Tax Efficiency of Treasuries

[Call_Me_Op]

I am interested in holding individual treasuries in my taxable account, so I performed a spreadsheet analysis to assess tax efficiency. What I compared are the following two scenarios over a time period of 20 years:

1.) Place 10 treasuries with 4% coupon into tax-deferred account, reinvest dividends, and pay taxes after 20 years at a combined federal plus state tax rate of 33% + 6% = 39%.

2.) Place 10 treasuries with 4% coupon into taxable account, reinvest dividends, and pay taxes each year OUT OF SEPARATE CASH ACCOUNT. Tax rate is federal only, at 33%. (Treasuries are state-tax exempt.)

I assumed that cash is yielding nothing to simplify the problem. (I paid taxes out of cash because that's how I always do it.) My results were as follows:

1.) Tax-deferred case: start $10,000......finish with $17,265.85

2.) Taxable case: start $10,000......finish with $17,980.53

Thus, with the assumptions made, I end up with more money placing the treasuries in the taxable account. This seems to go against the so-called "conventional wisdom." Comments welcome.

[Clive]

Its geographically and personally subjective.

In the UK for instance our tax efficient holdings are exempt from taxes (you can invest (add) around $16,000 each year into an ISA) so after 17 years net = gross potentially.

Our Gilts (treasury) are exempt from capital gains tax, but income is subject to income tax, so if held outside of tax-efficient and its costs effective to do so, selling just prior to x-div and buying back again is a reasonable choice.

We're allowed around another $16,000 each year in capital gains exemption, which when combined with tax harvesting makes holding stocks outside of tax-efficient perhaps the better choice (again all highly subjective). We even have spreadbet versions of stocks, gains from which are tax exempt (but you're holding a derivative rather than the actuals). UK gold Sovereign's and Britannia's (being legal tender) are tax exempt, other gold coins aren't.

Some states may have lower (or perhaps even no?) state taxes.

You just have to figure out what's best for you personally and go with that. As general advice however I think bonds in tax-efficient is probably the best overall average choice.

[Call_Me_Op]

Clive,

Yes indeed - my analysis applies to a particular set of circumstances, and each person should evaluate their own optimal location for this (or any) asset. However, I would bet that many people in the 33% marginal tax bracket and a medium tax state would have guessed that they should never place treasury bonds in a taxable account - based upon the prevalent dogma.

[Bob's not my name]

Dogma or math? Think hard about what you've left out of your model.

[Call_Me_Op]

I documented my assumptions, but of course may have missed something. Please help me out if I have.

I assumed that there is a separate cash account making nothing. This is realistic for me - but may not be for others.

[Bob's not my name]

Here's a strong hint:

1.) Tax-deferred case: start $16,393

[sscritic]

Where did the $10k come from and how did it get into each account? To get $10k into your taxable account, you had to make $16,393.44 (less 39% for taxes = $10k). If you had deferred that $16,393.44, what did you do with the extra $6,393.44 that you didn't use to buy treasuries? Keep it in cash inside your tax deferred account?

[Call_Me_Op]

Here's a strong hint:

1.) Tax-deferred case: start $16,393

I was thinking more of an after-tax IRA or annuity. I believe you are referring to the case of a 401K or other account where one can invest using pre-tax money. I agree that is a different case.

[Call_Me_Op]

Where did the $10k come from and how did it get into each account? To get $10k into your taxable account, you had to make $16,393.44 (less 39% for taxes = $10k). If you had deferred that $16,393.44, what did you do with the extra $6,393.44 that you didn't use to buy treasuries? Keep it in cash inside your tax deferred account?

See response to not-Bob. I did make an assumption that you are starting with a level playing field, which means you have already paid income tax on the money in both cases.

[Bob's not my name]

Yow. Then you are agreeing with the wiki that a non-deductible IRA not converted to Roth is not very attractive.

[Call_Me_Op]

Yow. Then you are agreeing with the wiki that a non-deductible IRA not converted to Roth is not very attractive.

Depends upon projected future tax rate. For the assumptions I made above, yes I agree.

[STC]

I have really enjoyed your recent posts. I like when people challenge the convectional with well supported arguments.

[sscritic]

I have really enjoyed your recent posts. I like when people challenge the convectional with well supported arguments.

I don't use the convection setting on my oven; I just use the traditional. Do you like your convectional oven?

