The following is in response to Joe Tomlinson’s article, Optimizing Asset Location: Is It Worth the Effort?, which appeared last week:
Dear Editor,
I disagree with Mr. Tomlinson's analysis of asset location (AL). His analysis continues the errors of the past, even though they have been identified and mostly accepted. My position is presented in this paper. It suggests a position part way between the two identified by Tomlinson.
Michael Kitces blog (used by Tomlinson to represent one position) has a number of problems:
- It models a situation where the asset-allocation (AA) dollars are initially set along with the placement of assets. After that there is no rebalancing between the asset classes. In real life, there will be some rebalancing, even if not done by a specific set of rules. This rebalancing will change over time the composition of assets within the different account types. The rebalancing will dilute the benefits of AL.
- It uses one set of assumptions for expected rates-of-return and tax rates. Tomlinson redoes the math with different assumptions, but that does not change the problem. Different individuals will have different assumptions, and those assumptions will change over time. When advice depends on underlying assumptions, a publicly available and easily adaptable model should be used.
- It treats $100 in a tax-deferred account as equal to $100 in a taxable or tax-exempt account. Many people will disagree with that, even while they totally reject Reichenstein's AL approach as well. The tax reduction on contributions to a tax-deferred account is the government giving you their money to invest for them. You must repay it, along will all the income it earned, on withdrawal. It is never “your money” and should not be included in AA calculations.
The Reichenstein/Meyer position (used to represent the other position) also has problems:
- It completely ignores the benefits of tax-deferred accounts. It does not measure the benefits realized or attempt to maximize them. Many people would assume that is the point of AL. Yet they throw away those benefits.
-
It models a reality where income taxes paid on investment income in taxable accounts are actually paid from those accounts. It assumes the resulting cash flow for taxes is seen, measured and felt emotionally – e.g., that a tax refund eight months later reduces the stress of equity losses today, and allows the investor to increase their AA to risky equities.
But does anyone actually separate out the portion of their tax return due to investment income? Does the tax payment eight months after the asset sale cause you to feel that your realized gains were not so big after all? Does anyone accrue a tax liability for capital gains that won't be realized for five years? Does anyone track their year-to-date performance on an after-tax basis? I don't.
The model adjusts for a differential in risks between accounts that exists theoretically, and may be understood by investors, but it is not reported or felt, and does not change their risk tolerance.
