What spending order is optimal for a retiree who is older than 59½ having three types of accounts: a taxable account, a Roth IRA, and a traditional IRA with no basis?

The default order is:

1. Taxable (preserves tax deferral in the other accounts)
2. IRA (no RMDs on the Roth, so this reduces future RMDs by taking the funds now)
3. Roth

In other words, the taxable account should be exhausted before drawing down the IRA which in turn should be exhausted before drawing upon the Roth. There are exceptions to the default order however:

1. To the extent the assets in the taxable account have a low basis and/or the owner’s life expectancy is short, it may be prudent to leave the assets untouched for a step-up in basis rather than sell them to live on.

2. After the taxable account is exhausted, if the tax bracket is abnormally high it may make sense to spend Roth funds ahead of IRA funds (or find some other way to get to next year and a more normal tax bracket such as tapping a HELOC temporarily).

Contrary to popular belief, if the tax bracket is abnormally low (or there is “room” in a moderate bracket) it is not optimal to withdraw extra funds.  Rather, (partial) Roth conversions to use those low brackets are a superior strategy.  (See this post for more on the analysis of Roth vs. IRA in general.)

The reason the Roth is last in the list is simply because, under current tax law, there are no RMDs on Roth IRAs (there are on Roth 401(k) accounts).  If all retirement accounts had RMDs (or no RMDs) they would all be equivalent (see this post for caveats though).

The real goal is to preserve tax deferral as long as possible (without triggering higher tax rates), so with that in mind, we can add a little complexity to the simple order above, this is a more complete ordering for a married couple who may also have inherited IRAs:

1. Taxable
2. IRA or Roth inherited by older spouse
3. IRA or Roth inherited by younger spouse
4. Older Spouse’s IRA
5. Younger Spouse’s IRA
6. Roth

This order assumes the spouses have similar ages.  It is possible that given a large enough age disparity options 3 & 4 could swap places temporarily.

What is the optimal mortgage an investor should have on their home? Should investments be liquidated to reduce the mortgage? Or, should the mortgage kept or even increased in order to increase the investment portfolio?

Though these are common straightforward questions, the answers provided are often simplistic and incorrect. I would like to discuss this in four parts. First, I will clarify the questions and address some common misconceptions. Second, I will address the analytical and rational analyses needed. Third, I will examine some emotional and behavioral constraints. Fourth, I’ll attempt to summarize, and make a few other observations.

Clarifying the Question. Whether considering immediately paying off a million-dollar mortgage from a portfolio or simply adding an extra $100 per month to the mortgage payment, the issue is the same. In both cases, the funds could either be invested or be used to reduce the mortgage, in effect “earning” the applicable interest rate. The easy answer here is, “Keep your mortgage; you’ll get a higher return on a larger portfolio.” Is the easy answer, however, the best for the investor? Each situation should be analyzed individually to determine the best solution.

(Note that, unless defaulting is a possibility, we are not addressing real estate exposure. The actual exposure to real estate is identical in both cases, and this decision merely affects the financing. To clarify, imagine that instead of having a mortgage, you have a debt identical in every way but unsecured – the real estate exposure is unchanged.)

Analyzing the Payoff. At its root, the question involves the asset allocation. Reducing the mortgage is effectively the same as purchasing bonds (technically it is a reduction in a short bond position). Conversely, taking out or increasing a mortgage is, in effect, issuing bonds. If the overall asset allocation already includes exposure to fixed income (as almost every portfolio does), what is the optimal way to determine that allocation?

Suppose a family has a portfolio of $1 million allocated 60/40 to stocks and bonds, along with a $200,000 mortgage. They thus have $600,000 exposed to equities. But while they are $400,000 long in fixed income, they are also $200,000 short in fixed income via the mortgage, for a net of $200,000. So, although the investment portfolio has a 60/40 mix, the allocation is actually 75/25 when looked at from a broader perspective.

