Robots Leave Money On The Table schools the robo advisors on ETF selection.

Director of Research
Reviewed by: Elisabeth Kashner
Edited by: Elisabeth Kashner schools the robo advisors on ETF selection.

This is the fifth blog in a multiple-blog series by’s Director of Research Elisabeth Kashner on the new “robo advisory” industry. The first was titled "Which Robo Advisor For My Teen?”;  the second was titled "Ghosts In The Robo Advisor Machine”; the third was titled "Inside Robo Advisor Asset Allocation”; and the fourth was titled "Rebooting Robo Advisors’ ETF Selection”.


In my last blog, I called out the robo advisors—all of them—for choosing the wrong ETFs.

They picked low expense ratio funds, sacrificing coverage or tax efficiency for a few extra basis points. Before I recommend any of these firms as managers of my son’s bar mitzvah money, I’d like to see them improve their ETF selection game.

Today I invite you to play robo chief investment officer for a day. We’ll take all the puzzle pieces and see if we can put them together sensibly. And I do mean all the pieces: opportunity costs, holding costs and trading costs, in that order, which is from highest to lowest impact for a long-term buy-and-hold investor.

Our puzzle picture is a map of the world. We’ll be selecting funds that get us closest to our goal of covering every stock in every country on the map.

To make sure we’re working from the same box, we’ll follow the robo CIO’s requirements to choose three funds, one each covering the U.S., developed-markets ex-U.S., and emerging markets. We’ll have to work with the available pieces; namely, existing broad-based, cap-weighted ETFs.

The U.S. fund is easy. and all the robo advisors agree that VTI is the best option. The non-U.S. funds are more difficult.

We have to pick the two non-U.S. funds as a pair, to avoid gaps or overlaps with South Korea. The non-U.S. puzzle pieces are arrayed in the table below, in order of inclusiveness:

IssuerDeveloped Ex-US OptionsEmerging Market OptionsIncludes CanadaSmall CapsS. Korea Treatment
BlackRock (core)IEFAIEMGNoYesEmerging
BlackRock (legacy)EFAEEMNoNoEmerging


Now, let’s get down to business.


Opportunity Costs: Lost Puzzle Pieces

My kids, even my teen, get disappointed when they can’t finish a puzzle because pieces have gone missing. They expect, after all that work, to see the whole picture. Yet investors who miss out on returns because of portfolio gaps often don’t see the gaps. But the costs of the missing pieces are real. Global investors who missed out on Canada or non-U.S. small-caps left up to 0.50 percent on the table in each of the past five years.

To measure the opportunity cost of excluding Canada and small-caps, I first compared 10 years of MSCI’s Canada index’s historical returns with those of MSCI’s EAFE, which includes all developed nations except Canada and the U.S. Then I compared the returns of MSCI’s developed ex-U.S. small-caps and MSCI’s emerging market small-caps with their MSCI Standard counterparts, which cover large- and midcaps.

See the opportunity costs table below for the results:

Canada Vs. EAFE Returns5 Years
10 Years
MSCI EAFE Standard10.25%7.40%
MSCI Canada Standard9.26%10.90%
Canada opportunity cost (market weight)-0.09%0.31%
EAFE Small cap vs EAFE Standard returns5 years10 years
MSCI EAFE Standard10.25%7.40%
MSCI EAFE Small Cap13.90% 
EAFE Small cap opportunity cost (market weight)0.51% 
Emerging Markets small caps vs Standard5 years10 years
MSCI EM Standard7.82%12.60%
MSCI EM Small cap9.49% 
EM Small cap opportunity cost (market weight)0.23% 

Data: as of July 30, 2014

Including Canada at market weights in a developed-market portfolio would have added 0.31 percent to MSCI EAFE’s returns each year over the past decade.

Adding small-caps in proportion to their market weight would have boosted MSCI EAFE Index returns by 0.51 percent/year and MSCI Emerging Markets Index returns by 0.23 percent/year over the past five years.

These performance differentials will surely be different in the future. Indeed, Canada would have been a drag on EAFE performance for the most recent five years. The small-cap premium could fade out or reverse, or even expand. The takeaway here is not that Canada is the investment of the century, but that there is opportunity cost to excluding these exposures.

Clearly, we want to include Canada and the small-caps in our global portfolio. Let’s look again at our puzzle pieces, and rank our choices according to breadth of coverage.

All else equal, our ranked preference would be:



  1. The SSgA pair—The SPDR S&P World ex-US ETF (GWL | B-96) and the SPDR S&P Emerging Markets ETF (GMM | C-87)—because they offer both Canada and small-caps
  2. The BlackRock ‘Core’ funds—the iShares Core MSCI EAFE ETF (IEFA | A-93) and the iShares Core MSCI Emerging Markets ETF (IEMG | B-99)—because of the small-caps, whose opportunity cost is higher than Canada’s
  3. The Schwab funds—the Schwab International Equity ETF (SCHF | A-95) and the Schwab Emerging Markets Equity ETF (SCHE | B-88)—which includes Canada but excludes small-caps
  4. The Vanguard funds or the legacy iShares funds—the iShares MSCI EAFE ETF (EFA | A-91) and the Vanguard FTSE Developed Markets ETF (VEA | A-91); and the iShares MSCI Emerging Markets ETF (EEM | B-98) and the Vanguard FTSE Emerging Markets ETF (VWO | C-90)—which offer neither Canada nor small-caps

Explicit Costs: Expenses, Tracking And Spreads

We don’t want to overpay for any puzzle piece, so we’ll have to balance the opportunity costs with fund expenses.

