Comparative costs of different index-tracking solutions


By Nerina Visser

20 Nedbank Logo

20 Nerina VisserThe costs of investments – whether products or services – have come under increasing scrutiny in recent times, and rightly so. Many attempts have been made to identify, justify, explain and expose investment costs, but it sometimes seems that the more we learn about these costs, the less we know and understand about the full impact of all costs associated with investment.

It should be acknowledged at the outset that ALL investments have costs associated with them – whether these are disclosed or not – and regardless of whom picks up the tab, the expenses are still incurred. The full extent will eventually show up in the performance achieved by the end client, rather than in the declared ‘total’ expense ratio (TER) or the selected charges quoted by different parties in the investment value chain.

By that same token, as much as indices are used as comparative performance measurements, it is not entirely an exercise that compares apples with apples, as an index is merely a mathematical calculation, reflects zero costs and applies theoretical ‘reinvestment’ actions that can’t be replicated by an actual fund, e.g. reinvesting dividends on the ex-dividend date at no cost, and at the closing price of the last-day-to-register (LDR) date, compared to the reality of the fund who can only re-invest such a dividend after the payment date at the then ruling price, incurring costs in the process. As such, the performance of an index is a theoretical calculation that can never be matched exactly by actual fund performance, neither passively nor actively managed.

Costs can be divided into three broad categories:

  1. Product costs, incl. the cost of management and administration of the product itself.
  2. Acquisition costs, incl. the management, administration and custody of the client assets and cash flow.
  3. Advice costs, whether for a financial advisor or consultant.

Product costs fall largely outside of the control of the investor, other than choosing a more cost efficient investment vehicle, or having access to certain preferential pricing models due to size, influence or the like. For the most part, product costs will depend on the investment process, product ‘wrapper’ and implementation strategy selected by the product provider. To this end the informed investor should familiarise himself with the benefits and (often unintended) consequences of a particular investment choice, to ensure that it is most suitable to his particular requirements.

Acquisition costs also have some components that are outside the choice or influence of the investor, but there is more that can be controlled by the client, and alternatives that can be explored and thus compared to select the most cost efficient alternative. Not all channels are equally available to all investors alike, and it is misleading to present institutional pricing models alongside those available to the retail investor, or even wholesale clients.

Advice costs are the components that are most within the control of the client, but that does not mean that it should be avoided – its value should not be underestimated. Of cardinal importance though is that for advice costs to be truly fit-for-purpose, it should be independent of both product choice and acquisition channel. Just like one would pay your doctor for his skilled assessment and advice, not for the medicine he prescribes, the same should apply to financial and investment advice.

For the purpose of this article, we shall limit our discussion to the first components, and in that way hopefully assist advisors and consultants, or private investors, to make more informed decisions regarding investment options.

This article concludes that TERs do not tell the full story of costs. The evidence presented by evaluating relative performance of portfolios following the same investment mandate (e.g. tracking the same index) is much more insightful. When we evaluate the closest possible options, i.e. two investments tracking the same index, we can start to ask some pertinent questions.

For example, for the three year period to 1 September 2014, ProfileMedia reported the following comparative returns after costs:

  • Satrix RAFI40 Total Return ETF (A): 18.7% p.a. vs. Satrix RAFI40 Index Fund (A1): 18.4% p.a.
  • BettaBeta Equally Weighted Top40 ETF: 18.3% p.a. vs. Satrix Equally Weighted Top40 Index Fund (A1): 17.5% p.a.
  • Satrix Divi+ ETF (A): 14.2% p.a. Satrix Dividend+ Index Fund (A1): 13.7% p.a.

Therefore the key question to ask is the following: “to what extent are trading costs included in the TER, and what impact does it have on total return?” Let us evaluate some of the reasons for these differences.

Product costs:

In an ideal world, the best way to determine explicitly what the impact of all costs have been on a particular investment product, would be to compare the actual performance of the fund, with the theoretical performance of the same fund if no costs were incurred. The closest we can get to this ‘ideal world’ is to compare the performance of an index-tracking investment product with the performance of the index it tracks. Unfortunately this type of analysis is impossible in the case of an actively managed portfolio, or even in a passively managed portfolio that does not employ a full physical replication strategy (e.g. sampling or synthetic replication), as we would never know what the return would have been if no costs were incurred. However, it is still useful to know that most of the costs identified in this article are applicable to both actively and passively managed funds, albeit not in the same magnitude.

