Friday, February 25, 2005

The post that gave my blog its name...

(Part 5 of 5)

I’ve got a great idea. The government can borrow money really cheaply, right? It can issue bonds backed by the “full faith and credit of the US Government”. Rating agencies classify its bonds as “riskless”. The 14th amendment even makes it unconstitutional to question the validity of the public debt of the United States.

So, I propose that the government issue a bunch of bonds, about a trillion dollars or so, and dump that money into the stock market. There is a good spread between the yield on government bonds and the return of the stock market. This spread has historically been around 5%. This spread would result in a nice income of around $50 billion per year to the government. We couldn’t lose.

OK, so it might not be the best idea. The stock market might collapse, leaving the government owing its creditors a trillion dollars and not having the assets that were supposed to back that loan. Further, governments directly owning stocks – and thus businesses – is a bad idea; just ask the people who lived in the Soviet Union.

So, maybe it’s a bad idea for the government to borrow a trillion dollars and put it into the stock market. Too bad that the government has already done exactly that.

Look back at Ted up there, with his traditional IRA. By putting $1000 into it, he deferred $250 in taxes. That means that the government had to borrow an extra $250 that year. However, in return, the government got a claim on 25% of any profits that Ted makes in his IRA. If Ted waits until he’s 59.5 and then cashes out his IRA, then the government gets $250 * g. Across a hundred million taxpayers, g probably represents something pretty close to stock market growth. However, that $250 that the government borrowed had to be paid off at government-bond rates. By deferring $250 in taxes, the government got to borrow money at government-bond rates and invest it at stock-market-return rates.

As of 2005, IRAs, 401(k)s, and similar tax-deferred plans have about $5 trillion in them. That represents about $1.25 trillion in assets that the government has a claim on when they’re cashed out. If that $1.25 trillion is growing at 5% faster than the bonds that the government had to issue when the taxes were deferred, that represents $62.5 billion of appreciation every year. Even if Ted does manage to drop into the 15% tax bracket a few decades later, the spread will more than make up for the difference.

The government has managed to successfully borrow over a trillion dollars and invest it in over a hundred million little mutual funds, each of them managed by an ordinary citizen of the United States. Those citizens invest that money as if it was their own, and the government gets to keep the stock market rates-of-return that are generated, but without the bread lines, gulags, Great-Leaps-Forward, and similar things that usually happen when the government gets too directly involved in the economy.

Not only is it growing, it’s a cushion that the government has if it ever needs to collect some extra tax revenue. If it wanted to, the government could simply declare that all IRAs and 401(k)s are being converted into Roth IRAs, and that all of these rollovers will be taxed at 25%. This would result in a one-time windfall of about $1.25 trillion in revenue, though the effects on the stock market would be…interesting. Because most people know that they’ll wind up paying taxes on their withdrawals in the end anyway, an announcement like this would probably only result in a minor outcry. Drop the tax percentage to 20% and you would probably make people happy, and still pull in a trillion extra dollars. In early 2005, George W. Bush proposed a 2.5 trillion dollar budget. It would appear that the federal government is following the 6-months-of-savings rule that any financial advisor will tell you is a good idea.

So, what public-policy implications does all of this have? The government is investing in over a hundred million little mutual funds, each managed by an ordinary citizen of the United States. Some people will handle their money well, others poorly. To maximize the spread, we should try to get as much tax-deferred money into the accounts of people who are good at growing their own money, and discourage less-capable stock-pickers from having IRAs.

One possible way to do that is to increase your ceiling on 401(k) and IRA contribution limits by a factor that is related to the amount of capital gains you had that year. This would have the effect that people with lots of capital gains would accumulate larger IRAs. To the extent that past performance predicts future results, the return in IRAs would slowly creep up as the good investors were allowed to have larger balances. Growing the spread by .5% would earn the government an additional $6 billion per year.

If the IRA contribution limit were raised from $3000 to ($3000 + net capital gains), then that would result in people who had capital gains putting more money into their IRAs. Right now, capital gains are taxed at 15%, so every $1000 in capital gains I earn would results in me owing $150 in taxes. If I could turn around and put $1000 into my IRA and deduct it from my normal inncome, I would be able to defer $250 in regular income taxes. The government would actually have a net tax loss (this year) of $100, but would in return have $250 growing at stock market rates.