[STC]

I have really enjoyed your recent posts. I like when people challenge the convectional with well supported arguments.

I don't use the convection setting on my oven; I just use the traditional. Do you like your convectional oven?

The pitfalls of the iPad and auto-correct. Lol. There are some really funny ones on the Internet. Give a google search if you want a laugh.

[Call_Me_Op]

I have really enjoyed your recent posts. I like when people challenge the convectional with well supported arguments.

Thanks STC. Appreciate your participation on the board.

[FNK]

One hole everybody noticed: a lot of confusion about tax deferred accounts.

Another hole: by paying taxes "from a separate cash account", you're effectively investing more. Should account for that.

Confusing (to me): is it really possible to get fractional bonds by reinvesting coupons?

[Call_Me_Op]

Confusing (to me): is it really possible to get fractional bonds by reinvesting coupons?

One can always use a fund to aid in rebalancing (to absorb the fraction).

[Random Walker]

This thread gives me some piece of mind. After reading all the BH books and being DIY for a few years, I ultimately went with an advisor. He put Treasuries in my taxable account and I've been questioning that all along. Thankfully I'm pretty sure my advisory firm knows more than me

Dave

[Bob's not my name]

This thread gives me some piece of mind. After reading all the BH books and being DIY for a few years, I ultimately went with an advisor. He put Treasuries in my taxable account and I've been questioning that all along. Thankfully I'm pretty sure my advisory firm knows more than me

Dave

That's a different question, addressed in this thread: viewtopic.php?f=10&t=106053 . For your peace of mind, the wiki supports the conventional argument that a non-deductible traditional IRA contribution may offer no advantage over taxable investing.

[halfnine]

Off the top of my head...

I think to have a fair analysis you'll have to look at what assets the treasuries would otherwise be replacing in your taxable account and the tax consequences of having those assets instead in your tax deferred account. For instance, those assets are now subject to income tax where otherwise they may have been subject to capital gains tax. Now, what I tend to do in these scenarios is not to assume a rate of return for those assets but mathematically find out what rate of return makes the two scenarios break even. Then gauge whether it appears more reasonable whether I would exceed that rate of return or not.

[Call_Me_Op]

This thread gives me some piece of mind. After reading all the BH books and being DIY for a few years, I ultimately went with an advisor. He put Treasuries in my taxable account and I've been questioning that all along. Thankfully I'm pretty sure my advisory firm knows more than me

Dave

That's a different question..

Different from what? It sounds exactly like the question addressed.

[Random Walker]

Call me op,
That's what I thought.

[Call_Me_Op]

Off the top of my head...

I think to have a fair analysis you'll have to look at what assets the treasuries would otherwise be replacing in your taxable account and the tax consequences of having those assets instead in your tax deferred account. For instance, those assets are now subject to income tax where otherwise they may have been subject to capital gains tax. Now, what I tend to do in these scenarios is not to assume a rate of return for those assets but mathematically find out what rate of return makes the two scenarios break even. Then gauge whether it appears more reasonable whether I would exceed that rate of return or not.

That's actually not the question in my case. I have a separate taxable account. (I keep the 401K separate for several reasons, one being convenience.) In any case, I do not have the option of holding treasuries in the bulk of my deferred space. If I want to hold them, they will be in taxable. The analysis was to see how the idea looks (from a tax-efficiency standpoint) for money that currently exists as after-tax cash. The analysis indicates it is a reasonable decision for the selected parameters.

Treasuries are uniquely attractive in portfolio construction, and provide strong diversification to equities and the other major asset classes.

[DSInvestor]

1.) Tax-deferred case: start $10,000......finish with $17,265.85

2.) Taxable case: start $10,000......finish with $17,980.53

I get slightly different numbers.

1) Tax Deferred case where 10K is initial IRA basis.
Value after 20 yrs = 10000 X 1.04^20 = 21,911.
Taxable amt of withdrawal = 21,911 - 10K basis = 11,911
Tax on withdrawal = 39% of 11,911= 4645
After Tax value = $17,265 (same as your number)

2) Taxable case.
4% after tax assuming 33% tax rate = 4% * 0.67 = 2.68%
Value after 20 years = 10000 X 1.0268^20 = 16,971 (lower than your number)

[Call_Me_Op]

1.) Tax-deferred case: start $10,000......finish with $17,265.85

2.) Taxable case: start $10,000......finish with $17,980.53

1) tax deferred case.
Value after 20 yrs = 10000 X 1.04^20 = 21,911.
Taxable amt of withdrawal = 11,911.
Tax on withdrawal = 39% of 11,911= 4645
After Tax value = 17,265 (same as your number)

2) Taxable case.
4% after tax assuming 33% tax rate = 4% * 0.67 = 2.68%
Value after 20 years = 16,971 (lower than your number)

Not correct. Your analysis assumes that bond interest is used to pay the taxes. I explicitly stated that taxes are being paid out of a separate cash account, and coupons are reinvested into new bonds.