The product costs of a physically replicated, index tracking fund can be divided into four categories:

  1. Initial costs
    1. primary market operations;
    2. secondary market operations.
  1. On-going costs
    1. rebalancing costs;
    2. implementation of corporate actions.

Let us discuss the implications and impact of these.

Primary market operations refer to the initial investment into a fund, or the disinvestment from a fund. In the case of a collective investment scheme (unit trust, or exchange traded fund), this refers to the creation or redemption of participatory units in the fund. These operations are equally applicable to an Exchange Traded Fund (ETF), a Collective Investment Scheme (CIS) or a Segregated Mandated Portfolio (SMP).

The transaction costs to invest or disinvest funds in a primary market operation, has three components – the brokerage payable to the executing stock broker, statutory charges payable to the JSE and Strate (e.g. investor protection levy, electronic settlement charges) and securities transfer tax (STT) payable to government (fiscus).

  • Brokerage is negotiable and can vary between a fixed percentage of the value of the trade (e.g. 5 bps, or 0.05% of the transaction amount), or a fixed rand amount per transaction (e.g. R50).
  • Statutory charges are not negotiable and vary on a sliding scale between a R12.45 and R62.23, incl. of VAT per ‘hit’, i.e. for each transaction – buy or sell – executed on a day.
  • STT of 25 bps, or 0.25% of the transaction amount, is not negotiable, but only payable in a buy transaction of securities, not a sell transaction and not payable when trading in ETFs or in Real Estate Investment Trusts (REITs), as these are trusts, not securities.

The total transaction costs, for a base case of R100m investment in an ALSI tracker fund (±165 shares), at 5 bps brokerage, is 31.1 bps for buying, but only 6.1 bps for selling. From the graphs in Figure 1 below, it is clear that on the buy side the costs are dominated by the STT charge, whereas the majority of the sell side costs are attributable to brokerage.

Fig. 1 Total transaction costs in basis points (bps) to invest (buy) or disinvest (sell) R100m in an ALSI tracker fund at 5bps brokerage


Source: Nedbank Capital calculations


From this base case scenario of R100m in an ALSI tracker at 5bps brokerage, the following cost sensitivities can be noted:

  • Sensitivity to trading costs: STT is non-negotiable, so it remains a fixed percentage for any investment made into securities (shares). The amount attributable to brokerage increases linearly as the brokerage rate increases, and the statutory charges increase with the number of trades (i.e. the number of ‘hits’ – therefore, should it take more than one day to implement a trade, the statutory charges will increase accordingly).
  • Sensitivity to investment amount: there is a non-linear decrease in trading costs on a percentage basis as the investment amount increases, although the statutory charges may increase if there is an increase in the number of ‘hits’ required to implement the total investment. This will also be influenced in the depth of the bids and offers of the underlying shares, i.e. if there is sufficient liquidity in all the underlying shares to complete the full investment in a single day.
  • Sensitivity to number of constituents: there is a non-linear increase in statutory charges as the number of constituents increase, but there could be an unexpected inverse effect on the number of ‘hits’ as constituents decrease, especially as the investment amount increases, depending on market depth.

Another sensitivity that can be evaluated is to the impact of a lump sum investment, vs. periodic cash flows, e.g. what is the impact of investing R100m as a lump sum on a single day, vs. investing R10m on 10 consecutive days? This is shown graphically in Fig. 2.

Fig. 2 Total transaction costs in basis points (bps) to invest (buy) a R100m lump sum in an ALSI tracker fund vs. 10 separate flows of R10m each


Source: Nedbank Capital calculations


The cost differential for this investment size (R100m) is not significant, however, as the investment amount decreases – to a lump sum of R10m on a single day, vs. investing R1m on 10 consecutive days, the differential becomes much bigger, as is shown in Fig. 3.

Fig. 3 Total transaction costs in basis points (bps) to invest (buy) a R10m lump sum in an ALSI tracker fund vs. 10 separate flows of R1m each


Source: Nedbank Capital calculations


It is clear from these graphs that a lump sum investment definitely attracts lower transaction costs than periodic flows of the same overall size. However, these increased costs relate entirely to statutory charges which is outside the control of the investor (and fund manager for that matter) – it all comes down to the number of ‘hits’, which relates directly to the frequency and size of investment flows.