This looks suspiciously like a subsidy to the wealthy, so getting it through Congress might be challenging. The editorials pretty much write themselves: “The government is actually PAYING the wealthy to have capital gains! When the rich earn money on their stocks, not only do they not pay the taxes due on them, they get 10% of the gain back from the taxpayers!”

Another thing to do is to make sure that people who are already in a high tax bracket don’t contribute to IRAs. Deferring 35% taxes now in return for 15% taxes later requires decades of a healthy spread (almost 18 years at 5% just to break even) before it’s a good idea. This is already done with phase-out limits on IRAs, though the rationale (at least publicly) is more along the lines of “keep the rich from having too many tax breaks”. Combining this with the previous suggestion of allowing capital-gains-earners to put more money in their IRAs might also be challenging.

Finally, I believe that we should eliminate the Roth IRA. It's almost never the best choice for taxpayers. It sucks money out of the economy, and it deprives the government of future revenue, with no corresponding benefit to taxpayers or anyone else.

Roth IRAs are bad public policy

(Part 4 of 5)

So, having established that there is no significant benefit to having a Roth IRA for an individual, is there any public-policy benefit to them?

They look tax-free, which might encourage people to save more than they might otherwise, and that’s usually good.

On the other hand, the taxes in a Roth IRA have been paid up-front, and if the owner holds the account until retirement, no further taxes will be paid on the balance. This has the effect of creating a large group of people who don’t care what happens to tax rates. This is not a good recipe for sound fiscal policy. Not only that, but I fear that because Roth IRAs are only the right thing to do when tax rates go up, people who have them will subconciously want that to happen.

A final, even more important note: Let’s consider Ted, Neal, and Ralph again.

Ted gets his $1000 and puts it into his Traditional IRA. There’s now an additional $1000 out in the economy, funding businesses, being loaned to homeowners, or doing whatever else savings do. Meanwhile, Ralph and Neal each give $250 to the government and put $750 into their accounts. There’s now only $750 out working in the economy while the government spends $250 on [pick your favorite government-sponsored boondoggle]. Multiply this by over a hundred million taxpayers, and this leads to slower economic growth. Roth IRAs first appeared in the late nineties, right before the economic slowdown of the early 21st century.

There is only one significant advantage to Roth IRAs...

(Part 3 of 5)

...and that is the ability to take the principal out early without penalty. If you want to cash out the entire thing, you are better off with a traditional or non-deferred account. However, if you only take part of the cash, the principal, a Roth is the best choice.

Consider Ted, Neal, and Ralph again. Imagine that they each make their $1000 contributions, so that T=1000, N=R=750. After a while, the amount in their accounts has doubled, and they decide they want $750 in their hands.

Ralph has $1500, removes $750, and still has $750 there. This is not a taxable event for him.

Ted has $2000. To have $750 in cash, he actually needs to withdraw ($750 / 65%), or $1154, leaving $846 in his account.

Neal has $1500. Half of whatever he withdraws will be capital-gains profit, taxed at 15%. To get $750 in hand, he has to cash out $375 in principal and $375 * 1.15 to cover the capital gains -- a total withdrawal of $806, leaving $694 in his account.

Assume that the remainder for each man doubles before they hit 59.5 years old. Ted will have a total balance of $1683, or a withdrawable balance of $1262. Ralph will have a balance of $1500, all withdrawable tax-free. Neal will have a balance of $1388, or a withdrawable balance of $1236.


  • Ted: $1262
  • Ralph: $1500
  • Neal: $1236

This assumes contribution, g = 2, withdrawal, and g = 2 again. If the two g-values are different, it can lead to wildly varying results.

In the situation where part of the money will be withdrawn early, a Roth IRA can be the best choice.

Thursday, February 24, 2005

For an individual, a tax-deferred account is at best only slightly better than non-deferred account, and a Roth IRA is almost never the best choice.

(Part 2 of 5)

In the United States today, there are three basic ways to save money.

Let’s imagine three people, call them Ted, Ralph, and Neal. Each of them has a $1000 bonus coming from his boss and decides to save it for retirement. Ted puts his $1000 into a traditional IRA, Ralph puts his into a Roth IRA, and Neal puts his in a non-tax-deferred brokerage account. To make this a bit simpler, I’m going to ignore state taxes, and assume that all three of them are in the 25% tax bracket. There is one more important rule:

Money placed into any of the three types of accounts will grow at the same rate; $1 placed into any of the three accounts will, after a certain amount of time, have grown to exactly the same amount (before any taxes and withdrawal penalties).

