2019-02-09

ETF vs Mutual Fund (ITOT vs FZROX)

Scope And Purpose Of This Post

This post will try to help inform people about the "ETF vs mutual fund" decision, and we're going to go through some examples, most notably ITOT (ETF) vs FZROX (mutual fund).

Terminology notes:
  • Percentage points are abbreviated "pp". Percentage points are for describing absolute amounts, not relative amounts (which we use percentages for).  A tax rate increasing from 5% to 15% is an increase of 200% (relative) and is also an increase of 10pp (absolute).
  • basis points (hundredths of a percentage point) are abbreviated ‘bp’.  So 3 bp is the same as 0.03% and 0.0003



End Result For Jacob

Rule of thumb: assuming my choices are competitive, ETF (or certain Vanguard mutual funds) in taxable accounts and mutual funds in tax-advantaged accounts.

For investing in the US equity asset class, I currently purchase ITOT in my taxable account and FZROX in my Roth IRA.  I think ITOT will probably be notably better than FZROX in taxable accounts and is pretty much the same as FZROX in tax-advantaged accounts.   In tax-advantaged accounts, I go with FZROX mostly for the extra ease/convenience and making sure I don't trigger wash sales (Wiki, Fairmark, tax code).



[Side note: To explain the wash sales thing; if I have ITOT in both my taxable account and Roth IRA, and if I buy ITOT in my Roth IRA within ±30 days of selling ITOT at loss in my taxable account (including automated dividend reinvestments), I’ll never be able to use that loss to decrease my taxes.  Automatic dividend reinvestments can trigger wash sales.]

If I were at Vanguard, I'd probably do VTSAX in both my taxable and tax-advantaged accounts (because VTSAX is unlike many mutual funds; it does not have capital gain distributions), and I'd make sure to turn off automatic dividend reinvestment in my tax-advantaged accounts.


Important Concepts At Play

I'd like to focus on a few performance-related aspect and the convenience aspect when thinking about whether to go with ETF or mutual fund, specifically ITOT vs FZROX.


Expense Ratio

Winner: FZROX, but the advantage is small compared to tax drag from capital gain distributions.

FZROX has an annual expense ratio of 0 bps, ITOT has 3 bps, which is to say they’re both basically zero.  One way to think of it is that after 33.5 years, the 3 bps expense ratio will have reduced your holding to 99% of what they would have been without the expense ratio.

If you stick to competitive, low-fee index funds from places like Vanguard, iShares, and Fidelity, expense ratio differences will not be very large, and other aspects will be more important.


Tax Inefficiency From Capital Gain Distributions (CGD)

Winner: ITOT in taxable accounts but no winner in tax-advantaged accounts; this is most important aspect (biggest difference) in this list.

FZROX is a non-Vanguard mutual fund, which means it has “capital gain distributions” (CGD) that incur taxes in taxable accounts; ITOT is an ETF and thus will probably never have tax-inducing CGD.

In a taxable account, CGD taxes are a recurring expense, and can be thought of as another annual expense ratio.  FZROX is a young fund without much history, so does not have much CGD history to look at and might behave differently as it matures.

Let's look at another Fidelity mutual fund to calculate its 2018 CGD tax drag, FXAIX.

The FXAIX Fidelity page lists dates, CGD amounts, and share price at time of CGD. Unfortunately, the breakdown between short term and long term CGD is missing; thankfully the FXAIX Morningstar page lists the following {Date, Share Price, LT CGD, ST CGD} info:
  • 2018-04-06, $90.99, $0.075, $0.012
  • 2018-12-14, $90.38, $0.268, $0.104
  • 2018-12-28, $86.36, $0.000, $0.106
For the following calculation, I'll assume that the short term marginal tax rate is 24%, that the long term marginal tax rate is 15%, and that taxes are paid at time of distribution.

April06 Drag: (0.15*0.075 + 0.24*0.012) / 90.99 = ~0.000155
Dec14 Drag: (0.15*0.268 + 0.24*0.104) / 90.38 = ~0.000721
Dec28 Drag: (0.15*0.000 + 0.24*0.106) / 86.36 = ~0.000295
2018 CGD Drag:  1 - (1 - 0.000155) * (1 - 0.000721) * (1 - 0.000295) = ~0.00117 = ~12bp

To summarize the calculation, FXAIX's 2018 capital gain distribution taxes functioned like an additional 12bp annual expense ratio for my hypothetical person.  This 12bp is far greater than ITOT's 3bp annual expense ratio, or FXAIX's 1.5bps annual expense ratio.

Here's an easier and less precise calculation:
  • Find out the short term and long term CGD totals for the year
  • Assume you paid taxes at year end or the time of the last CGD; choose whichever timing is easier to get the share price for that time
Simple 2018 CGD Drag using time of last CGD:
(0.15 * (0.075 + 0.268) + 0.24 * (0.012 + 0.104 + 0.106)) / 86.36 = ~0.000121 = ~12bp

Tax Inefficiency From Qualified and Nonqualified Dividends

Winner: Too soon to tell for ITOT vs FZROX; even if we had plenty of history, QDI ratios jump around a lot and usually don't differ a huge amount between similar funds.  For example QDI ratio differences between ITOT and VTI make an annual difference of ~1bp in tax drag.

