[This article previously appeared in
Exchange-Traded Funds Report
ETFs are a better mousetrap.
Thatâs the saying, at least. And while proponents of ETFs always
point to their low expenses first when discussing their advantages,
the second calling card is always tax efficiency. ETFs, weâre
told, are vastly more tax-efficient than mutual funds.
The data support this. In 2010, the largest ETF issuer in the
worldâiSharesâpaid out capital gains on exactly five of its 219
ETFs. PowerShares, another big player, went three for 112. Most
mutual fund companies would weep for those records.
That tax efficiency is no accident, proponents say. Instead, itâs
a core feature of the product, driven by the key underlying
features of how ETFs work.
ETF Tax Efficiency Explained
To understand the pro-ETF argument, think about what happens when
an investor sells out of two products:a mutual fund and an ETF.
When you sell shares in a traditional mutual fund, that fund
usually pays you in cash. If the fund doesnât have enough cash on
hand, it may have to go into the market and sell some of its
holdings to raise cash to pay you. If the securities it sells have
appreciated in price since the time that the fund bought them, that
sale will generate capital gains, which may have to be paid out in
a taxable distribution at the end of the year.
When you sell shares of an ETF, however, you sell them on the open
market to another investor. The ETF company doesnât know, and
doesnât care, that youâve sold. There is no reduction in the
assets for the fund, so there is no need for the portfolio manager
to sell shares to raise cash. If a lot of investors sell at once,
that will push the price of the ETF below its fair market value.
And thatâs where the secret sauce comes in.
When ETFs dip below fair market value, a group of institutional
investors called authorized participants (or APs) will âredeemâ
shares from the issuer. That is, they will buy up blocks of the ETF
(typically 50,000 shares) and turn them in to the ETF issuer. In
exchange, the issuer will give them the full, fair value of the
underlying securities. This is a risk-free arbitrage profit for the
AP, who buys the ETF shares at a discount and redeems them at fair
value. But itâs also a good deal for the ETF, because the ETF can
actually pick which share lots it will pass back to the AP. Not
surprisingly, it will tend to pick shares with the most capital
gains. That helps cleanse the fund of potential gains.
Mutual funds can do this too, but to a much lesser extent.
The Vanguard Debate
All this has created a conflict in the ETF industry, because the
third-largest ETF providerâVanguardâdoes not operate like the
rest. While most ETF issuers operate their funds as discrete pools
of assets, Vanguard operates its ETFs as share classes of existing
mutual funds. In other words, a fund like the Vanguard Emerging
Markets ETF (NYSE Arca:VWO) provides access to the same pool of
assets as the Vanguard Emerging Markets Stock Index Fund (NYSE
For most of the past decade, the two sides have railed at each
other about this share class structure. Detractors argue that
Vanguardâs share class structure pollutes the tax efficiency
argument; that is, if masses of investors redeemed money from the
mutual fund share classes, the ETF share class would be on the hook
for the capital gains that ensue.
Vanguard, for its part, has argued that the ETF tax efficiency
argument is overblown, and that the inherent tax efficiency of
index funds overwhelms any ETF-specific advantages. It has even
argued that its share-class structure may be more tax-efficient, as
it allows the firm to combine the best tax advantages of the ETF
structure with a few tax tricks that only work for mutual funds.
The firm put out a paper in June comparing the tax efficiency of
Vanguard, SSgA, and iShares ETFs, hoping to settle the debate in
IndexUniverse decided to compile its own data on the topic and see
how the data shook out. The goal was to answer the question of what
truly contributes to ETF tax efficiency.
To truly compare apples to apples, itâs important to control as
many factors as possible. Investor redemptions arenât the only
way that funds can generate capital gains, after all:Index changes,
corporate actions, or rebalancing schedules can also generate
To place things on an apples-to-apples basis, we began our study by
looking only at market-cap weighted domestic ETFs operating in the
same capitalization tiers and style categories. We examined all
ETFs that fit this description, looking back over the past 10
years, using the IndexUniverse ETF Classification System to divide
ETFs into the appropriate buckets.
Figure 1 shows payouts by fund years. A fund thatâs been in the
market for 10 years contributes â10 fund yearsâ to the
analysis. A new fund would contribute just one fund year. A gain in
a given year counts as one fund year, and a single fund can have
multiple years when it paid gains.
Looking at the results, the first thing you see is that capital
gains payouts are rare. A little more than 3 percent of ETFs
distributed gains in any given year, a paltry number that speaks
well to the ETF industry as a whole.
Growth and value funds were more likely to have payouts than
broad-based ETFs. Thatâs not surprising, as growth and value
funds tend to change components more than broad-based funds over
the course of a year.
