Jump to content

Any downside to converting 100% of a $5,000,000 IRA to a Roth IRA on J


Guest MTL-CPA
 Share

Recommended Posts

Art,

The questions you pose assume (A) that your analysis of the situtation is correct, and that (B) your situation under analysis represents the norm.

I've been trying to nicely suggest to you that it ain't necessarily so.

Your statement "I was hoping his response would explain in what way his client was somewhat unique" assumes that his client IS somewhat unique. There are those of us who believe his client represents the norm. Again, that's not the issue. The issue, which you seem to ignore, is that the answer for his client was based upon data that you and I are not privy to (so I guess his client is unique, just like everybody else).

You later state "But since reasonable assumptions for a similar situation show they may cost the children lots of money, I am curious what circumstances would reverse this conclusion.". This is troubling because you take a hypothetical from a specific set of facts ("MAY cost the children") and then immediately install it as a given ("reverse this CONCLUSION").

What is unique about YOUR situation (unique again being a poor word choice since every situation is unique) is that the IRA needs to be tapped for living expenses. That negates a large advantage of the Roth, that being no mandatory distributions. Obviously, who cares if you can legally leave the money untouched, if in fact you have no intention of leaving the money untouched. Change that aspect of the equation, and the result changes.

But most important, stop generalizing.

Barry Picker, CPA/PFS, CFP

New York, NY

www.BPickerCPA.com

Link to comment
Share on other sites

  • Replies 50
  • Created
  • Last Reply

Top Posters In This Topic

Mr. Olsen

Thank you very much for your kind offer. I'll take you up on it!

And a bit of clarification. You said "In the original scenario, it was stated that there would be "equal post-tax withdrawals annually". Later, it was explained that, in the Regular IRA scenario, there would be nothing left."

First, your second concern. There would be nothing left in the regular IRA at the parents death in their mid 80s because they planned to use their IRA funds exclusively to meet their income needs as long as it lasted. It was shown that the IRA funds lasted to about age 84 at which time the only funds (assets) remaining were in the taxable account's stocks.

The "equal post-tax withdrawals annually" were from each option's net after taxes, in-pocket funds to the daughter after the parents death. These withdrawals were shown for a period of 45 years AFTER the parents death. They were in a Roth account for the Roth option and in a taxable account (the stocks remaining from the estate after about half were sold to pay taxes) for the regular IRA option. Both analyses used the same pre-tax growth rate. The Roth withdrawals were shown to be tax-free. The taxable account withdrawals were a bit more each year than the Roth because they were taxed as capital gains after the cost basis was considered, but the after-tax amounts were the same each year for both options. So for several years some of the taxable account withdrawals were partially taxed at 18% until the basis was used up, then the full withdrawals were taxed at 18%. The Roth account ran out of money long before the taxable account (it started out with over 1M less). In my opinion this part of the analysis was a bit simplistic and slightly biased against the Roth, but not by enough to make a significant difference.

I hope this clarifies at least two of your concerns.

And thanks again for your kind offer.

Link to comment
Share on other sites

Mr. John G.

I'm no way an expert. I am a third party with a special interest in the one case but no personal ties to the other. Model builder? Yes, and I prefer N gage to HO.

Perhaps the post to Mr. BPickerCPA will remove some of the vagueness. My numbers in it are from a copy of the stuff I was asked to look at a while back.

I can understand your disagreeing with the conclusions I have; so has everyone else. But I don't understand the significance about half the conversion taxes being offset by tax savings. The numbers showed that the taxes were about 50% ,so you are correct and half of the conversion taxes are "recovered". But what about the other half ? Half a M isn't peanuts. And then what about the loss in growth of the full tax amount over 25 or so years. If the tax was about 1M on the conversion, wouldn't that amount have grown to over 4M in 25 years, or about 2M net after a 50% tax. This looks like a net loss of about 1.5M to me. However, I do realize that where the 1.5M is, i.e. in an IRA, or in a taxable account in stocks, or etc., is also very important.

I still don't feel anyone can practically do such a complicated analysis on an HP calculator. It's true that complicated models are prone to errors but that doesn't mean they have errors. Overly simplified analyses guarantee errors. That may be why most people don't use HPs for these analyses. Ask your mom about that .

Again I state I'm not trying to provide expert insight into anything - just asking a question about what seems to be two different conclusions for similar situations. There probably is a good reason for the difference.

You asked about the model source and user knowledge. I don't know, but I can see it's a question that I should have asked.

Link to comment
Share on other sites

Mr. BPickerCPA

I appreciate your patience. Unfortunately I can't be any more specific in what I have posted.

And I can understand that my cases may be the ones that are unique relative to using the Roth IRA funds rather after the conversion rather than never touching them. I must have misunderstood MTL-CPA's first comment about his client only having enough non-IRA funds to pay the taxes on a 2M conversion.

From this I assumed his client would either have to dip into the IRA for living expenses, like my case, or they don't spend much each year for people with a lot of assets. Their income would to be around 100K- or they couldn't do a conversion.

Thanks again

Link to comment
Share on other sites

"....I don't understand the significance about half the conversion taxes being offset by tax savings. The numbers showed that the taxes were about 50% ,so you are correct and half of the conversion taxes are "recovered". But what about the other half ? Half a M isn't peanuts. And then what about the loss in growth of the full tax amount over 25 or so years. If the tax was about 1M on the conversion, wouldn't that amount have grown to over 4M in 25 years, or about 2M net after a 50% tax. This looks like a net loss of about 1.5M to me. However, I do realize that where the 1.5M is, i.e. in an IRA, or in a taxable account in stocks, or etc., is also very important. "

REPLY: With all tax rates pegged at the max because of the magnitude of the assets, it is the inheritance tax offset that makes the Roth so attractive.

