Problems with using MaxiFi

As I was trying to use MaxiFi, I found there were problems with what it was trying to do. Maybe it cannot do what I want, or I am expecting the wrong result. But independent of that, there are problems I have with the way that it interacts with the user. This post is not addressing the content of MaxiFi per se, but rather how I interact with it.

For example, MaxiFi wants to know about my family. Specifically, my children. For each child, it wants to know its birthdate, so it can compute its age, and when it will graduate from high school, and off to college, and so on. You can add, or delete, children fairly easily. But MaxiFi doesn’t ask for a name for the child, it just labels it by its year of birth: “Child born 1994” and “Child born 1996” — you can’t give them names! I don’t know what happens if you have twins, or even just two children in the same year. (But I can add and delete children; let’s try! Answer: you just get two children labelled the same “Child born 1994”. Apparently children are all anonymous and interchangeable.) MaxiFi should allow children to have a name.

When entering data, most of the information is either a date or a number. Dates have to be in the form MM/DD/YYYY, even tho the slashes provide no information; you could give the same information as MMDDYYY. Numbers, on the other hand, must be purely numeric, you can’t have a “$” or “,”. In fact it specifically checks for these and instead of just ignoring them, has a special error message to complain about them and make you take them out. Input of dates and numbers should be more flexible.

When you are entering a date, during the actual time that you are typing it in, it is of course not in the proper format — not until you are all done entering it. But MaxiFi complains with every keystroke that this is an invalid entry. MaxiFi should not complain while data is being entered, only when it is complete.

When you enter a number, the field has been defaulted to zero. Unlike most things that erase the default when you start giving a value, MaxiFi just leaves the zero there. If you are trying to enter “500”, you get either “0500” or “5000” depending on which side of the default zero you enter your number. So you have to explicitly delete the 0 default. If you do that at the beginning, before typing your response, MaxiFi complains “Value is required” because, at least for the moment, the field is empty. MaxiFi should distinguish between “default” input and erase it when the user starts to fill the field.

There is no “undo” capability. If you do something by mistake, you have to figure out how to fix it. MaxiFi should have at least an undo for actions the user takes.

And some mistakes can apparently not be fixed. I am divorced, but living with a partner for the past 25 years. She doesn’t want to get married. And we both keep our finances separate. We work in the same general field (computers) with similar post-graduate degrees. But, since she is a woman, she has consistently been paid less than I for similar work — software development. Over time, I have accumulated half again more financial assets than she has. So I pay for all the “normal” joint bills — food, utilities, entertainment. We each pay for our own car, clothes, and personal projects. Joint purchases over $500, we split 2/3, 1/3. And while I tell her my finances and she tells me hers, to the extent it matters, we keep our finances separate — no joint accounts, we don’t share passwords. MaxiFi offered me two marital status options: married or not married. If not married, I could be “partnered”. Which is what I am, so I selected that. Only later did I find out MaxiFi then wanted a complete run-down of her finances, all of her Social Security earnings history — none of which I have. But I can’t just skip over them; it insists on knowing when she will retire, when she will claim SS benefits, and how much she will get.

The obvious thing to do is go back and change my marital status and delete her from the computations. But that is not possible. I can’t break up with her. I can’t delete her. I do have the option of marrying her, or keeping her as a partner, but that’s it. MaxiFi should allow marital status to be changed freely.

When entering “Regular Assets”, we add accounts for savings, and checking, and the brokerage account. After each one, you select either “Save and Add Another” or “Save and Mark Complete”. So as you enter one after another, you keep hitting “Save and Add Another”. Until the last one is entered, you hit “Save and Add Another” and then realize, there is not another one. But there is no “Cancel”. Your only choice on what to do is either “Save and Add Another” or “Save and Mark Complete”. But both of these will fail, wanting a name and an amount to “save”. MaxiFi should always allow a “Cancel” or “Mark Complete” action.

You can get back to the main Profile, to abandon what you were doing, but then it is not clear how to continue. Going from one topic to another works only until you stop. In general, the “guided tour” approach to filling in the data needed doesn’t work once it gets interrupted. If a kid needs help with his homework, and you go to do that, when you come back, you will be logged out (inactivity) and have to log back in, and then there is no real way to know where to resume what you were doing.

But I got all my data entered, and told it to run my Base Plan. It thought for a moment and then said

There was a problem with the Base Profile.
Your plan is unaffordable given the information you’ve entered.

Please return to the Base Profile and review all information to be sure it is accurate, especially large expenses such as Housing and Special Expenses. Also be sure your income and assets are entered correctly. If you still run into this error, you may want to reset all Settings and Assumptions.

Which really doesn’t help much. Something is wrong. You spent hours entering your data, but something is wrong. You figure it out. I have several times in my life had jobs where I was suppose to solve a problem, and all I got was either “it worked” or “it didn’t”. It took me weeks to solve those problems, with no feedback on what is wrong. MaxiFi should provide specific problems when it gives an error, not just “something is wrong”.

Remember how I mentioned that Children do not have names, just “Child born 1996” and you can actually have multiple children of that same label? This seems a generic problem with MaxiFi. I created a Roth optimized plan, and saved it as an alternative plan as “Max Roth”. Then remembered something, went back and changed my Base Profile, and then ran the Roth optimization again. It gave different results, so I wanted to save it and saved it under “Max Roth” again, not remembering that I already had one of that name. The result? I have two plans called “Max Roth”. No error message saying I already had one of that name, pick a different name, and no warning that the new plan would replace the old plan of that name. MaxiFi should make sure names are unique.

When you are running the Roth optimization, it puts up this Percent Completed gauge so you can see how it is getting closer to finishing the computations. But when you hit “Run”, it immediately puts up the gauge, and moves it to 84%, waits a bit, and then moves it back to 66% and then proceeds to gradually increase it until it is 100% and done. This seems to indicate someone can’t count correctly, and is not a good look. Any “completion” gauge should only move forward.

Also, I’m 75. I’ve already made my decisions about Social Security. There is nothing I can do now to improve my SS decisions. So when it gives me the option to “Maximize Social Security”, repeatedly, when there is nothing it can do to improve my SS, it’s gets progressively annoying. Even after I’ve run “Maximize Social Security” and it did nothing, it keeps asking. When it does the Roth optimization, it reminds me “Not optimized in this plan. In most cases maximizing Social Security can significantly increase your Lifetime Discretionary Spending. If you have not run this optimization yet, please consider doing so.” when it has already tried and failed to do any good. MaxiFi should not suggest or ask the user to do things that do not apply to their situation.

MaxiFi can support multiple profiles, but really only one base with variations in some parts. You can’t change the family, for example, so you couldn’t do a comparison of being married versus just being partnered. And everything is stored on their servers on the internet. You can export your SS earnings record, but you can get that from SS directly also. For those that don’t like their data stored “somewhere on the internet”, it seems they could provide an ability to export all the profile data to a file, and then upload it later. Maybe into an XML or Jason format. MaxiFi should allow all user generated data to be exported and imported.

Looking for Overlapping Mutual Funds

I have investments in mutual funds at T. Rowe Price. In one of their communications they mentioned “an online tool … to gauge overlap in your investment portfolio”. That seems like it would be useful. I have, in total, 20 different mutual funds, but in only 10 asset allocation buckets, so I may have funds which are effectively duplicates. It would be helpful to know if a Vanguard Midcap fund is almost the same as a Fidelity Midcap fund — they contain the same underlying set of stocks in roughly the same proportions.

The tool comes from Morningstar and is called Instant X-Ray. It requires a Premium account from Morningstar, so I arrange that. Then I have to type in all my mutual funds and select the X-Ray tool. They produce a lot of information, but we are looking, presumably, for how much overlap there is between funds. The closest looks to be the “Stock Intersection” tab.

Unfortunately, the “Stock Intersection” looks at things from a stock point of view, not a fund point of view. So it provides a return like:

which, if I am interpreting it correctly, says that 1.17% of my overall portfolio is Microsoft stock, held in 7 different mutual funds. Some funds have more Microsoft (like Jensen which has 6.6% of its holdings in Microsoft) and others have less (Fidelity Capital & Income only has 0.55% of its holdings in Microsoft).

And this goes on for hundreds of stocks and bonds (827 in my case), but only 152 of these are in multiple funds. Microsoft, our example above, is in 7 different funds. But even then it is only 1.17% of our holdings. So it seems we are mostly diversified.

But that is not what we wanted to know — we wanted to know if our funds were mostly the same or different.

While this report does not tell us what we want, it does have the raw data that would tell us. It says, for example, that Jensen has 6.60% in Microsoft, and T.Rowe Price Value has 3.36% in Microsoft. If Jensen and T.Rowe Price Value have a lot of other stocks in common, in similar amounts, then they are pretty similar.

If we extract from this report, for all funds, for all stocks, how much each fund has of that stock, we can re-sum those amounts by fund, rather than by stock. Specifically, for each pair of funds, we look at all the holdings that they have in common. Suppose we have two funds, one has 6.60% in Microsoft, and the other has 3.36%. Take the minimum of the two percents. They both have that much in common. In the example, both funds own at least 3.36%.

This is our measure of similarity. If we sum those minimums for two funds, over all the stocks they own, we have a number between 0.00% and 100.0%, saying how much they are the same fund. A value of 0.00% says they have no assets in common; a value of 100% says they have exactly the same stocks in exactly the same amounts (percentages).

This produces a lower-triangular matrix where the (i,j) element (or the j,i element) is the sum for two funds of the shared percentage for each stock. A large value means the funds have substantially the same holdings, and should then behave the same and generate the same level of return.

