An asset class is a group of investments that have similar behavior and characteristics. When selecting investments it is vitally important to have estimates of three statistics for each asset class:

1. Expected Return – the rate of return anticipated on average.
2. Standard Deviation – the variability likely in those returns.
3. Correlation Coefficient – the manner in which two classes move relative to each other.

Very safe investments tend to have very low returns and very low chances of much volatility in those returns. More speculative investments tend to have both higher average returns and a higher chance that actual experience will be vastly different from that average. The third factor, correlation, is extremely important but frequently overlooked. With investments that move in different directions from each other, the portfolio will have a lower standard deviation of returns than its components. This is the rare case of a free lunch – diversification works.

We believe there are only two primary asset classes – stocks (equity) and bonds (debt). There is overwhelming evidence that the primary determinant of gross portfolio performance will be the portfolio allocation between these two classes. Net performance can be greatly enhanced by accessing these classes efficiently, though this is not often done. Within these two main categories, a number of sub-classes, while they may be useful, are vastly less important than getting the “big picture” right. The precise sub-classes used and the level of use will vary by client. Following are some typical subdivisions:

Equities (Stocks). Some exposure to stocks is the only way most people can reach their financial goals. We believe the risk of a portfolio losing money in a given year is far less important than the risk that the portfolio won’t last as long its owner. The following are common ways to classify stocks:

• Geographic location: Domestic, Foreign, Emerging Markets, Frontier Markets. We believe foreign exposure may reduce portfolio volatility, but investors frequently prefer low exposures to foreign companies and are willing to endure potentially higher volatility to be more aligned with domestic markets.

• Market capitalization (size): Micro Cap, Small Cap, Mid Cap, Large Cap. Though it appears smaller companies may perform better than their larger brethren over long periods of time, the effect seems less pronounced now than it once was and may not exist at all after adjusting for the higher risk level. Because of the often high transaction costs with these smaller companies, it is important to get exposure to them as efficiently as possible.

• Valuation: Growth, Value. A fair amount of academic research indicates that value stocks have higher returns and lower volatility over time. This appears to be true because investors irrationally prefer growth stocks and therefore systematically misprice growth vs. value investments. This effect appears to be larger for smaller companies.  For this reason we tend to tilt portfolios toward value in general and small value in particular.

Fixed Income (Bonds). We believe the primary function of bonds in a portfolio is to reduce risk. For this reason we do not attempt to seek out incremental returns by adding risk in this area. We believe that if you need higher returns, the appropriate approach is increasing equity exposure rather than attempting to get higher returns from the fixed income portion of the portfolio.

• Duration: This sophisticated measure incorporates the magnitude, timing, and discounting of future cash flows to give a measure of volatility. It is related to, though not synonymous with, the more familiar “maturity.” Because we believe the purpose of the bond allocation is to reduce risk, we have a large bias toward short duration instruments. Long duration fixed- income investments have higher volatility levels but not commensurately higher returns. At the other extreme, the shortest duration is simply a cash equivalent.

• Taxability: Some fixed-income instruments (Municipal Bonds or “Muni’s”) are tax free. We tend not to use these investments much because 1) marginal rates appear to be in a long-term (not short-term) declining trend, 2) bonds are optimally held in tax-advantaged accounts, 3) Muni’s only make sense for taxpayers in the highest tax brackets, and 4) they are riskier than U.S. Government bonds.  However, for high tax bracket investors without the “room” in their tax-advantaged accounts to accommodate their fixed income holdings Munis can make sense as a portion of the fixed income holdings.

• Credit Quality: High Yield, Corporate, Agency, Treasury. We strongly advocate only investment grade bonds in the fixed income allocation of the portfolio and therefore consider High Yield (aka “junk”, “non-investment grade”, or “speculative bonds”) to be inappropriate.

• Inflation Protection: Treasury Inflation Protected Securities (TIPS) are a relatively recent innovation (1997) with no appreciable credit risk or inflation risk. We generally use these instruments in significant quantities.

This has been a much abbreviated and simplified overview of the two primary asset classes and some of the sub-classifications you will frequently see, but so-called “alternative” investments are becoming increasingly popular. My next post will delve into those areas.

