Good Analysis of Direct Indexing by Allan Roth

[Northern Flicker]

My synopsis of Allan Roth's very good article on direct indexing follows.

1. Direct indexing has a higher expense ratio than a low cost equity index ETF or mutual fund (.03% vs .4% in his comparison with an S&P500 ETF).

2. The value of tax-loss harvesting opportunities is purported to more than offset the higher ER.

3. Eventually, gains in the stock market should lead the stocks held having embedded gains. The embedded gains would be amplified by past tax loss harvests.

4. The tax loss opportunities will dry up, and the investor will pay the higher ER for many years to come.

5. There may be some investor-specific benefits like reducing the correlated risk of holding the stock of one's employer by not holding it in the direct indexing portfolio.

6. The benefit of tax loss harvests could be diluted if capital gains tax laws are changed so that LTCG's are taxed less favorably.

I'll add my own additional one:

7. The possibility of customization may lead to the investor making some active decisions or stock picking decisions, for better or for worse.

[Northern Flicker]

I have some observations.

Changes in tax laws that lead to LTCG's being taxed less favorably could go either way. Direct indexing will create more embedded gains and reduce TLH opportunities, but direct indexing also may overall enable more tax-efficient withdrawals to deal with a hypothetical less favorable tax rate on LTCG.

During accumulation, new contributions and reinvestment of dividends will re-seed TLH opportunities. But in decumulation, that will no longer be in play. Direct indexing may offer the opportunity for withdrawals that are not just a slice of the portfolio, but this may be unproven.

Direct indexing is a form of indexing by sampling. With sampling, an optimizer is used to find a portfolio other than a full replica that minimizes tracking error. I am not familiar with any of the direct indexing products, but the way I could see them working is just to enhance the optimizer to have a constraint on holding a given stock.

Suppose you wish to TLH Intel stock. Set the portfolio representation with the Intel holdings replaced with cash, and run the optimizer to invest the cash with an additional constraint of not holding Intel. This is just a simple example.

A concern I have is that after years of TLH and amplified embedded gains in assets held, could the portfolio anneal to a sample that once in decumulation cannot be modified without realizing embedded gains? This would mean that all withdrawals were just a slice of the portfolio (just like for a traditional index fund) and now the investor will have an ER drag on decumulation for no benefit. Moreover, changes in the behavior of individual stocks and the makeup of the market may then require realizing gains to track the market properly, but there would not be the techniques used by ETFs to manage the gains.

Companies offering direct indexing may have thought about these issues carefully, and may have simulated portfolios over years of direct indexing. I don't know how much of an issue these observations might be. But holding a low cost, tax-efficient index ETF seems like an easy, safe choice.

[the_wiki]

I think the biggest reason that some advisors/brokers are pushing direct indexing is because you are effectively trapped once you get a few years into it with some solid gains. You'll have a few hundred individual holdings. Sure you can ACATS out at any time, but trying to unwind all those positions without blowing up your taxes and maintaining your asset allocation is going to be a bigger challenge than most would be willing to take on.

[Taylor Larimore]

Allan Roth, Northern Flicker and the_wiki:

Thank you for a good analysis of the latest attempt by the financial industry to tempt us into purchasing costly and complex strategies.

Best wishes.
Taylor

Jack Bogle's Words of Wisdom: "Fund marketers favor the fads that are in the momentary limelight, with the expectation that investors will take the bait."

[Gaston]

I'll add my own additional one:

7. The possibility of customization may lead to the investor making some active decisions or stock picking decisions, for better or for worse.

I’ll add one more.

8. At present, it’s difficult to unwind, leaving the investor with potentially hundreds of individual stocks to manage, or a sizable tax bill if sold to invest the proceeds into a Boglehead-friendly ETF. Maybe one day it will be possible to move “in kind” from a direct indexed portfolio to an ETF with no adverse tax consequences.

[anil686]

I'll add my own additional one:

7. The possibility of customization may lead to the investor making some active decisions or stock picking decisions, for better or for worse.

I’ll add one more.

8. At present, it’s difficult to unwind, leaving the investor with potentially hundreds of individual stocks to manage, or a sizable tax bill if sold to invest the proceeds into a Boglehead-friendly ETF. Maybe one day it will be possible to move “in kind” from a direct indexed portfolio to an ETF with no adverse tax consequences.

This x 1000 IMO…. It was the same (separate from Direct Indexing) with Wealthfront and others before with the continuous TLH portfolios. Those were a mess after a year….