You don't see this because you confuse pretax dollars with after tax amounts. Lets get an apples to apples comparison.... the after tax value to the heirs in 25 years. I will make the following assumptions 40% income tax rate, 50% inheritance tax rate, annual investment return of 10%, 1M conversion tax cost, 2.5M conversion (thus 2.5 * 40% = 1M), parents die at 24.9 yrs.

First the BAU case. The 1M you did not pay in conversion taxes could over 25 years at 10% grow to $10.8M. (Try using that HP12C, your 4M would assume a 5.7% growth rate) Then add the unconverted IRA which would grow from 2.5M to 27.1M using 10% over 25 years. Total pretax assets would be 37.9M. Estate taxes cut this in half, then income taxes reduce that by 40% which gives the heirs an aftertax net of 11.4M.

What was the value of the Roth? The 2.5M Roth growing at 10% for 25 years creates 27M. But estate taxes will cut that in half, so the heirs would have 13.5M in after tax dollars.

Lets see, 13.5M the Roth way, 11.4M with no conversion. Apples to apples. Two million more apples (dollars) for the Roth conversion scenario. All with a HP12C. More importantly, a simple model with consistent assumptions.

The Roth scenario looks better when you consider that that the heirs will likely have the tax shelter for many years past the 25 yr snapshot. The Roth becomes more attractive with a higher investment return. An additional benefit is that the parents do not required distributions.

[This message has been edited by John G (edited 12-16-1999).]

Link to comment
Share on other sites

Mr. John G.

I think your recent posting helps explain why so many feel the recommendation I've seen for not converting is wrong.

You stated, "First the BAU case. The 1M you did not pay in conversion taxes could over 25 years at 10% grow to $10.8M. (Try using that HP12C, your 4M would assume a 5.7% growth rate. NOTE: I said 4M+ and I used 6% growth.) Then add the unconverted IRA which would grow from 2.5M to 27.1M using 10% over 25 years. Total pretax assets would be 37.9M. Estate taxes cut this in half, then income taxes reduce that by 40% which gives the heirs an aftertax net of 11.4M. (I assume your 11.4M net reflects the 10.8M growth of the 1M used to pay conversion taxes not getting hit with the the 40% income tax and any IRD effects.)

I agree with one exception. I'd edit it to read - "the unconverted IRA was depleted by living expenses, taxes and excess withdrawals which were invested in stocks over 25 years leaving the unconverted IRA balance $0 after 25 years, so there were no personal income taxes due which gives the heirs an aftertax net of only 5.4M (50% of the 10.8M growth of the 1M not paid in conversion taxes over 25 years) PLUS 50% of whatever the excess IRA withdrawals grew to in the taxable account." It's probably beyond my skill base to estimate this amount, but whatever it is must be added to the 5.4M, but I do give it a shot later.

Then you state, "What was the value of the Roth? The 2.5M Roth growing at 10% for 25 years creates 27M. But estate taxes will cut that in half, so the heirs would have 13.5M in after tax dollars." The problem I see with this is, for my case, some of the Roth must have been needed for living expenses if the BAU case spent some of the potential non-Roth IRAs 27M for living expenses. So it's a net to heirs is about ??M for the Roth case and ??M for the non-Roth case. ?? - ?? = ??

Therefore it appears that Mr. BPickerCPA's comment about the conversion being a poor choice if you have to dip into it and can't leave it to grow untouched is most pertinent. But is it?

Assume for both of the above cases the IRA owner had just enough money outside the IRA to live on for 25 years without tapping into the Roth. If the 27.1M in the BUA case IRA was used up for living expenses and taxes at 40% were applied to these withdrawals resulting in 10.8M in taxes, then about 16.3M after tax was needed to live on for 25 years. So it appears that about 1.8M had to be in a non-IRA account initially to cover the living expenses for 25 years and the 18% cap gains tax on stock sales. The 1.8M would grow to about 19.5M at a 10% annual growth rate if the IRA funds were used for living expenses rather than the 1.8M in the non-IRA account as in my case, or a net of 9.7M after 50% estate taxes. Then the total BAU case (what is BAU?) aftertax estate would be 15.1M (5.4M + 9.7M). For the Roth case the the 1.8M and it's growth would have been spent leaving only the untouched Roth account.

Lets see, 13.5M the Roth way, 15.1M with no conversion. Apples to apples. Almost 1.6 million LESS apples (dollars) for the Roth conversion scenario. All with a Sharp calculator.

But I admit there may be serious flaws in my logic and the calculations may be a too much for a me and a calculator, so my conclusions may be grossly incorrect. And I suspect there is a problem because the numbers I have show the Roth is only about 1M worse than staying with the regular IRA, not 1.6M worse.

Art

Link to comment
Share on other sites

Art, I gave you a very basic example to demonstrate how a substantial Roth conversion could be extremely benefitial in estate planning and provide heirs significantly greater after tax assets. I stand by that analysis, Roth wins over not converting the IRA (BAU case). You previously made some sweeping conclusions that Roth conversions were bad... my basic case shows how you were wrong by $2M in net after tax dollars for the heirs.

In your "Assume" paragraph in the very first sentence you pose the situation where the IRA owners can live on non-IRA assets and makes no withdrawals..... well that is exactly the case I presented you. Roth wins. The rest of that paragraph is a total hash. You are definitely mixing terms and perspectives and have not set up a consistent scenario. You talk about an 18% cap gain tax rate on stocks.... what is that? There are no cap gains on any transactions in either regular or Roth IRAs. You talk about a scenario with no living expenses, then you talk about funds used for living expenses. You can't just wing it when you are setting up a scenario, the assumptions must be clear AND consistent.