Worse, people are typically paying on the spread. The long rate (bonds) is lower than the short rate (mortgage). When most people pay off the mortgage, what they really do is adopt a more conservative asset allocation.

When determining whether to pay off the mortgage, it’s important to keep the desired aggregate asset allocation unchanged. The asset allocation should have been determined earlier in the planning process using a Monte Carlo simulation or similar analysis. In addition, the comparison should be with the rate of return on Treasury bonds with similar duration.

This calculation may seem strange at first. But remember that a mortgage, although risky to the issuer, is actually a risk-free opportunity for the investor. If the investor pays off the mortgage, they are guaranteed to save the interest that they would have paid. Also, the duration of the mortgage is shorter than the duration of a bond with the same maturity, since a portion of each payment is principal in the case of the mortgage. For simplicity, the yield on a 10-year Treasury will generally be close enough in duration to a 30-year mortgage for comparison.

Tax considerations, because they tend to be the same for both alternatives, are not relevant. If an individual has a mortgage at 6% and is in the 25% tax bracket (ignoring any state income taxes), the after-tax cost is 4.5%. Alternatively, in purchasing a bond yielding 6%, the after-tax return is also 4.5%.

Other tax-related factors might favor paying off the mortgage, however. These include the impact of state income taxes and whether the taxpayer itemizes on their tax return (or is over the standard deduction by less than the mortgage interest).

In other words, if the gross rate of return on the fixed-income investment is lower than the interest rate on the mortgage (almost always the case), no further analysis is needed. If the fixed-income gross yield is lower but close, more detailed analysis may be necessary.

Looking at the Psychological Factors. The emotional implications of the mortgage-payoff decision are equally important. How an individual feels about having a mortgage is significant, yet more difficult to quantify. Many people are happy to be debt-free. Yet they should consider at least three other issues before tearing up the mortgage.

Perceived volatility. If an investor pays off the mortgage yet keeps their global asset allocation unchanged (reducing fixed-income exposure by the same amount as the mortgage payoff), the portfolio will appear more risky even though the total risk has not changed. This is a big problem; at a minimum, the investor should understand that their finances will look substantially more volatile than they really are.

Going back to our earlier example, the person with a 60/40 portfolio now has a 75/25 portfolio though the aggregate asset allocation has not changed. Since investment portfolios (but not mortgages) are “marked to market,” the position may feel more precarious. For people who anxiously examine every monthly statement, this shift could be an important factor in the payoff decision. It will be less important for more financially sophisticated people who understand the situation and have a long-term view.

Ability to stay with the target allocation. Some people will react to the increased portfolio volatility by panicking in poor equity markets and improperly reducing their exposure to stocks. Others, however, may feel more secure knowing that, no matter what happens, they have their home paid for and will be more willing and able to tolerate volatility and an appropriate equity exposure.

Propensity to save. Since most people in this country don’t save enough, paying off a mortgage can be problematic. People may feel poorer after reducing their portfolios to pay off the mortgage, and so they should be motivated to save to get their portfolio back to where it was. Yet behavioral finance has shown the reverse to be true; as people have more wealth, they are more motivated to save.

More significant is the fact that an individual without a mortgage must still save the amount of the mortgage payment to remain in the same financial position. In a way, a mortgage is a type of forced savings plan. If the person simply spends the monthly amount that was previously being spent on the mortgage, they have effectively reduced their savings rate. For example, in one real-life case, while the investor understands this analysis perfectly, he prefers to keep a relatively large mortgage to restrain household spending.

Proposing an Answer (and some other considerations). First, many investment advisers and planners have a conflict of interest in giving advice on this issue and should fully disclose that conflict. Advisers, whether they receive commissions on investment transactions or fees for assets under management, will reduce their compensation by recommending the mortgage be paid off from the assets in the portfolio. The reduction in fees, however, in my opinion is eliminated in the long run as the adviser gains a reputation for doing the right thing regardless of the personal cost thereby garnering more than enough business to compensate for losing managed assets initially.