I will call this growth factor ‘g’. For example, if the underlying investments return 10% anually, then after one year, g=1.1, and after ten years, g=2.59.

Tax rates are unstable. Every few years, they move up or down a bit. Nobody knows what personal income taxes will be like 10, 20, or 50 years from now. For this analysis, I’m going to assume that they will be the same when our three guys hit age 59.5 as they are now.

Let’s start with Ted, who puts his money into a Traditional IRA (I’m going to ignore 401(k) accounts for the rest of this article. For tax purposes, they are almost identical to traditional IRAs.) If we imagine that his boss can put the money straight into his IRA, we can ignore the withholding and later tax refund, which cancel out. Over time, the balance printed on his monthly account statement is $1000 * g. However, if he wants to access his money after retirement, he has to pay 25% taxes on it, so the final withdrawable balance at retirement will be $1000 * g * 75%. If Ted withdraws the money early, there will be an additional 10% penalty, and the withdrawable balance will be $1000 * g * 65%.

Let’s move on to Ralph. He decides to put his money into a Roth IRA. Because Roth IRAs are funded with post-tax dollars, Ralph’s extra $1000 in income results in him having an extra $250 in tax liabilities, giving him only $750. At first glance, he seems to be at a severe disadvantage to Ted, because the balance printed on Ralph’s account statement is $750 * g. However, Ralph can access his original $750 tax-free. If he decides to access the entire account, the amount above $750 is considered taxable income. Further, if he accesses it in the first five years, there’s an additional 10% penalty. However, at retirement, the entire balance of the account will be withdrawable completely tax-free. Unfortunately, the various penalties make the formulas more complicated, but, Ralph’s withdrawable balance looks something like this:

Profit = 750 * (g – 1)
Withdrawable balance in the first five years = 750 + (Profit * 65%)
Withdrawable balance after five years = 750 + (Profit * 75%)
Withdrawable balance at retirement = 750 * g

The interesting thing to note is that Ralph’s withdrawable balance at retirement, (750 * g) is exactly equal to Ted’s, which is $1000 * g * 75%. At retirement, if Fred’s tax bracket at withdrawal is equal to Ralph’s tax bracket was when they received the $1000, then the actual value of their accounts will always be identical. The only difference between a Roth IRA and a traditional IRA is that the taxes are levied at the front instead of at the back. With no change in tax rates, the actual realizable value from the two accounts is identical.

Meanwhile, what happened to Neal? He doesn’t trust fancy accounting shenanigans, so he just puts his money into a non-tax-deferred account. Like Ralph, he can only fund his account with $750. Also like Ralph, the original $750 can be taken out tax-free. However, Neal has some important differences when compared to Ted and Ralph. There will never be any special penalties on Neal’s withdrawals. Nothing special happens to the true value of his account at retirement. Finally, the big advantage: Gains in Neal’s account can be taxed at the much-lower capital-gains rates instead of as regular income. If Neal buys stock with his $750, then his gains won’t be taxable until he sells that stock. Further, as of 2005, capital gains from any stock held more than a year, as well as all dividends, are taxed at only 15%. Because of this, the best option for Neal is to buy a non-dividend-paying stock and hold it until he is ready to withdraw the full balance. If he is able to do that, his numbers look like this:

Profit = 750 * (g – 1)
Withdrawable balance in the first year: 750 + (profit * 75%)
Withdrawable balance after one year: 750 + (profit * 85%)

Neal also has one other advantage: If g is negative, meaning that the underlying investments have lost value, then Neal gets to take that loss as a tax deduction. Neither Ted nor Ralph get to take losses in his account as a deduction.

Maximum Withdrawable Balance, Profit = ($750 * (g-1))

Ted (Traditional IRA)Ralph(Roth IRA)Neal (Non-Deferred)
After one year:$1000 * g * 65%750 + (Profit * 65%)750 + (Profit * 85%)
After five years:$1000 * g * 65%750 + (Profit * 75%)750 + (Profit * 85%)
After age 59.5:$1000 * g * 75%
(750 * g)
750 +(Profit * 100%)
750 + (Profit * 85%)

Again, this table assumes that tax rates will be stable, and that all three workers will stay in the same tax bracket, from deposit time to withdrawal time.