So who picks up the tab of these initial transaction costs? This is one of the biggest differences between ETFs on the one hand, and CISs and SMPs on the other hand. In the case of an ETF, the creation or redemption of units is an infrequent event and usually paid for by either the ETF issuer, or the market maker – therefore, both the transaction costs of the underlying basket of shares and any additional regulatory fees and taxes due (e.g. trustee fees, bank charges, etc.) are not payable by the investor. In the case of a CIS or SMP, creation or redemption takes place every time there is an investment flow (often daily), and although these initial costs are typically not included in the TER, the investor ends up paying these costs through an effective higher buying price or lower selling price of the underlying basket of shares.


Secondary market operations refer to the trading that takes place in existing participatory units in the fund. These operations apply mostly to ETFs. In the case where there is netting off of investment inflows and outflows into a CIS or SMP on a particular day, this may also be construed effectively as secondary market operations. In this case there would also be a difference between the buy and sell price of the CIS units, which effectively is the same as a bid-offer spread on the trading in ETF units, but more about this later.

Trading in existing units has no cost impact on the underlying fund value, i.e. the net asset value (NAV) of the fund. The trading costs are only incurred for the listed participatory units (which we will discuss in the next section on acquisition costs), i.e. brokerage and statutory charges, although as mentioned before, no STT is payable in the case of ETFs. Trading in the secondary market takes place around the live intraday NAV level, which includes the accrued distributable amount net of fees.

On-going costs relating to rebalancing activities refer to the trading activity that is required to ensure that the fund remains in line with its reference index. The frequency of this type of activity would mostly be the periodic rebalancing events (e.g. quarterly), at which time both constituents and weights may change, but there could also be ad hoc corporate actions that affect the structure of the index, and thus index-tracking fund, e.g. changes in the issued shares of constituent companies, available free float, capital repayment, rights issue, unbundling, M&A, etc. All of these rebalancing activities are applicable to index-tracking investments, regardless of product type – ETF, CIS or SMP.

The primary driver of costs related to rebalancing activities depends of the extent of churn, or portfolio turnover required. However, even at the base case of a R100m portfolio and 5 bps brokerage, rebalancing costs are still MUCH less than the costs of the initial investment. E.g. at the annual review in December 2013, the following was applicable:

  • All Share index (J203): 1.84% total churn ≡ 0.55 bps rebalancing cost;
  • Top 40 index (J200): 2.69% total churn ≡ 0.55 bps rebalancing cost;
  • SA Listed Property index (J253): 10.73% total churn ≡ 2.01 bps rebalancing cost;
  • equally weighted Top40 index (J2EQ): 15.59% total churn ≡ 2.92 bps rebalancing cost;

and at the bi-annual review in March 2014, the Dividend Plus index (J259), despite a total churn of 54.84%, cost only 10.14 bps to rebalance.

At this point it is important to note that the costs detailed here are actual cash flow expenses, but it excludes price impact ‘cost’, which although it would result in a tracking error of zero, could result in actual value destruction in the case of adverse price movements.

Costs relating to the on-going implementation of corporate actions mostly relate to cash receipts from corporate actions such as dividends, distributions and interest payments paid by the underlying securities. These would have no impact on tracking error relative to the price-only/capital index, but there would be an impact on the performance relative to total return index, also referred to as cash drag. The impact would be that the fund would underperform the index during rising/bull markets, and outperform during falling/bear markets.

Cash receipts from corporate actions are used to pay allowable expenses in the fund, e.g. trustee and custodian fees, asset management fees, regulatory and exchange fees, bank charges, rebalancing trading costs (also of complex corporate actions), audit fees etc. The remaining cash represents the net distributable amount payable to investors, usually paid out quarterly. These expenses incurred in the management of the index-tracking fund result in a “reduction in yield” for the ETF, rather than having to sell units to pay for costs, which is the normal practice in the CIS industry.

In addition to these actual expenses incurred, there are other reasons why the performance of the index-tracking fund will not match the performance of the total return index. Some of these relate to timing mismatches, as the index ‘reinvests’ on the ex-dividend date, at the closing price on the ‘last day to trade’ (LDT) and at no cost, whereas the fund can only reinvest once the funds are received, at whatever the ruling price is at that time, and it has costs associated with reinvestment.

