Financial Professionals Fall 2024

This is my quarterly agglomeration, intended primarily for my fellow financial professionals. It’s simply a way to share things of possible interest that I have read or thought about this quarter. Enjoy!

First, how much do Americans think they need (source)?

“Americans feel they’d need to earn over $186,000 to feel financially secure or comfortable”

“To feel rich, the average American feels they need to earn $520,000 a year”

Second, I have written about forecasting many times (see here for example) and even quoted Howard Marks. But he just dropped a new (short) memo on this that is well worth your time.

Third, Kitces, et al just put out a nice flowchart that I would call an Inherited IRA Cheat Sheet:

Fourth, AI developments are happening rapidly. This was posted months ago (but after I sent my last missive), but perhaps you haven’t seen the examples from OpenAI of what they call “Advanced Voice Mode.”

Fifth, good points about inequality here.

Sixth, good (and free!) Social Security calculator here.

Seventh, years ago, (when direct indexing first became a “thing”) I was predisposed to like it. I still am. But I have never seen an appropriate analysis of the strategy that didn’t “cook the books” in many ways. Since no one appears to be producing an analysis that is reasonable, I am led to conclude that the strategy mostly doesn’t make sense.

Before I get to the list of issues, let me explain, simply, what is happening. This is merely borrowing money. You get tax savings now and (absent a step-up or charitable donation) you repay it later. For simplicity we assume no change in tax rates over time, though if you think marginal tax rates will increase (which seems to be a common assumption) you should be prejudiced a little more against DI. In traditional tax-loss harvesting, the loan is essentially free (aside from maybe some tracking error, which is as likely to add value as it is to subtract value). In DI, the loan can be expensive, as I will show below. For example, I recently attended an event where they did a very brief overview of their “personalized indexing” product (i.e., direct indexing). The cost is 20 bps, but it could be slightly lower for very large accounts or relationships. I thought the speaker threw out an estimate of maybe 2%, 4%, or 6% of the value of the account being recognized as losses each year – in the early years. But I’m not sure those are accurate figures (both because I’m not sure exactly what he was stating or what it was compared to). I’m going to use them to illustrate the math, just for exposition purposes. As most people know, the portfolio “ossifies” over time since most things go up, so in the out years there isn’t anything to loss harvest. Again, I’m not sure those numbers are the net benefit of direct indexing – if the whole index is down, you can loss-harvest a regular ETF. (You are really long cross-sectional dispersion with these products.) They also don’t buy all 500 of the S&P 500 stocks (they buy three hundred and something – I didn’t write it down) so there is some tracking error, but, again, I have no reason to believe that error would be negative so I ignore it here.

[If anyone has better figures, I would love to have them. I would like to know the additional amount over “regular” loss harvesting of the index fund that could have been done. In other words, in down markets there are a lot more losses harvested in both direct indexing and regular loss harvesting of a fund or ETF. What is the typical marginal advantage of the direct indexing over loss harvesting a normal index fund by year since the original investment (no additional cash flows assumed)?]

I assume the index used is the S&P 500, and since the traditional Vanguard S&P 500 ETF (VOO) is 3 bps, you are paying 17 bps extra for loss harvesting. But you are paying that 17 bps on the entire account value. I’ll also assume that you harvest 6% of losses the first year, 4% the second year, 2% the third year and none thereafter. I have no idea if those figures are reasonable. I’ll further assume that the portfolio is not donated to charity or held for a step-up at death. Further, I’ll assume other gains against which to take the losses and a high-ish LTCG bracket of 30% which is constant through time.