A stream of withdrawals would not make the Roth conversion scenario worse, but it would reduce the benefits. I can not think of any reasonable scenario for a high asset family where the Roth would be a worse case. If you could construct the problem correctly, you will find that the Roth case would be attractive.

You need to stop posing as an expert at this site. Your understanding of IRAs and Roths is limited. You seem incapable of setting up a clean scenario with consistent math. A combination of four CPAs/financial planners have told you that your conclusions are wrong. Maybe you should read/listen and stop trying to enlighten.

[This message has been edited by John G (edited 12-17-1999).]

Link to comment
Share on other sites

JohnG.

While I found this exchange both interesting and informative, I must give up. We can't seem to avoid basing our comments on two very different situations.

You're looking at a part of the picture, i.e. what happens only to an untouched non-Roth IRA plus the growth of taxes saved by not converting, and to an untouched Roth.

You appear to ignore the fact that there must be some "other taxable account assets (stocks) available initially" to grow and be sold over 25 years for living expenses IF everything else is untouched. It may be quite OK not to consider the whole asset picture and to ignore this "other money". I don't know, but I'm curious if it is OK. I'm clearly no expert and don't profess to be, but even to me it's quite obvious and understandable that for the approach you described the Roth is indeed best. I'm not contesting that conclusion.

However, the numbers I have appear to be for a totally different approach than yours. One (big??) difference in my case is where those "other assets (stocks) available initially" that normally are used for 25 years of living are instead left invested to grow untouched, and it's the non-Roth IRA assets that are dipped into early and often (used to depletion). That is, it's a situation where (1) ALL of the initially available assets are addressed in the analysis, and (2) a totally different approach toward how/when the non-Roth IRA funds are used. Since we are comparing different withdrawal strategies, the fact our conclusions differ isn't surprising.

And before you tell me that spending the regular IRA first is a dumb move, although I'm not an expert, I do realize that this spend the regular IRA first approach is usually bad for people needing to spend all of their assets during retirement. But is it also bad for people wanting to benefit their heirs; people who don't plan to spend all of their retirement assets for living? That is, minimize the amount in their IRA at the time of death so as to minimize the estate taxes and the income taxes the heirs have to pay.

Also, I was finally able to contact the org who did the analyses I have copies of. The advisor is now retired, but he had some interesting comments about this very type of flap. He will try to e-mail me his copy of the program since he no longer uses it. He said in's only good till Y2K because yearly tax updates are needed.

So I'll rely on Mr. John Olsen to sort this out as he understands there are two different cases and he has the interest/curosity to examine the numbers I'll be faxing to him.

Thanks to all, I've enjoyed it and hopefully learned a little in the process.

Till Later - Art

Link to comment
Share on other sites

Art sent me the analysis he was referring to.

I confess that I cannot make head or tails of it. That is NOT any reflection on Art. It isn't his analysis, as he told me. And I am perfectly willing to concede that someone smarter might be able to decipher the reports. But I can't. In the first place, there is SO MUCH information on each page, and the print is so incredibly tiny, that I can barely read the legends. But, even reading them, I have no idea how the resulting figures were obtained. It's all summary. There are several graphs, and huge "detail" ledger, but without a code, describing each line, what it is supposed to represent - and, most importantly, how the values are obtained - it's just a matrix of numbers.

Art, who markets this analysis? Surely, there is some EXPLANATION for what it produces. I am not all that unfamiliar with sophisticated financial software. In fact, I am a beta tester for some of the best programs out there. Wealthtec's "Advanced Pro", for example, is VERY powerful, and considers elements not considered by most analyses. And, quite frankly, it's full analyses are VERY complicated and can be pretty confusing. But I can work my way through that program, because I can read the labels and ascertain what is being done with the data (even without formulas). This thing, however, has me just plain stumped.

------------------

John L. Olsen, CLU, ChFC

Olsen Financial Group

St. Louis, MO

314-909-8818

John L. Olsen, CLU, ChFC

Olsen Financial Group

St. Louis, MO

314-909-8818

Link to comment
Share on other sites

Mr. John Olsen - Thanks again for your time and efforts.

Happy the fax got to you but sorry and not too surprised that the small print wasn't readable. Fortunately I now have the program and the small print is readable when printed. I also read the help file which was very informative.

The help file explained the summary page with the chart. Almost all of the numbers surrounding the chart are for numerical value or Yes or No action inputs. The chart itself and the "Estimated FY Portfolio Values - 5 Year Increments" table under the chart are the calculation results. The very bottom row on the pages with the chart is the Net to Heirs After-Tax , i.e. what appears to be the subject under debate here. (Although these data are in a different location in my printouts.)

The 3 summary pages each with a chart are clearly for 3 different cases for R&J. The page with the chart showing "Usual" appears to be the case for the usual recommendation most people get or tend to do - to use their non-IRA assets before their IRA, except for mandatory IRA withdrawals after 70. The chart showing "2M Conversion" is for the case of converting 2M of their nearly 5M IRA to a Roth.

The results of these 2 cases are in total agreement with experts' comments. Converting 2M of a 5M IRA is clearly better for the heirs than the usual "use the non-IRA assets before using the IRA assets" approach, at least until about age 90. So I can see why they were so adamant about the Roth conversion being best, and why many people in similar situations were probably advised to convert.

But the person doing R&J's analyses apparently didn't stop after getting the results for the 2 cases. He looked at another option, probably many more before settling on the one selected.