Mutual funds and ETFs pass along dividends.  Qualified dividends will be taxed at your long term capital gains rate.  Nonqualified dividends will be taxed at your ordinary income tax rate.  Thus, you should prefer qualified dividends over nonqualified dividends.  If two funds cover the same index, they will likely have the same dividend yield, but they might have a significantly different qualified-dividend-to-total-dividend ratio.  This ratio is also called the Qualified Dividend Income (QDI) ratio.  "QDI" is usually the key term you want to do searches for.

Resources:
Here's a QDI ratio comparison between some iShares and Vanguard ETFs:

Year Fund QDI Ratio
US Total Stock Market US Small-Cap
ITOT VTI IJR VB
2008 1.0000 0.6768
2009 1.0000 0.7216
2010 1.0000 0.7331
2011 1.0000 0.8143 0.6907
2012 1.0000 1.0000 0.8824 0.7355
2013 1.0000 0.9435 0.8185 0.6336
2014 0.9965 1.0000 0.8221 0.6441
2015 0.9483 1.0000 0.7629 0.7171
2016 0.9042 0.9300 0.7935 0.6679
2017 0.9187 0.9478 0.8319 0.6679
2018 0.9296 0.9401 0.7143 0.6913

The table shows that VTI usually beats ITOT and IJR usually beats VB.  Let's imagine that for a pair of broad market funds (like ITOT and VTI), the QDI ratios usually differ by around 4pp, and let us calculate the resulting tax drag increase for a hypothetical person.  We will assume a 15% qualified dividend tax rate, a 24% nonqualified dividend tax rate, and a 2% dividend yield.

ExtraTaxDrag = QdiDifference * DividendYield *(NonqualDivTaxRate - QualDivTaxRate)
= 0.04 * 0.02 * (0.24 - 0.15) = 0.000072 = 0.72bp

A 0.72bp per year difference is not very big.


Improving Returns Via Securities Lending

Winner: too soon to tell; isn't necessarily consistent over time anyway; probably isn't bigger than 5bp in size.

Sometimes funds will loan out some of the shares to short sellers and collect interest on the loaned assets; for ITOT and FZROX, this securities lending profit completely goes back to the fund (usually), so it is possible for funds to actually beat their benchmark index because the fund experiences UnderlyingAssetReturns+SecuritiesLendingProfits-OperatingExpenses

This Vanguard pdf on securities lending says:
Securities lending can modestly enhance an investment portfolio’s return. In 2014, for example, index mutual funds and exchange-traded funds earned an asset-weighted average of about 3 basis points (0.03%) from securities lending (Rowley, 2016). (Because of their limited trading activity and extensive holdings, index funds are attractive sources of securities loans.) In smaller-capitalization and international stock funds, returns from lending can average 10 basis points or more. Profits also depend on the market environment. Some years are better than others.


TODO: Jacob to look at ITOT's and VTI's historical return improvements via security lending.


Trading Inefficiency And Other Sources Of Tracking Error

Winner: too soon to tell and differences are probably 3bp or less


Funds can not perfectly and costlessly match the index at every point in time.  It is my impression that Fidelity, Vanguard, and iShares are all pretty good at following their indices, but be aware that some funds have very large deviations from their index.  You can explicitly look at histories of "tracking error" but be aware that different places use different definitions for "tracking error".


Bid-Ask Spread

Winner: FZROX, but bid-ask spreads are tiny ~1bp and a one-time cost, making this the least important aspect.

With a mutual fund, you buy and sell at exactly the NAV (net asset value); with an ETF or stock, there are slightly different prices for buying (bid) and selling (ask), such that if you buy and then immediately sell, you'll lose some money.

ITOT's bid-ask spread is (at time of writing) ~1.7bp, which is basically zero.  Also, this is a one-time cost (you can think that you suffer half the bid-ask spread at purchase, then half the bid-ask spread at sale), not a recurring cost like many of the other aspects.

Convenience

Winner: FZROX; how big a deal this is depends on you

FZROX is a mutual fund with no minimum investment, so it’s very easy to say, “here’s $100 dollars, just give me $100 worth of FZROX” and you don’t have to do any of the calculations that you need to do when buying an ETF (see “How To Buy An ETF Using a Buy Limit Order” section of this article) and you aren’t restricted to buying integer amounts of shares.  It's also possible to set up automatic money transfers and purchases of mutual funds, but not ETFs (as far as I know, but I look forward to the day of automatic ETF investments).

I like the convenience of mutual funds, and do use FZROX in my Roth IRA mainly for this reason (remember capital gain distributions are not taxed in Roth IRAs).

Lots of mutual funds DO have investment minimums ($3K, sometimes $10K to get the best expense ratio), which is very inconvenient if you're just getting started.

1 comment:

  1. You can buy fractional shares of etfs through m1 finance now. Similar to mutual funds

    ReplyDelete