In addition, payouts increase in likelihood the farther down the
capitalization spectrum you go. This is intuitive as well, as
thereâs simply more movement across the small/mid boundary than
there is between the mid and large. GE likely wonât get ousted
from the S'P 500 anytime soon.
But what about weighting schemes? The IndexUniverse ETF
Classification System groups the 187 funds in our study by the
following weighting methodologies:Beta, Dividend, Earnings, Equal,
Fundamental, Market Cap, Momentum, Multi-Factor, Price,
Proprietary, Revenue, Tiered and Volatility (See Figure 2).
Of the 13 categories, ETFs using Dividend weighting strategies
had the highest percentage of years with capital-gains
distributionsâ8 percent of the available fund years resulted in a
distribution. Equal weighting did not fare much better, with 5.71
percent of payout years. Multi-Factor and Market Cap came in third
and fourth, with 3.33 percent and 3.10 percent, respectively.
Itâs worth noting that many of the weighting methodologies tested
here are new. Beta- and Volatility-weighted funds, for instance,
didnât actually have valid years in the study, and the number of
years for many other strategies was small.
Moreover, many of the payouts were vanishingly small. The
Dividend-weighted products, for instance, had an average payout of
just 0.22 percent of NAV â¦ hardly something to get worked up
In general, the average distribution for these funds was just 0.77
percentâa tiny number. The sample size is limited, but the
findings suggest that the ETF promise of tax efficiency holds.
Interestingly, in our study, the use of a non-market-cap weighted
methodology did not have a detrimental effect on tax efficiency.
Taken together, just 2.20 percent of non-market-cap weighted funds
paid out gains in a given year, compared with 3.06 percent among
What of Vanguard and the share class assessment?
To evaluate this, in Figure 3, we looked at capital gains
distributions on an issuer-by-issuer basis.
Vanguard indeed looks superb:With 97 tax years in the sample,
Vanguard never paid a capital gain.
Neither, however, did many ETF issuers with traditional ETF share
classes, including BlackRock (332 fund years with zero payouts),
PowerShares (92 fund years, zero payouts), and First Trust (50 fund
years, zero payouts).
Our findings here differ somewhat from the research Vanguard
presented in its June white paper. Thatâs because the data
reflected a small oversight, according to Vanguardâs Joel
Dickson, Ph.D., one of the authors of the paper:âIt turns out
that the data source that we used combined return of capital
distributions with capital gain distributions,â he said. Vanguard
said it plans to update its table to reflect the new analysis in a
future white paper.
More broadly, of the 19 issuers included in our data, only six had
capital gains to distribute over the 10-year period.
High percentages from ALPS and FaithShares are potentially
misleading due to small sample size. The Fidelity payouts are
concentrated in its one fund, the Fidelity Nasdaq Composite
Tracking Stock ETF (NYSE Arca: ONEQ).
The surprising data come from SSgA, with 16 payouts over 110 fund
years. Even here, the payouts are hardly earth-shattering:Seven of
State Streetâs 16 payouts in the 10-year period are concentrated
in a single fundâSLYV, the SPDR S'P 600 Small Cap Value ETF. SLYV
has historically had low trading volume relative to its creation
unit size (50,000 shares). That makes it difficult for the
redemption process to help mitigate capital gains, effectively
limiting the tax advantages of the ETF share class structure for
The other SSgA payouts were concentrated in the 2006â07 tax
years. Since then, the firm has a nearly perfect record in its
SSgA was not available for comment.
Taken together, does this suggest the Vanguard structure is
better or worse than its peers?
Vanguard has an exemplary record of managing its domestic-equity
funds for tax efficiency over the past 10 years. So do iShares,
PowerShares and a number of other providers. The tendency toward
payouts seems to be more idiosyncratic, tied to issuers and trading
volume rather than share class structure.
Looking at the numbers, one has to wonder if capital gains
distributions are quite the bogeyman theyâre made out to be.
âI think that people have been way, way too focused on capital
gain distributions as the measure of so-called tax efficiency,â
said Dickson. âSometimes we do get too focused on the actual
distribution and forget all of the other things that matter in
terms of the after-tax return that an investor receives.â
Dickson says pretax return and tracking error (including the
expense ratio of a fund) should play a far bigger role in
determining tax efficiency than monitoring the occasional cap-gains
payout. We would agree.
Taken together, the data suggest that ETFs are extraordinarily
tax-efficient, regardless of structure. There was no substantial
difference between issuers like iShares and Vanguard, suggesting
that the difference between the share class structure and the
non-share class structure simply has had little impact on the
realized tax efficiency of a given ETF, at least within the
domestic equity market.
Don't forget to check IndexUniverse.com's ETF Data
2011 IndexUniverse LLC
. All Rights Reserved.