That other option is the third chart showing "no Conversion". It's for the case where R&J would withdraw a fixed percentage of their IRA assets annually from age 61 - 70, then drawdown the remaining IRA over only 15 or so years (even to the extent of taking withdrawals much greater than needed or required toward the end of the 15 years) and investing the after-tax excess each year in stocks in their taxable account. The results show this approach is much better for the heirs than the 2M Roth conversion, by about 1M if R&J die at age 85. The Roth is not quite as bad if the if they go earlier; but it gets worse if they live longer.

Obviously if there are errors in the program the above results and conclusions are incorrect. The results of the first 2 cases seemingly in agreement with 2 experts' opinions indicate the program may be OK. The guy I got if from indicated he/they had checked it. So I'm more comfortable now than I was last week that the recommendation against the Roth conversion is correct for R&J's situation, but not 100% satisfied if it were for me.

Also, the problem of not being able to duplicate the info I sent you is solved. The program calculates from today's date forward. With the same starting assets and annual incomes/expenses, the results from Feb. to the end of 1999 and onward will obviously be different than the results from now to the end of 1999.

The 2 pages faxed to you each with about 100 rows of SO MUCH information in small print were what I assumed to be all the details. But I now find that there are over 500 more hidden rows of data. The 100 rows are likely the more important data, but there are a lot more. I can see and print the full 600+ rows, but only the numbers, not the formulas.

In the fax I asked how does one check out such a program to determine if it's results are valid. When I thought there were only 100 or so lines and I could pick a few random years, I considered checking the numbers manually over a couple of days. But 600 lines for each of a few years to do by hand??? Maybe the principle of all this isn't worth the effort; it's not my money. Hopefully there's an easier way. Maybe trying to prove the program has errors is easier than trying to validate it. Plug some numbers with known valid results from another program into this program? Use my numbers in a program known to be OK? Punt?? I'd appreciate any suggestions - about how to reasonably check out the program results that is .

Art

Link to comment
Share on other sites

Art,

If you've the time and inclination to try to verify the numbers that program produces, more power to you. I showed the output to Kit Mueller, the developer of "Second 1/2" software, and his response was similar to mine. We can't figure out what the thing is trying to tell us, and are pretty sure we wouldn't believe it if we could figure it out. It's unclear - INCREDIBLY unclear. Personally, I can't abide that in a program which is supposed to CLARIFY complex problems for users.

As to the strategy of taking very large distributions from the IRA - earlier and larger than required - that is one of the BEST moves possible in a LOT of scenarios.

If the client takes those early distributions, pays tax on 'em, and gifts the net to an Irrev Trust, which then buys life insurance on the client, the additional wealth tranferred to heirs can be significant.

But, as good as that technique can look, it can look EVEN BETTER if the client FIRST converts that IRA to a Roth, and then proceeds with the "end run" technique described above.

If you're looking for a software package that will model a LOT of distribution alternatives, check out Ocaso's "Second 1/2". (www.ocaso.com) The brand-new version has a lot of improvements and is more flexible than Brentmark's "Pension and IRA Analyzer", although the output isn't nearly as customizable.

Best of luck (and I hope you've got a BIG bottle of Excedrin!)

John Olsen

John L. Olsen, CLU, ChFC

Olsen Financial Group

St. Louis, MO

314-909-8818

Link to comment
Share on other sites

One of the reasons for this string of responses was the following first comment from Art E:

"What excapes me is why convering $2M to a Roth is good for either the owner of the IRA or his/her heirs. The $2M conversion will make the taxes due at conversion the highest level. My progran shows convering $2M of a $5M IRA will cost the heirs from $1M to $2M after tax."

The statement seems fairly simple, but assertion is totally wrong. The conclusion has never been backed up by any comprehendable math. A large asset household that is still eligible for a Roth conversion may find Roth conversion attractive. Reason: a large part of the conversion taxes are offset by reduced estate taxes. Those rules may change in the future, either favorably or unfavorably to this scenario, but that is how the math works right now. Lets not get confused or distracted about the most fundamental point.

Art E also recently said:

"That other option..... It's for the case where R&J would withdraw a fixed percentage of their IRA assets annually from age 61 - 70, then drawdown the remaining IRA over only 15 or so years (even to the extent of taking withdrawals much greater than needed or required toward the end of the 15 years) and investing the after-tax excess each year in stocks in their taxable account. The results show this approach is much better for the heirs than the 2M Roth conversion, by about 1M if R&J die at age 85. The Roth is not quite as bad if the if they go earlier; but it gets worse if they live longer"

The difficulty with the above scenario is that payouts to the parents must be valued in the same terms as some snapshot of net residual estate. Classic economic studies would typically convert all of these amount to "present value". That is how the economic value of bridges and airports are calculated. But Roth models tend to focus on income streams and assets at the end of some study period. You can not compare two different scenarios if the payouts and residual asset value are not evaluated consistently. I don't believe Art E's conclusion makes sense and suspect the problem is in how the program evaluates amounts in different classes over time.

A Roth conversion for a high income family essentially allows you to buy a Roth at "half price", that is for half the conversion taxes. For this to be a bad idea, the family must have a way in the future of taking very large payouts at half their normal tax rate. Not very likely. Other conclusions are probably due to math errors or the problem not being set up consistently.

We should end this stream of comments. Art is telling us about some model that seems to support (I conclude) his bias against Roths. He is not the creator of the model. He did not set up the example, nor is he the client for whom the model was constucted. He was not trained in the use or mis-use of this model. He does not apparently have access to the code or equations. At this point, a working version of this model is not available to others. The financial planners and CPAs have voiced plenty of skepticism.... but this is like trying to disprove an some vague "urban legend" that happened to someone in some nearby town. It is like reasoning with the flat earth society, which persisted for many decades apparently just to tweek the establishment not to enlighten. I have a good mallet, does anyone have a wooden stake?