In our case, we found most of my large cap growth funds were very similar

0: overlap of T. Rowe Price Blue Chip Growth (TRBCX) and Fidelity Contrafund (FCNTX) is 44.10
1: overlap of Fidelity Contrafund (FCNTX) and Fidelity 500 Index (FXAIX) is 32.67
2: overlap of Vanguard Real Estate Index Admiral (VGSLX) and T. Rowe Price Real Estate (TRREX) is 28.20
3: overlap of T. Rowe Price Blue Chip Growth (TRBCX) and Fidelity 500 Index (FXAIX) is 27.91
4: overlap of Vanguard Windsor II Admiral (VWNAX) and T. Rowe Price Value (TRVLX) is 23.35
5: overlap of Vanguard Windsor II Admiral (VWNAX) and Fidelity 500 Index (FXAIX) is 22.41
6: overlap of T. Rowe Price Blue Chip Growth (TRBCX) and Jensen Quality Growth J (JENSX) is 20.20
7: overlap of Jensen Quality Growth J (JENSX) and Fidelity 500 Index (FXAIX) is 19.30
8: overlap of Jensen Quality Growth J (JENSX) and Fidelity Contrafund (FCNTX) is 17.59
9: overlap of Vanguard Windsor II Admiral (VWNAX) and Fidelity Contrafund (FCNTX) is 17.50

The similarity of the two Real Estate Investment Trusts (REIT) was not expected, but VGSLX says explicitly it is an index fund; if TRREX is also an index fund, then maybe it is not so unexpected — two funds following the same index.

There are some limitations of this. While Morningstar says they are analyzing the contents of each fund, if we look at the number of the data they present never has more than 50 stocks in any one fund. Some (like JENSX) have less (only 27 given for JENSX) but 15 of my 20 funds show exactly 50 stocks in the Morningstar report, which suggests they are only looking at the top 50 holdings of each fund. A more complete analysis then says we should skip Morningstar as the source of our data and go directly to the funds to find their holdings, in terms of percentages of their total fund.

A Model for Retirement Savings in Retirement

My Father worked for 40 years for the same company and got a pension. He dabbled a bit in stocks, but his retirement finances were simple: a check arrived every month.

I, on the other hand, worked for many companies, and at retirement had an array of 401(k) accounts, 503(b) accounts, an IRA, a Roth 401(k), two Keogh accounts (one profit sharing, one money purchase), as well as several investment accounts, with a (balanced) set of mutual funds and stocks. How should I think about all these assets, and how to structure them to create an income stream? I needed a simple mental model, and from that a simple set of accounts.

The first step was to recognize that almost all my retirement accounts were just variants on a simple Traditional IRA, and could be consolidated into one Rollover IRA. The 401(k), 403(b), Keogh, and IRA could all be rolled over (custodian to custodian) into one account. The Roth 401(k) could be rolled over to a Roth IRA. While that simplified things a lot, it still left me with 3 different types of retirement assets: savings (the brokerage investment accounts), a Traditional IRA, and a Roth IRA.

These 3 different accounts have significant differences, in terms of my retirement:

• Savings. Easily accessed.  I can add or subtract at any time.  Mutual funds and stocks may throw off dividends, which can be reinvested, but are taxable as ordinary income.  Selling, either for income or just to rebalance my asset allocation, can create capital gains (or losses).  Inheritance is governed by my will.
• Traditional IRA (tax-deferred).  Adding assets is limited by income and a maximum amount each year (currently $6000), but any income tax on contributed funds is deferred.   Dividends, and buying and selling to rebalance, has no tax implications.  Funds withdrawn from an IRA are taxed as ordinary income.  Required Minimum Distributions (RMDs) start at 70.5.  Cannot be accessed before 59.5.  Inheritance is by beneficiary.
• Roth IRA (tax-free). Same as for the Traditional IRA, but contributions are taxed when they are put in, and there is no tax on withdrawals. Cannot be accessed before 59.5. Inheritance is by beneficiary.

This is a somewhat simplified description.  Funds in IRAs can be accessed before 59.5 under special circumstances. Earnings on Roth contributions are only tax-free after the account is open for 5 years.  Some capital gains (such as buying and selling gold or collectibles) is treated as ordinary income.  RMDs are required for inherited IRAs.  And so on.  Life is complicated.

The result of all this is that every time that funds are needed, a decision has to be made as to which of the accounts it comes from.  There are occasional articles written describing how to make that decision.  Some say to always draw on savings first, then the Traditional IRA and finally the Roth.  Others suggest a mixture, depending on tax brackets and “Break-Even Cost Basis Percentages”.   As we age, it becomes even more difficult to balance the pros and cons of financial decisions.

I propose a different approach — a Roth centric structure — based upon the observation that, in the best of all worlds, all your money will be in a Roth, and the other two types of accounts will be empty.   Money in a Roth account is better than money in any other account.

Compare a Traditional IRA and a Roth.  The main difference is that a Roth is tax-free while a Traditional IRA is tax deferred.  That means that if you have $1000 in a Roth, you have $1000 while if you have $1000 in a Traditional IRA, you really only have $1000 minus the taxes that you will have to pay if you  actually use it.

A friend wanted to use his IRA money as a down payment on a house.  He needed $300,000 (it was a very expensive house).  But in order to have $300,000, he needed to withdraw $300,000 plus the taxes that would be due for that.  He was in a 28% marginal tax bracket, so he needed to withdraw an additional $84,000 for the taxes.  But that meant he was withdrawing $384,000 and needed not $84,000 to pay the taxes, but an extra $23,520 to pay the taxes on the $84,000 he neede for the taxes on the $300,000.  And that $23,520 pushed him into a higher marginal tax bracket.  The result was that to get $300,000, he had to withdraw $472,880 and pay $172,880 of it in taxes. (And this was just Federal income taxes).

With a Traditional IRA, eventually you have RMDs, whether you need them or not, and the RMDs will create a tax liability, possibly pushing you into a higher marginal tax bracket.  The more money you have in the Traditional IRA, the larger the RMDs, the more taxes.

And this is also true for an inherited IRA, although the tax liability will be on your heirs rather than on you.  Suppose you leave your IRA to your children, or grandchildren.  With current life spans, it is likely that your heirs will inherit your Traditional IRA near their peak earning years.  Which means that their RMDs will come on top of an already high income, creating a substantial tax liability.

A Roth IRA has none of these.  If you need $300,000, you can just withdraw $300,000 — it is tax-free money.  While your heirs may have RMDs, it will not add to their tax liability.

A Roth is also superior to plain Savings.  With Savings, you would want to keep the bulk of your money invested, and well-invested money grows.  If the investments throw off income (dividends or interest), even if those are re-invested, you have an on-going issue of additional ordinary income taxes, every year.  In the best case, your investments might be tax-efficient, and create little ordinary income, instead growing in value. But  when you want to spend the money, you need to sell the investments, creating a capital gain, so even in the best case, you then have long term capital gains taxes to consider.

It is also more difficult to manage Savings investments than Roth.  All transactions in a Roth are tax-free, so you can sell and buy, to re-balance your assets, with no tax implications.  In Savings, if you sell something with a capital gain, even if you use that money to buy something else, you owe capital gains tax on the gain over your cost basis.  A friend neglected her investments for about 10 years until it became obvious that she needed to reblance her asset allocation to get her risk under control.  It had been a good 10 years, and just to rebalance would require selling and buying close to a $1,000,000 with a capital gain of close to $400,000 creating a $80,000 tax liability.

The financial press seems to suggest that you offset your capital gains with capital losses, but if you are doing it correctly, you don’t have (many) of those, so that advice is pretty useless.

Now Savings does have the value of the cost basis “step-up” when they are inherited.  An asset purchased for $40, which grows to $70 (creating a $30 capital gain for you) is stepped up to a basis of $70 (the current value) when you die, so your heirs inherit no capital gains (or losses) when they inherit the asset, but any further growth in the value of the asset, after they inherit it, is their liability.  With a Roth, there is no step up in cost basis, but there are no taxes due either before or after inheritance.

So, all things considered, the best place to have your money is in a Roth.

If you have money (surplus income) that you are going to save, if you are going to simply invest it, it will be better to contribute it to a Roth, and then invest it.  If you simply invest it as Savings, you will pay ongoing taxes (either ordinary income or capital gains) on any earnings from that investment.  In a Roth the earnings are tax-free.  If you are in a high income tax bracket, putting it into a Traditional IRA (or equivalent 401(k), etc.) you can temporarily avoid (defer) the taxes on that income, but eventually you will have to pay the taxes on both the deferred income and all earnings, as ordinary income.  If you are in a high tax bracket now, deferring the taxes until you are in a lower tax bracket makes sense.

But there are limits on how much you can put either directly into a Roth or into a tax-deferred Traditional IRA account, each year.  So while you have an income and particularly if you are in a high tax bracket, you may want to accumulate assets in both Savings and Traditional IRA accounts.  But at some point that may stop.

In my case, for example, I had unexpected heart surgery at 60, and retired.  After a lifetime of working, I had Savings and tax-deferred accounts, but no longer had an income. And, if I planned carefully, I could avoid having an income until I was 70 (when Social Security kicked in), or 70.5 (when I had to take RMDs).  Other friends found themselves laid off or with other health problems or simply were tired of “working for the man” and wanted to retire.

But having low income, it creates an opportunity to move to a Roth centered financial account structure.

With low income comes low (marginal and actual) tax rates, allowing money to be converted from a Traditional IRA to a Roth IRA.  A Roth conversion moves money from a Traditional IRA into a Roth.  This creates a withdrawal from the Traditional IRA, creating a tax liability for the amount of money converted (you do not have to convert everything at once).  So each conversion means taxes (at ordinary income rates), but from then on the money is in the Roth and will grow tax-free.

Note that you don’t have to do this.  A friend had a financial planner who said (as they all seem to say): “Never pay taxes that you don’t have to.”   And as a result, for years, she lived off Savings, and had (effectively) no income, and paid almost nothing in income taxes.  Meanwhile her Traditional IRA grew.