Second, investors should almost never take funds from tax-advantaged accounts to pay off the mortgage. They should also maximize their contributions to these accounts before increasing their mortgage payments to pay down principal.

Third, in the case of a minister who needs housing expenses to preserve the tax-free treatment of his housing allowance, a mortgage may be preferred.

Fourth, an individual with a great deal of inflation risk (such as a retiree with a very large pension that has no COLA) may be well-served by a long-term, fixed-rate mortgage to function as an inflation hedge.

To recap, the traditional advice to keep the mortgage and collect a higher return from the portfolio is too simplistic; it compares a risk-free return with a risky return. Reducing the portfolio by paying off the mortgage is usually the correct answer from a logical standpoint because the rate of return on the mortgage is frequently (though not always) higher than the equivalent fixed-income investment opportunity.

Retirement spending rates are very important. Even for those not yet retired, this report should be useful because it indicates how to set an appropriate goal to enable comfortable retirement.

We generally try to use the phrase “achieve financial independence” rather than “retirement” since some folks don’t plan to retire; but, for the purposes of this paper, we’ll say retirement because we are talking about potentially spending down a portfolio which implies actual cessation of work.

When working with clients, we are actually trying to maximize happiness given an uncertain future. To compare retirement spending strategies, we use a mathematical technique known as Monte Carlo Simulation (MCS), which is used when attempting to model something with high levels of uncertainty (such as the future returns on financial instruments). A successful plan doesn’t run out of money and yet allows the maximum spending possible. (As many people have observed, optimal financial planning would have your check to the undertaker bounce.) These two objectives are obviously opposed to each other, and, to complicate it further, few people know how long the funds need to last (i.e. they don’t know their date of death in advance). MCS can help us analyze the probabilities of various outcomes. As the comedian Henny Youngman observed, “I have all the money I need provided I die by 4 o’clock this afternoon.” How much do we need?

Let me give the general conclusion first and then mention some items that may influence the conclusion. Many studies have been done on what are called “sustainable withdrawal rates” from a portfolio given a number of parameters. We have replicated the research and are comfortable with the conclusions. First, there are assumptions about the future returns and volatility of the portfolio. Sometimes purely historical figures are used; sometimes expectations of the future are used. Second, there are assumptions about what the spending pattern will be. Generally, not only are people uncomfortable with their spending having to potentially decrease; they would actually like it to increase over time to keep up with inflation. Third, people would like a high probability (85-90% chance) of not running out of money.

Let me talk about that last point for a moment. The reason you don’t seek a 100% success rate is that it leads to plans that are so conservative you could maintain your spending rate through two consecutive market crashes and great depressions, but wouldn’t get to enjoy yourself much in all the other situations. Remember, the goal is to maximize happiness by steering a middle course between 1) not running out of money and 2) enjoying your accumulated wealth (by spending it rather than merely looking at it or counting it).

Using those assumptions, for a typical person entering retirement (typical age, life expectancy, reasonable investment portfolio, etc.), the sustainable withdrawal rate is approximately 4%. In other words, if you desire $40,000 each year from your portfolio (in addition to Social Security and any pensions), you need a $1,000,000 portfolio. If you want $100,000 each year from your portfolio, you need $2,500,000. A simple way to compute this figure is to multiply your desired spending from the portfolio by 25 to get the nest egg needed. Alternatively, as you approach retirement, you can divide your portfolio by 25 to see what your initial spending might be.

Caveats and other points:

1. The figures mentioned previously are gross of taxes. In other words, the figures should be reduced by the taxes. If you are in a 25% marginal tax bracket and the funds are coming out of a traditional IRA or 401k, you would only net 3% from a withdrawal of 4%. If the funds are coming out of a Roth IRA, no reduction is necessary since the proceeds are tax free.

2. Many people with relatively modest lifestyles needs may have much of their need covered by Social Security. On the other hand, for younger people, it may be wise to discount the possibility of Social Security. Also, Social Security is very regressive and will replace a much higher proportion of a lower income than it does a higher income.