Whatever savings you capture now you repay later when the position is sold. The only other variable is holding period, so with those assumptions, we can solve this as a TVM problem. Assuming the portfolio does not go up over time and the starting value is $1,000, for example, in year one the expenses are $1.70 ($1,000 times the marginal 17bps cost) and the savings are $18 ($1,000 times 6% times 30% tax bracket). If you sold that year, you would pay back the $18 and just be out the expense ratio. That’s silly of course. Here are the cash flows if you sold at the end of year four:

            Year 1: $18 savings less $1.70 of costs equals $16.30
            Year 2: $12 savings less $1.70 of costs equals $10.30
            Year 3: $6 savings less $1.70 of costs equals $4.30
            Year 4: no savings, $1.70 of costs plus another $36.00 of tax recapture for a total of minus $37.70.

In other words, it’s like you borrow $16.30 in year one, $10.30 in year two, $4.30 in year three, then pay off the loan with $37.70 in year four. That is a cost of 8.6% annualized. You would need to earn a rate of return higher than that on the tax savings to make this make sense. Since it is a deferred tax liability, I would suggest that using short-term treasuries or a high yield savings account as the appropriate benchmark. You could also think of this as the cost of borrowing to have money now vs. later. In other words, would you borrow at these rates? Since most people have fixed income in their portfolios they can essentially “borrow” by selling some bonds. (The rate of return is the forgone interest.) You could also use the interest rate on your HELOC as your marginal cost of capital.

I have tried to make the assumptions reasonable, but in favor of the direct indexing strategy. In actual practice, I would expect these returns to be too high. Remember I assumed the portfolio doesn’t go up over time? If it does, then the $1.70 annual cost figures increase and get increasingly off over time. Here is the same table, but I assumed 10% annual increases in the portfolio (and hence the expenses) each year:

TL;DR: If (a big if) those magnitudes of losses harvested in the early years are right (and I’m not at all certain they are). I don’t think direct indexing works given the current expense ratio (and current interest rates), unless you are expecting to get a step-up in basis or donate to charity in the future – then it absolutely would.

Ok, but why am I fiddling with this when so many papers have been done. Because every paper I have seen cooks the books (to varying degrees) in favor of the DI solution. This leads me to believe that with accurate assumptions there is little to no value to the strategy. I can’t prove it, but it’s like the dog that didn’t bark – why is there no good analysis? Here are the issues I have with most of the studies that purport to quantify the value of DI:

1. They book the savings of taxes each year, but do not account for the repayment later. Of course, if you get to count the benefits and ignore the costs it looks attractive!
2. They assume there are STCG against which to use STCL generated. For tax-aware advisors this is generally not true. (And I don’t think they account for the strategy causing some dividends which would have been qualified becoming non-qualified.)
3. They assume without DI you would not harvest at all. But of course, a high-quality advisor would harvest the ETF or fund if there is a loss to be taken at that level. (In other words, the papers use a straw man argument where they compare to an idiotic alternative of never harvesting anything at all.)
4. They assume investing the proceeds in the market. This may seem appropriate, but it isn’t. You are borrowing money. You should look at the cost of borrowing, not muddy the analysis with what you might do with the proceeds. (Further, is it even reasonable to assume that lower tax bills will lead to 100% of that to be invested in U.S. stocks? Much of the time, I assume, it would be spent, but even if it isn’t, how many clients have a 100% S&P 500 allocation in their portfolio? The savings would be invested in the overall asset allocation (e.g., 60/40), not just in the S&P 500. If you look at the cost of borrowing – as I would argue is appropriate – then you can determine if you want to get funds from DI, or whether it might be cheaper to use margin, a HELOC, sell some fixed income, etc.)
5. They assume ongoing additions to the portfolio. This makes no sense! Each contribution should be considered its own lump sum. Why in the world would I decide what to do with the current lump sum by mixing it with all the future (smaller) lump sums I might add to it? I would submit it’s merely because it makes DI look much better. (And some of these assumptions are truly stupid and astounding. One paper assumes contributions to the portfolio equal 1% per month of the current portfolio value. This is … odd? It’s a momentum strategy? I’m sure they applied it to the benchmark portfolio too, but it gives a lot of potential for harvesting since there are constant (large) contributions. Let me quantify how silly the 1% savings rate assumption is. I’ll use the 1995-2018 period (the last of theirs in the paper). I don’t know what they used for a starting value, I’ll use $1,000,000.  This means the savings rate is $10,000/month to begin. That seems plausible. But if I just build a quick spreadsheet with CRSP1-10 returns (S&P 500 would be the same, but I had monthly CRSP handy) and apply that strategy, the assumed savings at the end is $1.7 million – per month!!!!  (That was worth using my lifetime allotment of exclamation points!)  You can knock off a decimal place, but that’s still going from a savings rate of $1,000/month to almost $170,000/month. I am astounded that anyone could cite such ridiculous assumptions in favor of the strategy. But they do.)