There are two paths that can be taken here: to convert the IRA to a Roth, or not.  Let us assume that the funds in the IRA and the Roth can be invested in the same way, so will return the same investment results. Assume that over 10 years, the value of the investments will double.  Then $100,000 in the Traditional IRA will grow over 10 years to $200,000, if we do not convert it.  If we assume we have (for example) a 20% marginal tax rate, then if we convert the Traditional IRA to a Roth, then we will pay 20%, or $20,000, in taxes now, and have only $80,000 in the Roth.  That will then double in 10 years to $160,000.  Clearly less than the $200,000 in the Traditional IRA.

However, to actually use the money in the Traditional IRA, we have to withdraw it and pay taxes on it, at ordinary income tax rates, so we have to pay 20% (our marginal tax rate) or $40,000 in taxes, leaving us with only $160,000.  So as long as the marginal tax rate does not change, converting from a Traditional IRA to a Roth is equivalent.

Except for two additional factors.  First, note that if we do not convert to a Roth, we are dealing, not with $100,000 at 20% marginal tax rate, but at $200,000.  That much larger number could well mean that we are pushed into a higher marginal tax bracket and will pay more taxes (eventually, on the larger amount) if we leave it in the Traditional IRA. And, of course, this is assuming that tax rates do not change.  You are gambling on the difference between the current tax rates and the tax rates 10 years from now.  (Even in the 2017 Trump “tax cut”,  tax brackets for a single tax payer earning $157,500 to $191,650 went up from 28% to 32%)

Plus, we assumed that you converted $100,000 and paid $20,000 of that in taxes, leaving you with only $80,000 in the Roth.  But the taxes, while they do have to be paid, do not have to come from the money being converted.  The taxes can be paid from other sources, that is, from Savings.  You can convert $100,000 from a Traditional IRA to a Roth IRA and pay the taxes from Savings.   That results in $100,000 in the Roth, and $20,000 less in Savings, effectively moving $20,000 from Savings into the Roth.  Now, after 10 years, that $100,000 in the Roth will have grown to $200,000 all completely tax free.

So for low-income years, it makes sense to convert assets from Traditional IRA to Roth IRA, especially if you can pay the taxes with Savings.  This increases you assets growing tax-free.  It decreases your assets growing tax-deferred.  And it decreases the amount subject to RMDs.

In fact, you should never voluntarily withdraw funds from a Traditional IRA; you should always convert them to a Roth, and then withdraw them if you need to.  The tax liability will be exactly the same.  But if you pay that tax liability with Savings, you are effectively moving those Savings into the Roth, tax-free.

And if for some reason, you do not end up spending (all) the funds that were withdrawn, they can be reinvested to grow tax-free.  A friend needed money from his Traditional IRA for a trip to Europe, but ended up not being able to go because of unforeseen health problems.  If he had converted that money to a Roth, it would now be growing tax-free; instead by simply withdrawing the money from the IRA, it is now in his Savings, with all growth taxed (either as ordinary income or capital gains).

So during my 10 years of low-income (no income), from 60 to 70, I have been converting assets, each year, from Traditional IRA to Roth IRA, paying the taxes from Savings.  I convert as much as I feel comfortable paying taxes on.  (In my case that means staying below a 30% marginal tax rate.)  I have exhausted my Savings (paying the taxes on the conversions and living expenses for 10 years), but still have half my Traditional IRA left to convert.  The plan is to continue conversions, probably at a slower rate, as necessary, until it is empty and all my funds are in the Roth.

It may be possible to speed up this process, but it depends upon the market.  In general, and like most people, I prefer the economy to grow and the market to go up.  However, at the moment, I have assets that I want to convert from a Traditional IRA to a Roth.  I could just convert everything, but that would push me into the highest marginal tax bracket, 37%.  I originally deferred taxes at an average of 30% (so for every $100 I put into my 401(k) or IRA, 30% of that was taxes I did not pay at that time).

Suppose the market were to suddenly drop substantially.  For example, in the 2007-2008 recession, my investments were down almost 50%.  That would mean that I could have converted everything for half the taxes I would have paid before (at 100%) or after (when it got back to 100% and beyond).  While it might still be a 37% rate (it might drop back to a 35% rate), the actual amount of money would be half as much.  As long as I believe that eventually the market will come back, this is a good deal.

 

 

 

 

 

Timing a Roth Conversion

A tax-deferred account can be converted to a Roth account by paying the taxes on the amount of the conversion.  The amount of tax due depends upon the amount converted, and if it is invested in stocks and bonds, the amount invested varies from day to day.  Does it matter when we do the conversion?

Obviously the actual amount of tax paid does matter, but while we are not rabid about paying less tax, we do want the amount of money that is left for us to spend to be as large as possible. If we convert our investments at $100 today, we will pay less tax than if we convert it tomorrow at $105, but we will have more left for us at $105 than we would at $100.  On the other hand, we will have more at $100 than we would at $95.  The market can go either up or down.

But in the long term, we expect the market to go up.  So is it better to convert when the market is down, and reap the investment returns of a bull market tax-free in a Roth, or do we get the same effect by letting our money grow tax-deferred, and then converting later?  In particular, does that suggest we should “convert on the dips” — that is convert from a traditional IRA to a Roth IRA when the market is down?

We actually have an example of this situation with the 2008 financial crash.  The S&P 500 index fell from 1557 on Oct 5, 2007 to 683 on March 6, 2009, before climbing back to 2872 on Jan 26, 2018.  Would we pay less tax if we converted from a Traditional IRA to a Roth IRA at the low point in the market?  (We understand that it is not possible to know when the low point is, but it may give us some guidance anyway.)

In my particular case, having been employed as a professional doing research in computers, I worked at several employers over my career, both private and public.  Each employer had a 401(k) (or equivalent) program, and when I moved from one job to another, I rolled my 401(k) into a self-directed Traditional IRA.  On Dec 8, 2007, I had a balance of $952,821.11.  On March 7, 2009, it had fallen to $491,851.29.  More recently, on January 27, 2018, it had grown back to $2,021,155.18. This is the equivalent of a 17% annual increase for 9 years.

(A minor side note — my IRA is invested according to a simple asset allocation approach, and I rebalance it annually to keep it at my desired asset allocation.  The result is that while the S&P 500 fell to only 47% of it’s previous value in 2009, I fell to 52%, doing somewhat better in a down market.  On the other hand, from it’s low, the S&P has increased by 4.2X while I have only increased 4.1X.  Diversification reduces both my downside risk and my upside potential.  Your experience is probably similar.)

Suppose I had converted my entire Traditional IRA to a Roth at the low point of $491,851.18.  Assuming no other income, being single, using the tax rates for 2009, I would have owed $146,559 in taxes, paying at a 35% marginal rate (but only a 29% actual rate).  While it seems that would leave me only $345,252 in my Roth, I would have been able to pay the taxes using other money from a taxable account, so that I would still have had $491,851 in my Roth, which, assuming my actions would not have changed global financial history, would have grown to the same $2,021,155, and the growth would have all been tax free.

Now, by contrast, having not done that, since no financial advisor would have suggested I voluntarily pay $146,559 in taxes, at the bottom of the recession, I have instead a Traditional IRA of $2,021,155.  I could convert it now.  Presumably, I will do much better, since we just had a massive tax cut.  If I were to convert the entire $2,021,155 now, again as single with no other income, I would owe $709,077 in taxes, paying at a 37% marginal rate (an actual rate of 35%).

The $146,559 that I would have paid in taxes in 2009 has grown by 4.1X (assuming the same growth rate as the rest of my investments) to $600,891.  So I would now owe $108,186 more in taxes than if I had done the conversion in 2009.  Why is this?

1. Delaying the conversion meant that the amount in the traditional IRA grew (4.1X), which then pushed me into a (much) higher tax bracket.  This accounts for about $80,000 of the increase.

2. The “tax cut” recently passed was actually a tax increase for those of us with income in these ranges.  This accounts for the remaining $30,000 of the increase.

Of course, the numbers would be different if I could not have paid for the taxes of the conversion from other funds, but the result is pretty much the same.  There are two possibilities:

1. Paying $146K in taxes from taxable savings, resulting in $146K more in the Roth, growing tax-free, or
2. Paying $146K in taxes from the Roth, resulting in $146K more in taxable savings, with all growth taxed as appropriate.

Paying the taxes with other funds is effectively just moving normal taxable savings into the tax-free Roth, the equivalent of a very large contribution to my Roth, and means that the growth in those savings will then be tax-free.

Of course, it is not possible to know when the absolute bottom of a correction or recession occurs, so this is only a best case solution. That suggests that “dollar cost averaging” may be a better approach to conversion.  Converting a fixed amount at regular intervals means converting more assets when the market is down and fewer assets when the market is up.

A word of acknowledgement.  The tax computations where done on the moneychimp.com tax calculator web site and are grossly simplistic, but they capture the essense of the problem.

 

 

How to save for retirement: 4 different mechanisms

Assume you want to save for retirement.  How do you do that?  There are lots of ways to do it: IRAs, both traditional and Roth, 401(k) accounts (again both traditional and Roth), or just plain banking or brokerage accounts.  But despite the details, it basically comes down to:

• a taxable brokerage (or bank) account
• a tax-deferred account (IRA or 401(k))
• a Roth tax-free account (IRA or 401(k))

Which of these mechanisms is the best way to save for retirement?  Well, what do we mean by “best”?  I suggest that the best plan would provide the most money in retirement.  So do the different mechanisms produce different results?

The standard suggestion is to put as much money as possible into a tax-deferred account — an IRA or a 401(k).  Financial advisors repeatedly suggest that you max out your 401(k) at work.  But the money that comes out of a tax-deferred account is taxed as “ordinary income”; maybe you could do better by investing in a standard (taxable) brokerage account in a tax-efficient investment which would create value which would be taxed as “long-term capital gains”, which is generally a lower tax rate than “ordinary income”.

To figure this out, in reality, you have to know your return on your investments over time, your tax rates, both now and in the future, and your income, both now and in the future, all of which are unknowable in advance.  So, to get at least a hint as to how the savings mechanisms will compare, let’s simplify the situation.

