them into actual shares on a later day."

Cash component. Being open-ended, index funds have to keep a fraction of their corpus ready in the form of cash. This cash component adds to tracking error. For illustration, assume an index fund with 4 per cent of its corpus in cash. If the index rises by 5 per cent, this fund’s NAV can rise only by 4.8 per cent. The tracking error will be directly proportional to the cash component of an index fund.

Frequent trading by punters. Index funds are usually favoured by long-term investors, but punters often trade in and out to profit from short-term fluctuations. Frequent trades increase transaction costs, thereby increasing the tracking error.

All index funds have entry and exit loads to deter frequent trading. IDBI Principal charges an entry load of one per cent. UTI and Templeton funds do not charge entry loads, but have an exit load of one per cent for withdrawals before six months from the date of entry.

How is tracking error measured?The tracking error is the difference between the return generated by an index fund and those by the index it is based on. Assuming that the return generated by an index fund on a given day is 5 per cent and that by the underlying index is 5.13 per cent, the fund’s tracking error for that day would be 0.13 per cent. These errors are squared and added up for the last one year; the square root of this sum is the tracking error value for the full year. Another way to arrive at the tracking error is to calculate the standard deviation of the daily errors. This standard deviation is then annualised by multiplying it by the square root of 250 (on the assumption that there are 250 trading days a year)