Here’s how I would design the research, for clarity I’m going to call the “regular” loss harvesting strategy (where we do it at the overall fund level) the ETF strategy:

1. Use the S&P 500 or the largest 400 stocks, or whatever (market-cap weighted of course). Direct index vs. ETF version is what we compare.
2. Run a direct indexed version with loss harvesting at the position and lot level vs. the ETF version where you harvest also but only at the index lot level, not the individual holdings. (i.e., best practices now vs. best practices with direct indexing, not direct indexing vs. doing nothing at all)
3. Use rolling periods starting each month. Do three time-horizons, 10, 20, and 30 years. For example, you could do 10-year periods that go from 1980 to 2020 (starting each month from 1/1980 to 1/2011) which would be 360 different runs, or 240 periods of 20-year returns, etc. (I just arbitrarily picked 1980, use the earliest period for which you can get data)
4. Make sure you liquidate both positions (direct index and ETF) at the end of the 10 years, 20 years, etc.
5. Use the same tax rate throughout. I would use 30% which would be 23.8% plus 6.2% state just to make it a nice even number. Even better would be to do a FL version and a CA version (or whatever).
6. Assume the tax savings is invested in risk-free treasurys (since the tax savings will have to be repaid at the end of the periods) that would have been available at the time. This eliminates the effect of leverage, but does account appropriately for the time value of money.
7. Ignoring transaction costs is fine to start; they shouldn’t be that high for a U.S. large cap universe.

Report the typical statistics for each of the scenarios (bell curves would be great too, I like pictures):

• Mean alpha (and if it is statistically significant at standard levels)
• Median alpha
• Standard deviation of the alpha
• Skewness of the alpha
• Kurtosis of the alpha

This would tell us if this is a viable strategy and for what holding periods. My guess is that the overall alpha is trivial (or negative), but I’d be happy for someone to prove my guess wrong.

Eighth, this is very related to a point in the previous item, but I wanted to break it up and this is worth its own section. I’m repeating some of the above but expanding it. Everyone shows Direct Indexing performance by assuming investing the proceeds in the market. This may seem appropriate, but it isn’t. You are borrowing money.

It’s like quantifying the value of doing accelerated depreciation by assuming the extra taxes saved up front are invested and showing the difference in ending value (net of tax) when the property is sold. It just muddies the waters by combining unrelated things.

It’s just a tax strategy. I want to know the estimated tax savings by year (so I can compare to costs). Not what those might compound to over some period of time if I invested them in the S&P 500. And the fact that no one shows those figures makes me highly suspicious.

Some other examples:

• You could argue for a HELOC strategy where you borrow from your HELOC, invest in the market and after X years liquidate and pay it back.
• You could take SS at 62, invest the payments up until 70 in the market and then use them to supplement your retirement income.
• You could show the value of doing a cost-seg analysis where the up-front tax savings are invested in the market and then used to pay the deferred liability later.

Any of those strategies, might, on average, show an “alpha” but it’s just leverage. DI is not different. (In the case of “normal” loss harvesting it’s free or very close to free so it’s great – but it’s still leverage if those savings are invested.)