[DSInvestor]

Not correct. Your analysis assumes that bond interest is used to pay the taxes.

My calculation uses the after tax return on 4% interest at 33%. How did you account for the other money in the separate cash account to pay the taxes? Did you start with 10K and some amount of extra cash to cover the tax over 20 years?

I get your number for taxable as follows:

Pretax value after 20 years = 10,000 X 1.04^20 = 21,911
Amount of interest received in 20 years = 21,911 - 10000 = 11,911
33% Fed Tax on interest = 3930
After tax value = 21,911 - 3930 = $17,980 (Same as your taxable number).

If this was the way it was calculated, I don't agree. This works for the tax deferred calculation because no taxes are due until withdrawal. In the taxable case, taxes are due every year and it has to come from somewhere. Your taxable account increases each year by the amount of interest received, but your other account will be reduced by the amount of tax paid from the other account.

[Call_Me_Op]

Not correct. Your analysis assumes that bond interest is used to pay the taxes.

My calculation uses the after tax return on 4% interest at 33%. How did you account for the other money in the separate cash account to pay the taxes? Did you start with 10K and some amount of extra cash to cover the tax over 20 years?

Correct. I assumed that the investor has a separate cash allocation, from which he or she pays taxes. (I have always done it this way.) I also assumed (simplistically but realistically) that the cash account yields zero. Of course, as you have shown, a separate cash account is a key assumption.

[Epsilon Delta]

Call_Me_Op:

You are assuming that 20 year zeros are yielding 4% while all shorter zeros (19 years and less) will yield 0%. This is a very strange yield curve.

[halfnine]

Off the top of my head...

I think to have a fair analysis you'll have to look at what assets the treasuries would otherwise be replacing in your taxable account and the tax consequences of having those assets instead in your tax deferred account. For instance, those assets are now subject to income tax where otherwise they may have been subject to capital gains tax. Now, what I tend to do in these scenarios is not to assume a rate of return for those assets but mathematically find out what rate of return makes the two scenarios break even. Then gauge whether it appears more reasonable whether I would exceed that rate of return or not.

That's actually not the question in my case. I have a separate taxable account. (I keep the 401K separate for several reasons, one being convenience.) In any case, I do not have the option of holding treasuries in the bulk of my deferred space. If I want to hold them, they will be in taxable.

Treasuries are uniquely attractive in portfolio construction, and provide strong diversification to equities and the other major asset classes.

If you can't put the treasuries in tax deferred then the comparison is irrelevant. Because, if you can put them in tax deferred but otherwise choose not to then one would have to look at what I indicated above to make a reasonable assessment.

Along a similar line of reasoning if you put your treasuries in taxable it is taking taxable space from another asset class. So you are either changing your asset allocation to accommodate this or if you are not changing your asset allocation then to be fair you have to take into account what changed in the rest of you portfolio and the tax consequences for that.

The analysis was to see how the idea looks (from a tax-efficiency standpoint) for money that currently exists as after-tax cash. The analysis indicates it is a reasonable decision for the selected parameters.

One can certainly look at it from that standpoint. But, looking at an asset in isolation from a tax-efficiency standpoint isn't much different than looking at an asset in isolation. One needs to look at the tax efficiency of the whole portfolio. The treasuries are replacing something.

[Call_Me_Op]

Call_Me_Op:

You are assuming that 20 year zeros are yielding 4% while all shorter zeros (19 years and less) will yield 0%. This is a very strange yield curve.

Not quite. All re-invested dividends are earning 4% yield in my model. I did collapse the yield curve at zero years for simplicity (that is, the cash account yields zero).

[Call_Me_Op]

One can certainly look at it from that standpoint. But, looking at an asset in isolation from a tax-efficiency standpoint isn't much different than looking at an asset in isolation. One needs to look at the tax efficiency of the whole portfolio. The treasuries are replacing something.

