Putting the ETF Tax Efficiency Debate To Rest

October 25, 2005

Our study of ETF and traditional mutual fund tax efficiency begins with the S&P 500 - foundation of both the first retail mutual fund and the first ETF.

 

I've been working on an article for Financial Advisor magazine about ETF tax efficiency. ETF managers have long promoted the tax efficiency of ETFs. And the assertion that ETFs are more tax-efficient than traditional mutual funds has been largely unchallenged, but also undocumented. Only now have most ETFs had the five years of track record (through both bullish and bearish markets) that are really necessary for us to make any legitimate comparisons.

If you're looking for the greatest amount of data, the best index to look at is the S&P 500. So it was appropriate, I thought, that some recent Morgan Stanley research (Paul Mazzilli's group) chose this index to make the distributions comparison:

I looked at this chart and found it shocking, as might many investors. But comparing SPDRs (or iShares S&P 500 fund, IVV, for that matter) to the field of S&P 500 index funds with expense ratios ranging to over 200 bp (2 percent) is a bit like comparing the Yankees and Red Sox to every other baseball team in the country, from the major leagues down to tee ball. In fact, if we compared, say, Vanguard to the field of S&P 500 funds, we might get similary skewed results. In fact lets do that:

Capital Gains Distributions as a Percentage of NAV*

 

Vanguard 500 (VFINX)

S&P 500 Index Fund Ave.

1995

0.23

4.76

1996

0.30

2.43

1997

0.63

3.72

1998

0.27

2.24

1999

0.40

1.79

2000

0.00

2.20

2001

0.00

0.51

2002

0.00

0.11

2003

0.00

0.14

2004

0.00

0.50

Average

0.18

1.75

*Year-end NAV. (In Vanguard's case at the date of distribution).

You would be correct to point out that the derogatory expense ratio comment in the above Red Sox/Tee Ball comparison should have little to do with capital gains distributions. So why does the average S&P 500 index fund look so different from the Vanguard 500? In a word, mismanagement; essentially, most index funds are overly passive in how they handle cash and shares. And keep in mind that we haven't witnesses heavy redemptions during the above period. In an environment of heavy redemptions, even a well-run traditional mutual fund would suffer.

When traditional funds see significant net fund redemptions from investors, the fund must pay capital gains on the shares that have been sold to pay back the investors. I don't believe the Vanguard 500 has ever had a year in which it has suffered net redemptions, as Vanguard investors tend to be the buy-and-hold type. All the capital gains distributions in the Vanguard funds have been the result of index changes:When companies go out of the S&P 500 (which has around 10 percent turnover a year, normally), those shares must be sold and taxes must be paid on any gains. So why zero gains in recent years for Vanguard? The fund has realized losses from the same mechanism. And it must be said that the very time that traditional funds are most likely to suffer a wave of redemptions is likely also the time they are most insulated from capital gains, with realized losses compensating for gains.

So in short, in an enviroment that was about as bad as it could get (with enormous gains taken by funds in the 1990's go-go years followed by a hard turn in 2000 and 2001), the worst of it for the Vanguard 500 was a 63 bp tax distribution (16 bp of it short-term gains) at the end of 1997. And it's not too hard to do the math on that. In 1997, capital gains rates stood at 20 percent for those in the top tax brackets and 10 percent for those in lower tax brackets. In 2003, those numbers changed to 15 percent and 5 percent, respectively. So in its rawest form, even under the higher tax regime, the 1997 distribution cost wealthier taxable Vanguard 500 investors about 14 bp of returns. And in the current tax environment (see assumptions below), that number would be closer to 9.5 bps.

Most importantly, the proof is in the pudding. And I like vanilla, but with a little cinnamon on top. Vanguard (as well, increasingly, as SPY and IVV) manages its index funds with the cinnamon of good cash flow and index effect management. It all comes down to what you make in the end. And Gary Gastineau has long held that the tax efficiency of ETFs should on average give them about a 50bp annual edge for taxable investors. Lets put this to the test.

And of course, marginal tax, even when it was at 39.4 percent in 1997 (it's currently 35 percent) wasn't really 39.4 percent. The average rate even at the top bracket of 39.4 percent was actually 29.9 percent, while the average for someone in the 31 percent bracket was 19.5 percent.(1) That's a discussion for another day, but it sheds some light on the difficulty of calculating after-tax returns in any sort of uniform manner.

2005 Tax Brackets

I began looking at these numbers just thinking that I would use Morningstar's numbers for pre- and post-tax returns at 10-year annualized levels (available for S&P 500, SPDRs (SPY) and the Vanguard 500); and the 5-year pre- and post-tax annualized returns for many of the other asset classes. But reviewing the data…while Morningstar does show us numbers, some of them suffer from clearly missing or erroneous data that has skewed the post-tax numbers. Not to mention that they show the effect of total distributions (including dividends).

For our purposes, we'll just keep it simple and just look at year-end capital gains distributions as a percentage of fund NAV, and use an assumption of the 28 percent tax bracket for short-term distributions (which we'll asign the 14.3 percent average rate that taxpayers in the 28 percent bracket actually pay, according to the above referenced Center on Budget and Policy Priorities study). And for long-term capital gains, we'll just use the current higher 15 percent rate. In fact, lets just make it 15 percent for all capital gains distributions.

Effect of Capital Gains Distributions on S&P 500 ETF and Mutual Fund Returns

 

S&P 500 Index Returns

SPDRs (SPY) Returns

Vanguard 500 (VFINX) Returns

iShares S&P 500 Fund

SPY After Tax**

Vanguard 500 After-Tax

S&P 500 Index Fund Average Distrib.

