Asset allocation is tough enough without introducing the complicating effects of taxes into the equation. So, why do it? Because reality demands no less, says Stephen Horan, the recently appointed head of private wealth at the CFA Institute. Ignoring the 800-pound gorilla in the room risks drafting an asset allocation plan that's less than it could be, if not just plain misleading.
Horan is no casual observer of such matters. Taxes and investing are his specialty, as his published paper trail reveals. Among his latest: "Applying After-Tax Asset Allocation," forthcoming in The Journal of Wealth Management.
Adding taxes to the asset allocation process gets you "closer to a better after-tax result," he told Wealth Manager in a recent interview. No one ever said better was easier, of course. Indeed, after-tax asset allocation is a relatively embryonic field. Even if you're inspired to craft asset allocation strategies with taxes in mind, there are no turnkey software programs to lean on. This is still a do-it-yourself niche.
Nonetheless, interest in after-tax asset allocation is rising, says Horan, who's worked in the finance industry and was a finance professor at St. Bonaventure University just before he became head of CFA's private wealth division in January.
The proof that taxes matter to the bottom line comes with the recognition that they ultimately change portfolio allocations, Horan advises. One simple example can be seen in the table on page 70, which is based on Horan's calculations for comparing how asset allocation shifts in pre- and after-tax analyses.
Factoring in taxes is vital for every individual's investment strategy, but it's particularly crucial for wealthy investors, who endure higher marginal tax rates. "The wealthier you are," Horan reminds, "the higher the dollar stakes."
Just how big are the stakes? Our conversation with Horan takes a stab at an answer.
What's the rationale for building an asset allocation plan on an after-tax basis?
Asset allocation is typically about thinking of how to divide assets between stocks and bonds. When it comes to figuring out how much spendable wealth is invested in stocks and bonds, you have to factor in taxes. And that means factoring in the types of accounts in which the securities are held.
For example, if I have $1 million in a 401(k) plan, I won't have $1 million to spend after I take the money out and pay the taxes. The anticipated tax liability affects the spendable wealth, and hence the effective asset allocation.
Is after-tax asset allocation (ATAA) widely used in the investment advisory profession?
No, but its use is growing. It's still not the usual case for an advisor to understand and appreciate that those assets in tax-sheltered accounts ought to be viewed differently than the assets in taxable accounts.
Prof. William Reichenstein of Baylor University has been developing models on asset allocation, and he's been involved in general after-tax issues for a long time. He and I have been sort of working in the same field. Another guy in this area is Jarrod Wilcox in Boston, president of Wilcox Investment Inc. He wrote a monograph recently that was published by the Research Foundation of CFA Institute's private wealth management series. He's been focusing on mean variance optimization, which is how to put together different assets in the most efficient way to minimize risk for a given level of expected return. What Bill and Jarrod have done is come up with a way of putting taxes into the asset allocation framework because taxes affect the efficient assembly of stocks and bonds.
Is ATAA a formalized strategy, or is it practiced on an ad hoc basis?
I'd say that it's ad hoc. It's not yet mainstream finance, which focuses more on institutional issues, and so much of asset allocation is focused on the pre-tax basis. Introducing taxes into the [asset allocation] optimization framework creates a fair amount of complexity. You don't start with that, but now that all the pre-tax stuff is more or less figured out, we can address after-tax asset allocation.
In fact, there's been a growing demand for after-tax asset allocation models. Part of the demand is coming from the financial planning profession. The wealth management business recognizes that the pre-tax framework doesn't apply to their clients, who are heavily taxed--and taxed in different and complex ways [compared to institutional investors].
One example is the time-horizon difference between a pension fund, which can have an infinite time horizon, and a private client. The individual's finite time horizon changes the focus and the analysis. For individuals, you're concerned about end wealth because you have an end period by which you want to achieve a certain result. Monte Carlo simulations [for crafting asset allocation strategies] are prevalent in the private-client framework because of the finite time horizon.