No - the treasuries are not replacing anything. They are without substitute. And tax-efficiency is only one factor considered - and not the most important by the way.

If I could take my cash and buy treasuries in a tax-deferred account (e.g., traditional IRA), the analysis says there is no real benefit. That was the main point of the exercise. I was addressing the common myth that treasuries should not be placed in a taxable account. The real answer is - it depends.

[Bob's not my name]

common myth that treasuries should not be placed in a taxable account

citation needed

[Call_Me_Op]

common myth that treasuries should not be placed in a taxable account

citation needed

I don't want to search through all of the Boglehead posts that refer to treasuries as tax inefficient or just refer to bonds in general (except munis) as tax inefficient. Maybe using "common myth" somewhat overstates it, but I do not believe it is widely appreciated that treasuries are quite acceptable (maybe even preferred) to be located in a taxable account.

[Epsilon Delta]

Call_Me_Op:

You are assuming that 20 year zeros are yielding 4% while all shorter zeros (19 years and less) will yield 0%. This is a very strange yield curve.

Not quite. All re-invested dividends are earning 4% yield in my model. I did collapse the yield curve at zero years for simplicity (that is, the cash account yields zero).

So your assuming you could invest the money you did not paid in taxes at 4%, but you choose to have it earn 0%. Why does it surprise you that this less than optimal?

[jdilla1107]

Not correct. Your analysis assumes that bond interest is used to pay the taxes. I explicitly stated that taxes are being paid out of a separate cash account, and coupons are reinvested into new bonds.

I don't get it. How does this make any sense? Does this mean I can earn 100% return by moving money from account A to account B? You are comparing after tax returns to before tax returns? What if you used the money you are using to pay taxes and put it into treasuries?

[grabiner]

I am interested in holding individual treasuries in my taxable account, so I performed a spreadsheet analysis to assess tax efficiency. What I compared are the following two scenarios over a time period of 20 years:

1.) Place 10 treasuries with 4% coupon into tax-deferred account, reinvest dividends, and pay taxes after 20 years at a combined federal plus state tax rate of 33% + 6% = 39%.

What you may have missed here is that you got a tax deduction by putting the Treasuries in a tax-deferred account; if you are in the same tax bracket now as when you withdraw, the effective tax cost of anything in a tax-deferred account is zero.

However, you might already have the tax-deferred account, and be comparing the effect of holding Treasuries to holding stocks. In that case, there is an error in your math, as you pay tax on the full account value, not just on the gains (unless you use a non-deductible IRA).

Also, the tax rate should be 37%; you can deduct the 6% state tax from federal tax.

2.) Place 10 treasuries with 4% coupon into taxable account, reinvest dividends, and pay taxes each year OUT OF SEPARATE CASH ACCOUNT. Tax rate is federal only, at 33%. (Treasuries are state-tax exempt.)

This is also unreasonable, because you won't be holding extra money in a cash account now in order to pay the taxes you will owe in 19 years. A more reasonable assumption is that you have a Treasury fund earning 4%, and you reinvest the after-tax dividends.

So, here are my numbers.

If you put $10,000 in Treasuries in a tax-deferred account, you will have $21,911 after 20 years, and when you pay 37% tax, you will be left with $13,804. If you put $15,873 in Treasuries in a tax-deferred account, paying $10,000 out of pocket, you will have $21,911, just as if you put $10,000 in Treasuries in a Roth IRA.

If you put $10,000 in Treasuries in a non-deductible IRA, you will pay 37% tax on the $11,911 gain, and you will have $17,504 after tax.

If you put $10,000 in Treasuries in a taxable account, you will earn 2.68% per year after tax, for a final value of $16,971.

Thus you get a very slight gain from putting Treasuries in a non-deductible IRA, and a much larger gain from putting them in a tax-deferred account.

But tax efficiency is relative. For comparison, consider a stock index fund which earns 2% in dividends and 6% in unrealized gains, and assume you pay 19% tax (15% federal + 4% state after deduction) on dividends and capital gains.

If you put $10,000 in stock in a tax-deferred account, you will have $46,610 before tax, which is $29,364 after tax. (Again, if you put $10,000 in a Roth IRA, or $10,000 out of pocket for $15,873 nominal in a tax-deferred account, you will have $46,610 after tax.)