Many comparisons of vehicles with differing tax treatments are faulty because they don’t start at the beginning and go all the way to the end.  To perform a proper analysis, it is important to start with “Pat earns a dollar” and go all the way to “Pat spends the proceeds.” If the comparison starts with “Pat puts $X in the accounts,” or stops with “the account values are $X,” the conclusions are likely to be erroneous.

Here is a quick mathematical comparison of a Roth vs. a traditional (deductible) IRA.

Assume a 25% marginal ordinary income rate and a 15% capital gains that remains constant.

Scenario 1:

• You earn $1,000.
• You pay $250 in income taxes on it.
• You put the remaining $750 in a Roth.
• The investment grows until it doubles to $1,500.
• You take it out and get to spend $1,500.

Scenario 2:

• You earn $1,000.
• You put it in a deductible IRA (no taxes because it is tax deductible).
• It also doubles and thus becomes $2,000.
• You take it out and pay 25% ($500) in taxes.
• You get to spend $1,500.

As you can see, there is no difference between the two vehicles under those assumptions.  Here are the confounding factors:

1. If the marginal tax bracket is lower in retirement, it will favor a deductible option; if it is higher, it will favor the Roth option.

2. The analysis above assumed that you did not max out the IRA. If you contribute the maximum, it will favor the Roth.  Because the Roth takes after-tax funds, and the limits are the same, you can essentially get more earnings into a Roth.  An example may clarify this:

---

Using the same assumptions as above, assume you are going to save pre-tax wages of $8,000 while the maximum contribution level is $6,000 (this doesn’t match current limits of $5,500 or $6,500 if 50 or older, but makes the calculations simple to follow).

Scenario 1:

• You earn $8,000.
• You pay taxes on it ($2,000 in the 25% bracket).
• You place the remaining $6,000 in a Roth.
• It doubles to $12,000.
• You withdraw the funds and get to spend $12,000.

Scenario 2 – IRA Portion:

• You earn $8,000 and put $6,000 in a deductible IRA, (no taxes on that portion).
• The $6,000 in the IRA doubles to $12,000.
• You withdraw the funds paying $3,000 in taxes.
• You get to spend $9,000.

Scenario 2 – Remaining Portion:

• You pay $500 in taxes on the other $2,000 (of the original $8,000).
• You invest the remaining $1,500 in your taxable account.
• It doubles, but has some drag due to taxes (see below).
• At withdrawal, the after-tax amount available to spend is less than $2,775.

(The $1,500 in the taxable account will do less than double because of taxes on dividends, interest, and turnover; but even if we assume it did the full double, it would become $3,000 with a basis of $1,500.  At a 15% capital gain rate, the taxes would be $225.  So, even with a perfectly tax-efficient investment, you would have only $2,775.)

Adding the $2,775 to the $9,000 proceeds from the IRA is $11,775 which is less than the $12,000 from the Roth.  In reality, the difference would be somewhat greater.  (Note:  If you use very efficient investments like an index fund or ETF that has little turnover and small distributions and hold it till death, it receives a step-up in basis to your heirs, and the overall effect is similar to a Roth.  The higher the turnover and distributions, the more inefficient it becomes.)

---

3. Some people with access to a qualified plan may still be able to do a Roth even though their earnings are too high to deduct contributions to a traditional IRA.

4. The traditional IRA may be a little better if you are concerned about your children squandering their inheritance. Psychologically, having to pay taxes on discretionary withdrawals may be just the thing to restrain them from taking more than the RMD (Required Minimum Distribution) each year.

5. A major assumption is the tax rate in the future. This depends not only on your financial situation but also on tax policy at that time.  There has been a long-term trend toward lower marginal rates on a broader tax base.  While it seems likely that will reverse at least slightly in the short run, due to modern understanding of the Laffer Curve, rates will probably not rise to the levels many fear.  (For a tutorial, see http://www.youtube.com/watch?v=fIqyCpCPrvU.)  In addition, the deduction for a traditional IRA happens now, while the Roth is merely a future promise.  I am skeptical of government promises, and while it may seem inconceivable that Roth proceeds will incur taxes, at one point it was also inconceivable that Social Security benefits would someday be taxed.

