Charts with Chartolini: Thematic ETFs are Similar, But Ultimately Different

I like to listen to music anytime I sit down to write.?I throw in my AirPods and block out the noise.?Absolute focus is the goal, like a coder mining some Bitcoin.?And depending on the weather, I will wear a hoodie (no Patagonia Wall Street vest for me).

Because songs that I know all the lyrics to could distract me and disturb my writing process, I throw on my Spotify Indie Rock mix and vibe to tunes that I enjoy, but do not know by heart.?Beck, Black Keys, and Wombats alongside some lesser known bands. Songs from a similar genre, but not all from the same artist.?Similar, but different. All of this makes me think of thematic exchange traded funds (ETFs).

The Thematic ETF Playlist

Thematic ETFs are like a Spotify playlist, a collection of funds (songs) that exist in a similar area of focus (genre) that may strike a different tone from one another – much like how I just heard The Adults are Talking by the Strokes followed by Banquet by Block Party.???

Principally, and contrary to some critics’ opinion, thematic ETFs are not all alike.?There are some similarities, but ultimately the level of differentiation across thematic ETFs is greater than that of traditional sectors, styles, and country exposures.?

Understanding these nuances, as well as some of the drivers of returns, can be helpful in performing due diligence on these funds that seek to offer exposure to innovative firms driving technological change.?

Given that more strategies have come to market over the years, and track records are being built, using historical holdings and return time series can help dissect these nuances.?

The Notes of Dispersion

Dispersion, the difference between the best- and worst-performing strategy, can help illustrate the differences among thematic ETFs – and how they compare to more traditional exposures such as GICS sectors, industries, broad styles, and major countries.

The universe of thematic ETFs is based on our classification system and consists of 172 ETFs broken out across 12 distinct thematic subs-sectors, such as Intelligent Infrastructure and Future Security.?Now, some of the ETFs in our universe do not have a three-year track record, but 50% of them do – leading to a diverse universe that expands over time (as more strategies have been launched more returns are populated and included in this analysis).??

A back-of-the-envelope measure of the spread between the best- and worst-performing thematic ETF in 2021 (78%) can illustrate the disparate nature of returns, and thus the differentiation among thematic ETF investment objectives.?Yet, a more robust analysis can strengthen this case.?For instance, the median level of weekly return dispersion, as shown below, for thematic ETFs over the last three years is 12.89%.?For traditional segments, it is far less.?In fact, only when we expand to the very granular GICS level three industries (68 segments) does the dispersion level come close to that of thematic ETFs.?

Given that there are 172 thematic ETFs but only 11 GICS sectors, 9 styles, and 10 major countries, the dispersion should be greater.?Empirically, the more segments analyzed, the higher the probability of wider dispersion.?

As a result, to further refine the dispersion analysis, the dispersion within each of the 12 thematic sub-sectors was calculated.?An average of those 12 figures was created and then compared to the traditional markets.?As shown below, the average sub-sector dispersion is still greater than that of sectors, styles, and countries – illustrating how thematic ETFs don’t all follow the same sheet music, and the difference between the best and worst fund can be quite large.?

Similar Chorus, Different Compositions

Another common criticism is that thematics are just tech exposures with bells and whistles, akin to overlaying autotune onto a well-known chorus.?But that claim is not entirely accurate (the tech claim, not autotune).?The median tech exposure within the 172 funds observed is 24%.?The average, which would account for a few outliers, is still only 28%.1?

For reference, the market cap-weighted Russell 3000 Index has 27% allocated to tech stocks.?So, thematics, on average, are really no more exposed to tech stocks than the general market is. And in fact, 35% have less than 20% allocated to tech stocks and another 20% of the funds have no tech exposure at all.?

A market cap segmentation analysis can also add credence to the notion of similar but different.?While this category of funds skews toward smaller companies, roughly 30% of the funds have more than 80% allocated to large caps.2

Overall, there is some uniformity from a sector and market cap perspective, but the composition of these portfolios can be noticeably different.

Styles the Same

While return dispersion is wide and underlying holdings are differentiated, style (factor) exposures are similar. After all, Beck and the Black Keys do fall into the same style even if the rhythm on Where It’s At and I Got Mine are not similar.??

For starters, thematic ETFs are, on average, a higher risk exposure.?The 80 thematic ETFs with a three-year track record, on average, have a high beta (1.08) to the Russell 3000 Index,3 reflecting higher market risk.?Additionally, the same group of thematic ETFs has a higher standard deviation of returns over the past three years relative to the broader market (27% versus 18.3%, respectively)4 – reflecting higher systematic risk.?In fact, 85% of the funds have a registered a higher standard deviation of returns than the market.

Yet, there are more similarities among this group than this.?Particularly, when performing a factor-based analysis the 80 thematic funds’ returns.?To measure thematic ETFs’ sensitivity to traditional factors, the five factor Fama-French model, plus momentum, was used.?A multivariate regression was run on the daily returns of the same 80 thematic ETFs to the daily Fama-French factor returns.

While a three-year return period does not cover a full market cycle, using daily returns does produce over 700 observations per fund, creating some robustness in this analysis.?And as shown below in the box and whisper chart, there is some consistency among factor betas.?Note the median X mark, the outliers, and number of funds in a specific quartile (the boxes).

Thematic funds’ returns have a high positive sensitivity to small cap stocks (Small Minus Big, or SMB), a negative exposure to value stocks (High Minus Low, or HML), and typically they do not hold quality stocks as evidenced by the negative beta exposure to both Robust Minus Weak (RMB) and Conservative Minus Aggressive (CMA) factors.?Momentum is neutral, which may surprise some. From a significance perspective, SMB, HML, and CMA are the factor betas where more than two-thirds of the funds had a p-value below 5%.?

Grouping the same 80 funds into one large portfolio where each ETF is equally weighted every quarter shows similar results when running a return attribution over the last three years.?While this equally weighted theoretical portfolio outperformed the Russell 3000 Index over this time period, 50% of the outperformance was driven by style effects based on the Bloomberg US Equity Fundamental Model.5?And this is consistent when viewing the returns on a calendar year basis, as in both 2020 and 2021 more than 60% of the return difference can be attributed to specific style tilts.6?

Overall, style tilts matter for thematics and investors should know that – much like how my wife knows the music I like to do work to is not one should want to hear while she does her work (a reason why I wear headphones at the home office).

Look up the Lyrics

When I really want to “know” a song to better understand its meaning or just to sing-along to, I look up the lyrics. Investors need to do the same with thematic ETFs. Dig into the methodology, analyze the holdings, scrutinize the returns, and run scenario analysis.?

Overall, sizable differences exist among thematic ETFs, even though there may be some similarities.?And understanding what is driving the returns is incredibly important.?

Factor sensitivity analysis as well as performance attribution can be extremely helpful, particularly to see if the alpha generated is statistically significant and not just as a result of implicitly earning a well-documented premia.

1. Bloomberg Finance L.P. as of 09/10/2021, based on SPDR Americas Research Calculations?
2. Bloomberg Finance L.P. as of 09/10/2021, based on SPDR Americas Research Calculations?
3. Bloomberg Finance L.P. as of 09/10/2021, based on SPDR Americas Research Calculations?
4. Bloomberg Finance L.P. as of 09/10/2021, based on SPDR Americas Research Calculations?
5. Bloomberg Finance L.P. as of 09/10/2021, based on a time period from 12/31/2018 to 08/31/2021?
6. Bloomberg Finance L.P. as of 09/10/2021, based on a time period from 12/31/2018 to 08/31/2021

?