[This message has been edited by John G (edited 12-21-1999).]

Link to comment
Share on other sites

I was working on a case last night that may be of some interest in this connection. Clients are couple age 60 and 58. Total estate - 3 million, of which 1 mill is qual plan. Non qp assets include rental real estate, which is producing a nice income. Goal is to maximize wealth transfer to heirs.

I ran four scenarios through Brentmarks' "Pension and Roth IRA Analyzer".

A: Do nothing. Take only RMDs from QP.

B: "End run". Begin taking distribs from QP now, pay tax on same, gift net to ILIT, which buys 2nd to Die life insurance on clients.

C: Convert entire 1 mill to Roth (by first rolling to traditional IRA)

D: Convert to Roth and then do "end run" "B", except pay premiums from NON-Roth assets.

There are a LOT of variables interacting here. But holding tax rates, desired income, return on investments, life expectancies, etc. constant, the "best" scenario is either the Roth conversion or the Roth conversion Plus End Run. When maximum deferral was checked (in which it's assumed that 2nd generation heirs will take Roth money only on RMD schedule), the Roth looked best for about fifteen years, with the Roth Plus End Run looking best after that point.

Changing ROR, income needed, sources of income, tax rates, etc. all had the effect of changing the "winners'" pattern.

My take on this, having done similar analyses more than a few times, is that GENERALLY the Roth conversion will beat no conversion, provided (a) there is a desire to stretch out deferral to the max (B) there is an Estate Tax liability © the income tax rate is not GREATLY lower after retirement than at present © conversion tax is paid from outside assets.

The End Run scenario, by itself, OFTEN beats the "do nothing" Present Case plan, BUT, MANY TIMES, ONLY FOR A PERIOD OF YEARS, following which it becomes "more expensive".

The End Run Plus Roth scenario has, in several cases I've done, produced the best numbers.

I would NEVER presume to suggest that this is a Rule Of Thumb.

... for what it's worth...

------------------

John L. Olsen, CLU, ChFC

Olsen Financial Group

St. Louis, MO

314-909-8818

John L. Olsen, CLU, ChFC

Olsen Financial Group

St. Louis, MO

314-909-8818

Link to comment
Share on other sites

John O, I totally agree with your very good

summary:

"... GENERALLY the Roth conversion will beat no conversion, provided (a) there is a desire to stretch out deferral to the max (B) there is an Estate Tax liability © the income tax rate is not GREATLY lower after retirement than at present (d) conversion tax is paid from outside assets."

Anyone with significant assets or a complicated situation clearly needs professional help to run the numbers. If a couple of million are involved, it might make sense to have two different professional teams working on the problem.

Link to comment
Share on other sites

Mr. J. Olsen

Thanks to your comment about insurance, I may have accidentally found the parameter that can make the Roth better or worse than the no conversion case. While it wasn't the insurance, it was what I had to do to examine the effects of insurance that gave the answer. I think spendable income may be the stake JohnG. was looking for because everybody can say we were correct; X is better than Y.

You said

I looked at your - takes early distributions (I used 10K each year) paying taxes and gifting the net for insurance - as being similar to spending an annual 10K withdrawal's after-tax net over X years to buy insurance each year and the insurance payoff is not in the estate?

However, I was unable check out this exact case. I'm pretty sure that I can't input pre-tax expenses into the program. It appears it only allows after-tax expense inputs. It calculates what the pre-tax income has to be to provide that net expense and then subtracts this pre-tax amount from income, assets or whatever.

But I did look at a situation where an inflated, after-tax 10K went to buy insurance annually. This was essentially the same as increasing R&Js spendable by 10K and not worrying about the insurance payoff. The payoff isn't in the estate so I assume it will provide the same amount to a heir in either case because the same amount of insurance was purchased in both cases.

The numbers were very interesting. Buying the insurance "cost" the heir 400K to 450K. So the insurance payoff must be close to or more than this amount for insurance to benefit the heir as opposed to just investing the 400K in a trust. I suspect the payoff is a lot more because over 400K after-tax was spent over 25 years to buy the insurance.

But the big thing was that while the no conversion case was still better than the Roth , it was now only better by about 500K, not 1M. So a relatively small increase to 210K from 200K in annual spendable over 25 years had a big effect on the difference between the two options.

BINGO - When I raised the initial annual spendable to 250K from 200K the Roth turned out be the best option. The no conversion case was now only 200K better for the heir than the Roth at the parents' age 85. And 200K more in stocks in a taxable account (initially) underperforms a Roth account (with 200K less initially) over the long haul.

So it appears for R&Js 2M by 5M IRA situation, and probably most others, one critically important factor in evaluating a Roth conversion is the actual spendable income desired by the IRA owner. And this makes some sense because an increased spendable will increase the effective income tax rate which hurts the no conversion case, and visa versa. So I don't see how you can do a Roth evaluation without using the spendable income as an input/factor.

I had looked at other factors, like increasing the investment appreciation as JohnG commented and increasing the assets available so the Roth wasn't touched as BPickerCPA commented. But neither of these changes had much of an effect on the net to a heir difference at age 85. But for each case I ALSO changed the spendable to reflect the increase in appreciation and initial assets. The rationale for the increased spendable was that I thought if a person expected to have or had a lot more money, they'd probably spend more. Which is probably valid, but it clouded the picture because I changed two variables. I'll eventually do the increased appreciation and increased asset cases without bumping up the spendable just to see how significant the effects of these two variables are on a case with a fixed spendable.