Modern portfolio theory focuses on maximizing return and minimizing risk. How do taxes fit into that paradigm?
It's still a two-dimensional framework of risk and return. The basic modification is changing the standard inputs. Your expected returns are different because taxes affect returns. Taxes also affect volatility. By sharing your returns with the government through taxes, the government actually reduces your risk. If the tax rate is 50 percent, your range of outcomes has shrunk by that proportion. Through taxation, the government shares in some of the investor's risk.
Take the extreme scenario: the government taxes 100 percent of all your investment returns. That would stink, but we know the end result: You'll make zero, meaning that there's no variance of outcome. There's no risk. Of course, there's no return either.
The basic inputs for mean variance optimization are returns, volatility and correlation. So, you're saying that all three are affected by taxes, but otherwise they remain intact for building optimized portfolios.
Yes, but rather than correlation I'd say co-variance. Correlations are probably not affected [by taxes], but the co-variances are. Correlations measure how likely investments move together or apart from each other. Co-variance captures that along with the magnitude of the co-movement. Correlation, in other words, is part of what comprises co-variance.
But it's not just a matter of [taxes] changing the inputs. A bond, for instance, is a different after-tax asset depending on the type of account it's held in. If I'm holding a Treasury security in a Roth IRA and a Treasury security in a taxable account, those are two different after-tax assets.
Because the returns are taxed differently, and so their risk profiles are different. Factoring in the after-tax optimization process exponentially multiplies the number of investment opportunities. As such, the optimization process will analyze if the bond should go into the Roth IRA or the taxable account. That's what we call asset location.
ATAA, as a result, introduces greater complexity into the task of building and analyzing portfolios. Is there a risk that the additional complexity hurts rather than helps when it comes to the end result?
You always run the risk of getting very precise results that are no more valid than the assumptions on which they're predicated. The same certainly is true for the traditional Markowitz optimization model. Many people eschew it on the basis that if we can't trust the inputs, we can't trust the outputs. The after-tax asset allocation framework is subject to the same criticism. However, the tax issues are no less certain than what we think an asset's expected return is, or what the expected volatility or co-variance will be.
For example, my tax rate may be 28 percent. Do I know that for certain? No. Maybe it will end up being 33 percent; maybe it will end up being 25 percent. But I'm pretty certain that there will be taxes, and so there's more certainty about the tax issue than what the returns are going to be.
True, but doesn't introducing additional variables raise the risk that the model becomes unwieldy and therefore more unreliable?
What I'm saying is that the additional variables that you're adding are more predictable than the ones that were there in the first place. You're not resolving any of the existing ambiguity [by introducing taxes into asset allocation], but you're not introducing that much more ambiguity, either.
The second thing to keep in mind is that it's a rare breed that takes the output from these models as gospel and implements them in a rigid, prescriptive way. Rather, this type of analysis provides strong guidance about optimal asset location. If you're going to hold bonds, for instance, hold them in your tax-sheltered accounts because they're tax inefficient. That may seem like an intuitive, obvious result, but not everyone agrees with it.
Are there studies showing that, all things equal, investors obtain superior results with ATAA versus the traditional pre-tax asset allocation process?
There are certainly hard studies on this. It's clear that when I run Markowitz optimization in both pre-tax and after-tax environments I get two different results. To the extent that adding the additional tax variables gets you closer to reality, closer to what produces the best result, you're getting an improvement--a better after-tax result.
Do portfolios informed by pre-tax asset allocation look radically different from those designed on an after-tax basis?
No, but it's substantial enough to take notice. [The difference] means that you've mis-estimated the risk exposure.
Are there software products that integrate the after-tax issues into asset allocation modeling?
That's a good question, and I get it all the time. As far as I know, it's all pre-tax. Part of the problem is that there's a fair amount of inputs that have to go into the model. And every investor's tax situation is different. That's part of what makes after-tax asset allocation so challenging.
James Picerno (email@example.com) is senior writer at Wealth Manager.