• Assume we are going to save $10,000 a year every year for 30 years.  That would be possible in either a 401(k) or a brokerage; it actually exceeds what you could put into an IRA.
• Assume a constant return on investment of 8% per year.  In reality, the return will vary, and depending on what you invest in, it may be more or less, but 8% has traditionally been a reasonable return.
• Assume a constant marginal 28% tax rate.  Depending on your income you may pay more or less.  You may be able to manipulate your income from year to year to define what tax bracket you are in, but if you make enough, you could well be in a 28% tax bracket, so that additional income is taxed at 28%.

To figure out what would happen, we use a simple spreadsheet to see how investments will compound over time.  To start, we have an invested balance of $0.  Each year (one row in the spreadsheet), we invest a given amount ($10000).  We may (or may not) have to pay income taxes on that, leaving a possibly smaller amount to be invested.  That amount is added to the account balance.  That balance generates a return, based on the account balance and the amount added this year.  We assume an 8% return on the account balance and a 4% return on the amount added each year (which is an 8% return assuming the income is added uniformly over the year).  The growth in the account balance may (or may not) be taxed, meaning the actual account balance will increase by the possibly smaller post-tax amount.  If the income or growth is taxed, then it adds to the basis of the account, that amount which has already been taxed.

For a 401(k), we invest $10K each year.  We pay no tax on this $10K, so we can add the entire amount to the account balance.  In addition, the growth is not taxed, so the entire growth is added to the account.  After 30 years, we have contributed $300K, but at 8%, our account balance will have grown to $1,178K almost 4 times as much as what we invested.

If we invest in a Roth 401(k), we invest $10K each year, but we have to pay the tax first.  At a 28% tax rate, we pay $2800 in tax, and have only $7200 to add to the investment account.  But this grows tax-free, so after 30 years, our $300K has been divided into $84K in taxes and $216K in investments.  The $216K in investment grows (at 8%) to $848K (again almost 4 times).

So it seems that the simple 401(k) beats the Roth 401(k).  But remember that the 401(k) is tax-deferred, while the Roth 401(k) is tax-free.  We can take all the Roth 401(k) with no further taxes.  As we take money out of the traditional 401(k), we have to pay taxes on it, at ordinary income tax rates (which we assume to be 28%).  We can lower that rate by taking out money at a lower rate (how much each year), but if we take enough out each year it will push us into a 28% bracket (or possibly higher).  If we pull out our entire 401(k) balance of $1,178K and pay a 28% tax rate, we will pay $330K in taxes and have $848K left — exactly the same amount as in the Roth case.

Although we have less money in the Roth scenario, we have already paid all the taxes on it.  In the traditional 401(k) scenario, we have more money, but we haven’t paid the taxes on it, and that hidden tax burden reduces the actual amount of usable money to the same amount as the Roth scenario after we pay the taxes.

Note that 401(k) versus IRA is not relevant; both are effectively the same kind of tax-deferred accounts. You can roll your 401(k) account balance over to an IRA; you can merge and split IRA accounts.  But it’s all tax-deferred money. Roths are different; they are already taxed and all the growth is tax free, so there is no hidden tax burden inside a Roth account.  The difference is whether the taxes are paid going in, or coming out.

At a constant tax rate, and because of the commutative property of multiplication (both for the tax rate and the investment return), it doesn’t matter which is done first — tax and then growth or growth and then tax.  The catch is the constant tax rate.  If you put money into a Roth at a young age, before you make very much, you can do it at a low tax rate and have more of money to grow tax free.  If you put money into a 401(k) during your peak earning years, when you have a very high marginal tax rate, and can pull that out when you are retired and have no other income (so that you are in a low tax rate), then you can do better with one or the other scenario, by varying the tax rate.  But who knows what your income, and/or the tax rates will be in the future?

An alternative to tax-free and tax-deferred accounts are taxable investment accounts.  You can always just take a part of your income, pay the taxes on it, and then invest it, typically in a brokerage or bank.  Investments grow in two ways:  some throw off dividends and some grow in value by an increase in their share price, and some do both.  Dividends are taxed as ordinary income, while an increase in the share price is not taxed until the shares are sold.  The difference between the original purchase price and the sales price defines how much is taxed, and that is taxed as a “capital gain” (or loss, but we are assuming a gain over all over the long term).  If the investment is held for at least a year, it is a “long-term capital gain”, and long-term capital gains are taxed at a different (currently lower) rate than ordinary income.  For ordinary income taxed at a 28% rate, long term capital gains are taxed at 15%.

A pure tax-efficient investment will minimize taxes by providing all of its growth in long-term capital gains.  A pure tax-inefficient investment will provide all its growth in dividends.  Most investments are neither purely tax-efficient or tax-inefficient, but a mixture, returning some growth as dividends and some growth as capital gains.

If we take our $10K a year that we are saving, and invest it in a purely tax-efficient investment, it will grow over time.  How does this compare to the tax-free and tax-deferred scenarios?  Each year, we invest our $10K, but before we can do that, we have to pay taxes on it.  At a 28% rate, $2800 goes to taxes and the remainder goes to our investments.  Since it is tax-efficient, it throws off no income (dividends) as it grows, and so will grow at 8% per year.  After 30 years, we have taken $300K and paid $84K in taxes, while the remaining $216K has grown, at 8% a year, to $848K.

This should sound familiar, because it is exactly the same behavior as for the Roth 401(k) (or IRA).  We pay our taxes on the money on the way in, and our investments grow tax-free.  For a Roth investment, it is tax-free by definition; for this after-tax scenario, we are assuming we can find a tax-efficient investment which will grow create long-term capital gains by increasing the price of our investments, not by dividends.  Either way, we pay no tax on the growth while the investment is growing.  A tax-efficient investment is effectively a tax-deferral mechanism.

The difference comes then after our savings, when we try to withdraw the money for retirement.  Just as with the traditional 401(k) (or IRA) scenario, we must pay taxes on the money that is withdrawn from our account.  As we sell our investments, to withdraw the money, we generate capital gains and must pay long-term capital gains tax on the growth (not on our original contributions — our basis).  Our money has basically increased by 4X over the past 30 years, so our basis is about 25% of our funds, and we will pay 15% capital gains taxes on the growth, 75% of our funds.

This is where the Roth and the tax-efficient taxable investment scenario differ.  For the Roth versus traditional 401(k) scenarios, the traditional 401(k) scenario generates a larger sum which, after the 28% ordinary income taxes, is the same as the Roth tax-free amount.  For the tax-efficient scenario, we start with the same amount as the Roth tax-free amount, but then reduce it by the capital gains rate on all the growth, ending up with 11% less money to spend.  And the difference increases over time as the money remaining in the account continues to generate (tax-efficient) growth as we slowly withdraw funds during retirement. Even if we withdraw at a sustainable rate so that our withdrawals are no more than the investment growth, each withdrawal removes a portion of our basis.  Asymptotically, we will be paying 15% on effectively all of our withdrawal.

So the tax-efficient scenario is always worse than the Roth or traditional 401(k) scenarios, by 11% to 15%, the amount of the capital gains tax on the growth.

A fourth scenario is to put all of our investments into tax-inefficient investments, which return their growth via dividends.  In this case, each year we contribute our $10K. 28% of that is taxed, leaving us with $7200 which we invest.  The growth on our investment is 8%, but this comes in the form of dividends, which are taxed each year, as ordinary income, 28%.  This means that we have an effective growth rate of only 5.76% (72% of 8%, since 28% is lost due to taxes.)  This (substantially) lower growth rate means our total $300K investment, over 30 years grows only to $560K, less than twice our investment.  On the other hand, all of this amount is part of our basis, since it has all been taxed.  Still, we pay a continuing 28% a year on the return on our investments during retirement, so it grows slower, and still is taxed more than any of the other scenarios.  The constant drag of taxes reduces the effective rate of growth of our investments, and clearly shows the advantage of tax-deferral.

In summary,

(a) A traditional IRA or 401(k) will allow you to invest more money which will grow to a larger account balance.  But the account balance is misleading since it contains an embedded tax burden.

(b) A Roth IRA or 401(k) invests less money, since you pay the taxes first, but the entire account is tax-free.  The commutative property of multiplication means that the usable amount is the same as for the traditional IRA or 401(k).

(c) A tax-efficient taxable account grows just like a Roth account during the accumulation phase, but the growth must pay at least long-term capital gains as it is withdrawn, so you end up with (asymptotically) 15% less money.

(d) A tax-inefficient taxable account effectively grows at a much lower rate than any of the other approaches, due to the constant drag of taxes, resulting in less than half as much money in retirement.

 

 

 

 

 

 

How to convert to a Roth (over time)

We know that keeping a traditional IRA versus converting to a Roth is a wash (tax-wise) if rates and return stays the same.  If we have $100K in a traditional IRA in a 20% effective tax bracket, with an 8% investment rate of return, we end up with the same amount by leaving it in the traditional IRA or in converting it to a Roth:

In the traditional IRA, we start with $100K, it grows by 8% to $108K, and then we pay 20% taxes on that ($21.6K) to leave us with $86.4K.

If we convert to a Roth first, we pay $20K in taxes, to leave us with only $80K, which grows at 8% to $86.4K.

In both cases, we have the same result.  But the exact result depends upon both the rate of return being fixed and the tax rate remaining unchanged.  For example, notice that in the traditional IRA case, we paid taxes on a larger amount ($108K) than in the Roth case ($100K).  If this larger amount were to push us into a higher tax bracket, then we would pay more taxes (and have less left over for us) in the traditional IRA case.

Assume then that we have a more complex, but possibly more likely, problem to solve:  We have (after 30 years of contributions and rollovers) $1 million in our traditional IRA and we want to convert to a Roth over a ten year period (so that the traditional IRA is empty when required minimum distributions start).  Is this conversion also a wash?  And if not, what is the best way to do the conversion?