[Gaston]

This x 1000 IMO…. It was the same (separate from Direct Indexing) with Wealthfront and others before with the continuous TLH portfolios. Those were a mess after a year….

Interesting. Did not know that.

[the_wiki]

This x 1000 IMO…. It was the same (separate from Direct Indexing) with Wealthfront and others before with the continuous TLH portfolios. Those were a mess after a year….

I'd have a hard time calling any of them "a mess" when you just get the ETF portfolios. They are still a good portfolio of low cost index funds. Worst case you get like a 10 fund portfolio with most of the funds from Vanguard or iShares. I'm surprised you mentioned Wealthfront as their default portfolio is just VTI, VWO, VEA, BND and VIG.

[kxl19]

I've been considering direct indexing, and reading the academic research on direct indexing, and I think it's a potentially valuable product for certain situations

1 - High tax bracket - increases the post-value of the capital losses
2 - Continuous capital gains from year to year (ie, FAANG-like employee with annual refreshes of options/RSU that have been held to LTCG), so you could use the capital loss offsets
3 - Already have a recurring periodic investment into an index fund - so that there'll always be new opportunities to TLH from direct indexing.

I share similar concerns about what happens in decumulation stage - you won't have new funds coming in, but will be making sales, so maybe there's still opportunities to capitalize on the losses.

Excellent paper by Andrew Lo's team (well known finance professor at MIT) - showing a roughly consistent 1% tax alpha for direct indexing during various market periods. Their sensitivity analysis shows tax alpha increases with greater periodic contributions (my point #3). and also increases with higher tax brackets (point #1).

https://alo.mit.edu/wp-content/uploads/ ... -Alpha.pdf

Even If there's a 40 bps expense ratio, direct indexing could still have positive alpha.

What's not known is if one did a monthly TLH strategy "by hand" of ETFs, how much more alpha is obtained by direct indexing?

[aristotelian]

Tax loss harvesting is nice but it's a marginal benefit. Optimizing tax loss harvesting seems like a marginal optimization of the marginal. The maximum benefit to anyone is $3,000 deduction per year. Anyone who has enough money for direct indexing to make a difference is going to have big enough losses to max out the deduction through tax loss harvesting with ordinary index funds. Just seems like a solution in need of a problem.

[retiredjg]

The 1099 tax form on my little $5,000 direct indexing experiment is 86 pages!

Wow

Direct indexing is obviously not a good choice for someone who values simplicity.

[CletusCaddy]

Tax loss harvesting is nice but it's a marginal benefit. Optimizing tax loss harvesting seems like a marginal optimization of the marginal. The maximum benefit to anyone is $3,000 deduction per year. Anyone who has enough money for direct indexing to make a difference is going to have big enough losses to max out the deduction through tax loss harvesting with ordinary index funds. Just seems like a solution in need of a problem.

There are many other uses for tax losses than just the $3k ordinary income offset. I own a home in the Bay Area. With a very conservative assumption of 3% CAGR, the home will appreciate to a $1.5M capital gain in 20 years when I retire and might want to downsize or move to LCOL. Homestead exemption is $500k, leaving me a $1M taxable capital gain that I could wipe out if I had enough tax losses. Hardly marginal.

The 1099 tax form on my little $5,000 direct indexing experiment is 86 pages!

Wow

Direct indexing is obviously not a good choice for someone who values simplicity.

DIY tax loss harvesting is not simple. At least not for people who get paid twice a month and send their paychecks to their brokerage for immediate investing. It’s been my experience that you actually need to juggle as many as three separate funds:

1. Start out with VTI
2. Two weeks later, your paycheck deposits and VTI has experienced a loss. However you can’t just sell VTI because that would be a wash sale. You also can’t buy more VTI with your new paycheck because that would reset the clock on the wash sale. So you buy ITOT.
3. Two weeks later, you get another paycheck and the market has gone down even more. Now you can sell VTI for a tax loss. But you can’t buy or sell ITOT yet, again, that would reset the clock. So you buy SCHB.

What’s more complicated, monitoring all of the above every two weeks, or plowing in your regular paychecks into the direct indexing account, having the service TLH for you, and clicking a button at tax time for TurboTax to load in the 80 page 1099?

