December 18, 2009. I’m working on a big case for a Roth conversion and contemplating my usual recommendation: a big vertically segregated Roth conversion. A vertically segregated conversion is isolating specific asset classes into separate Roth accounts. If one or more asset classes declines, you can ‘recharacterize’ those particular Roth accounts back to your traditional IRA as if you never converted in the first place and not pay income tax on those amounts. You even have the potential to reconvert it back to a Roth at a later date (after waiting at least 30 days). The vertically segregated Roth conversion is a great plan if you want to cherry-pick your investments and potentially save taxes on the overall conversion. I’ve written on recharacterization before under the heading of ‘Multiple Mulligans’.
However, while looking at this scenario, another notion hit me about recharacterization. If recharacterizing a segregated Roth account when it declines in value is good, then not recharacterizing a segregated Roth account when it has increased in value is equally good. Hence, our latest plan, the horizontally segregated Roth conversion or the ‘Layered Roth Conversion’. Here’s how it works:
You determine in advance two major things: a) how much you want to remain in a conventional IRA and b) what asset allocation you prefer to use in the combined conventional and Roth IRAs. You then determine what the maximum market gain your asset allocation could return during the allowable recharacterization period. For our purposes, presume you are converting in 2010, so the recharacterization period is from the date of conversion to the due date of your 2010 return, including extensions, or October 15, 2011.
Next step: You set up multiple Roth conversion IRA accounts. One is a ‘base’ conversion Roth with about 75-80% of the converted amount (this is the ‘base’ amount that will be recharacterized under most circumstances). The next group are ‘layers’, of increments of potential market gain. For purposes of illustration suppose I think the maximum gain I can expect on a balanced portfolio for the conversion period from 01/02/10 to 10/15/11 is about 25% (I hope I’m wrong on the low side). I’d set up a base Roth conversion with 80% of my assets and 8 separate Roth conversions of 2.5% each (the “Layered Roths”). So if I had a $1,400,000 IRA that I wanted to stay in conventional form, it would look like this:
Now, here’s where we have some fun. Here are the rules, and note that even though we wanted to keep $1,400,000 in our conventional IRA, we converted the whole thing.
If the market goes down or stays flat: We recharacterize the whole thing, and we are back where we started. No income taxes are due!
If the market rises: For every 2.5 -3% increment the market goes up, we leave one of the Layered Roths in their Roth form. So if the market rises by 2.5%, we recharacterize the base and layers 1-7, leaving layer 8. If the market goes up by 10%, we recharacterize the base and layers 1-4, leaving layers 5-8 in their Roth form.
What we’re doing should be obvious: we’re taking the profits and turning them into Roths. If the market rose 10% during the recharacterization period (I think that’s reasonable), our happy little family of segregated Roth IRAs would grow to $1,540,000. If we recharacterized the base and layers 1-4, we’d pay tax on the $140,000 basis in levels 5-8. We’d still have $1,386,000 in our conventional IRA (because we recharacterized it), and now have $154,000 in our Roths IRAs. Risk? If the market goes down, we recharacterize the whole thing and call it an exercise in paperwork management. Market goes up, we turn the gains into tax-free money with all the tax advantages of a Roth.
Why so darn many Roths? Why don’t you just have one and partially recharacterize it? Simple reason: you can’t. Roth recharacterization is done at the account level, and you can’t partially recharacterize an individual Roth. Coverting to only one Roth account leaves you an ‘all or nothing’ approach. Layering allows you to keep the gains and return the original amount.
You could carry this to the extreme and layer at 1% levels. You can also combine this strategy with the vertical segregated Roth strategy. For example, you could take the part you know you want in a Roth and segregate it into asset class conversion Roths and then take the rest of the IRA and layer it. You’ll drive your custodian crazy, but you save taxes on downswings (with the asset class conversions) and earn tax free gains on the upside (on the layered conversions). This gets complicated, but stay tuned for my next piece on ‘Matrix Segregated Roths’.
How much to layer? Given you have the opportunity to seize gains tax free and recharacterize the rest, I’d go long; Mulligans are free and unlimited.