An investor should look at the cost of borrowing, not muddy the analysis with what they might do with the proceeds. (Further, is it even reasonable to assume that lower tax bills will lead to 100% of that to be invested in U.S. stocks? Much of the time I assume it would be spent, but even if it isn’t, how many clients have a 100% S&P 500 allocation in their portfolios? The savings would be invested in the overall asset allocation (e.g., 60/40), not just in the S&P 500. If you look at the cost of borrowing – as I would argue is appropriate – then you can determine if you want to get funds from DI, or whether it might be cheaper to use margin, a HELOC, sell some fixed income, etc.)

Again, I don’t know that anyone in real life has any firm tie between their tax liability and their investment rate in U.S. Large Cap stocks. But even if they did, you shouldn’t combine the two to determine what makes sense. My guess is that increased or decreased taxes are more correlated with discretionary spending. But the magnitude probably matters. Huge tax bills or refunds probably affect portfolio size, smaller ones maybe spending. Probably client specific too.

So why does everyone assume investment of the tax savings in 100% U.S. Large Cap? That probably is never a remotely accurate assumption. Just show the actual or expected losses generated vs. a “generic” loss harvesting strategy (i.e., of the index fund or ETF).

Ninth, are we in another housing bubble? I plotted two housing indexes vs. CPI:

(I’m annoyed that FRED doesn’t let me show the origin point on the y-axis. I hate chart crimes.)

Tenth, I always encourage reading the classic literature; here is an overview.

Eleventh, I saw a nice little vignette that illustrates how active vs. passive works here.

Twelfth, I don’t know how to describe this, but it’s very interesting. (The title is Are We Now Too Impatient to Be Intelligent? but I don’t think that really describes it well.)

Thirteenth, I just came across a paper from 2007 on understanding risks in new asset classes that I hadn’t seen. Good stuff.

Fourteenth, Carl Sagan predicted the future back in 1995:

“Celebration of Ignorance” could be the motto of our time!

Fifteenth, a great quote from Naval Ravikant: “People who live far below their means enjoy a freedom that people busy upgrading their lifestyles can’t fathom.”

Sixteenth, I’m not sure people realize what an epic market we have had lately (though it’s been very concentrated in large domestic growth companies), and how phenomenal the 80s and 90s were. The 80s were less surprising because valuations were extremely low to start. In 2009, they came down, but not to very cheap levels. Ben Carlson lays it out here.

Seventeenth, there is a good literature review on Private Equity here.

Eighteenth, I was at a conference recently and Jackie Wilke of First Trust had a few lines that I liked. I’ve shared before that I like to use the “Tell me about…” in client/prospect meetings (as in “Tell me about your estate planning” or “Tell me about your retirement plans”); Jackie is a fan of “I’m curious about…” and I like that too. Particularly good for a follow-up question: “I’m curious about why you did that” or “I’m curious about how you feel about that now” or whatever. She also pointed out that:

• Heritage is better than inheritance.
• Values are better than valuables.

Finally, my recurring reminders:

J.P. Morgan’s updated Guide to the Markets for this quarter is out and filled with great data as usual.

Morgan Housel continues to publish valuable wisdom.

That’s it for this quarter. There were fewer items than usual, but I hope some of the above was beneficial.

If you are receiving this email directly from me, you are on my list of Financial Professionals who have requested I share things that may be of interest. If you no longer wish to be on this list or have an associate who would like to be on the list, simply let me know.

We have clients nationwide; if you ever have an opportunity to send a potential client our way that would be greatly appreciated. We also have been hired by some of our fellow advisors as consultants to help where we can with their businesses. If you are interested in learning more about that arrangement, please let us know.

We also offer a monthly email newsletter, Financial Foundations, which is intended more for private clients and other non-financial-professionals who are interested. If you would like to be on that list as well, you may edit your preferences here.

Finally, if you have a colleague who would like to subscribe to this list, they may do so from that link as well.

Regards,
David

Disclosure