3. Some retiree costs, particularly healthcare, generally increase at rates higher than the overall inflation rate. On the other hand, many retirees gradually reduce their consumption (of items other than healthcare) as they age. One wag has termed this the three stages of retirement: go-go, slow-go, and no-go. To the extent a retiree does not have good health coverage (for long-term care for example) they may want to reduce their withdrawal rate slightly to have more available for that contingency.

4. The figures above are for someone beginning retirement at normal retirement age. Obviously someone who is 90 years old could consume much more than 4% of his or her portfolio each year with relative safely.

5. In the 10-15% of cases where the projection indicates running out of funds, the retiree shouldn’t literally run out. The problem should be evident early and lead the retiree to take action before actually depleting all of their resources. The cases where failure of the plan occurs are generally where very poor returns happen right at the beginning of retirement (someone retiring at the end of 1999 or 2006 for example). The options at that point would include 1) reducing spending (downsizing possibly), 2) generating more income (part-time job for example), or 3) a reverse mortgage (to tap home equity). This last option is why our retired clients generally have no debt. It leaves the home equity as an emergency back-up in extreme situations.

6. In many cases, spending can increase in later years by more than the initial plan indicated if portfolio returns have been good. This methodology in essence provides for all but the bottom 10-15% of cases working out. If, in fact, the retiree is simply in the middle case (the 50th percentile), they are in very good shape.

One final note: While many people have “don’t spend principal” as a rule of thumb, it isn’t the best way to look at your portfolio for three reasons:

1. It ignores the effects of inflation over time. If inflation is 10% and interest rates are 12%, spending just the 12% “interest” will deplete the purchasing power by 10% per year.

2. It can lead to poor investment decisions as high-yielding investments are selected over investments that might be safer or have a higher return but through growth rather than yield. Many individuals have bought bonds trading at a premium or extremely poor quality investments due to the high current yield. Buying exclusively high yielding investments will also increase exposure to interest rate risk.

3. For younger retirees, focusing on maintaining the real (inflation adjusted) value of the portfolio would make sense, but for someone older the portfolio can decline somewhat in value without serious risk. For example, imagine a 90-year-old with a $1,000,000 portfolio. He or she could spend $50,000 per year of the principle and it wouldn’t run out until age 110.

In conclusion, most people should plan to spend 4% (gross) of their initial portfolio balance in retirement and increase that slightly every year to keep up with inflation. About 10-15% of the time, an adjustment of some sort will be necessary, but the other 85-90% of the time the plan should work out just fine.

Periodically, the financial press discusses costs imbedded in various financial products, especially mutual funds. Some costs are explicit and fully disclosed, but many are not. We fully support disclosure of all costs so investors have information upon which to make good decisions.

Quality financial professionals do not try to make less than their competitors, but they do try to keep their clients’ costs extremely low. That may sound like a contradiction, but many products are designed to pay the salesperson without the client noticing the payment. Furthermore, these products are generally very expensive to own. Following, in no particular order, are some costs investors face. In most cases a good advisor can significantly lower these expenses.

Advisor Compensation. Financial professionals (including us) should be adequately compensated. However, the financial services industry often lacks a correlation between the amount of compensation and the quality of the advice. Among reputable, quality firms, advisor compensation is generally 100 basis points of the assets under management per year. (A “basis point”, or bp, is 1/100^th of a percent, thus 100 bps is 1.00%.) This is true regardless of the method of compensation. However, we believe some methods of compensation align advisor and client interests better than others. Methods of compensation, ranked by popularity, are discussed below along with potential conflicts of interest for each method.