6. Despite the preceding paragraph, given future uncertainty, a traditional IRA should probably be the default choice. If you achieve more financial success and are thus in a higher tax bracket, using a Roth will have been preferable.  In less rosy scenarios, the traditional IRA would be superior because the tax bracket upon withdrawal is modest.  Optimizing the good scenarios isn’t as important as mitigating the less favorable futures, thus favoring the traditional IRA.

7. If you desire to leave funds to charity upon death, a traditional IRA will give a tax deduction now for that future contribution. In other words, a tax deduction is received for putting funds into the traditional IRA.  A charity may be named as beneficiary (or contingent beneficiary) thus essentially receiving a tax deduction for that future bequest.  Not so with the Roth IRA.  Conversely, since charitable deductions that are limited by adjusted gross income (AGI) may only be carried forward for five years or until death, whichever occurs first, doing a conversion to create taxable income and thus avoid losing the deduction may be prudent.

8. If you have a taxable estate, a Roth is more advantageous since the income taxes paid upfront will no longer be in the estate, though there is an offsetting deduction to heirs upon recognition of the IRD so this issue is smaller than many assume. This factor is more important to the extent that heirs will stretch out their distribution and/or have state or local estate taxes since there is no offsetting deduction in those cases.

To recap, when both a deductible IRA and Roth IRA are available, Roth IRAs are superior when:

• Tax rates will increase for you in the future (either because of your personal situation or because tax rates in general change).
• You are trying to save more than the limits (or have other taxable investments to pay the taxes).
• You have a taxable estate.
• You do not want to be forced to take RMDs after age 70½.
• You trust the tax-free treatment of Roth IRAs will continue in the future.
• You will not leave the funds to charity.
• You will leave a portion to heirs who are in a higher income tax bracket.

Traditional IRAs are superior when:

• Tax rates will decrease for you in the future (either because of your personal situation or tax rates in general change).
• You are not trying to save more than the limits (and don’t have other taxable investments available to pay the taxes).
• You don’t have estate tax issues.
• You will need to withdraw funds to live on in retirement anyway.
• You don’t trust the Congress not to change the tax treatment of Roth IRAs in the future.
• You will leave the funds to charity.
• You will leave a portion to heirs who are in a lower income tax bracket.

You and your advisor will have to weigh the relative importance of these factors to decide which vehicle is right for you.

Roth conversions may be analyzed in exactly the same manner as above, and will frequently be advisable to the extent the taxes on conversion can be paid out of funds that are not currently tax advantaged, and the conversion will not place you in a higher income tax bracket.  It may be advisable to perform “opportunistic conversions” by converting the maximum amount each year that will not cause migration into a higher marginal income tax bracket.

Finally, a non-deductible IRA is inferior to both the deductible and the Roth unless you intend to convert it to a Roth in the future.  It should be compared to a taxable account where, given a very long time horizon or very tax-inefficient investments, the IRA has advantages.  However, because of the lack of flexibility (early withdrawal penalties), more difficulty harvesting losses, ordinary income rather than capital gains tax treatment, and loss of the step-up in basis on death, I would generally choose a taxable account over a non-deductible IRA unless there are significant tax-inefficient holdings in the portfolio that exceed the amount of tax-advantaged space.  In other words, your high returning, ordinary income property is being held in your taxable account because there isn’t room in the IRA or Roth.

It may be advisable to initially convert larger amounts than you actually intend to convert in order to exploit the opportunity to recharacterize the “losers.”  For example, you have $100,000 in an IRA and want to convert $20,000 of it to a Roth.  Converting all $100,000 in January into separate Roth IRAs for each asset class may be advisable.  Suppose, in this case, $20,000 is placed in each of five Roth IRAs, each invested in a different type of investment.  Prior to October 15^th of the following year (the tax due date with extensions) the four Roth IRAs that appreciated the least can be recharacterized back to an IRA leaving only the “winner” converted.

My last post mentioned at the end that many people purchase annuities for the guarantees.  I thought I would expand on that by posting here an analysis I did a few years ago of some examples of “market linked” or “equity linked” products with some sort of guarantee.  These “guarantees” are attractive propositions to those who are understandably gun-shy after watching the market plummet by about half.  They aren’t outright theft in the way that schemes perpetrated by Bernie Madoff, Allen Stanford, and others are, but they are harmful to investors all the same.  These gimmicks are extremely popular in insurance products and perform similarly.

