Jack Bogle on All-In Fund Costs

A Nickel Here, a Dime There
Jack Bogle has a forthcoming article [no link, as it has not yet been published] for Financial Analysts Journal that looks broadly at mutual fund costs, as opposed to merely the official expense ratio. The paper is the best treatment yet of the subject.

The key table comes at the end of the paper.

For the first time, the six elements of mutual fund costs appear in a single table. They are sorted into two bins: costs that are inextricably attached to a fund, so that they cannot be avoided by any shareholder; and costs that vary by investor. The distinction is important. Investor costs should not be ignored, but they are different than fund costs and should be treated as such.

The three fund costs:

Expense Ratio
Bogle cites an average of 1.12%, taken from the category of large-blend U.S. stock funds. He uses that category, as opposed to all funds, because his reference point for a (nearly) costless investment is  Vanguard Total Stock Market Index (VITSX). Fair enough. His decision to equal-weight expenses is similarly sound. For measuring the typical fund, rather than aggregate investor experience (a subtle but important distinction), equal-weighting is the way to go.

Nevertheless, aggregate investor experience is also worth examining. That story is somewhat happier. As investors favor lower-cost funds, the asset-weighted average expense ratio for actively managed stock funds of all flavors (not just large-blend U.S. funds) is about 90 basis points (that is, 0.90%). Actively managed balanced and bond funds are at 80 and 65 basis points, respectively.

Transaction Costs
Transaction costs are a wild card. Tangible brokerage costs can readily be measured, as they are provided in a fund's Statement of Additional Information, but they are a secondary matter. The bigger issues are the cost of paying the bid-ask spread and the possibility that the fund pushes the security's price because of its transaction. Neither item can be found on a fund's financial statements. (The second of the two is largely a mystery to the fund itself.)

In the absence of data, estimates of transaction costs vary widely. Bogle mentions one academic study that arrives at 60 basis points per year in transaction costs for the average equity fund, a second that reaches the far higher conclusion of 144 basis points, and a third source, Morningstar's Don Phillips, who states the amount as being something north of 30 basis points. Taking a conservative approach, Bogle splits the two lower estimates and plugs in 0.50% for yearly transaction costs.

The task is too perilous for me to criticize. An accounting for transaction costs needs to be done, so something must be put into that row. Fifty basis points seems no worse than many other numbers--and better than many.

I have no idea what a reasonable estimate is for the transaction costs of bond funds, and I doubt that anybody else does, either.

Cash Drag
If stocks gain 6 percentage points per year more than cash, and the typical U.S. actively managed stock fund carries a 5% cash position, then the fund loses 30 basis points per year in performance because of its cash. As Bogle acknowledges, many funds "equitize" some or all of their cash positions by holding stock market futures, which gives them stock exposure without owning the stocks directly. Bogle therefore halves his initial estimate and places 0.15% in the row for annual cash drag.

No argument again. It's a small item, and that seems a fair approach. For bond funds, the cash drag would be smaller yet, perhaps 5 basis points.

​The three investor costs:

Sales Charges
This one is dicey.

Once again, Bogle has little data. He can see how many front-end load A shares were sold but not whether investors paid the full front-end load for those shares, only part, or none at all because the sales charge was waived. Similarly, he can sum the 12b-1 fees levied by mutual funds, but he does not know how much money was collected by financial advisors for asset-based services. As a result, Bogle makes a very rough estimate of 0.50% per year, somewhat in the spirit of splitting the difference between the low range of zero and the high range of 1%.

That works for me. (Better roughly right than precisely wrong will be written on my tombstone.) However, I'm not happy with the notion that in contrast to those 50 basis points, index-fund investors face no sales charges. They do if their index funds are DFA's (yes, I know that DFA now maintains that its funds are active ... whatever), and increasingly they do if their index funds are somebody else's, bought with the assistance of a financial advisor. The line separating actively managed load funds and passively run no-load funds has become blurred.

Sales charges belong on the ledger, and they are indeed a cost borne by the investor (who presumably also receives the benefit of a service in return). The 50-basis-point total for actively managed funds, however, is not 50 basis points more than those paid by index-fund owners.

Tax Inefficiency
Tax efficiency matters only to those who hold funds in taxable accounts and who have reasonably high tax rates. The latter condition holds for most assets held in mutual fund retirement accounts (although, to reprise an earlier point, it's not necessarily true of most investors), but the former does not.

Within this context, Bogle's assumption of a 0.45% tax bite is sound. Using Morningstar data that assumes the highest federal income-tax bracket, Bogle finds that actively managed large-blend funds shed about 75 basis points per year to taxes over the past decade, while Vanguard Total Stock Market Index was at 30 basis points.

This tax gap is likely similar between other stock index funds and their actively managed competitors. It is smaller for bond funds, which realize fewer capital gains.

In fact, low-cost bond index funds compete at a tax disadvantage, as mutual fund income is distributed after fund expenses are paid. All things being equal, the higher a fund's expense ratio, the lower its investors' tax bill. This would not be a valid argument for buying an expensive fund, however!

Investor Behavior
The final investor cost is that of buying high and selling low. Once again, Bogle uses Morningstar data--the calculation of Morningstar Investor Returns, which measures how much money a fund makes (or loses) for shareholders. A fund may be highly profitable on paper and show imposing total returns in its prospectus, but if it held few assets when its net asset value rose and many assets when its NAV fell, then it had poor investor returns.

Over the trailing 15 years, investor returns on actively managed large-blend funds lagged their official total returns by 191 basis points per year. Bogle cuts this estimate to 120 basis points because the size of the gap between investor returns and total returns is unstable, depending as it does on the time period's ending point. That works for me. The figure should be somewhat smaller for bond and balanced funds, but not greatly so.

Once again, this cost is also borne by indexes. Bogle cites Morningstar data that for the trailing 15-year period, Vanguard Total Stock Market Index had no behavior gap (in fact, the fund's investor returns were slightly better than its total returns). That is true--but it is also true that New Millennium investors heavily bought into  Vanguard Growth Index (VIGIX) on the way up and bailed when it went down.

Also, there are many narrowly focused index funds, particularly among exchange-traded funds, that elicit different investor behavior than does a very broad fund such as Vanguard Total Stock Market Index. As Bogle surely would concur, such specialized index funds tend to have rotten investor returns, as they attract hot monies.

In Summary
The sound-bite totals for the costs of actively managed funds are 1.77% for fund costs and 2.15% for investor costs, giving a total of 3.92%.

Fund costs require few adjustments. They could be lowered by 30 basis points by using the entire fund universe and by asset-weighting rather than equal-weighting, but 1.77% is close to the mark by any measure.

The figure lines up nicely with observed performance. Looking at one long-term time period (20 years through December 2012), Bogle finds that actively managed U.S. large-blend funds trailed Vanguard Total Stock Market Index by 180 basis points per year--an almost-perfect match.

I ran the numbers over a different long-term period (15 years through last week) and found a slightly smaller gap of 143 basis points. Let's split the difference and do some rounding, as Bogle is apt to do, and call all-in fund costs 150 basis points.

Investor costs are harder to interpret. They depend highly on the situation. They also do not lend themselves to a simple distinction between actively managed funds and index funds. A buy-and-hold investor purchasing directly from a fund company for a tax-sheltered account will have low investor costs, even if using actively managed funds. In contrast, those who like to trade, who pay for investment advice, and who are using a taxable account will have higher investor costs. That is so even if they own index funds.

The concept of investor costs is very useful, however, as is Bogle's attempt to quantify those amounts. 

John Rekenthaler has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.