While the program may have errors, I don't think I'll need that bottle of Excedrin. First, I'm satisfied the program is probably OK. Second, I read about how it handles a little thing like SS taxes on incomes. For salary incomes it taxes at a 7+% rate. For self employed income it taxes at a 15+% rate and then accounts for half of this in calculating income tax liabilities. I realize that accurate tax liability determinations are very important in Roth evaluations, so if it gets down to this level in determining taxes it's results are probably OK. Third, I don't think I really care how tax credit invertments effect the conversion.

PS regarding your latest post. What do you mean by holding taxes and changing taxes?

Art

Link to comment
Share on other sites

Mr. JohnG's

Perhaps my post to Mr. Olsen reconciles our differences.

In response to your comment I think I understand your point here and believe the value of the payouts are addressed in the program's analysis. For each of the 3 cases the payouts to the parents were the same amount each year in CY, FY or whatever dollars. That is, for each case they "spent" the same amount each year and the cum amounts spent were the same. So the value to the parents of these payouts appear to be identical for each case, i.e. none of the cases benefitted from any payout value difference.

And I'm not biased against Roths as I often said they are no doubt good for some people. I even ran my numbers and found it optimum to converted about 23% of my IRA to a Roth over 4+ years. It would be good for ME by over 300K when I'm 90. But from this exercise, I assume that there have been a few, some or many Roth conversions done that should not have been done and some may have been very expensive errors.

Art

Link to comment
Share on other sites

[[but from this exercise, I assume that there have been a few, some or many Roth conversions done that should not have been done and some may have been very expensive errors.]]

Art,

From my own experience reviewing people's conversions, I have to agree with you.

But I have found far more people for whom a Roth conversion WAS advantageous, yet they were advised against it by advisors who were either too lazy or incompetent to weigh ALL the relevant factors, and due a proper analysis. For many people, it is now too late, and so they've lost a great opportunity.

Barry Picker, CPA/PFS, CFP

New York, NY

www.BPickerCPA.com

Link to comment
Share on other sites

Mr. John Olson

You knew I couldn't pass this up.

Used your numbers, but had to make some assumptions about your Mr.60/Mrs.58 case.

From you - couple age 60 and 58. Total estate - 3 million, of which 1 mill is qual plan. Non qp assets include rental real estate, which is producing a nice income. Goal is to maximize wealth transfer to heirs.

The added assumptions I needed to make for inputs were

- Neither individual is blind.

- 1M in rental property with no debt on it and no depreciation left, value growing at 70% of inflation rate with about 1.5K property tax now growing at 1.2%

- 1M in taxable accounts, i.e. 915K stocks with pre-tax 6% growth and 0.7 dividend/dist'n/etc., 30K in munis, 30K in US bonds and 25K in Corp. bonds yielding pre-tax of 5, 5.2 and 6 % respectively.

- 1M in IRA also growing at 6.7% tax-deferred.

- Inflation 3.5%

- Spendable income goal 100K case 1, 70K case 2, both growing at inflation rate.

- Rental net income 70K pre-tax, growing at only 3% annually.

- SS incomes beginning at ages 62 (calculated by program to be 14.8K him and 7.2K her), growing at 100% of inflation.

- No other incomes, i.e. say he retired say a couple of years ago (I would have) and she never worked, no pensions.

- Today's Federal tax rules apply adjusted for inflation at 100% (could but didn't bias this % up and down).

- State income taxes while living on the high average side, also program calculations (not a fixed % option).

I think these are the only inputs I had to guess. I recognize that the investment growths may be low and SS income growth at inflation rate may be a bit high. But some of the asset amounts, spendable and rental income are no doubt way off and need to be changed too. Which I'd do if you'd tell me what to use, even if he/she is greedy and still working and has incomes of ?? to age ??, etc.

As I noted earlier, I can't/don't know how to do the insurance end run thing. But it seems that its results should be about the same as bumping up the spendable by the same amount for each case. The new net to heir results will each be equally low by the amount of the insurance. Yet it would seem the insurance amount to be added to each net to heir result will be the same for each case - since the same amount of insurance was bought and received in each case. So the net to heir differences would be the same.

Anyhow - the numbers I see for the 100K spendable case show the don't convert but do diddle with early/big IRA withdrawals is always best. A Roth conversion is better than doing nothing only until about age 75 when it falls to worst. I did the Roth conversion over 2 years as it's a little better for the Roth than doing it all in one year. Couldn't do it over a longer time as the program indicated their AGI or something was too high to convert after his SS income started coming in at age 62.

For the 70K spendable case, the don't convert, but diddle is best until between age 85 - 90 when the do nothing case becomes best. The Roth holds second place again until about age 75 when it falls to worst case.

For the 100K case the best to worst case difference is roughly 100K at age 70 and it increases about 100K every 5 years to age 95. For the 70K case the differences are roughly half that.

It's easy to see that there are several other options that should be run. Start SS later so the conversion can be spread over more years, different growth, inflation, etc. rates, + and - tax rule biases, etc. etc. etc. Who said we are moving toward a paperless office???

Art

Link to comment
Share on other sites

Mr. BPickerCPA

You said <-------- (didn't) do a proper analysis.>

It's easy to agree with you here because of the number of variables likely needed to do a proper analyses as well as, and no offense JohnG, probably a relatively sophisticated computer program. I can see how it would take a day or two to run the necessary analyses and evaluate them all to consider the "other factors". And that's after gathering the piles of info needed for inputs.