For example, we could convert the entire $1 million dollars at once, and pay a 32.4% average tax rate (around $310K in taxes), or we could convert $100K a year for ten years, and pay less than 19% in taxes.  Of course over that ten year period, the funds in the traditional IRA (as well as the funds that have been converted) will continue to grow at our investment rate of return, so we may have to convert more than $100K a year (at least for some years) to take the traditional IRA balance to zero at the end of the ten years.

And during that ten years, our rate of return will vary, possibly by large amounts.  And our tax rates are likely to change.  And we have an additional degree of freedom in how much is converted each year.

So real life poses a much more complex problem than the simple conversion scenario we started with.  And there may not be just one answer.  It is not even completely clear what our goal is.  By varying the way the conversion is done, and getting different rates of returns and different tax rates, we may end up with more (or less) money in the Roth, and we will pay more (or less) in taxes.  In the simple case, we ended up with the same amount of money in the Roth, but paid more taxes by leaving it in the traditional IRA until the end.

I suggest that we are only interested in our final Roth balance, and not in how much tax we pay.  If we pay $1 in taxes, or $100K in taxes, as long as we get the same amount in our Roth, we don’t care.  The government might care, but that’s their problem, not ours.  In fact if we end up with more in our Roth and pay $100K in taxes, versus less in our Roth and pay only $1 in taxes, we want to take the approach that will cause us to pay more taxes.  Not that we want to pay more taxes, but because we end up with more in our Roth.  We don’t want the tail to wag the dog, and the dog, in this case, is our Roth balance; the tail is taxes.  So our goal is to (a) reduce the balance of the traditional IRA to zero, and (b) have the largest possible Roth balance when we finish.

Let’s make some assumptions to help narrow the problem done:

1. Assume that we pay the taxes for any conversion from the funds being converted.  Clearly if we convert $100K, we will owe $19,091 in taxes.  We can pay that from the $100K, leaving only $80,909 left to invest in the Roth.  But we could also come up with the taxes from some other source of money (savings, for example), and put the entire $100K into the Roth.  If we do that, we are effectively taking $19,091 of savings and putting it in the Roth.  The value of this is that this $19,091 (which used to be outside a Roth, and hence generated taxable returns), now is inside the Roth and its returns are tax-free. While we may have other funds for the taxes, we may not.  For comparison purposes, we should assume that the conversion has to pay the taxes that result.
2. Assume that the current tax rates and brackets remain the same for the ten year period.  It seems unlikely that this will be true, but changing either the rates or the brackets introduces a large amount of variability and probably makes the problem unsolvable.
3. Assume we have no state or local tax issues, only Federal taxes.  For Texas that is the case.  In a state with a state income tax, we would want to adjust the rates and brackets accordingly.
4. We will need to live off something for the ten years we are doing the conversion.  We could live off the IRA or Roth IRA balances, but that would change the results, depending on how much we spend.  We would assume that we have a base income from other sources of $A.  This effectively offsets the tax computations for converting $X to a tax cost of  “taxes($A+$X) – taxes($A)”.  That is, if we did no conversion (X = 0), we would still have our base income of $A, and have taxes of taxes($A).  We are really only interested in the incremental tax cost for the additional income resulting from converting $X from the traditional IRA to the Roth IRA.  If we had no base taxable income (A = 0), say because we were living off savings, or from the income produced by tax-free muni bonds, then our first $1 of Roth conversion would be taxed at the lowest bracket (10%), but if we had, say $34K in base income to live off of, our first $1 of Roth conversion would be in the 25% bracket.  For now, let’s assume A=0.
5. The result is going to depend upon both the taxes paid, and the rate of return.  Over a 10 year period, we really need to consider a vector of returns.  Let’s assume a constant rate of return per year.  There’s no reason that we couldn’t use a rate of return per month, but if we do the conversion only once per year, an annual rate of return is sufficient.  This points out the assumption that we only do conversions once per year, effectively at the beginning of the year.
6. Once a conversion is done (in some amount), it will then grow, in both the Roth and the traditional IRA, at some rate of return for the remainder of the year.  The rate of return will depend upon the investment.  We can abstract that to the type of investment (bonds or stocks; domestic or foreign; large, medium, or small cap; short, intermediate, or long term bonds; and so on).  Let’s assume that we use the same investments, or at least the same investment classes in both the Roth and the traditional IRA, so that both get the same annual rate of return.
7. Over the past 25 years, my rate of return has varied from -33% to +37%, but with only 5 years of negative returns and 20 years of positive returns.  Different rates of returns will result in different final balances.  We could use fixed returns or historical returns or we could generate artificial random returns according to a similar distribution.  None of these is necessarily what will happen, so we will need to compute results for a variety of vectors of annual rates of return and see what the distribution is of those results.

With these assumptions, we have narrowed the problem down to converting from a traditional IRA to a Roth IRA with a given tax computation, given a vector of the rates of return for each of the 10 years.  The remaining variable is how we much we convert each year.  There are a number (probably an infinite number) of ways to convert.  We can consider the following:

• All Early: Convert everything at the beginning (year 1).
• All Late: Convert everything at the end (year 10).
• Front Load: Convert more at the beginning than at the end.
• Back Load: Convert more at the end than the beginning.
• Fixed: Convert a fixed amount (say $100K) each year.  This would work if the average rate of return was zero, but since the balance of the traditional IRA may go up or down over time, we have to convert the complete final balance in the last year, and stop converting if the balance goes to zero.
• Proportional: Convert a portion of the balance that represents the number of years still to come.  So we convert 1/10 the first year, 1/9 the second year, 1/8 the next, …, 1/2 the 9th year, and the complete balance the last year.

If we compute the results for each of these conversion schemes, we will have a distribution of results (depending on the return rate vector), and we will need to compare the distributions.  The hope is that one of the result distributions will clearly be better.

Are Tax Deferred Savings a Good Idea?

Introduction

My father retired with a pension, so he received a check every month in retirement to pay expenses.  But it is becoming difficult to find jobs with pension plans; most companies have switched to a 401(k) system, where they may provide some funding, but the responsibility for retirement is left to the employee.  In my case, I have no pension.  Rather I have a set of tax-deferred savings plans.  The theory is that if I save while I am working, the savings grows tax-deferred, and eventually, when I retire, I can withdraw those savings as my retirement income.

All the financial advice that I read says that tax-deferred accounts are the way to go.  A common principle is “don’t pay taxes (now) if you can avoid it”.  And so I have persistently put money into tax deferred accounts, so that now I have $1,395,349 in tax-deferred retirement accounts.  This would seem to assure a comfortable retirement.  I mean, more than a million dollars!

But this number is misleading.  This is the sum of my tax-deferred accounts, but it’s an inflated number.  Not all that money is really mine.  Some amount is deferred taxes, and I will have to pay the taxes in order to get at the money.  Any money from these accounts is treated for tax purposes as ordinarly income. If I were to withdraw all the money, now, all at once,  I would owe $466,016 in taxes, leaving me with only $929,333 — fully a third of my tax deferred accounts would be a “hidden” tax liability.

I can lower that tax liability by taking the money out slowly, over a longer period of time.  If I withdrew only $40,000 a year, I would owe only 15% taxes.  But of course, it is unlikely that I would be able to access very much of the money I’ve saved.  Even assuming the money does not grow, it would take 35 years to withdraw $1,395,349 at $40,000 a year.  I may live that long, but it’s doubtful.

I would assert that saving all that money all these years was a waste, if I can’t get at the money.  If I die with a substantial amount of money still in my retirement accounts, I should have just spent the money long ago and enjoyed it then, rather than saving for retirement and then not being able to spend the money.  And I can access the money if I am willing to pay a high enough tax rate.  So the question becomes:

Is putting money into a tax deferred account really a good retirement strategy, or would it be better to pay the tax when the money is earned, and then invest it outside a tax-deferred account?

Note that this does not consider a Roth IRA as an alternative — I make too much money to be allowed to have a Roth.  And, of course, the correct decision depends on many, many variables — tax rates, investment returns, life expectancies, and so on.  We cannot know what these will be in the future, and so the question cannot be answered in any absolute way for the future.

However, we can look at what actually did happen in the past to see if it holds to our expectations and supports, or contradicts, common financial advice.  In my case, I have access to only one set of data, but I can analyse that data and see, in retrospect, if the decisions I made with respect to tax-deferred savings were, or were not, good decisions.   We want to see, for my particular case, if I saved more in the taxes that I didn’t pay than I will pay to get my money out of my retirement accounts.

To determine this, let us look first at how much I actually saved in taxes over the years.  Those tax savings, tucked into my tax-deferred accounts, grew over time, so we want to compute the growth of these savings, and then compare those saved taxes with what I have to pay now.  We should be able to determine one of three results:

1. No matter how I withdraw the money, I will always pay more taxes now.
2. No matter how I withdraw the money, I will never pay more taxes now than I saved.
3. If I am careful, I may be able to pay less taxes now than I saved.

What did I save?

After years of school, college, and graduate school, I entered the work force in 1974, as a faculty member at a public university.  But it was not until 1983 that I started explicitly saving for retirement using tax-deferred accounts.  My university position allowed me to publish and consult which after ten years began to return income rivalling my university salary.  This was, of course, not dependable income, so I contacted a financial advisor to help me save and invest this “extra” income, while we lived off my normal salary.

The first thing we did was to open an IRA and start making the maximum contributions to that.  Of course in 1983, the maximum was $2000, plus $250 for my non-working spouse.  I have continued to contribute the maximum possible to my IRA despite the fact that after the first 4 years the contributions were non-deductible.

We created a Keogh account, which allowed me to tax shelter 25% of my outside income. That allowed a substantial amount of my total income to be tax-sheltered over the year.

Eventually, I left the university and went thru a series of jobs with various research companies, hi-tech start-ups, and large companies. These allowed me to contribute to 401(k) retirement plans.