If you put $10,000 in stock in a non-deductible IRA, you will pay tax on the $36,610 gain, and have $33,034 after tax.

If you put $10,000 in stock in a taxable account, the account will grow at 7.62% (with 0.38% lost to taxes), so it will be worth $43,437. You will have $7109 of reinvested divdends, so you will owe tax on a capital gain of $26,328, for a final value of $38,434.

Thus putting stock into a non-deductible IRA causes you to lose value, and putting it in taxable rather than tax-deferred is a gain of $9070. Another way to look at it is that the stock returns 6.96% in taxable versus 8% tax-free, and the bonds return 2.68% in taxable versus 4% tax-free; thus the stock loses less of its return if you put it in taxable.

[Bob's not my name]

common myth that treasuries should not be placed in a taxable account

citation needed

I don't want to search through all of the Boglehead posts that refer to treasuries as tax inefficient or just refer to bonds in general (except munis) as tax inefficient. Maybe using "common myth" somewhat overstates it, but I do not believe it is widely appreciated that treasuries are quite acceptable (maybe even preferred) to be located in a taxable account.

Again, you are mixing issues. The "common myth" is that bonds are tax inefficient relative to stocks, and therefore it is better to place your bonds in your pre-tax 401k or TIRA or post-tax Roth IRA and stocks in taxable. But, as I said,

That's a different question, addressed in this thread: viewtopic.php?f=10&t=106053 .

I believe it is not at all common for bogleheads to advise opening a non-deductible IRA (neither post-tax nor pre-tax, but merely a way station) that will not be converted to Roth to hold any type of investment. But I found some examples for you:

You are close to retirement. If it was me, I would simply use my taxable account.

One rule of thumb:

For bonds, non-deductible traditional IRA is better than taxable.
For stocks, taxable is better than non-deductible traditional IRA.

In my opinion, there are two reasons to use a non-deductible traditional IRA.

-as the first step of a back door contribution to Roth IRA

-if you can't do back door, but do need more space for bonds or REIT

Only if you do a back-door Roth now or in the near future with the contribution. That is, you convert to a Roth IRA almost right away.

it may be better to avoid the non-deductible contributions and invest outside of tax advantaged accounts instead

What you will generally find in these threads is the comment that a non-deductible IRA is not a bad idea because there may be an opportunity to convert to Roth in a low bracket in the future. What you also generally find is a general statement on bonds, rather than Treasurys specifically. Since Treasurys are exempt from state tax, they are more tax-efficient than corporates. Speaking of which, besides your assumptions of a cash side-car and a high tax bracket in retirement, I think another weakness of your model is assuming that state tax will apply to IRA withdrawals. Many states exempt some or all IRA withdrawals from taxation. Examples (corrections welcome, since obviously I'm not a taxpayer in all these states):

• Illinois, Pennsylvania, New Jersey, Mississippi, and Tennessee exempt all IRA withdrawals in retirement (and some before retirement)
• Arkansas, Colorado, Delaware, Georgia, Iowa, Kentucky, Michigan, New York, North Carolina, South Carolina, and Oklahoma exempt IRA withdrawals subject to certain age limits (some fairly young) and/or dollar limits (some very high).
• Alaska, Florida, Nevada, New Hampshire, South Dakota, Texas, Washington, and Wyoming have no state income tax.

[Call_Me_Op]

My state does not exempt IRA withdrawals. And yes, there are certain assumptions made in the analysis - all clearly stated so that someone can decide whether they are applicable in their individual case.

I don't think anyone knows what their tax rate will be in retirement. Given that tax rates are generally heading up and not down, I believe assuming your current tax rate for retirement is reasonable.

When people use the generic term bonds, they generally include treasury bonds. In fact, "Total Bond" holds a large portion in treasuries. Therefore, I do not think it is a stretch to claim that many wrongly believe that treasury bonds are tax inefficient. Hence this post.

[Bob's not my name]

there are certain assumptions made in the analysis - all clearly stated

The fundamental problem with your thread is that your assumptions are not clearly stated. This is why it continues to be misunderstood. I suggest you edit the OP to state that you are considering whether to make a non-deductible contribution to a traditional IRA that will not be converted to a Roth.

I don't think anyone knows what their tax rate will be in retirement. Given that tax rates are generally heading up and not down, I believe assuming your current tax rate for retirement is reasonable.