Product One

An advisor at one of the largest investment firms emailed the pitch below to his clients recently.  The firm’s name has been removed to protect the guilty, but it is otherwise unchanged:

[Firm Name] is rolling out our market linked CD today:

(Get market performance on your insured CD without risk of losing $)

* Must be purchased by September 8, 09.

* Linked to S&P

* Approximately 5.5 yr term

* Performance Cap: 45-65%.

* FDIC Insured?  Yes up to $250K

* Downside risk: None, if S&P has a negative return at end of term, 100% of principal is returned.

* Minimum purchase: $ 4000.

The blue line was blue in the original, presumably for emphasis, even though it is clearly inaccurate.  “Market performance” is simply not available without risk in this product or any other.  The performance cap ended up being 60%, as it had been in earlier offerings, which is obviously not “market performance.”  Now, here are the problems:

1. The tax treatment is horrible. Owning stocks gives capital gains.  This product delivers all ordinary income, which is taxed at a higher rate.  Worse, the investor is taxed on the income during the 5½ year term even though nothing is received until the end.  (This is called “phantom income” in the tax world.)
2. This product is massively illiquid. During the term, there is no active market in which to sell the CD and get the money back, and the investor is not able to cash it in prematurely.  Hedge funds don’t even have 5½ year lockups.
3. The “S&P” mentioned is the S&P 500 price only index. The S&P 500 was created in its current form on March 4, 1957.  Using all of the monthly returns from April 1957 (the first full month after its creation) through August 2009 (the last full month before this CD was to be purchased), the average annual return for the S&P 500 was 9.65%.  Without the dividends, it was only 6.13% – a not insignificant 3.52% annual difference.
4. The gain is capped at 60%. In other words, compared to investing in the S&P 500 outright, in exchange for guaranteeing investors won’t lose money over a 5½ year period, the firm is taking 3.5% off the top (from the dividends) and then 40% of the rest of the return.

Of course, if investors are spared from some truly horrific results, it still might be worth it.  But since the investor is only getting 60% of the returns, what if he or she simply put 60% of the funds in an S&P 500 index fund and the other 40% in something conservative like 5-year government bonds?  On average, the CD would have returned 4.86% (annualized), the 60/40 portfolio 9.42%.

In short, the investor who bought into the pitch gave up an average of 4.55% per year (difference due to rounding) for the downside protection.  Again, that may have been worth it if the protection was frequently needed or if when it kicked in it averted catastrophe.  There were 564 rolling 5½ year periods from April 1957 through August 2009 (when I did this analysis).  The CD outperformed the simple 60/40 portfolio … once – the period ending in September of 1974.  Here it is graphically:

As you can see, generally it isn’t even close.

Product Two

Here is the description from the documentation:

The Notes Linked to the S&P 500^® Index due August 29, 2014 are senior unsecured debt securities of [Wirehouse] that provide (i) the possibility of a return, subject to a cap, if the ending level of the S&P 500 Index (the “Index”) increases from the starting level, (ii) the possibility of an 8% contingent minimum return so long as the ending level of the index does not decline by more than 40% from the starting level, (iii) return of principal if, and only if, the ending level of the Index is not less than the threshold level and (iv) full exposure to decreases in the level of the index from the starting level if the ending level is less than the threshold level.  If the ending level is less than the threshold level, you will lose some, and possibly all, of the original offering price of your notes. [To their credit, emphasis in the original.]

Here are the issues:

1. It is massively illiquid with a four-year lockup of the investment.
2. The “8%” return is holding period return (henceforth HPR) over four years – not annualized.
3. The upside is capped at 60-70% HPR, or 13.34% annualized (using the midpoint of 65% for the calculations.)
4. The tax treatment is ordinary income and is probably phantom income during the term (the IRS position is unclear).
5. The investor is an unsecured creditor of the company issuing the note during the term since he or she doesn’t actually own the underlying investments.

Given those issues, ideally this is held in a tax sheltered account, the firm stays in business, the investor doesn’t need or want the funds for four years, and the S&P 500 performs as it has historically.  Using month-end data for every rolling four-year period since the inception of the S&P 500 (March 1957), never (out of the 593 rolling periods) was the ending value of the S&P below 60% of the starting value over a four year period, so the massive downside never occurred and thus isn’t isn’t reflected in the analysis to follow, and the investment got “capped” at the 65% return 101 times (17 percent of the periods).