The introduction of a 15% dividend withholding tax (DWT) in South Africa has also introduced additional complexities when it comes to comparative performance. This is further aggravated in the case of dual listed and inward foreign listed stocks that declare and pay dividends in a foreign currency, partly due to exchange rate considerations and extended time lapses between ex-dividend and payment date, but also due to delays in refunds under dual taxation treaties. In the case of market cap weighted index-tracking funds, these stocks unfortunately represent a sizeable proportion of the fund.

It is also worth noting that CISs and ETFs are exempt from DWT within the fund (tax only becomes payable by the end investor once a dividend is declared from the CIS or ETF), but this is not the case for SMPs, making the latter less tax efficient than those funds which fall under the Collective Investment Scheme Control Act (CISCA).

Acquisition costs:

The costs of accessing any investment relates to both the costs of buying the investment (or again selling it when no longer wanted), as well as keeping it in safe custody. Different investment platforms or stock broking accounts offer various cost structures, and sometimes these charges are combined into a single platform fee, but regardless of the offering, it is worthwhile to understand the different components to better determine if the particular cost structure offers good value for money for the specific requirements of the individual client.

The acquisition costs of a physically replicated, index tracking fund can be divided into the following four categories:

  1. Direct costs
    1. trading costs, incl. brokerage and statutory charges;
    2. custody charges.
  1. Indirect costs (bid-offer spread)
    1. bid-offer, or buy-sell, spread;
    2. market (price) impact costs.

Let us discuss the implications and impact of these.

The direct trading costs incurred when investing include brokerage and statutory charges as discussed before, but in the case of trading in ETF participatory units, there is no STT payable. As before, brokerage rates are negotiable, whereas the statutory charges payable to the JSE and to Strate, are not. In the case of CIS units and an investment into an SMP, there are no trading costs levied, although there may be an initial fee payable which will be used in part to defray the costs associated with implementing the investment, i.e. the primary market operations discussed in the previous section. Trading costs can be considered an event-based fee, i.e. these costs are only incurred when a transaction takes place.

Custody charges are often included in the annual (or other periodic) fee payable to investment or stock broking platforms, and can be considered an on-going fee, i.e. these costs are incurred on a periodic basis irrespective of transaction activity.

The bid-offer, or buy-sell spread is where the market maker is remunerated. This service offered by the market maker involves taking on the risk of carrying a position – a supply of units – on his book. These positions are usually hedged by the market maker, and he is compensated for this in part by this spread. It is considered an indirect cost for the investor, as it is not noted as a cash flow expense to him; rather, he buys at a slightly more expensive price and sells at a slightly reduced price.

The bid-offer spreads of ETFs listed on the JSE vary between ±7 bps for the BettaBeta Equally Weighted Top40 ETF (BBET40) and ±1.3% for the NewFunds Shari’ah Top40 ETF (NFSH40). Fig. 4 details the range of average spreads as observed over the month of August 2014.

Fig. 4 Average bid-offer spread in basis points (bps) of JSE-listed ETFs as observed during August 2014


Source: Nedbank Capital calculations

Some interesting trends that are worth noting:

  • More liquid ETFs have more variability in spread, but not necessarily narrower, e.g. the Newgold ETF (GLD) has a spread of only ±10 bps on average, but with a standard deviation of 13.4 bps.
  • Many ETFs start the (SA trading) day with a wide double which narrows during the afternoon session. This is particularly evident in international ETFs (db x-trackers) and in the commodity ETFs (gold, platinum and palladium). One of the widest doubles is for the db x-trackers Japan ETF (DBXJP) at 60-65 bps a side, due to limited overlap of trading hours with the SA market.
  • The average bid-offer spreads shown in Fig. 4 are all ‘screen doubles’, i.e. what is offered on the electronic trading systems. Tighter spreads may be available when dealing directly with the market makers.

This brings us to the potential impact of volume, or size on the market, or price impact costs. When trading in individual securities, demand or supply of large quantities can often have a negative impact on the average price at which the overall transaction is concluded, ‘moving the market’ away from the ruling price before the transaction was initiated. However, in the case of ETFs, size often works in favour of the investor, as the market maker will be inclined to trade ‘inside the double’ if there is a larger volume to trade.