As is discussed frequently here, a general trending upward of tax rates does not mean that an individual's tax rate in retirement will be the same or higher vs. in his peak earning years. On the contrary, your tax rate in retirement is typically lower, especially in early retirement and late retirement.

[Call_Me_Op]

there are certain assumptions made in the analysis - all clearly stated

The fundamental problem with your thread is that your assumptions are not clearly stated. This is why it continues to be misunderstood. I suggest you edit the OP to state that you are considering whether to make a non-deductible contribution to a traditional IRA that will not be converted to a Roth.

I don't think anyone knows what their tax rate will be in retirement. Given that tax rates are generally heading up and not down, I believe assuming your current tax rate for retirement is reasonable.

As is discussed frequently here, a general trending upward of tax rates does not mean that an individual's tax rate in retirement will be the same or higher vs. in his peak earning years. On the contrary, your tax rate in retirement is typically lower, especially in early retirement and late retirement.

On your first point, I stated that the difference in the two cases is tax-deferral versus pay taxes as you go out of a separate cash account. That completely describes the set of assumptions (along with the numerical tax rates). This can apply to a non-deductible IRA, but that is not the only applicable vehicle.

On the second point, nobody knows their tax rate in retirement (assuming they are not already there). Someone who believes they know can easily run a similar analysis and plug-in the assumed rate to see if it changes the basic conclusion. I think you are assuming that I set-out to show more than was my intent. I was giving an example - with parameters that apply to my situation and may apply to others - not claim that this applies in all cases.

[Bob's not my name]

My state does not exempt IRA withdrawals.

But a majority do. If your model applies to a minority of states and the minority of taxpayers who see their tax bracket go up when their income goes down and the minority of investors who are ineligible for a back door Roth IRA, you have compounded minorities. An example based on compounded minorities does not rebut a "common myth" or "conventional wisdom," even if such existed, which I dispute.

[Call_Me_Op]

My state does not exempt IRA withdrawals.

But a majority do. If your model applies to a minority of states and the minority of taxpayers who see their tax bracket go up when their income goes down and the minority of investors who are ineligible for a back door Roth IRA, you have compounded minorities. An example based on compounded minorities does not rebut a "common myth" or "conventional wisdom," even if such existed, which I dispute.

Again, this was an analysis based primarily on my situation. I am in a state that taxes IRA distributions (even though we are not necessarily talking about an IRA). In addition, my income in retirement may be comparable - and as I stated tax rates are likely to rise in the foreseeable future.

As far as the "conventional wisdom", we may have to agree to disagree on that.

[grabiner]

I think another weakness of your model is assuming that state tax will apply to IRA withdrawals. Many states exempt some or all IRA withdrawals from taxation. Examples (corrections welcome, since obviously I'm not a taxpayer in all these states):

• Illinois, Pennsylvania, New Jersey, Mississippi, and Tennessee exempt all IRA withdrawals in retirement (and some before retirement)
• Arkansas, Colorado, Delaware, Georgia, Iowa, Kentucky, Michigan, New York, North Carolina, South Carolina, and Oklahoma exempt IRA withdrawals subject to certain age limits (some fairly young) and/or dollar limits (some very high).
• Alaska, Florida, Nevada, New Hampshire, South Dakota, Texas, Washington, and Wyoming have no state income tax.

The Wiki article State income taxes has date for some states. From the states on your list:

Michigan's exemption is of very limited value for high-income taxpayers born after 1952; taxpayers at least 67 can exclude either a personal exemption plus Social Security, or $20,000 of retirement income not indexed to inflation. (It might be useful for three years if you take IRA withdrawals at 67 and wait for Social Security at 70.)

New Jersey has only a limited exemption, $15,000 single/$20,000 joint for taxpayers over 62 making less than $100,000. However, NJ taxes all IRAs as non-deductible, and does not tax Treasury bonds even in IRAs.

Individual Income Tax Provisions in the States summarizes tax rules for all states, but it doesn't identify states which tax IRAs differently from 401(k)s and pensions. (In Maryland, for example, IRAs are fully taxed, while pensions and 401(k)s are exempt up to a limit.)

[afan]

Forgive me if I have missed it, but I think the flaw in the original analysis lies in the source of the funds used to pay taxes.