So, what does this table tell us? As noted above, once they hit age 59.5, Ted and Ralph have exactly the same withdrawable amount. However, I should note that Ralph and Neal have the advantage of being able to take the whole thing at once, whereas if Ted tries to take too much at once, it might push him into a higher tax bracket.

The next thing to notice is that Neal’s withdrawable balance is always higher than Ralph’s up until retirement, and always lower afterwards.

To recap: Before retirement, a non-tax-deferred account is better than a Roth IRA, and after retirement, there is no benefit to a Roth IRA over a Traditional IRA. The only time that a Roth IRA is better than the others is if tax law changes, and the withdrawals from the other accounts become more expensive. Assuming static tax rates, there is no situation where a Roth IRA will have the highest actual value of the three account types.

Finally, if g is high enough, Ted will have more money than Neal, even if they withdraw their money before retirement. With 25% tax rates, the crossover point comes at g=9, which will happen after 25 years at an 8% growth rate. If both Ted and Neal invest in an investment that will be worth 10x what it was before, then Ted’s higher starting position will cancel out his higher taxes, even before retirement. However, the difference between the accounts is very small. If (before retirement) g = 3, then Neal’s withdrawable balance will be less than 4% higher than Ted’s. On the other hand, if g=20, then Ted’s withdrawable balance will be barely 1% higher than Neal’s.

So, assuming reasonable investment growth rates and static tax rates:

  • Before retirement, The withdrawable balance in a non-deferred account will be very close to the withdrawable balance in a traditional IRA; N ~= T

  • Before retirement, the withdrawable balance in a Roth IRA will always be lower than the withdrawable balance in a non-deferred account; R < T

  • At retirement, the withdrawable balance in a traditional IRA and a Roth IRA will be identical; T = R

  • At retirement, the withdrawable balance in a Roth IRA will always be higher than the withdrawable balance in a non-deferred account. R > N

Or, again assuming static tax rates and reasonable growth:
  • Before retirement, R < T and T ~= N
  • After retirement, N < R and R = T

Unless tax rates go up in the future, a Roth IRA is almost never the best choice. There is one very important exception, which I'll discuss in my next post.

The way we evaluate tax-deferred accounts in fundamentally flawed

(Part 1 of 5)

When using Microsoft Money to tally my net worth, it adds up my asset accounts, subtracts my credit accounts, and gives me a total. This is an unrealistic way to figure the value of any non-liquid asset; my house might be worth $120,000, but (even without a mortgage) I could not easily convert it to $120,000 sitting in my checking account.

Tax-deferred accounts, even if they’re filled with nothing but cash, are similarly non-liquid. While I can cash out an IRA or 401(k) on a few days’ notice if I really want to, doing so is very expensive. The bank who holds it is required to withhold 20% of any withdrawals. If I’ve got a balance of $1000, the largest (unrestricted) check they can give me is $800. Since I’m a middle-class 25%-rate taxpayer, I will incur an additional $50 liability to the federal government and $78 to my state (Idaho) government. Since I’m not yet 59.5 years old, there will also be a $100 early-withdrawal penalty. That means that while the balance printed on my IRA statement is $1000, I can really only convert it to $572 in spendable cash.

I’m not planning to access the money immediately. However, when I do access it, I hope to be earning enough income that I’m still at least in the 15% tax bracket. As of 2005, that means having less than $29,050 in taxable income. Even if I have stay below that (inflation-adjusted) boundary during retirement, I’ll still have to pay at least 22.8% of any withdrawals to one layer or another of government.

The true value of an IRA, 401(k), or similar tax-deferred account is the amount of cash that can be extracted from it. For an Idahoan like myself, that is somewhere between 57.2% and 77.2% of the balance printed on the monthly statement. Unless you’re planning to be significantly poorer in retirement than you are during your working years, it’s not really fair to say that you get to “keep” your money by putting it in an IRA. At best, you get to hold onto it for a little longer before giving it to the government.

Going back to my $1000 above, if my IRA doesn’t grow, and I am in the 25% tax bracket at retirement, then putting money into my IRA means that I trade $672 now for $672 later. The benefits of retirement accounts only appear when the money in them grows, or tax brackets change. Even then, those benefits are not as big as they might seem.

First Post!

Welcome to my blog. This blog focuses on economics, politics, and personal finance, plus whatever else strikes my fancy.