There's another "fancy" scheme like this that can be used if any of the accounts decline in value. The idea is that if you convert to Roth, and the account subsequently declines in value, you can reverse the conversion and then convert again to a Roth. Since the account value is lower for the second conversion (which is now the first conversion acknowledged by the IRS), the taxable amount is also lower. This can be done repeatedly until the account stops declining.
There are time windows that must be followed to keep everything legit with the IRS. I have never done this for myself or a client (and hopefully never will), so anyone interested will need to dig into the details on their own.
Caveat 1 There are also rules limiting the frequency of conversions, recharacterizations, and reconversions. You may not make a Roth conversion, "unconvert" it and reconvert the same IRA money in the same year. [Footnote 4] Even if you straddle different calendar years, you must still wait 30 days before reconverting a Roth IRA that you had previously converted and "unconverted." [Footnote 5] See
formatting link
among other sites and cites. Caveat 2 The amount re-characterized will depend, inte alia, on how the Roth IRA's entire contents have performed. For example, if 1) one first does a partial conversion from Trad IRA to a Roth IRA; (2) that partial conversion is a small fraction of the Roth IRA; (3) the original amount in the Roth IRA increases in value while the small partially converted amount decreases; then re-characterizing is not necessarily going to get one very far and may even be a setback.
Right. It's a little stricter and less feasible than your post would lead one to believe IMO.
I'm probably still being dense, but if the intent is to convert at the lowest tax rate, why would you reverse the conversion at all? Or is the presumption that the converter only has $10k of headroom in the current tax bracket?
I actually did this back in 2001 or so during the bear market. In fact, I think I did it two years in a row... The rules have changed since then adding stricter caveats, but I can look up the details for my case if there is interest.
Skip the interim step for a moment. Just consider my last thought above. "what would you rather, having it split $11K IRA/ $13K Roth or the reverse?" $13K in the regular IRA would mean $2K more that would be taxed at conversion time. In my example, the second year only $11K is converted and taxed. Within the constraints of Elle's link details at
formatting link
and the limits of one's patience for details, this can squeeze out a bit of tax savings overall, not just if one is bumping their bracket's top. Again, I've not used this strategy at all, I've always targeted the 'bracket top off' when doing conversions. JOE
Again, just being dense, but I'm still not following you. I probably don't understand your example, so I'll try to reiterate it. At the beginning of year 1, I convert two $10k traditional IRAs to Roth. One gains 10% and one gains 30%. I undo the conversion on the account gaining 10% and reconvert it when the rules allow at something like $11k in year 2. So over two years I'm taxed on $21k of conversions. But if I had not undone the conversion, I would have been taxed on $20k. If this is not a tax bracket headroom strategy, then I'm drawing a blank.
Will, I think I see your point. Yes, this works if one has the need/desire to convert only so much per year. And the sucess of this strategy is based on the difference in returns of the split accounts. You are right, my example has $21K taxed, so if the investor is able to convert all at once this strategy is less beneficial, unless of course, the highest fraction goes up a great amount, and the remaining ones average a loss. See that link Elle offered, I don't know that I can explain it any better. JOE
I thought that, while one may have multiple Roth IRA accounts at multiple institutions, the law sees the Roth IRAs' owner as having exactly "one Roth Individual Retirement Account."
So for example I think it is the owner's responsibility to ensure the sum of all contributions to the multiple Roth accounts each year does not exceed $5000 for a single person under age 50, etc. Likewise with Rule 72(t)'s application (per joenospam's post) and conversion/recharacterization maneuvers (per joetaxpayer's post). I think one has to compute recharacterizations based on the total of all the Roth accounts the person has.
Generally, it the total has declined by "a lot," recharacterizing may make sense. But one cannot treat each institution's Roth IRA separately for the purpose of computing recharacterizations.
I got this little lesson when I recharacterized a year ago and the figure was not what I expected. I asked my broker/custodian (Fidelity) "what's up?" and the Fidelity tax rep explained it to me in something like two sentences. What she said seems consistent with the law. Congress (or maybe financial industry lobbyists?) really does (do) not want people monkeying around too much with recharacterizations. From my reading, the main purpose of IRA recharacterization rules is to help those folks who found themselves in a higher tax bracket than originally anticipated when the conversion took place. They did not want to help people "work the system" willy nilly.
Maybe I am misreading you all, or maybe this helps somewhat with Will's question. I too was a little confused when I first read the two Joes proposals here.
I th "More than one IRA. If you have more than one IRA, figure the amount to be recharacterized only on the account from which you withdraw the contribution."
I guess you wouldn't even need to start with 10 tradtional IRAs, you could probably just roll different assets from one tradtional into 10 Roths and accomplish this? Then recharacterize 9 back to the single traditional? But I would think that this trick is only useful in a few edge cases.
