Introduction

ETF selection is the most underappreciated challenge of being an investor in the rapidly developing world of exchange-traded funds (ETFs). Investors in Europe have over 2,500 ETFs to choose from, according to data from ETFbook, which currently house over $1.8trn assets under management (AUM).

Not only are there a whole host of strategies to wade through but the selection criteria vary in importance depending on the area of the market an investor wants exposure to.

For example, liquidity will not be the most important factor when selecting an S&P 500 ETF, however, it could be if an investor was eyeing the micro-caps space or a niche thematic ETF.

Index

When building a multi-asset portfolio, everything starts with the exposure an investor wants and subsequently, the underlying index.

It is crucial to take a look under the hood to determine the sector and country breakdown and how this will interact with the rest of the portfolio. 

If multiple ETFs are tracking the same index, then tracking difference – the difference between an ETF’s performance and its underlying index – becomes an important factor.

While the majority of S&P 500 ETFs will have the same tracking difference, this can start to vary in more complex markets where other factors such as a portfolio manager’s skill come into play. 

Replication methodology

An ETF’s performance can also be driven by several other factors. A fixed income ETF that tracks an index with thousands of securities will employ a sampling methodology where the ETF only invests in a proportion of the bonds to limit transaction costs.

However, this can lead to tracking difference depending on the securities a portfolio manager selects when they try to mirror the index.

The Bloomberg Global Aggregate Bond index, for example, is made up of over 25,000 securities, however, the number of bonds each ETF issuer buys to track the index can vary depending on the size of its ETF.

Furthermore, physical vs synthetic replication has been a major talking point in the European ETF market, in particular, over the past decade.

While the majority of ETFs were synthetically replicated in the 2000s, the Global Financial Crisis (GFC) spooked many investors on the issue of counterparty risk and demanded an increase in the number of physically-replicated ETFs. 

There are advantages to both types of replication methodologies and very much depends on the exposure.

For example, a US equity ETF that is synthetically replicated does not have to withholding tax on dividends as the substitute basket is restricted to non-dividend paying stocks while physical ETFs domiciled in Luxembourg pay 30% withholding tax on US equity dividends while Irish-domiciled ETFs pay 15%.

All else being equal, this guarantees the outperformance of synthetic S&P 500 ETFs, for example, which caused even BlackRock – a formerly staunch proponent of physical ETFs – to launch a synthetic S&P 500 ETF last year.

Liquidity

Furthermore, the total cost of ownership is a crucial factor to consider. While the headline total expense ratio (TER) is important in determining this, it does not provide the full picture.

An ETF can have a low TER but if it does not trade often or offers exposure to companies with low average daily volumes, wide spreads will drive total costs higher. 

The bid/ask spread of an ETF will also provide insight into an ETF’s overall liquidity. An ETF’s liquidity stems from two sources: the liquidity of the ETF itself and the liquidity of its underlying shares.

ETFs with higher average daily trading volumes and more AUM tend to trade at tighter spreads than ETFs with less daily trading or lower assets. But even ETFs with limited trading volume can trade at tight spreads if the underlying securities of the ETF are liquid.