In addition there were 3 extraordinary events.  The University had a form of a defined contribution retirement plan which they contributed to, and eventually they just gave me that.  Rather than take it (and pay taxes plus a penalty), I put it into a Rollover IRA.  One of the research companies, in the early days of 401(k)s did the same, and when I left them, I also put that into the Rollover IRA.  And eventually one of the large corporations switched from a standard pension plan to only a 401(k), and when I left, the remains of that pension plan was put in my Rollover IRA.  None of these represented any of my savings, but by rolling them over, I avoided income taxes (and a penalty for early access), so I effectively deferred the taxes on these.

Over time the outside work went away, and I closed down the Keogh plan, and rolled it over into the Rollover IRA.  This has made keeping track of things much easier — I have only my original IRA, my current 401(k) plan, and my Rollover IRA.

One other major event affects all this.  In 1993, I went thru a nasty divorce.  Basically we split all investments down the middle.  At first glance, that meant that I suffered a major loss in investments and assets in 1993, but a better way to look at it seems to me to be that all during this time, the investments were never really “mine”, but rather “half mine” and “half hers”.  With that perspective, the $12,800 that was put into the Keogh in 1984 was only $6400 mine, and accordingly I only have to account for that part and it’s tax savings; the other half of both the deposit and the tax savings were hers.

Accordingly the savings numbers before 1993 that are presented below are half the actual numbers — my half; I am not tracking her half.  The taxable income numbers, of course, are total taxable income, since that is what determines how much was saved in taxes. Before 1993, tax rates are determined by “married, filing jointly”; after 1993, tax rates are determined by “single” or “head of household”.

The table below, then allows us to compute exactly how much was saved using the various tax deferred savings plans: the IRA, the Keogh, the 401(k) plans, and the extraordinary rollovers.

Year	IRA		Keogh	401K	extra	Total Tax-deferred
1983	1,125	d				1,125
1984	1,125	d	6,400			7,525
1985	1,125	d	7,500			8,625
1986	1,125	d	10,500			11,625
1987	1,125	n	8,750			9,875
1988	1,125	n	7,550			8,675
1989	1,125	n	7,350			8,475
1990	1,125	n	6,350	1,806	30,496	39,777
1991	1,125	n	8,000	2,903	39,702	51,730
1992	250	n	6,700	3,112		10,062
1993	2,000	n	2,711	7,069		11,780
1994	2,000	n	4,712	7,488		14,200
1995	2,000	n	3,726	7,822		13,548
1996	2,000	n	3,585	8,610		14,196
1997	2,000	n	3,804	9,156		14,960
1998	2,000	n	4,010	9,629		15,639
1999	2,000	n	2,768	9,287		14,055
2000	2,000	n	9,646	10,339	92,362	114,347
2001	2,000	n	7,400	10,567		19,967
2002	3,500	n		12,000		15,500
2003	3,500	n		14,000		17,500
2004	3,500	n		16,000		19,500
2005	4,500	n		18,000		22,500
2006	5,000	n		19,990		24,990
2007	5,000	n		20,500		25,500
2008	6,000	n				6,000
2009	6,000	n				6,000
Totals	65,375		111,462	188,277	162,560	527,675

In this table, the first four years of IRA contributions were deductible (d); the remaining years were non-deductible (n).

In 2008, my employer offered a Roth 401(k), and I switched all my contributions to that, so there were no tax-deferred 401(k) contributions for 2008 or 2009.

How much tax was saved?

With this table of how much I put into each of the tax deferred savings plans, we can now compute how much tax was saved.  We go back to our old tax forms, and look for line 39 of form 1040: Taxable Income.  This is that amount that determines how much tax is to be paid, and has been lowered by the amounts that were put into tax-deferred accounts.  If we then add the tax-deferred amounts back into taxable income, and compare the tax with and without the tax-deferred amounts, we can compute exactly how much tax was actually saved.

To compute the taxes due, we had to go to the 1040 Instruction booklets for the years back to 1983 and look for the “Tax Rate Schedules”.  Luckily all these are available (as PDF files) from irs.gov.

Year	Filing Status	Taxable Income	Actual Taxes	Tax-deferred	Income with TD	Taxes with TD	Tax Savings
1983	J	38,295	7,707	1,125	39,420	8,101	394
1984	J	64,730	17,155	7,525	72,255	20,315	3,160
1985	J	89,981	27,377	8,625	98,606	33,109	5,732
1986	J	103,792	33,109	11,625	115,417	38,340	5,231
1987	J	113,132	33,496	9,875	121,882	36,865	3,369
1988	J	110,508	29,005	8,675	118,058	31,496	2,491
1989	J	132,772	36,048	8,475	140,122	38,474	2,426
1990	J	112,397	28,952	39,777	151,049	41,707	12,755
1991	J	120,357	30,426	51,730	170,962	46,114	15,688
1992	J	87,454	19,862	10,062	97,266	22,903	3,041
1993	HOH	59,718	12,873	11,780	69,498	15,611	2,738
1994	S	80,531	20,354	14,200	92,731	24,136	3,782
1995	S	77,028	19,146	13,548	88,576	22,726	3,580
1996	HOH	138,925	36,617	14,196	151,121	41,008	4,391
1997	HOH	89,944	21,025	14,960	102,904	25,043	4,018
1998	HOH	124,184	31,452	15,639	137,823	35,680	4,228
1999	HOH	127,731	32,430	14,055	139,786	36,167	3,737
2000	HOH	145,393	37,778	114,347	257,740	78,140	40,362
2001	HOH	133,394	33,344	19,967	151,361	38,824	5,480
2002	HOH	111,117	25,440	15,500	123,117	29,040	3,600
2003	HOH	124,451	27,594	17,500	138,451	31,514	3,920
2004	S	130,831	31,260	19,500	146,831	35,744	4,484
2005	S	142,641	34,446	22,500	160,641	40,011	5,565
2006	S	150,241	36,399	24,990	170,231	42,768	6,369
2007	S	198,449	51,556	25,500	218,949	58,321	6,765
2008	S	173,386	42,968	6,000	173,386	42,968
2009	S	172,428	42,044	6,000	172,428	42,044
Totals		3,153,812	799,863	527,675	3,620,611	957,169	157,306
				30%			16%

In this table, the Filing status was either Married filing jointly (J), Single (S), or Head of Household (HOH).  Taxes took 25% of taxable income.  By putting 15% of my income into tax-deferred accounts, I was able to lower my taxes by 15%.  Roughly 30% of the money put into tax-deferred accounts was tax savings.

The marginal tax rate for my income, which is the rate at which the tax-deferred contributions would have been taxed varied from a low of 28% (in 1993, due to the divorce) to a high of 45% (in 1984).

How did it grow?

Over time, the tax savings in the tax deferred accounts grew as they were invested and increased in value, either thru interest, dividends, or capital gains.  How much did it grow?  To determine this, we should look at the return on investment of my entire investment portfolio.

In addition to the tax deferred savings accounts, I also saved and invested non-tax-deferred money.  I managed all of my investments, both tax-deferred and non-tax-deferred, as one large pool, using a standard asset allocation policy.   I decided on the following asset allocation:

20%	Large Growth
10%	Large Value
5%	Mid Growth
5%	Mid Value
10%	Small Growth
10%	Small Value
20%	Foreign
20%	Domestic Bonds

Once a year, I would compute how far it had drifted from the desired asset allocation and bring it back into alignment, either by buying and selling, or by focusing where new money should go.  Most investments were in mutual funds.  I tried to stay with no-load funds which were in the top quartile of their peer group.

Looking at the complete set of investments, then we can compute the following table which shows, for each year, the total investment value at the end of the year, and the amount that was put in (or taken out) during that year.  From that we can compute an average return on investment for that year.

Year	Contributions	Investment Value	ROI
1983	–	–	–
1984	47,625	48,054	2%
1985	9,625	59,456	4%
1986	41,750	104,516	6%
1987	14,625	121,882	3%
1988	12,925	131,618	-3%
1989	10,475	158,641	13%
1990	45,927	210,767	4%
1991	55,584	292,847	13%
1992	10,026	320,833	6%
1993	11,158	372,841	13%
1994	(3,298)	363,138	-2%
1995	13,830	457,841	22%
1996	(29,314)	500,143	16%
1997	83,089	674,597	18%
1998	28,011	805,698	15%
1999	16,276	1,029,456	26%
2000	42,422	1,096,843	2%
2001	110,479	1,126,824	-7%
2002	7,038	970,641	-14%
2003	22,751	1,266,875	28%
2004	27,669	1,470,311	14%
2005	51,781	1,672,695	10%
2006	38,154	1,983,636	16%
2007	53,574	2,240,046	10%
2008	45,314	1,535,796	-33%
2009	(28,306)	2,075,572	37%

Returns varied from a low of -33% to a high of 37%; averaging about 14% a year.

We can then apply the annual return on investment (ROI) for my mix of investments to the tax savings computed for each year due to putting money into tax deferred accounts.

Year	Tax Savings	ROI	Growing Tax Savings	Total TD Investments
1983	394	–	394	–
1984	3,160	2%	3,560	7,363
1985	5,732	4%	9,424	17,243
1986	5,231	6%	15,179	31,896
1987	3,369	3%	18,946	43,737
1988	2,491	-3%	20,942	48,467
1989	2,426	13%	26,001	67,307
1990	12,755	4%	39,772	113,599
1991	15,688	13%	60,460	182,357
1992	3,041	6%	67,208	207,122
1993	2,738	13%	78,504	253,056
1994	3,782	-2%	80,937	268,477
1995	3,580	22%	102,542	349,235
1996	4,391	16%	122,973	415,444
1997	4,018	18%	149,456	502,279
1998	4,228	15%	176,523	593,807
1999	3,737	26%	225,718	781,946
2000	40,362	2%	271,554	843,863
2001	5,480	-7%	257,104	764,932
2002	3,600	-14%	223,463	667,190
2003	3,920	28%	290,344	870,918
2004	4,484	14%	335,111	1,004,504
2005	5,565	10%	375,001	1,141,167
2006	6,369	16%	442,526	1,368,364
2007	6,765	10%	494,541	1,561,859
2008	–	-33%	329,058	1,022,048
2009	–	37%	450,774	1,395,349

So we can see that $450,774 of the $1,395,349 in my tax deferred accounts is the result of tax savings, or the growth of those tax savings.