Total Expense Ratio (TER):

Despite its name, the “Total Expense Ratio” (TER) – although intended to be reflective of the ‘total’ expenses of an investment product, fails in this intent, as not all expenses – neither direct nor indirect – are included in the TER. As such, investors are advised to scrutinise investment costs carefully, and endeavour to identify additional costs that are not included in the TER before making investment decisions based only on costs.

The different cost components included in the TER are detailed in Table 1.

Table 1. Components of the TER for an index-tracking fund – illustrative example

Expenses Illustrative example Type
Fixed costs (independent of AUM)
Trustee fee R 120 000 p.a. Mostly fixed
Bank charges R 12 000 p.a. Mostly fixed
Audit fee R 120 000 p.a. Mostly fixed
STRATE fee R 24 000 p.a. Mostly fixed
Index calculation fee (e.g. FTSE/JSE) R 60 000 p.a. Mostly fixed
Variable costs (based on AUM)
Asset manager 10.0 bps AUM
Manco/issuer fee <30.0 bps AUM
Index tracking fee (e.g. FTSE/JSE) 3.0 bps AUM
JSE listing fees 1.0 bps AUM

Source: Nedbank Capital calculations

Due to the nature of the fixed costs items, the size of the fund is a significant factor when the costs are expressed in basis points, as is the case with the TER. Using the illustrative example in Table 1, these fixed costs represent 33.6 bps in the case of a R100m fund, whereas this drops to only 3.4 bps in the case of a R1bn fund.

The TER of an investment product often only includes the on-going or annual fees incurred to manage the product. According to the official guidelines, transaction costs – both on initiation and conclusion of the investment, and those incurred in trading activity inside the fund – do not have to be included as part of the TER.

However, when we compare the cost structures of different funds, the expenses attributable to managing the on-going cash flows and corporate actions are often more than the annual fees. This is evident from the relative performance of different funds with the same investment mandate and management style, such as an ETF and a CIS that offer an index-tracking fund over the same index, as detailed in the introduction of this article.

Table 2. Comparative components of the TER of different funds – illustrative example

Expenses Index-tracking ETF Index-tracking CIS Index-tracking SMP Actively managed CIS
Fixed costs p.a. (independent of AUM)
Trustee fee R 120 000 R 120 000 1 R 120 000
Bank charges R 12 000 R 12 000 R 12 000 R 12 000
Audit fee R 120 000 R 120 000  R120 000 R 120 000
STRATE fee R 24 000 R 24 000 R 24 000 R 24 000
Index calculation fee – FTSE/JSE R 60 000 R 60 000 R 60 0002
Variable costs bps (based on AUM)
Asset manager 10.0 10.0 10.0 >>10.0
Manco fee <30.0 <30.0 <30.0 <30.0
Index tracking fee – FTSE/JSE 3.0 1.5 <1.5
JSE listing fees 1.0
Trading costs3 minimal low-medium4 low-medium4 medium-high5
Scrip lending income – used to offset costs high potential high potential income seldom accrues to investor less potential
  1. Although there is no trustee fee payable, an SMP also does not qualify for tax exemptions such as DWT.
  2. Although index fees are payable on all types of index-tracking funds, it is more difficult to regulate this in an SMP.
  3. Trading introduces additional variable costs, e.g. bank charges, STRATE fees, trustee fees, etc.
  4. The extent of the trading costs depends on the size and frequency of cash flows.
  5. The extent of the trading costs depends on the strategy of the fund and the resultant turnover.

When we compare cost structures of different funds, it is easy to identify the real differentials in expenses, as can be seen in Table 2. The biggest contributors to the cost differences are the asset manager fees (typically much higher in the case of an actively managed CIS) and trading costs (minimal in the case of ETFs and highest for actively managed CISs). Finally, there is also the potential to offset costs through scrip lending income with index-tracking products having the highest potential, although in the case of SMPs, this income often does not accrue to the investor.

In conclusion, it is clear that TERs do not tell the full story of costs – this would be the information as offered by the fund providers. There is a lot more to be gleaned from the evidence of the impact of costs, as can be observed by looking at relative performance. Once one understands all the different components of the costs incurred in managing investment funds and acquiring those funds, the investor is in a better position to analyse and identify the fund construct and acquisition channel best suited to his particular investment needs.



  1. Hi there, this is quite an interesting article, surely its utmost important for the investors to know the actual cost which they will need to bear in investing in this so called low cost funds

Comments are closed.