Art

Link to comment
Share on other sites

Art E, you misunderstand my point about modeling. Early in my career, I created a number of models with thousands of lines of code in the era before spreadsheets. In the PC era I have created some rather large spreadsheets. A substantial part of my work has involved verifying/refuting the large scale models of others. I kind of like big models. They can answer complex questions. But complicated models have a high probability of error, and it gets harder to detect these errors as model complexity increases.

Look at you problem set up. Looks like a lot of clutter to me: rental property growth rates? property tax growth rates? Details on types of bonds and catagory growth rates. My OR friends would call this a false level of precision, details that do not enhance the accuracy of the result. The more clutter you pack into a model, the less likely the results are correct.

Data problem 1: One of the most important assumptions you made was a 6.7% growth rate for investments! That is probably the lowest assumption I have ever seen used in a financial plan, even for senior citizens. For comparison, lets look at the average annual return for what I think are the five oldest mutual funds at Fidelity: Fidelity Fund (since 1930) 11.3%, Puritan (1947) 12.4%, Trend (1958) 13.6%, Magellan (1963) 22.2%, and Equity Inc (1966) 14.6%. A variety of stock dominated profiles with between 34 to 70 years history. {Fidelity is a convenient source of benchmarks, which is the sole purpose of using the data here)

My grandfather, who is extremely conservative, would fire his investment advisor if 6.7% the long run result. Your assumption could probably be achieved with a 90/10 bond/equity asset mix. Seems unrealistically low to me. Over 20 years the difference between 10% and 6.7% is makes about a 2x difference in assets. Why use such a conservative assumption for assets that will be held for decades?

Data problem 2: You are using a long run inflation rate of 3.5% which is not consistent with your bond rate assumptions. The bond rates look a lot more like current yields, which are in a much lower inflation period. Looks like your effective bond yield (bond%-inflation) would be only 2%. On equity investments you imply about a 3.2% net yield which would be historically very low. For example, the oldest two Fidelity money market accounts (1974, 1979) have averaged 7.6% and 7.4% over 20+ years and these invest in short term instruments.

Whose model are you using Art? You have been asked multiple times and have yet to disclose the source. My experience with complex models is that the builder rarely checks out each feature for accuracy but rather focuses just on some core equations. Have you been trained to use this model? Do you have documention?

What is the "theory" behind any case where to Roth conversion is worse? To make these unussual results believable, you need to explain how it plays out.

[This message has been edited by John G (edited 12-23-1999).]

Link to comment
Share on other sites

Mr. JohnG

I may have misunderstood your comments on 12/23 about sophisticated models, but I still don't see how anyone can do a complicated Roth vs. regular IRA comparison on an hand calculator.

I understand precison; 8 x 3.2 = 24. So I disagree that considering more details is necessarily false precison, especially if these details are relevant to the answer.

And my post said "I recognize that the investment growths may be low and SS income growth at the inflation rate may be a bit high."

I agree some of the growths are probably low, but that's what R&J wanted to use and when I did the thing with Mr. Olsen's basic inputs I didn't bother change the numbers that already were there. If he was interested there were no doubt other numbers I used that would also have to change.

The bond yields are the actual averages for R&J's bonds. They bought them a few years ago when yields were lower. So the yields are correct for their case. But the value of the bonds are the face amounts, so they may be too high if R&J had to sell now, which they don't. Using the face amount seems reasonable if they hang on until they come due.

I did find the name etc. in the help file and an internet address. Since it probably isn't illegal for me to be using the program that was given to me -- the address is roarsoftware.com.

I've often said the model may be totally bad. But as you may have noted, I did finally come up with a case where it showed the Roth was best. So it's probably OK. I'm not interested enough to manually check it out because as Mr. Olsen said, it looks like a very big job. I did offer to plug in some numbers from a known good program to see if the results were the same but got no responses (due to no known good programs??).

It's clear I haven't been trained in the program so it's also probable I am doing something wrong. Although it seems simple enough to use and the answers are easy to understand. Enter a number or a Y or N, run it, and up comes the answers. But I did read the entire help file. Which was maybe overkill because a lot of it discusses another program that's there. You can use the other program to put in all sorts of investment details and it apparently sorts the stuff out into proper piles, averages, etc. and copies its summary results to the right places in the main program.

And I have no "theory" why the Roth is worst or best - yet. But I may be getting very close to at least having an opinion.

Art

[This message has been edited by Art E (edited 12-28-1999).]

Link to comment
Share on other sites

Art, you are the 2nd person to comment on this site that has used the roar software. (unless you are the same person using an alias) The last fellow used the roar support email address as his own. He also kept coming up with odd results.

At that time, I examined four scenarios he generated with their model. I also visited their web site, read the FAQ and the model features. This model has 150 input parameters. I am not impressed. Their marketing material claims "ease of use" and "understandable output". I agree with the prior comments from Olsen and Mueller that the output, labeling and documentation are unclear and confusing. The web site also makes a silly statement that one of the primary benefits of this software is that it "doesn't require users to guess at one future tax rates". That is because the model "guesses" for you. They built the current tax code into the model. Ho hum.

The documentation warns that the roar model does not apparently have checks for out of range data (they use the example mistaking 4.5% vs .045), a huge deficiency. No evidence that the model has been tested. Skimpy documentation. No information on the experience of the company who created the model, they just say they are better. No phone number to ask questions... unless you sign up and pay the higher professional fee.

Am I too harsh? Readers can check for themselves. Go to the roarsoftware site and pull up the sample input/output page. Do they really need 18 parameters on house purchases and sales? Looks like a NASA approach to modeling, throw in everything.

My first impression when looking at their sample page was "Oh no, another Club of Rome study". Do you remember them? They were the elete team that published "Limits to Growth" around the 1960s. Wonderfully complex charts, massive computer model. Just one problem, they were totally wrong. The built in model biases always gave a chicken little answer.