This is actually only an approximation.  Both ROI and the computation of growth over time, for a given ROI, depend upon both the balance at the beginning and end of the year, but also upon how much new money is added (or withdrawn).  But to accurately compute these values, we need to know the actual dates when the new money is added.  For example, if an account has $100 and we add $50, and it grows to $200 by the end of the year, we could have a ROI of 33% or 50%, depending on when the $50 is added.  If we add it at the very beginning, we have $150 growing to $200 over a year, or a 33% return.  If we wait and add the $50 at the end of the year, then the $100 grew to $150 over a year (50% return), and then we add the $50 to get our final $200.  The variation in ROI is nowhere near this extreme in our real life computations, but it will vary depending on when the contributions and withdrawals are made.  The extremes would be the contributions added or withdrawn at either the beginning or end of the year.

I don’t have the computing tools to do this exactly correctly, but the number above is a reasonable estimate.  Depending on how the ROI is computed and when the contributions to the tax-deferred accounts were made, the growth in tax savings could vary from $434,910 to $481,648.  With a total in tax-deferred accounts of $1,395,349, this gives a range of 31% to 34%.  If I can withdraw the money from the tax-deferred accounts and pay less than 31%, I will be better off than if I had not used tax-deferred accounts.  With 2010 tax brackets, a total income of up to $171,850, stays in a 28% marginal bracket, so if I had no other income in retirement, I should be able to easily access my tax-deferred accounts, paying less in taxes in retirement than I would have paid if I had not put money into the various tax-deferred accounts.

A Plan For Retirement

The objective of this blog is to arrive at a plan for retirement.  It appears that the plan is coming together.  This is not a detailed plan, but a structure that directs how to approach the decisions that will come up over retirement.  So it may be more a retirement policy than a plan.

First, when we retire, we will sell the company stock  and put that money in our brokerage account.  We will roll the company 401(k) plan over into our rollover IRA at Fidelity.  That should simplify our problems by limiting the number of different accounts to handle.  The principle goal is to reduce our finances to a reasonable number of things to handle.  We have already started by asking to have our Health Savings Account rolled over to our IRA.  It never amounted to much, and we may as well just do away with it.

While we want to reduce the number of accounts, we will continue to keep both Vanguard and Fidelity.  We also keep accounts at two credit unions, and we both a Discover card and a Master card.  I think it wise to always have a back-up, just in case something goes wrong.

These moves will reduce us to just four accounts:  brokerage and IRA at both Vanguard and Fidelity.  In addition we will create one new account, a Roth IRA.  We have some Roth 401(k) money and we will roll over into a Roth IRA.  This Roth IRA is to be the main focus of our retirement savings.  The objective is to convert as much of our regular IRA money into Roth IRA money as we can.  There are two reasons for that:

• The regular IRA money will need to be accessed.  Either we need it for living, or we can leave it there until minimum required distributions (MRDs) require that we take it.  If there is still money in the IRA when we die, it will pass to the children, but they will then be required to take it out.  The government agreed to postpone the taxes due, but they have structured it to assure that taxes will be paid on it.  We can’t just leave it alone, so we should plan how to withdraw it, on our terms, to hopefully minimize taxes (and any other costs).
• Given that the money must be removed from the IRA, it is always better to roll it into a Roth, rather than just taking it.  If we remove X from the IRA, we have to pay taxes on it, so we get a smaller amount (X-taxes).  If we keep it, and put it in a brokerage account, any growth or returns on it will then be taxes too.  On the other hand, if we roll it into a Roth, the growth is tax free.  Of course, if we spend it, then we have no issues with any earnings on the money.  But we can withdraw the funds from the Roth and spend it too, tax-free.

We have two bond funds — both at Vanguard.  One is our traditional IRA, which is invested in Investment grade Corporate Bonds.  It’s only $160K, so for now we’ll just leave it alone, and assume it continues to grow at about 4% a year.  It’s been doing that for some time.

The other bond fund is a recent attempt to lower taxes by investing in a tax-free municipal bond fund.  Tax free muni bonds can be a reasonable way to provide a dependable income stream.  But I have no experience in buying specific bonds, and individual bonds raise the risk of loss thru default, so I figure a bond fund eliminates the need for me to buy the individual bonds and spreads the risk of default over a much larger pool.  Unfortunately, bond interest rates are way down, so it’s not returning much nor growing.  So again, I’ll just leave it and try to learn from it.

That leaves the Fidelity IRA, which will include the company 401(k), and the Fidelity Brokerage account, which will be increased by the sale of the company stock.  These are our main accounts to work on.  The Fidelity IRA will be about $1.2M.  The objective is to drain the account and roll as much as possible over to a Roth IRA.  We could pull the entire amount out in one fell sweep.  That would cost about half a million in taxes, which seems excessive.

The problem with IRAs and taxes is that even if you have no other income, if you pull too much out of an IRA in one year, it will push you into a very high tax bracket.  The alternative is to pull a smaller amount out and spread it over many years.  If you pull no more than say $40K a year, your taxes will be only about $6K a year — 15%.  But of course if you start with $1.2M, and get a decent return on your money, say 8%, the account will grow by more than the $40K you withdraw, putting you further behind.

It looks like I can withdraw about $150K a year from the IRA, and after 10 years, be left with a much smaller amount — between $160K (assuming a 6% return) and $400K (assuming an 8% return).  Removing $150K a year from the IRA will cost about $33K in taxes — a 22% average tax rate.  That’s a bit higher than I might like, but to reduce that to only 20% reduces the withdrawal rate to about $110K a year, which would not be enough to make a sizeable dent in the the IRA over 10 years.

We want to roll this $150K a year over into a Roth IRA.  That allows the Roth to grow.  Assuming an average of 6 to 8 percent return a year, the Roth should grow to around $2M in ten years, assuming we withdrew nothing from it.

But we will need to pay living costs and the taxes, if we want to roll the entire amount over.  We can start by paying those costs from our Brokerage account. If we assume living expenses of $40K a year, and taxes of $33K, that means we will need $73K a year.  Again, assuming a reasonable rate of return, our Brokerage account should last for 7 years or so.  At that point we will need to either tap the Roth (meaning it won’t grow to $2M) or one of the other accounts.  That’s far enough out that I’m not going to worry about it.

Click on the following image to see a full size clear image of the initial Plan.

Initial Financial Plan

How to get money out of an IRA

For 35 years I’ve put money into my IRA. I put money into a retirement plan at the University that I rolled over into an IRA; I put money in 401(k) plans that where rolled over into an IRA; I put money into a Keogh that was rolled over into an IRA. In addition, I put money directly into an IRA (after taxes). And I’m continuing to put money into a 401(k) at work now, which I will roll over into my (rollover) IRA.

By rolling all of these funds into one rollover IRA, I can easily manage it as one large pool of investment funds, and, now, it shows the magnitude of the problem that I face. I have ended up with about $1.5M in tax-deferred retirement accounts — my traditional IRA, my rollover IRA, and my 401(k). I would expect to rollover my 401(k) into my rollover IRA after I retire, to minimize the number of different accounts to worry about.

Except for a very small basis in my traditional IRA ($60K), all of this money has not been taxed; as I withdraw it, I will pay ordinary income tax on it. So if I were to pull the entire $1.5M out all at once, I would pay about $499,371 in income taxes. (All tax computations are using http://www.moneychimp.com/features/tax_calculator.htm). So that’s clearly a 33% total tax rate.

If, on the other hand, I took out only $60K a year (and that was my only income), taxes would only be $8,844 or 14.7%. Since ordinary income taxes are graduated, the more I pull out, the more tax I pay. There are two ways to look at this: the marginal tax rate is the cost of pulling out each additional dollar. At $60K a year, the marginal tax rate is 25%. But the first dollars taken out have lower tax rates (0%, 10%, 15%, 25%, 28%, 33%, 35%), so the total (average) tax rate for a sum is lower than the marginal rate. So to pay the lowest taxes, I want to withdraw the lowest amount each year.

But there are two complications:

1. If I only take out $60K a year from $1.5M, that would be 4%. If the market goes up 6%, or 8%, the IRA will continue to grow. Eventually, I would die with more money in my IRA than I have now. (Of course, a good recession can fix that problem).

2. Eventually, for an IRA, the government requires you to take the money out, in the form of Required Minimum Distributions (RMDs). These start at age 70 (more or less) and initially the requirement is small (3%), but the percentage you must remove grows as you age. Eventually you have to take out 32.2% each year (of course you have to be 110 for this distribution rate)

But what’s the worst that could happen? If there is still money in the account when I die, it goes to my girlfriend or my kids, and then it’s not my problem! But that’s really irresponsible. If I just pull all the money out now, it’s not a problem any more either. The other possible problem would be that I withdraw the money too fast, and run out. But that seems unlikely.

So the basic problem is

(a) If I pull the money out too slowly, the account continues to grow and I (eventually) have to pay even higher taxes as I pull ever larger amounts out (under RMDs in the worst case), or

(b) I pull the money out too fast, and pay too much in taxes. (Of course “too much” is a debatable and vague term). In the worst case I could end up paying more in taxes than if the money had not been put in a tax-deferred account in the first place.

I’m 60 now; if I want to pull all the money out by the time I’m 70 (before RMDs kick in), I could pull out 10% a year. That ignores the compounding effect, but it seems unlikely that the markets will return 10% a year for the near future. Actually, Vanguard has a planning service run by Financial Engines that can analyze my investments and determine when I will (probably) run out of money, which is exactly what I want to do in this case.

While that seemed like a good idea, I’ve just spent hours and have been unable to even get Vanguard and Financial Engines to have the right starting point. Let’s just assume for now, that we will need to pull $150K per year out of the IRA. Of that $33K will go to taxes (a 22% average tax rate).

