The old adage that everything you read on the Internet must be true took a major hit today, as a flawed tracking error study comparing ETFs that lend securities to those that do not made its rounds.
In the past two years, IndexUniverse has gone to great lengths to ensure the data we use to analyze ETFs is both clean and trustworthy. While that may seem like a simple endeavor, in reality it has been a painstaking process that has cost us hundreds of staff hours.
At the heart of our mission, however, was being able to effectively and accurately measure ETF tracking error. In doing so, we found it can be easy to misinterpret tracking differences. One common mistake, which showed up in the blog post from the Securities Litigation ' Consulting Group, was that comparing total return NAV to a price return index would produce misleading results.
Now, this is not an indictment of SLCG, per se. After all, some issuers don't publish total return versions of their indexes. That said, the majority of index providers do provide total return index levels, so any study analyzing tracking error that does not compare total return NAV to the total return version of the underlying index will produce erroneous results.
Furthermore, there are some issuers who fair-value their NAVs for ETFs holding international securities, which will cause the published total return NAVs to deviate (sometimes greatly) from the index. The cause of this is simple to understand:An index simply reflects the prices (adjusted for distributions in the case of total return indexes) of the securities it tracks, and when foreign markets like, say, Hong Kong or Hanoi close, the index will not update during the portion of U.S. market hours that those markets are closed.
Fair-valued NAVs, on the other hand, use a model to estimate the price of the portfolio using things like futures, ADRs and highly correlated securities. The reason issuers do this is to try and dampen intraday premiums and discounts in the ETF market. This disconnect will cause tracking differences to appear larger than they (likely) actually are.
There is also another little-known quirk in indexing and NAV calculation that can be easily overlooked that pertains to international ETFs. There are some index providers that strike their currencies at the local market close, while the issuer of funds tracking those indexes strike their currencies during U.S. market hours. This creates another potential wedge between NAV and the index level that cannot be fully captured in a simple tracking error calculation.
As such, any robust study of tracking error will need to take all of these things into consideration in order to ensure you're comparing apples to apples.
To get a true gauge of the efficacy of securities lending, I went through the process of screening ETFs as described above. First, I eliminated any funds that are not allowed structurally to lend securities (unit investment trusts). Then I eliminated any leveraged and inverse funds to eliminate as much noise as possible from my study. Next I removed any ETFs that either fair-value their NAVs or do not have a total return version of their underlying indexes. Finally, I settled on a sample of funds that had more than two years of history.
After I had my sample of 433 funds, I split them into funds that lend portfolios securities and those that do not. Of those 433 funds, 287 lend securities and 146 do not. The table below shows the results of my study after backing out the expense ratio for each ETF.
|Tracking Error for ETFs|
|Securities Lending||No Securities Lending|
|(Asset) Weighted Average||0.05%||-0.09%|
After accounting for expense ratios, ETFs that lend portfolio securities did a demonstrably better job tracking their index. In fact, ETFs whose issuers actively engage in securities lending tended to claw back 5 bps over a one-year holding period. This figure aligns well with the tracking difference when weighted by assets.
ETFs that do not lend securities tended to lag their index by an additional 9 bps beyond their expense ratio over a typical one-year holding period. When you weight by assets, ETFs that do not lend shares lagged their indexes on average by the same 9 bps. That is a gap of 14 bps, a figure that may seem small but that will add up quickly over time.
Once you get inside the numbers, there are some very interesting takeaways. Some much smaller ETFs, like the iShares MSCI Emerging Markets Financials Sector Index Fund (EMFN), which has just $7 million in AUM, had huge index outperformance stemming either from very profitable securities-lending revenues or heavy index optimization.
EMFN outperformed its index by over 4 percent on a median basis over the past two years after accounting for its expense ratio. These outliers speak to the challenge of trying to sequester the impact of securities lending from other forces. Said another way, making concrete claims about the impact of securities lending can be a fool's errand.
But that will not stop me from trying. A complete analysis of the impact of securities lending should take into account all of the things I listed above—TR index availability, fair-valuing of NAVs and differences in foreign exchange rate strike prices—in order to weed out any noise.
And that noise is likely all the data from the SLCG study shows. Because the study compared total return to the price return, the mirage of superior tracking from ETFs not lending securities was likely a product of portfolio yield as opposed to any substantive "outperformance." As my nonno always told me, don't believe everything you read.
At the time this article was written, the author held no positions in the securities mentioned. Contact Paul Baiocchi at firstname.lastname@example.org .
Permalink | ' Copyright 2013 IndexUniverse LLC. All rights reserved