• Commissions – The advisor is paid a commission for a sale. Commission-based fees often cause a tendency to focus on the product instead of the process, to turn over the portfolio too frequently, and to neglect the client unless a sale is imminent. The larger the account, the greater the commissions can be, encouraging a bias toward making the account as large as possible even if it is not in the client’s best interest. For example, rather than building a larger account, it may be better for the client to pay off the mortgage, enjoy spending some accumulated funds, or gift to family members for estate tax purposes. Also, commission-based sales people are just that – sales people. As such, they are held to a lower legal standard because sales people are not fiduciaries. Frequently, significant commissions are hidden from the client in the product cost particularly in insurance products (including annuities), so-called “alternative” investments, and B-share mutual funds.

• Asset-Based Fees – The advisor charges a fee based on assets under management. This is our primary method of compensation, but it is not without its flaws. There can be the same tendency to keep the size of the account high to the detriment of the client (see above), and the advisor gets paid even when not “doing anything.” Even though fees were instituted to reduce conflicts of interest, one of the largest U.S. brokerage firms pays their advisors (at least they did the last time I saw their fee schedule) three times as much to recommend riskier equities (150 bps) over safer fixed-income products (50 bps) in a fee-based account. To remove this bias, it is preferable to have the advisor charge the same fee regardless of the product.

• Hourly Fees – The advisor charges a fee per hour worked. One potential problem here is doing more work than the situation justifies. Also, this method, in effect, causes the client to feel that “the meter is running,” lessening the likelihood of having a deep enough conversation to get all of the questions answered. In addition, this method is less conducive to ongoing management of a portfolio.

• Project Fees – The advisor charges a flat fee for a specific project. We do planning on this basis. It is very similar to the previous method, but removes the “meter running” feeling as well as giving the client certainty about the final cost.

• Retainers – The advisor charges a fixed annual amount. While a few firms have moved toward charging annual retainers, it is not common yet. Since the retainer is normally based loosely on the amount of assets under management and the estimated number of hours of work for the coming year, in practice it works like a hybrid of the previous two methods.

Taxes. Few advisors pay attention to managing a portfolio with an eye toward maximizing after-tax return. If all of the investments are in tax-advantaged accounts like 401(k) accounts or IRAs, this doesn’t matter, but if taxable accounts are involved, the impact could be significant.

Brokerage Costs. When mutual funds make trades, they frequently “direct” those trades to a specific firm, and the commission is paid by, though generally not disclosed to, the owners of the fund. In return, the firm that executed the trade frequently kicks back (though they would never use that term) what are known as “soft-dollar” perks and freebies such as research.

Trading Costs. Each trade in a portfolio generates a cost from covering the bid/ask spread. As with brokerage costs, this is a hidden cost paid by the owners of the fund.

Expense Ratio. Within a product such as a mutual fund, there are fees for the fund management, marketing expenses, etc. The average mutual fund charges about 125 bps (per Morningstar). It would be higher for actively-managed stock funds and lower for bond funds or index funds. As noted above, insurance products, hedge funds, etc. can be significantly higher.

Turnover Costs. Though not a separate cost, high turnover will generally increase the tax costs, brokerage costs, and trading costs listed above.

Conclusion. Investors typically lose significant amounts through costs they are largely unaware of. Decades of research has shown that these higher costs reduce net portfolio performance. It should be noted, however, that while we believe keeping these costs down is not the most important function we serve (the financial planning process, asset allocation, etc. are vastly more important), in many cases selecting more efficient investments for a client can more than cover the advisor compensation.

The title of this post may be surprising considering one would rarely call term insurance an investment. I don’t sell insurance; however, I think there is investment opportunity in certain situations using a term policy with a return of premium rider (ROPR). Essentially, this is a normal 20- or 30-year level-term life insurance product, but you pay extra premiums and at the end of the term get a refund of all of the funds invested. This means that the time value of money has paid the cost of the insurance. It is also a non-taxable event because, from a tax perspective, you simply received your money (basis) back.

Let’s look at an example. This is very close to being a best-case scenario. It is not typical, but it should get your attention. Assume a 25-year-old male in great health needs $1,000,000 face, 30-year, level-term insurance. It will cost him $770 per year, but he could also get it with ROPR for $1,095 per year “investing” the difference of $325, which is put in at the beginning of each period. At the end of 30 years, he will get back the total premiums or $32,850 ($1,095 x 30 = $32,850).

