Exchange-traded funds (ETFs) have much to recommend them. They are well regulated and transparent instruments that are easy to trade. They offer exposure to diverse assets that would otherwise be out of reach to everyday investors. ETFs are mutual funds that are traded on a public exchange. They can be bought and sold, long or short, on an exchange throughout the trading day. The funds usually track a benchmark equity index, such as the Hang Seng Index, but the instrument can track a range of assets such as currencies, commodities and bonds. Having decided on an asset or index, an investor has to figure out which among a range of ETFs offers the best value. Most people will look first at the historical returns of a fund. But past performance is no indicator of future returns. Then they will look at the management fee but, as we'll see, this figure does not tell the whole story. The obvious starting place is for an investor to decide which market they want to buy in. 'The primary question should be: which index do I want to track and which provider is going to offer me the most expertise and liquidity?' says Nick Good, head of iShares, Asia-Pacific, an ETF provider. For most Hong Kong investors, the index question is an easy one: they want A-share ETFs that track a major mainland equity benchmark. As Hongkongers cannot directly own A shares, ETFs act as a popular means of accessing the security. More recently 'frontier market' ETFs tracking equities in countries such as Vietnam and Indonesia have become popular. An investor might then look at the management fee levied by the various A-share ETF providers. This is the annual charge that ETF providers levy on investors for managing a fund. Management fees and other administration costs compose the total expense ratio, which is a more comprehensive measure of the cost of an ETF to investors. Hong Kong ETF expense ratios tend to fluctuate in a tight range between 0.5 per cent and 1.4 per cent. This is a low range - certainly compared to the 3 per cent to 5 per cent management fee typically seen on unlisted mutual funds. There is more. Because an ETF is exchange traded, buying it will involve a brokerage charge, usually 25 basis points. Each fund will also have a bid-offer spread. Just like the spread seen on a stock, an ETF will have a lower selling price than a buying price, which acts as a cost to trading ETFs. For example, the CSI Rafi 50 ETF was quoted on Bloomberg at a bid of HK$26.15 and an offer of HK$26.65. The spread implies a fee of 1.89 per cent to trade the fund. This is a one-time cost but not an insignificant one, and this spread varies greatly among the Hong Kong-listed ETFs (see table). As a rule of thumb, the more liquid the securities underlying the fund, the easier it is to replicate and the tighter the spread. An ETF built on more exotic, hard-to-trade securities will usually come at a higher spread. 'If an institution wants to buy US$100 million of an ETF that tracks Indonesian equities, that could move the market. The difficulty in buying and selling the underlying shares will be reflected in the bid-offer spread of the ETF,' says Joseph Ho, Credit Suisse's head of ETFs for Asia-Pacific. The next consideration is the tracking efficiency of the fund. This is simply a measure of how well an ETF follows the performance of the index it is linked to. ETF providers subtract their total expenses from the fund returns. So in a perfect world the fund will track the index minus the total expense ratio. All well and good. But mixed into that picture is tracking error, which is a measure of the ability of the fund manager to faithfully replicate the index day in and day out. Tracking error always enters into the picture because on any given day it will be challenging for a fund manager to own all the securities that exactly match the index. Shares have to be bought and sold in ways that don't overwhelm the market. Dividend income may come in that cannot be easily reinvested, etc - tracking an index is difficult. A Morgan Stanley report on ETFs in February 2010 found tracking error was routinely larger than the fees and expenses of a fund (or expense ratio). For example, the report found that, in 2009, the SPDR S&P Emerging Asia Pacific ETF underperformed the index by 7.46 per cent, or 12.4 times the fund's expense ratio of 60 basis points. Tracking error can act as a cost to investors. Moreover, a big tracking error shows a fund manager to be doing a bad job. Indeed, to the extent that fees and expenses create a slight underperformance of the fund to the index, and this differential in turn is reflected in the fund's tracking error, tracking error is a kind of meta indicator for an ETF's performance. 'The management fees are part of the total expense ratio, which is itself part of the tracking error. This means the tracking error is the most comprehensive measure of the cost of a fund,' Marco Montanari, head of Asia Deutsche Bank's db X-trackers ETFs, says. Investors may not pay attention to tracking error because, frankly, it is a complicated point. For example, tracking error could easily see a fund outperform an index, in which case investors will not see any problem. As noted, the tracking differential also encompasses the fees of the fund. It may not be obvious the extent to which tracking error represents legitimate fees or errors committed by the fund manager. There is another complication: most ETFs traded in Hong Kong are synthetic. This means the funds do not physically hold the underlying shares, usually because they are restricted from doing so for regulatory reasons. An ETF provider usually handles this by signing a swap contract with an institution that can own A shares. The firm will commit to replicating the index tracked by the fund, and paying out cash returns based on that performance. The swap deal means the job of tracking the index is handed to another party. That in turn means the tracking error can be cut to almost nil. 'With synthetic ETFs, all the noise of tracking error is contained by the counterparty, which means tracking error before costs is effectively eliminated,' Montanari says. However, swap costs can be expensive. The more fees that a fund manager has to pay the other party, the higher the total expense ratio of the ETF. Where does this leave a prospective ETF investor? Once a person has decided which market they want to buy in, they should compare relevant ETFs on the key criteria of total expense ratio, bid/offer spreads and tracking efficiency. The accompanying table presents all this data. Tracking efficiency is a sensitive measure in that it is an expression of the skill of the fund manager to replicate an index and total expense. The bid/offer spread is a big potential one-off expense that should be monitored. The tighter the spread, the better. ETFs are a simple concept but they involve some surprisingly complex inner workings. They are a difficult product to compare like for like. But with a little investigation on the above parameters, an investor could see themselves saving substantial sums by buying the best ETF. Synthetic ETFs Synthetic exchange-traded funds are ETFs that do not hold the physical securities they are meant to track. Often this is because an ETF provider cannot own the asset for regulatory reasons (for example, mainland rules prohibit offshore investors from owning A shares). Typically, an external party will hold the securities on behalf of the ETF provider, promising to pay out the financial return on the investment. This arrangement involves counterparty risk - the external party may not honour its commitment.Hong Kong requires all synthetic ETFs to be plainly labelled as such, to make sure investors know they are taking on this risk.