The study included 163 participants who invest in ETFs and smart beta.
ETF Stream: What is the purpose of the study?
Felix Goltz: We get a picture of the users of ETFs. It's a growing market and we can see some of the main trends. More ETFs are emerging in the finer subcategories such as smart beta, but we don't know the drivers for this.
The survey asks institutional users of ETFs for their views, so we can get an idea of the weak and strong points of ETFs. We've been doing the survey for more than 10 years, so we can see the trends over time. We find that ETFs are becoming more widely used. Two out of three respondents have an investment in smart beta which is a strong growth over the last four years. We learn the key criteria for why investors choose ETFs are costs and replication quality. 89 per cent of investors see costs being a main criterion for choosing ETFs and 83 per cent said replication is the main driver.
A key focus for the survey was on smart beta strategies and we had two interesting findings. First being the transparency of the strategies. Respondents have a strong motivation to use smart beta, with 80 per cent being motivated to manage the performance. The other finding was a point of concern with only 20 per cent of investor's portfolios are comprised of smart beta.
What potential is there for smart beta ETFs in the fixed income space?
Looking at the opinion of the respondents, the potential is very pronounced. Respondents show they would be interested in smart beta and factor strategies within fixed income. But there's not enough research on fixed income and factor investing strategies to produce this supply.
We haven't really seen any smart beta ETFs explicitly chase the 'profitability' and 'investment' factors. We've only really seen generic 'quality' ETFs. Why do you think there haven't been any profitability or investment ETFs listed?
In the area of quality factors, they do not employ the standard academic definition of profitability quality based on investments. They follow quality strategies that focus on the earning stability or the leverage (debt to equity). These are the most popular factors and I think this is because the product development by index providers focuses on capturing stock picking strategies using these factors. They do this instead of trying to develop indices that capture the academically grounded factors that have been documented in the topic's literature.
Practitioners often complain that the academic factor literature doesn't take transaction costs and tax seriously enough. What evidence is there that factor premiums survive tax and transaction costs?
Research by product providers doesn't usually include transaction costs. Index providers don't provide any information on what transaction cost levels are in their indices. On the other hand, if you look at the academic literature on factors, there's different studies that focus on the robustness of factor returns to transaction costs. We have researched focusing on the impact of transaction costs on factor returns so there are available methodologies to get transaction cost information. To the question "why don't I as an investor get information from the providers of the level of transaction costs I should expect?" One feedback from the survey is that investors complained there's not enough transparency and there's a big gap between information they require and information that is available because providers only publish the methodology. They provide information on how the index is constructed but don't show the implications of risk including transaction costs. There are however lots of studies by academics that apply transaction costs measures to standard factors and the methodology could be used quite widely. This is from providers not providing information that investors need.
Many are saying that the value factor is dead - and that it has been killed by software, changing accounting standards or whatever else. Do you think it's dead?
All factors are probably seen as dead. Even if they deliver a long-term premium, factors can still have short-term losses or minimal returns over a 5-year period. There's always debates on whether premiums are dead. A previous debate was if the size of premium was dead because it was very weak. Over the past decade the value returns have been very weak. Regarding value premium, research shows from 1926 to 1938 there was no significant value premium, so it was dead back then. There are other debates about size and momentum. These factors provide premia over long time periods and over a 5 to 10-year, period these factors don't provide any significant premium.
ETFs tracking EdHec indexes seem to be having a hard time. In the US, the Global X Scientific Beta US ETF (SCIU), which tracks an EdHec index, has underperformed the Vanguard Total Stock Market ETF over its lifetime. While in France, the EdHec-backed Amundi ETF Global Equity Multi Smart Allocation Scientific Beta ETF (SMRT) has underperformed the Amundi MSCI World ETF. Do you have any idea why this is?
The ETFs mentioned target a multi factor exposure and targets high level of diversification, therefore, you avoid taking on stock specific risks. So that's something that attracts a risk adjusted return. Over short time periods, you can have strong effects unrelated to the actual factors of a strategy which could be factor effects. Diversifying the portfolio different to the index and tilting towards factors can get factor exposures and show differences in market exposure.
Multifactor indices, over short periods, can underperform due to factor effects and biases in market beta. In strong bull markets or dominant sectors, you wouldn't expect losses. Investors decide whether they would accept these deviations in term of exposure to markets and factors. That's why scientific beta controls these risks. We also have indices where these exposures aren't controlled.
What we do with the scientific beta indices is offer investors to select the index which best suits them and likewise with ETFs that control these exposures. These would be a trade an investor would make between the long-term risk adjusted performance and would accept the deviation in market and factor risk. Alternatively, they'd focus on short term control of under performance and then the sector and market risk control could reduce that risk.