From the link you posted earlier: "The rules allowing recharacterization of conversions apply to each individual Roth IRA separately. So if you did five separate conversions, you could then choose to recharacterize a subset of the conversions."
When one converts from their IRA, the pre and post-tax money may not be allocated, it's treated as one IRA per person regardless of the number of sepaerate accounts containing them. But it appears that each comversion may be undone on an account by account basis. To Will's point earlier, the gains may be minimal and not worth the effort, again, depending on the size of the total original IRA portfolio. JOE
Will Trice, I see your citation too. I also note that in Pub
590 page 31 worksheet 1-3, line 3 is the Fair Market Value of the IRA before the conversion plus or minus (as appropriate) any transfers. The inclusion of transfers minimizes monkeying with multiple accounts to one's advantage.
Re Joetaxpayer's ten Roth IRAs example: For maximum tax advantage, and while still getting some amount converted each year, the owner should recharacterize yada all the Roth IRAs that show losses. But if there are no losses, then by my reckoning the multiple Roth IRAs strategy has no advantage.
On the third hand, the other popular criterion for how much one's net total Trad-to-Roth conversion should be is to keep one handily out of the next tax bracket. By recharacterizing, one
I think that summarizes this concept pretty well. Recharacterizing the accounts with losses makes sense. The multiple account strategy can allow the converted money to be the sum with the best jump in value, but it's the bracket issue that would suggest that an account that went up still be re-characterized.
Practically speaking, if one had say $30K and only wanted to convert $10 due to the bracket issue, I see no harm in a three account strategy. Again, this is not mainstream advice, just one of the many odd strategies out there. (and whoever said it - I agree this was not congress' intention, they can easily close this loophole by stating a 'one Roth regardless of number of Roth accounts' rule) JOE
I think I finally get what you're saying about how one could "game" the Roth-Conversion-Recharacterization system with multiple accounts, all with gains after the conversions. The smaller gains would have happened anyway if the conversion amounts had instead just been left in the Trad. IRA. But since one does not know in advance which allocation will do best, one can divide the Roth IRAs according to allocation, and see which does best.
Sorta still begs other questions like whether to guess where the peak of the run-up in the Roth IRAs occurs in the maximum of 15.5 months one has to recharacterize.
If it really can be considered a loophole, it's a mighty tiny one.
Still not sure about Rule 72(t) but will buy (without repeating) Joenospam's report until I read otherwise.
Also, it probably bears mentioning again (as I believe Rich C. often reminds) that "the next tax bracket" is not as big a threat as people often take it to be, since it's only every dollar above the limit of the tax bracket that gets taxed at the higher rate.
Joenospam's 72t scenario is both correct and common. The IRS pubs are somewhat unclear, but Private letter rulings 9050030 and 9525062 uphold that 72t calculations are based on individual, not aggregate, accounts.
I have 3 "early retirement" clients that have done this for many years with no trouble. Accurate record keeping is important, however, as some less meticulous brokerage companies fail to accurately report box
7 of the 1099, which results in the client having to file a form 5329.
As far as I can tell, it's a little messier than that, though the point stands that from a bankruptcy protection point of view, there may be a benefit in keeping the IRA accounts separate rather than commingling. In particular, it appears that the Bankruptcy Abuse (etc) act of 2005 puts a $1 million cap on IRA contributions and earnings on those contributions. However, rollovers from company plans are exempt from the bankruptcy estate completely.
In theory, if one's contributions (and the growth of those contributions) was worth over a million, and you commingled the funds with rollover funds, it appears possible that in a bankruptcy, one would have to try to disentangle those to see what part is capped and what part is not.
Of course, at least up until recently, the chances that contributed amounts (and growth on them) could be over a million bucks, given the long-time $2000 annual contribution cap (and even given the higher current cap). (In the article I read, it pointed out that if you had contributed the maximum to an IRA from 1975 to 2005, your total contributions would still have added up to only $63,500). But over time, eventually, with the larger contributions and the catch-up contributions - and the magic of time - contributed portions of an IRA could in fact exceed that $1million cap.
Anyway, it doesn't appear that commingling the funds actually affects the cap or the protection - but could make for seriously messy business if/when it becomes an issue.
Having not actually read the details of the law (egad!) but only an article or two on it, it's not entirely clear to me where this protection of pure rollover IRA funds comes from, since once the money's rolled from a
401k into an IRA, it's no longer covered by ERISA.
If you're really concerned about creditor protection, it appears the ultimate in safety is leaving it in your employer's 401k. That's one reason not to roll over to an IRA, regardless of whether one commnigles or not.
For what it's worth, I actually keep a separate IRA account which was funded only with non-deductible contributions, but I started doing that well before they lifted the ban on rolling over commingled funds. Being somewhat conservative in these things, even though it appears that the reasons to keep some of these things separate are no longer as pressing as they used to be, I think I'd still advise anyone considering a rollover to do so to a separate IRA account from any contributory IRA.
BeanSmart website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.