James Osborne of Bason Asset Management published a piece about tax loss harvesting on 3/17/17.
Most of the robo-advisors claim that their aggressive TLH strategy will add nearly 1% per year (right now on Wealthfront’s website they are claiming a whopping 1.55% per year!) in additional risk-free return.
Okay, time for some math. I pulled a handful of client account that had been around for at least a few years and looked at what amount of losses were harvested. For the random accounts sampled, in 2015 I harvested between 2% and 4% of the account value, and in 2016 between 1.5% and 4%. The variance depended primarily on asset allocation and inception dates. So in each of 2015 and 2016 you could estimate that we were taking around 3% of account value in harvested losses ($3,000 on every $100,000 in the portfolio). If you want to use some of the ridiculous math supported by certain robo advisors, you can get pretty close to that translating to 1.5% in tax alpha. But I think that’s absurd.
What you really get is a $3,000 current year ordinary income tax deduction, which is valuable. Yay! And then you get to bank any losses over and above that amount to offset future gains. Gains that you might take from rebalancing or from freeing up cash in a portfolio for retirement expenses. Again, this is a very good thing. You’re creating tax deferral in a taxable environment. Big win. But we arrived at these figures without constant daily trading of ETFs to pick up $10-$15 losses, risking bid/ask spreads, portfolio drift, trade execution risks and paying brokerage commissions. These figures were the result of reasonable, intra-year harvesting of substantive losses. No 40 page 1099-Bs required. Tax loss harvesting is great, but the presumed benefits of aggressive daily loss harvesting over a more “traditional” strategy are likely highly overstated.
I have an additional comment on this, and I know it is an easy one to criticize. None of our client accounts have taxes automatically withdrawn. To the best of my recollection, once in the last 10 years has a client made a withdrawal from an investment account in order to pay a tax bill, and it was an amount far exceeding the tax bill generated by the investments (the withdrawal was due to a cash flow issue). Additionally, clients are making contributions that are planned in advance, not based on on their tax bills. My point is that, while the taxes paid are certainly real, they are not really having any impact at all on returns within the investment accounts. Yes, this means that taxes impact consumption today, and yes, that is important also. And I know that money is fungible and the client’s overall financial position is what counts, and that this tax adjustment to returns is the most fair way to look at the impact taxes have on the clients. But it’s not an trivial point to make, that if no cash flows to the investment account are impacted, then all this churning in the robo accounts will actually incur a loss in the investment accounts compared to accounts that do not generate all these fees and other trading friction. Even if the total cash flows are net positive to the client. So clients will actually have less money in the investment accounts in the future when they want to use that money, unless they start contributing their tax savings into their investment accounts.