I think you have fallen in love with the tool. It comes out in your comments on how all sorts of extraneous details are essential to study Roth conversions and how you don't see how this can be done with a calculator.

Well, lots of folks can use a calculator or one-page spreadsheet (lets not forget the abacus) and get a decent answer. When you boil down the big IRA asset scenario we have been talking about the key parameters are initial IRA assets, conversion amount, period, investment yield, income tax rate and inheritance tax rate. In this high wealth scenario, about 1/2 of the conversion tax cost is offset by reduced inheritance taxes. This a fundamental element of the scenario. For the Roth to be a bad idea, the IRA withdrawal tax rate in the no conversion case must be less than 1/2 the tax rate applicable for conversion taxes.

You implied that you only recently found one case where a Roth conversion was better. That seems pretty strange to me. You have never explained how the conceptual structure or theory I outlined above would not apply to this large asset scenario. Why should anyone believe your black box output from a third party model that was based upon odd and inconsistent assumptions?

[This message has been edited by John G (edited 12-29-1999).]

Link to comment
Share on other sites

[[in this high wealth scenario, about 1/2 of the conversion tax cost is offset by reduced inheritance taxes. This a fundamental element of the scenario. For the Roth to be a bad idea, the IRA withdrawal tax rate in the no conversion case must be less than 1/2 the tax rate applicable for conversion taxes.]]

I have to disagree with this statement because it appears to ignore the income tax deduction for the estate tax on the IRA. The advantage of a roth conversion to a high net worth client, is really limited to the state death tax paid. No small potatoes, but not the 50% you stated.

Barry Picker, CPA/PFS, CFP

New York, NY

www.BPickerCPA.com

Link to comment
Share on other sites

In addition to avoiding the problem with the § 691© deduction not covering state death taxes that Barry Picker pointed out, the Roth conversion has lots of other benefits.

By paying the income tax out of other assets, you are effectively putting additional assets into your IRA.

No required distributions at 70 1/2.

You have a tax-paid IRA to leave in trust for your grandchildren and to allocate your GST exemption.

The only problem is that most clients don't qualify for the Roth conversion.

------------------

Bruce Steiner, attorney

(212) 986-6000 (NY office)

(201) 862-1080 (NJ office)

also admitted in FL

Bruce Steiner, attorney

(212) 986-6000

also admitted in NJ and FL

Link to comment
Share on other sites

Mr. JohnG

Earlier you asked "What is the "theory" behind any case where to Roth conversion is worse?"

I have no "theory" but I do have an opinion based on trying to compare the results I have to those from some programs designed to analyze Roths.

First I tried comparing it to some of the Roth programs on the internet because it seemed easy to do. But that was a waste of time. While I was armed to the teeth with a lot more info than those programs bothered ask for, they all asked a couple of inputs that I couldn't answer, and I don't see how anyone can.

I looked into the 2nd 1/2 program Mr. Olsen recommended and was fortunate enough to talk briefly to Mr. Mueller. I may have misunderstood some of his comments, but he indicated his program's forte is more into estate and distribution planning than Roth vs. regular IRA evaluations.

I also searched for the other program mentioned by Mr. Olsen (and also suggested by Mr. Mueller) and found BM has an internet site. They have a demo of their Pension and Roth Analyzer program. Unfortunately the demo wouldn't allow me to change its IRA dollars, ages, etc. So I had to change my numbers where I could to match their demo's numbers and use my tax rate numbers in their program.

Again, BINGO. Both programs showed similar results, i.e. it was better to stay with the regular IRA and spend it first rather than convert to a Roth. At 15 years out, their program showed the regular IRA approach was better by about 370K after-taxes. My numbers showed it was better by only about 290K. Also, both results showed the regular IRA advantage kept increasing with time.

However, these results were for a case with a 1.5M IRA and 800K in other assets where the Roth IRA had to be used for living expenses, and as Mr. BPickerCPA noted, it is likely to be bad for the Roth. So I tried another case by bumping up the Other Assets to 2.2M so the Roth wasn't touched. Again the BM results showed staying with the regular IRA was best by about 250K.

I will concede that for both of these cases over a few decades in the future, the initial 250K - 370K advantage to a heir for the regular IRA in stocks in a taxable account will probably decrease to nil or maybe less compared to having a lessor amount in a Roth account.

It's also probable that I may have made some errors in using both programs. And these few cases may not be typical, etc. etc. I did play a bit with the BM program by inputting various tax rates and found I could get about any answer I wanted.

But for the first cut, the BM analyzer program using essentially the same inputs produces results that are very similar to my numbers, i.e. a big Roth conversion may not be the proper choice (for a few, some, or perhaps most people??). This is another data point favoring the regular IRA for big dollar IRAs assuming you don't have similar concerns about the BM program and no input, use, or data interpretation errors on my part. (Granted, the latter is a big assumption.) That's 2 out of 2. So it's not clear why questioning the wisdom of a 2M of 5M conversion was so wrong. That big conversion may indeed be the right move; it just isn't obvious why it is.

My opinion is that taxes are probably the single most important factor in a Roth analysis. Items like the size of the IRA, how big the conversion is, other incomes, income needs, non-IRA asset types and characteristics, etc. ALL become important because collectively, they dictate what the taxes will be over time. They also dictate the size and type of piles left at the end to which estate and income taxes are applied. No answer will be right, but everything else being equal, the closer you get to the right taxes the closer you'll be to the right answer to use for making a Roth or probably any financial decision.

Art

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×
×
  • Create New...