After-tas effect on return

1995

37.53

37.23

37.45

 

37.23

37.42

4.76

-0.71

1996

22.95

22.74

22.88

 

22.72

22.83

2.43

-0..36

1997

33.35

33.08

33.19

 

33.08

33.10

3.72

-0.55

1998

28.58

28.35

28.62

 

28.35

28.60

2.24

-0.33

1999

21.04

20.86

21.07

 

20.86

21.01

1.79

-0.26

2000

-9.10

-9.15

-9.06

 

-9.15

-9.06

2.20

-0.33

2001

-11.88

-11.85

-12.02

-11.96

-11.85

-12.02

0.51

-0.08

2002

-22.09

-22.12

-22.15

-22.15

-22.12

-22.15

0.11

-0.02

2003

28.67

28.39

28.50

28.53

28.39

28.50

0.14

-0.02

2004

10.87

10.75

10.74

10.77

10.75

10.74

0.50

-0.08

Average

13.99

13.83

13.92

 

13.83

13.90

1.75

-0.26

*Year-end NAV. (In Vanguard's case at the date of distribution). After tax returns, are looking only at capital gains distribution effect on taxable returns. Dividends are not included in this calculation. Per methodology mentioned in the article, current tax rates are assumed, and a 15% rate is used for all distributions, long and short-term.

Summary

For our first segment of data, covering the S&P 500, the index with the longest history both for ETFs and mutual funds, the results show that in the case of the granddaddy of all ETFs, and the granddaddy of all mutual funds, that the tax issue is overwhelmed by other factors in the index management process. And the end result is that SPDRs, which is basically capital gains distribution-free, gains a couple bps annually on the Vanguard fund, but still trails in performance, despite the Vanguard 500's higher expense ratio (18 bps as opposed to the SPDRs 10 bps (actually 12 bps for all or most of this time period)). We've inserted the date for the iShares S&P 500 fund (IVV) for interest, because it is an open-ended mutual fund (unlike SPDRs, which is a trust) and has a bit more flexiblity in terms of share lending and use of derivatives. We show only total returns for IVV, but they've also only paid one small capital gains distribution in their history.

Also included in the above table is the effect that the average S&P 500 index fund capital gains distribution (and mind you SPY, IVV and VFINX are not average funds in any sense) would have had on return even even under the current tax regime. And the numbers, like the expense ratios of many of these funds, are truly scandalous. And of course the effect of losing that 50 or 70 bps of return in one year less an expense ration likely to be in the same ballpark or higher, compounds over time.

The verdict: In the case of the S&P 500, in any event, ETFs do have a clear but practically negligible tax advantage among the "major league" S&P 500 funds, If you put the average (minor league) S&P 500 fund into an ETF structure) it is unclear that they would enjoy the full tax efficiency benefit of SPY or IVV since most of these funds are small and would have little redeption activity to clear off low-cost lots. But at the same time, the "redemptions" they did have from the investors in individual ETF shares (which would come in the form of selling the shares on the secondary market) would have no effect on capital gains distributions to tee fund.

In down markets, one expects that the opposite would occur, with traditional index mutual funds enjoying an edge over ETFs. The existing data bears this out, in particular with the widespread 2002 iShares distribution, and with the nonesistant Vanguard distributoins through the same time period and very low distributions even among the average funds during the 2000-2002 bear market.

Gary Gastineau notes that investors must above all beware of the "capital gains overhang" factor. The late 1990's had overhang the likes of which might never seen again in the S&P 500...and it really came to not much of anything for Vanguard 500 investors. The Vanguard 500 suffered no net redemptions, and did a good job at managing its cost lots, so the capital gains distribution problem was minimized there. Given a wave of redemptions and significant capital gains overhang, the negative impact could be significant. Traditional mutual funds have their tax efficiency strenth in down markets, when they are able to hoard capital gains losses through the cash redemption process in a way that protects even routine index changes from capital gains exposure. ETF managers handle redemptions in kind, so just as they don't get hit with gains when they move out low-cost lots, they also can't take losses when redemptions come in. And it should also be noted that traditional mutual funds can and do occasionally do in-kind exchanges of underlying equities for fund shares with large investors. But that is the exception. And all factors (and the data) considered, it's pretty clear that ETFs are more tax efficient overall. In the case of well-run S&P 500 funds, though, the numbers are relativlye insignificant, and in fact the Vanguard 500 traditional index fund has overcome the slight tax disadvantage to outperform post-tax anyway.

For an analysis of the rest of the asset classes, you'll need to read the full article when it comes out. And we might expect that more volatile asset classes like small cap may be areas where ETFs might shine. Stay tuned for the article that will appear in the December issue of Financial Advisor magazine…we'll also certainly post the results here on www.indexuniverse.com as well.

One thing that there appears to be little doubt of is that ETFs, at least those currently on the market now - there could always be T-Ball ETFs in the future - have demonstrated remarkable tax efficiency thusfar in their brief history. Of the iShares funds, for example, only the country funds (formerly Webs), which must be rebalanced for regulatory reasons to mainain diversification, have conisistently had significant capital gains distributions. Other than those (20 or so of around 100 total U.S.-trading iShares), the iShares have only paid distributions (relatively small ones) in 2002, the D-Year so far of iShares capital gains distributions.

In the case of the S&P 500 funds, the data shows a clear winner in tax efficiency by ETFs, but the gain only amounts to a couple of basis points annually. In the case of the S&P 500, the data supports the importance of looking at other factors beyond just the tax efficiency issues in selecting a fund. In short, if you're taxable, you should not just be looking at fund structure, but also at tracking error and expense ratios. Other factors that must be examined when deciding whether to choose an ETF include the amount and frequency of the investment, the level of redemptions, and the level of capital gains distribution the funds have had in the past (particularly in rising markets). More data - covering all of the other asset classes is forthcoming.

 

Find your next ETF

Reset All