[Note: It is important to compare the best policy without the rider to the best policy with the rider. Many times the company that offers ROPR doesn’t have the best rates on the insurance without the rider. The “investment” should be the differential between the insurance that would actually be purchased in each case, not a single policy with and without the rider.]

On a financial calculator, the calculation is simply:

In beginning mode (because premiums are paid at the beginning of the period)

N=30 (the number of years)

PV=0 (there is no value initially)

PMT=-325 (this is the amount saved each year, it is negative because it is paid out)

FV=32,850 (this is the amount received at the end, positive because it is coming in)

Solving for I, you get 7.0%

In Excel, the same calculation looks like: =RATE(30,-325,0,32850,1)

This is a guaranteed, after-tax return (assuming you live and pay the premiums) and should be compared to the rate on investment-grade municipals.

Another way to look at it is to assume the funds would be put in CDs instead. If the taxes were paid out of the fund and you are in the 25% federal bracket and a 6% state bracket (combined a 29.5% rate assuming they itemize), the rate needed on the CD to accumulate the same amount of wealth is 9.9% [7/(1-(.25+(1-.25)*.06))]. If you are in the top federal bracket (35%), it would be 7/(1-(.35+(1-.35)*.06)) or 11.5%.

A few caveats:

1. Obviously that is significantly higher than current rates, but it is a 30-year program and is thus very illiquid.
2. A Roth or IRA would be better if you could get 7% on a guaranteed investment in them.
3. Even if you can’t quite get as high a guaranteed return in a tax sheltered account, if the rate difference were small, I would be inclined to use the retirement account for the improved flexibility.
4. You must need the insurance anyway for risk reduction. The numbers are abysmal if the insurance is not needed. In that case, the entire amount of the premium is the investment rather than just the differential over the cost of pure insurance.
5. In general, the younger and healthier you are, the higher the return.
6. Your asset allocation target should be kept in mind.  The asset allocation should drive the investment selection rather than the reverse.
7. Of course, if the insured dies, he/she paid too much for the insurance, but the probability of dying with a term policy is so low that including that eventuality (and weighting it appropriately) shouldn’t change the results more than a very trivial amount.

You may wonder how the insurance companies can afford to have implied rates this high. I can think of two reasons: 1) lapses allow the insurance company to pay better; and young (healthy) customers are more likely to lapse their coverage, 2) the purchaser of this product has perhaps “signaled” to the insurance company that he/she is a good risk – poor risks wouldn’t tend to buy it. In other words, the purchase of the rider may give useful mortality information to the insurance company that can’t be captured another way.

I ran calculations a few years ago for a male in excellent health at ages 25, 35, and 45; and for 20-year and 30-year time horizons (i.e. six different scenarios). The 20-year policies were not as good relatively speaking, and in one case (35-year-old), a company had a 20-year term policy without the rider priced so attractively that buying the policy with the ROPR would actually be worse than just saving the difference in municipal bonds.

I was also curious if the same results could be achieved with traditional insurance. Because of the much larger contribution amounts, this would be a significant strategy if the numbers were similar. To make equivalent comparisons, I used a non-participating whole life policy so the returns would be just as guaranteed as the previous calculations. I also liquidated the policy at the same 20- and 30-year horizons and compared the results. Unfortunately, the non-participating whole life strategy underperformed buying term insurance and investing the difference in munis by almost 2% per year on average. It also underperformed the term with ROPR strategy by about 2.6% per year – again averaged across the six combinations of age and term. Throwing out the one case where the ROPR shouldn’t be used (see previous paragraph), the difference actually averaged 2.8% annually (again, net of tax).

If an individual is relatively young and healthy, needs life insurance coverage, and can afford to save in an illiquid vehicle, these products are certainly worth investigating.