The rest would be rolled over to a Roth IRA. As near as I can tell, it is ALWAYS better to roll IRA money into a Roth than anything else. Which is to say, I only know of two options:

1. Pull the money out of the IRA to a non-tax advantaged account — just a normal bank or investment brokerage account. Pay ordinary income taxes on the amount pulled out.

2. Roll the money over from the IRA into a Roth IRA. Pay ordinary income taxes on the amount rolled over.

These look like the same thing, except in one case the money is completely unconstrained and in the other it’s still in an IRA (albeit a Roth IRA)! But let’s consider the future.

We expect to live on $30K to $60K a year. So $60K to $90K will be left-over. If we put that money in the Roth IRA, all of its earnings will be tax-free. Tax-free, not tax-deferred. If we put it in a bank or brokerage account, any earnings will be taxed, at best as capital gains, at worse as ordinary income. And we can always pull money out of the Roth — tax-free — since it has already been taxed.

In conclusion, we expect to pull $150K a year from our rollover IRA, for 10 years (or more if we have to) in order to drain it (or at least get it down to the point that we don’t have to take as much out of it). This will mean paying an average tax rate of 22% (under current tax rates) while moving it to a Roth IRA.

And stepping back and looking at the big picture, we can pay our $30K to $60K living expenses and the $33K income taxes on the Roth rollover from our current savings/brokerage funds, which will let us put the entire $150K into the Roth each year. If and when we run out of savings, we can start to pull from our hopefully fat and happy Roth IRA.

The Contribution of Social Security

My “full” retirement age for Social Security is 66 or 67, but I can file for early retirement at 62, or I can wait until 70. Claiming benefits at 62 provides the lowest monthly benefit; it goes up for each year (actually for each month) that I wait to claim benefits, until 70. After 70, benefits don’t increase. So I can take them as early as 62 and as late as 70. When should I file to take them?

There are political issues (the laws governing Social Security will certainly change since it is already running a deficit), and social issues (if I don’t “need” Social Security, should I even file?), but let’s ignore those for now and concentrate on just the financial aspects.

The Social Security web site gives the formula for how benefits change depending on when you start to receive them. In the case of early retirement, a benefit is reduced 5/9 of one percent for each month before normal retirement age, up to 36 months. If the number of months exceeds 36, then the benefit is further reduced 5/12 of one percent per month. After my full retirement age, a delay of a year increases benefits by 8% a year.

You can easily put together a spread sheet that has columns for the year that you claim benefits, and the rows are the amount that you receive cumulatively, as each year passes. If you want to maximize the return from Social Security, you then look for the largest amount, in a row. That tells you when to start Social Security benefits. It basically depends on how long you live.

If you die at 66, you should claim at 62. That gives you 4 full years of (reduced) benefits. If you live to 100, you should wait until 70 to start benefits; the loss of (reduced) benefits from 62 to 70 is more than made up by the 30 years of much higher benefits after 70.

In general, my analysis of the spread sheet results says that you should claim Social Security at 62 only if you don’t live past 75; if you can live to 85 or beyond 70 is the best age to start claiming. Between 75 and 85, the age varies from 64 to 69 (63 and 66 never are the best, but otherwise each year gets 2 years of being tops. Live to 76 or 77, and you should claim at 64, etc.)

Of course it’s difficult to know when you are going to die, but being optimistic, I figure I’ll probably live to about 90. That’s just the current working hypothesis.

And all this assumes that I can survive without Social Security benefits until I’m 70. That seems a safe bet, but obviously if you had to choose between reduced early Social Security benefits or no income at all, take the early benefits. But if you can comfortably wait, and expect to live beyond 85, and the rules don’t change between now and then, then your total Social Security benefits will be greatest if you wait until 70 to file for benefits.

The Details

The table below shows the results of such a spread sheet. In this table the columns select the year to begin taking Social Security benefits, from 62 to 71. The first row under the column headings, labeled “Base” shows the definition of benefits according to Social Security, in per cents. So my normal full retirement is 66; this is 100%. For each year after this that I wait, until 70, the amount goes up 8% — 108%, 116%, and so on. (Note that, as I understand it, this is 8% of full retirement, not 8% of the previous year, so it does not compound.) Notice there is no benefit to waiting until 71; benefits do not go up after 70. If I file for benefits early, the amount goes down, according to the formula provided by Social Security, so that at age 62, I would get only 75% of what I would get at 66.

In my case, I should get about 1783 at 62, 2414 at 66, and 3214 at 70. These percentages are slightly off what I am using (which is based on the Social Security web site).

The rows after that indicate the total amount received from Social Security at the end of each year. If we start taking benefits at 62, at the end of the first year we would have 12 months at 75% of the “full” monthly benefit, or 900. At the end of the next year (at 63), we would have two years, 24 months at 75% of full, or 1800. And so on. This first column shows the total amount that would be received, assuming we started taking benefits at 62.

The second column shows that if we wait until 63 to start taking benefits, we get nothing at age 62. At the end of age 63, we would have one year, but at a higher rate — 80% of “full” benefits, which gives us 960 at the end of age 63. Here you have the tradeoff between starting earlier, and having reduced benefits for a longer time versus starting later and having higher benefits, but for a shorter time. Starting at 62 gives a full year of benefits (900) before you even get started at age 63. Starting at age 63 give you higher benefits, 60 more each year, but you start off 900 behind. So for the first 15 years, you are behind, if you wait until 63. After 15 years, the two amounts are equal (14400 each), and after that starting at 63 returns more than starting at 62.

The other columns are computed similarly. If you wait until 66, “full” retirement, to claim benefits, you get nothing until you are 66, and then accumulate payments at 1200 per year.

Each row then represents, for a given age, how much you will have accumulated for each starting age. In each row one, or more, columns will be the largest value. There are times when two different ages will produce the same, maximum, accumulation. I’ve highlighted the maximum value in each row in red, to make it easy to spot the maximum.

Now deciding when to claim Social Security, so that your total accumulated benefits is maximized, is just a matter of figuring out when you are going to die. If you are going to die before you reach age 75, you clearly want to claim benefits at 62; if you are going to live to 85 or beyond, you want to wait until age 70 to start Social Security benefits.

Of course, there are lots of complications we have ignored in this table. We’ve left out the effects of inflation. The assumption is that Social Security benefits are indexed to inflation, so all the columns will go up by the same inflationary adjustment, but that ignores that the later payments will be in lower-value inflated dollars. It also ignores the potential investment returns from the earlier payments. You can only wait until 70 to claim benefits if you have some other source of income to keep you alive until you turn 70. If that’s true, then the earlier payments, if you start at 62, can be invested and produce a larger sum by the time you are 70 than the simple sum given. (Of course, your investments could also go down, so the accumulation could be small than shown too.)

All of these factors change the numbers somewhat, but they don’t change the basic idea or the basic pattern. And, to some degree, none of it matters unless you have a firm sense of how long you will live. I believe I will live to be 85 or 90 or 95. But it only takes one accident, one cancer, or one ill-timed heart attack to prevent that. My Father only lived to 82, but my Mother is approaching 88.

Age	62	63	64	65	66	67	68	69	70	71
Base	75	80	86.66	93.33	100	108	116	124	132	132
Sum after 62	900
63	1800	960
64	2700	1920	1040
65	3600	2880	2080	1120
66	4500	3840	3120	2240	1200
67	5400	4800	4160	3360	2400	1296
68	6300	5760	5200	4480	3600	2592	1392
69	7200	6720	6240	5600	4800	3888	2784	1488
70	8100	7680	7280	6720	6000	5184	4176	2976	1584
71	9000	8640	8320	7840	7200	6480	5568	4464	3168	1584
72	9900	9600	9360	8960	8400	7776	6960	5952	4752	3168
73	10800	10560	10400	10080	9600	9072	8352	7440	6336	4752
74	11700	11520	11440	11200	10800	10368	9744	8928	7920	6336
75	12600	12480	12480	12320	12000	11664	11136	10416	9504	7920
76	13500	13440	13520	13440	13200	12960	12528	11904	11088	9504
77	14400	14400	14560	14560	14400	14256	13920	13392	12672	11088
78	15300	15360	15600	15680	15600	15552	15312	14880	14256	12672
79	16200	16320	16640	16800	16800	16848	16704	16368	15840	14256
80	17100	17280	17680	17920	18000	18144	18096	17856	17424	15840
81	18000	18240	18720	19040	19200	19440	19488	19344	19008	17424
82	18900	19200	19760	20160	20400	20736	20880	20832	20592	19008
83	19800	20160	20800	21280	21600	22032	22272	22320	22176	20592
84	20700	21120	21840	22400	22800	23328	23664	23808	23760	22176
85	21600	22080	22880	23520	24000	24624	25056	25296	25344	23760
86	22500	23040	23920	24640	25200	25920	26448	26784	26928	25344
87	23400	24000	24960	25760	26400	27216	27840	28272	28512	26928
88	24300	24960	26000	26880	27600	28512	29232	29760	30096	28512
89	25200	25920	27040	28000	28800	29808	30624	31248	31680	30096
90	26100	26880	28080	29120	30000	31104	32016	32736	33264	31680
91	27000	27840	29120	30240	31200	32400	33408	34224	34848	33264
92	27900	28800	30160	31360	32400	33696	34800	35712	36432	34848
93	28800	29760	31200	32480	33600	34992	36192	37200	38016	36432
94	29700	30720	32240	33600	34800	36288	37584	38688	39600	38016
95	30600	31680	33280	34720	36000	37584	38976	40176	41184	39600
96	31500	32640	34320	35840	37200	38880	40368	41664	42768	41184
97	32400	33600	35360	36960	38400	40176	41760	43152	44352	42768
98	33300	34560	36400	38080	39600	41472	43152	44640	45936	44352
99	34200	35520	37440	39200	40800	42768	44544	46128	47520	45936