MANY ETF ISSUERS, MOST NOTABLY THE BIG THREE—BlackRock, SSgA and Vanguard—engage in securities lending. At its core, securities lending is an admirable—or at least lucrative—endeavor: Issuers lend securities from their portfolios to institutions looking to short those securities, earning fees in the process that are refunded (in varying amounts) to ETF shareholders. This is a symbiotic relationship: Investors who want to short shares have to borrow them in order to sell them short, while passive ETF managers have positions in these firms vis-à-vis their ETFs that are mostly static outside of marginal changes stemming from rebalances or index changes. In a perfect world, the lending of these portfolio shares that are sitting “idle” helps dampen holding costs, and can even completely offset the management fee if done well.
However, not all ETF issuers rebate the full amount of those collected fees to the funds. When the issuer keeps a portion of that revenue, it can be seen as “free riding” on the backs of the shareholders. After all, the real owners of those shares are ETF investors, not the issuers, so any income generated by these lending activities is money earned by investors. The counterargument to that is securities lending is not a simple endeavor, and the firms that do it best have to employ intelligent, talented people to manage securities-lending programs. That overhead has to be paid for, and as we all know, the major banks and asset managers in this country are not in the business of charity.
How much is fair for issuers to keep is the subject of much debate—and a recent lawsuit against BlackRock. In recent weeks, however, that narrative has changed course to focus on the value-add of securities lending. Unfortunately, much like the coverage of the lawsuit itself, there seems to be a wholesale misunderstanding about tracking error and how it’s measured.
One Internet story at the end of February ran on the website of the Securities Litigation & Consulting Group. It examined the tracking error of ETFs that lend securities and those that don’t. To measure this, they compared the tracking difference of each group of funds in aggregate. Their (incorrect) conclusion was that securities lending did not dampen tracking error, even going so far as to say that ETFs that do not lend shares actually showed better tracking. The flow of that article was a common one—they compared the price return of the indexes with the total return of the ETFs. Thus, any ETF tracking an index that paid significant dividends looked fantastic, and any fund tracking a low-yielding index looked like a goat.
The point here isn’t to lambaste a now-gone blog post, but to highlight the fact that this kind of analysis trips up many sophisticated investors, so it’s worth going through how to do it right.
THE STEP-BY-STEP PROCESS
To fully capture the real tracking difference between an ETF and its underlying index, all comparisons have to be on an apples-to-apples basis. This may sound like a cliché, but in reality it’s a mistake all too commonly made by analysts and pundits attempting to discuss tracking error in a nuanced manner.
At their core, passive ETFs are designed to provide the return of their underlying index, net of fees. When the constituents of an equity or fixed-income fund make distributions, in the form of dividends and coupon payments, respectively, the price of those securities will adjust to reflect those payments. It follows then that an index made up of those securities will also adjust to reflect those distributions. This is where the distinction between total return and price return comes into play. Just as the index itself will adjust to changes in price stemming from distributions, so too will the price or NAV of the ETF.
The versions of the ETF’s NAV and its underlying index that take into account these distributions reflect the total return—changes in price as well as any income from the securities held. Any tracking comparison must start with the correct version of both NAV and the underlying index.
There are hiccups, however, and the first has to do with “fair valued” net asset values. It works like this: Imagine you have an ETF tracking only Japanese stocks. The net asset value of that ETF is flat all day during U.S. trading hours, because the Tokyo stock exchange is closed. That means that if you compare the price of the ETF on the open market at, say, 4 p.m. ET, it will look like it’s trading at either a premium or discount to where Tokyo closed. To circumvent this perceived premium or discount, issuers like Global X or Vanguard publish NAVs that attempt to estimate the value of the ETF portfolio based on futures, ADRs and highly correlated securities.
While this is an imperfect science, the goal of mitigating ETF price distortions is admirable. The problem is that the indexes these ETFs track do not have corresponding fair value prints, so the premium and discount problem is essentially transferred to the tracking error reading of the fair-value NAV versus the index. As such, any robust tracking error analysis will need to either ignore these funds or get access to a nonpublic, nonfair-valued NAV series directly from the issuer.
A similar disconnect between ETF and index can happen just because of currency exposure. Since currency markets never “close,” index providers and issuers have to decide when to strike their exchange rates for non-U.S.-dollar securities—often 4 p.m. GMT or 4 p.m. ET. If the index provider and the issuer pick different times, this will drive a wedge between the underlying index and the NAV of the fund, even if both entities hold 100 percent perfect mirrors of the index.
The remaining considerations in analyzing true tracking difference are less obscure, but just as important. Some funds simply aren’t allowed by legal structure to lend their securities (such as any ETF organized as a unit investment trust). Second, it’s useful to exclude inverse and leveraged funds, as decay and compounding problems will ultimately cause more noise. Finally—and this may seem obvious—active funds are impossible to measure for tracking error because they do not actually track an index.
With all of these factors taken into account, any tracking difference should be explainable by three things: expenses, which are easy to account for; optimization of the portfolio; and securities lending. It’s the latter that we’re concerned with in this analysis.
Here at IndexUniverse, we take a nuanced approach to analyzing ETFs that we believe is unique to our firm. That is perhaps most evident in the way we approach tracking error. To effectively measure the variance and persistence of tracking differences between ETFs and their indexes, we developed a measure of rolling tracking difference that is a key tenet of our ETF Analytics tool.
What we do is measure the tracking difference between a fund and its index (total return, of course) over rolling one-year periods over 24 months. These one-year snapshots give us an intuitive measurement of the difference between the ETF and its index over time. Once we have these rolling one-year statistics, we calculate the range and median for each fund to convey the distribution of these tracking differences.
Figure 1 summarizes all ETFs that have more than two years of history, do not fair-value NAVs and are allowed to lend portfolio shares. We have split them into those that are currently lending portfolio shares and those that are not. Currently, there are just 99 funds that are allowed to loan securities that do not do so, and 432 that do.
Based on Figure 1’s sample, it’s clear that securities lending has added significant value for shareholders. Regardless of how you measure it—simple average, median or asset-weighted average—ETFs that lend securities tend to claw back a decent chunk of their expense ratio thanks to the revenue from securities lending. What’s more, those ETFs that do not lend portfolio shares have tended to lag their indexes by 13 basis points more than their stated expense ratios. In other words, even if you believe that issuers like BlackRock should rebate more of their revenue to shareholders, it’s impossible to argue that simply engaging in securities lending—regardless of the revenue split—does not improve the holding cost of an ETF.
And therein lies the rub. With so much furor over whether issuers are profiting from the assets of shareholders disproportionately, what gets lost in the shuffle is the fact that these programs add major value to shareholders. If you want to condemn issuers for lending portfolios, be my guest, but make sure you have your facts straight and be ready to fight a losing battle.