I put together a simple holding-cost estimate, which combines’s median tracking difference and trading spreads. You can see them in the table below. I also included’s more robust “Efficiency” and “Tradability” scores, which measure dozens of cost and risk factors:

IssuerDeveloped Ex-US OptionsEmerging Market OptionsAnnual Holding Cost EstimateCombined Efficiency ScoreCombined Tradability Score
BlackRock (core)IEFAIEMG0.13%9486
BlackRock (legacy)EFAEEM-0.25%9189

Data: as of July 30, 2014

The Annual Holding Cost Estimate equals’s Median Tracking Difference plus 10 percent of’s median trading spread. I was comfortable with taking only 10 percent of the spread because Betterment’s Behavioral Economist, Dan Egan, models 10 percent annual portfolio turnover. Betterment’s emphasis on tax-loss harvesting suggests that 10 percent is most likely an aggressive estimate. In any event, spreads are minuscule for all these funds except the SSgA pair.

Median tracking difference measures the differential in one-year returns between the fund net asset value (NAV) and its underlying index. It’s the “kitchen sink” metric, combining fund expenses like the management fees and trading costs with optimization error and securities-lending revenues.

The Annual Holding Cost Estimates line up well with’s Efficiency and Tradability scores. The Vanguard funds lose a few Efficiency points for failing to disclose their portfolios daily.

Now, back to our ranked choice list. It’s time to put our puzzle together.


We can’t justify the cost of holding our No. 1 choice, the SSgA suite, even though it’s the only one of the bunch to offer both small-cap and Canadian exposure. SSgA’s pair’s annual holding cost of 0.97 percent is higher than Canada’s combined maximum 10-year historic opportunity cost of 0.31 percent.

Worse, the SPDR S&P World ex-US fund (GWL | B-96), the one with Canadian stocks, is the more costly of the two, with an annual median tracking difference of -1.17 percent. We’d be better off avoiding the SSgA funds and adding iShares MSCI Canada ETF (EWC | A-92) to our global equity suite.

Our second choice, the BlackRock “Core” suite, works much better.

From its inception through July 30, 2014, IEFA actually outperformed its underlying index by a median 0.18 percent/year, no doubt because of fortunate optimization and aggressive securities lending. But even if IEFA were to revert to expectations and trail its index by the amount of its expense ratio, we would be willing to pay the minuscule cost differential between the BlackRock core suite and the Schwab or Vanguard funds, in order to access the small-caps.

The BlackRock Core suite, IEFA and IEMG, are the clear choice. We get access to the full market spectrum across the globe, except for Canada. We also have excellent cost control.


So why did five out of the six robo advisors choose the Vanguard suite, the one without exposure to Canada or non-U.S. small-caps? When I asked the robo CEOs and CIOs, “How do you select ETFs?” they all began their answer with “the expense ratio.” Many added other factors, and Betterment even made sure I spoke directly to the analysts who designed their TACO (total annualized cost of ownership) model.

It’s telling that none of them had an actual metric for portfolio breadth, or the opportunity cost that arises from portfolio gaps.

Let’s look at what they measure, and at’s corresponding data:

 WealthfrontBettermentFuture AdvisorCovestorWiseBanyanInvessence
Expense Ratio
Daily Volumes  
Portfolio Breadth 
Tracking Error  
Portfolio Turnover    
Assets Under Management   
Client-Friendly Securities Lending   
Market Impact    
Commissions on Custodian's Platform     

Data: Robo Investment Firms, summer 2014



Developed Ex-US FundsExpense RatioMedian Daily Dollar Volume ($,M)Tracking ErrorMedian Tracking DifferenceAUM ($,B)Securities Lending RebateSpreadsUnderlying Volume/Unit (Market Impact Proxy)
Emerging Markets Funds        

Data: as of July 30, 2014

VEA’s and VWO’s low expense carried the day, at all five shops. The secondary criteria—mostly regarding liquidity—must have knocked out the actual low-headline-cost leader Schwab funds.

A second possibility is that the robo CIOs wrote off funds with comparatively low daily volumes.

That’s a pity, because ETF daily volumes don’t matter that much to market makers, who earn their keep by assisting investors—even robo advisors—to make large ETF trades. If robo advisors are to grow, they’ll have to get comfortable working with these liquidity providers, who often assess portfolio liquidity rather than share volume.

When I pointed this out to Burt Malkiel, Wealthfront’s CIO, he quickly conceded the point. Indeed, all the robo masters I talked to were eager to work with to refine their fund selection process, to measure holding costs more effectively and to understand the opportunity costs of their exposures more thoroughly.

Practical Advice

It’s early innings for the robo-advisor field. Even Wealthfront, the current asset leader, crossed the $1 billion in assets under management line only a few months ago, in June. There’s a competitive advantage to whoever can demonstrate the smartest fund selection process.

Those who offer tax-loss harvesting, or who operate on multiple platforms, have a path for transitioning existing clients, and can use data about exposures and opportunity cost to rerank their primary, secondary and even tertiary fund choices.

But even the firms that don’t offer tax-loss harvesting can revisit their fund selection process. Smart fund selection is a win for clients; explaining this well is a marketing advantage.

May the best robot win.

At the time of this writing, the author was stuck in a low-basis position in EEM. Contact Elisabeth Kashner, CFA, at [email protected].


Elisabeth Kashner is FactSet's director of ETF research. She is responsible for the methodology powering FactSet's Analytics system, providing leadership in data quality, investment analysis and ETF classification. Kashner also serves as co-head of the San Francisco chapter of Women in ETFs.