Essentially, this is a simple options strategy:

1. Buy zero coupon bonds that at maturity will be worth 108% of the original investment.
2. Buy a four year S&P 500 call with a strike price 8% above the current value.  (8% was already received from the bonds; this gets the rest of the return.)
3. Sell a four-year, European-style, call (that means it can only be exercised at expiration) with a strike price at 165% of the market value (this limits the upside to the cap).
4. Sell a four-year, European-style put with a strike price at 60% of the market value (this captures the downside risk that exists with this product).

It looks pretty good at first glance.  It never lost money; the worst return was 1.94% annualized – the 8% HPR – and it had average annual returns of 6.99%.  While it certainly underperformed stocks, that isn’t an appropriate comparison since stocks experienced far more downside even over a four year period.  In fact, the return on the note during that period is exactly the same as five year treasuries have earned, so it is far more similar to fixed income.

So what would be a fair comparison?  I mixed investments in the S&P 500 (total return) and five year constant-maturity treasuries to get a portfolio of equivalent risk.  A 25% stock/75% bond mix never had a return lower than 9.19% – higher than the 8% “guarantee” from the note, and of course this investment is much better in terms of liquidity, taxes, default risk, etc.

The returns on the 25/75 mix are one percent higher on average as well.  So, when purchasing the note, essentially the investor gives up 1% on average in exchange for worse ancillary features.  In addition, the risk, as measured by standard deviation, of the 25/75 mix is lower.  This isn’t a great comparison because the note has the ends of the distribution truncated, which is why I used minimum return as the risk metric.  The index data I used doesn’t have expenses, but since SPY (the S&P 500 ETF) has an expense ratio of 9 bps and an equivalent treasury ETF (the 3-7 year iShare) would be 15 bps it doesn’t change the conclusion at all.

Also, keep in mind that although the “knock-in” if the market is down in price by more than 40% (excluding dividends) never happened in the last 50 years of data used here, if it did happen the investor would get killed in this note versus the 25/75 mix.  And “killed” is not an exaggeration (though it is clearly metaphorical).

To make the comparison fair, and incorporate that downside risk, the investor would buy the 25/75 mix and then sell four-year naked puts with a strike at 60% of the current price.  I don’t know of many advisors who would think that was a prudent strategy.  It would make the historical returns even better because the investor received the option premium and yet was never exercised upon.

Here is the graph of hypothetical historical performance, without including the extra income on the blend strategy from selling a naked put (essentially the red lines would be even higher):

Product Three

An advisor was extolling the virtues of “Reverse Convertible Notes” to me, so I asked him to send me the prospectus.  Here is what I found:

The investment is a 6-month note that pays 10.15% (annualized).  But, if over that six months Exxon stock (the reference security used in this case) trades below 85% of its initial value, and is below 100% of the initial value at the end of the period, the investor gets the stock value rather than the original investment.  (This is called being “knocked in.”)  Overall this sounds attractive, but…

Using data back to 1970 (which is all I had handy) on XOM (Exxon), the investor would have been “knocked in” 22% of the time.  Thus 78% of the time the investor got (annualized) the 10.15%.  The other 22% of the time the investor gets the (annualized) 10.15% plus either the original investment (if the stock is back over the original price), or the value of the stock.

It may seem surprising, but since the upside is capped while the downside is unlimited, the average 6 month return from the strategy is a loss of 0.71%  Thus investing in this note underperforms burying the money in the yard (on average).  Obviously it has huge negative skewness and excess kurtosis which is fatal in combination.  Here is the frequency distribution:

Clearly these are a poor investment.  It is equivalent to buying fixed income while selling naked puts.  If the stock performs as it has for the last 40 years, the compound return from this strategy (annualized) is -1.99%.  So why was the advisor so excited about something that on average loses money?  Well, most of the time (80%), it did fine so the client is happy, not realizing they have done the equivalent of buying a coastal home with no insurance that gets washed away periodically.  The other years it seems great!