[km91]

DIY tax loss harvesting is not simple. At least not for people who get paid twice a month and send their paychecks to their brokerage for immediate investing. It’s been my experience that you actually need to juggle as many as three separate funds:

1. Start out with VTI
2. Two weeks later, your paycheck deposits and VTI has experienced a loss. However you can’t just sell VTI because that would be a wash sale. You also can’t buy more VTI with your new paycheck because that would reset the clock on the wash sale. So you buy ITOT.
3. Two weeks later, you get another paycheck and the market has gone down even more. Now you can sell VTI for a tax loss. But you can’t buy or sell ITOT yet, again, that would reset the clock. So you buy SCHB.

This is quite extreme though, taking tax losses quarterly or yearly is much simpler to implement and doesn't require constant trading between multiple funds. Choose a pair of funds, like VOO and IVV and on a quarterly basis swap between them to lock in the tax loss

[CletusCaddy]

DIY tax loss harvesting is not simple. At least not for people who get paid twice a month and send their paychecks to their brokerage for immediate investing. It’s been my experience that you actually need to juggle as many as three separate funds:

1. Start out with VTI
2. Two weeks later, your paycheck deposits and VTI has experienced a loss. However you can’t just sell VTI because that would be a wash sale. You also can’t buy more VTI with your new paycheck because that would reset the clock on the wash sale. So you buy ITOT.
3. Two weeks later, you get another paycheck and the market has gone down even more. Now you can sell VTI for a tax loss. But you can’t buy or sell ITOT yet, again, that would reset the clock. So you buy SCHB.

This is quite extreme though, taking tax losses quarterly or yearly is much simpler to implement and doesn't require constant trading between multiple funds. Choose a pair of funds, like VOO and IVV and on a quarterly basis swap between them to lock in the tax loss

How is this quite extreme? Literally everyone I know gets paid twice a month or every two weeks. Assuming you are investing in taxable at all then you should be investing with every paycheck. And if you do so then you are forced into the three fund approach I described above or else you would also find yourself in a wash sale situation.

If you lump sum your taxable investments once per month then you don’t have this problem I agree, but then what are you doing with your other paycheck mid month?

[km91]

How is this quite extreme? Literally everyone I know gets paid twice a month or every two weeks. Assuming you are investing in taxable at all then you should be investing with every paycheck. And if you do so then you are forced into the three fund approach I described above or else you would also find yourself in a wash sale situation.

If you lump sum your taxable investments once per month then you don’t have this problem I agree, but then what are you doing with your other paycheck mid month?

You don't need to TLH every two weeks, you could do quarterly TLH trades and achieve the same benefit. I make contributions every 2 weeks and have never needed to juggle between 3 funds

[CletusCaddy]

How is this quite extreme? Literally everyone I know gets paid twice a month or every two weeks. Assuming you are investing in taxable at all then you should be investing with every paycheck. And if you do so then you are forced into the three fund approach I described above or else you would also find yourself in a wash sale situation.

If you lump sum your taxable investments once per month then you don’t have this problem I agree, but then what are you doing with your other paycheck mid month?

You don't need to TLH every two weeks, you could do quarterly TLH trades and achieve the same benefit. I make contributions every 2 weeks and have never needed to juggle between 3 funds

Ok I see. You wouldn’t “achieve the same benefit” however, you’d miss out on all the TLH opportunities mid quarter if the market dips and then recovers.

[km91]

Ok I see. You wouldn’t “achieve the same benefit” however, you’d miss out on all the TLH opportunities mid quarter if the market dips and then recovers.

Fair, but there's a trade off between operational complexity and TLH that needs to be made somewhere

[CletusCaddy]

Ok I see. You wouldn’t “achieve the same benefit” however, you’d miss out on all the TLH opportunities mid quarter if the market dips and then recovers.

Fair, but there's a trade off between operational complexity and TLH that needs to be made somewhere

But that’s the point with direct indexing. Zero operational complexity for the investor and maximal TLH benefit in exchange for the 0.4% expense ratio.

[Northern Flicker]

The 1099 tax form on my little $5,000 direct indexing experiment is 86 pages!

Wow

Direct indexing is obviously not a good choice for someone who values simplicity.

Not to mention the increase in tax preparation cost.

[nedsaid]

The 1099 tax form on my little $5,000 direct indexing experiment is 86 pages!

Wow

Direct indexing is obviously not a good choice for someone who values simplicity.

I don't know, it just seems to me that you can do tax loss harvesting with a whole lot less turnover. I think the algorithms harvest a lot of small capital losses. I am sure there is a logic to this but I am not impressed with a tax statement with pages and pages and page of transactions. There is a point where it just seems like churning.