For the tax deferred investment , the OP appropriately deducts the taxes from the total return. However, for the taxable investment taxes are paid with other funds. To make a valid comparison, one would have to pay taxes on the deferred account the same way, from this separate account. Then compare the final values including deducting taxes paid no matter the source of the funds used to pay them. As it is, the original example acknowledged that taxes would be due on the taxable account, but then compared final values as if this were not the case

A simple way to see the effect would be to ADD to the total for the deferred account the amount paid annually in taxes for the taxable account. Even ignoring the time value of money, inappropriate when talking a 20 year horizon, one gets very different results.

[Call_Me_Op]

I disagree. The analysis determines the net gain in both scenarios. It accounts for the loss associated with paying the taxes. If the taxes are paid from cash in the tax-deferred case, the results are the same.

[FNK]

Op,

If everyone is confused by your presentation, arguing that it's crystal clear to you is not going to advance the argument. Maybe you want to restate your analysis more clearly.

[JamesSFO]

Forgive me if I have missed it, but I think the flaw in the original analysis lies in the source of the funds used to pay taxes.

For the tax deferred investment , the OP appropriately deducts the taxes from the total return. However, for the taxable investment taxes are paid with other funds. To make a valid comparison, one would have to pay taxes on the deferred account the same way, from this separate account. Then compare the final values including deducting taxes paid no matter the source of the funds used to pay them. As it is, the original example acknowledged that taxes would be due on the taxable account, but then compared final values as if this were not the case

A simple way to see the effect would be to ADD to the total for the deferred account the amount paid annually in taxes for the taxable account. Even ignoring the time value of money, inappropriate when talking a 20 year horizon, one gets very different results.

I tend to concur, the biggest flaw is the source for the tax payment that the op keeps handwaving away. If it takes $X to pay the taxes, that $X is still around in the tax-deferred case, so instead of $17,265 vs. $17,890, op should show $17,265+X and X is going to be more than the $700 spread.

[Call_Me_Op]

I am not sure why there is so much confusion about this. Let's say I have $10,000 and I want to decide whether to put it in a tax-deferred account (say an after-tax IRA) or keep it in taxable. In both cases, I plan to invest in a 4% treasury bond. And, to eliminate the apparent source of confusion, in BOTH cases I will pay taxes out of a cash account that is yielding nothing. I will reinvest at 4% for simplicity.

1.) In the tax-deferred case, before I pay taxes (at 39%), the sum grows to $10,000*(1+.04)^20 = $21,911.23. After I pay taxes (from cash), I am left with $17,265.85. That is my net gain.

2.) In the taxable case, the total tax on interest is $3930.71 (you can check this yourself, using 33%). This is paid as we go along - but out of the cash account. So the amount in the brokerage account still grows to $21,911.23. After taxes are netted-out, the net gain is $21,911.23 - $3,930.71 = $17,980.52.

Thus, my net gain is greater in the taxable case. In both cases, the amount paid in taxes (all out of cash) has been subtracted from the accumulated amount to result in the net gain. The only difference between these two cases is that in the deferred case, taxes are all paid at the end. In the taxable case, the taxes are paid as we go along, but the taxes total to a smaller sum because the bond interest is state tax exempt.

The key here is I am allowing the investment interest to compound in both cases. So the only difference is what is paid in taxes, which must be smaller in the taxable case due to the tax exemption.

[FNK]

OK, replace "total interest" in case 2 with "total taxes" and you're golden.

So you've discovered that, all other things being totally equal, 33% tax is better than 39% tax.

That's a good point, I guess.

The problem, of course, is that in equalizing other things you've thrown away all advantages of tax-advantaged accounts.

[Call_Me_Op]

OK, replace "total interest" in case 2 with "total taxes" and you're golden.

So you've discovered that, all other things being totally equal, 33% tax is better than 39% tax.

That's a good point, I guess.

The problem, of course, is that in equalizing other things you've thrown away all advantages of tax-advantaged accounts.

More importantly, I have shown that if you pay out of cash, tax-deferral in and of itself is essentially useless. (There is a minor issue of interest on the cash, which I eliminated by eliminated cash interest rate.) All that really matters is the relative tax rates. If you feel very confident that your tax rate in retirement will be lower than during your working years, tax deferral may make sense.

As far as throwing way all the advantages of tax-deferred accounts, that's the whole point. There is no fundamental advantage (not counting pre-tax contribution) unless your tax rate in retirement will be lower - with the exception of earning interest on the cash issue. Now if there were no cash from which to draw, the analysis would look different.