More importantly, and all too typically, the advisor got paid a lot to not look at it too closely.  The commission on this product is 1.75%.  So, since this is a six month deal, if they keep rolling them (which is what this advisor did), the brokers are getting 3.5% every year!  Commissions of that magnitude should have been a blindingly obvious sign that this is a poor deal for the investor.

In conclusion, these types of products are complicated, gimmicky, and ubiquitous, but the lesson is old: “There ain’t no such thing as a free lunch.”

To make a case for annuities, advisors often compare the most expensive, tax-inefficient mutual funds to the most inexpensive variable annuities.  However, an objective analysis, will find that annuities appear to be poor choices in most (but not all) cases.

To fairly compare the two (and remove compensation considerations), I used a 0.30% marginal cost gleaned from the cost of a Vanguard “no load” annuity.  The 30 basis points are in addition to the underlying cost of the funds – vastly lower than the additional costs of most annuities that are sold.

Annuities made a lot more sense when the ordinary income and capital gains rates were the same.  Now, annuities have to overcome both higher rates (ordinary income vs. capital gains) and higher expenses.  Given a long enough time horizon, the 100% tax-deferral theoretically may overcome these disadvantages.  Seeing if that is true is the point of this exercise.

Six factors impact the decision:

1. Rate of return – higher returns favor annuities; lower returns favor taxable accounts.
2. Ordinary income tax bracket – lower ordinary income tax brackets favor annuities; higher ones favor taxable accounts.
3. Capital gains tax bracket – higher capital gains rates favor annuities; lower ones favor taxable accounts.
4. Tax efficiency of taxable alternative – tax-inefficient, high-turnover investments favor annuities; low-turnover approaches favor taxable accounts.
5. Time horizon – long time horizons favor annuities; short horizons favor taxable accounts.
6. Annuity cost – the lower the marginal cost of the annuity over the comparable mutual fund investment, the better it will compare (obviously).

Also note that I am not talking about immediate annuities which can have a place in a portfolio as insurance against the additional costs of a long life.

Here is one example:

1. Investment rate of return – assume 10%, the long-term stock market average.  (In reality, it would be somewhat lower due to the expense ratio of the subaccount/fund, ad due to lower expected returns currently but using this relatively high number should favor the annuity.)

2. Tax bracket – assume the most favorable case for an annuity, 25% ordinary income and 15% capital gains. (This is most favorable because the spread between the rates is lowest.)

3. Tax efficiency – assume a 50% portfolio turnover.  This is extremely inefficient and would favor the annuity.  I have assumed that all gains are long-term, however.  This implies an advisor would not be foolish enough to select funds for a taxable account that throw off short-term gains.  (In reality, I think a 10% turnover in a taxable account is more reasonable for a competent advisor.)

4. Annuity cost – as mentioned earlier, I used a 30 bps marginal cost to the annuity.  This attempts to remove compensation confusion from the analysis.  In other words, if the total expenses are 1.15% for a fund, the annuity would be 1.45%.  In this example, the net return ends up being 10.0% for the fund and 9.7% for the annuity.

In short, I have tried to use reasonable factors that would favor the annuity.  Using the numbers above, we solve (using my spreadsheet calculator here) for the time horizon necessary to make the annuity a better investment than the taxable alternative.  In this case the breakeven is 26 years.  In other words, a rational investor should not place in an annuity any funds he or she will need within the next 26 years.  If we change any one assumption, it just gets worse.  For example:

1. If the portfolio turnover is 10%, a 49-year time horizon is required to favor the annuity.
2. If the net investment is 8% instead of 10%, the breakeven becomes 34 years.
3. If the ordinary income tax bracket is 35%, the breakeven is 42 years.
4. If the ordinary income tax bracket is 15% or lower, the breakeven is never (because of a 0% capital gains tax rate).
5. If we make a conservative assumption that the market will return 8%, and our alternative is a passively managed investment with a 10% turnover (in essence combining 1 & 2 above), the breakeven is 62 years.

Some other factors:

1. In the case of death, the heirs are vastly better off with a taxable investment because of the step-up in basis.  The odds of dying in the early years (when the investor would be likely to have losses) are trivial vs. the odds of dying in much later years (when the odds are in favor of large embedded gains).  Remember, if there aren’t big gains, the taxable investment will be better; it is therefore irrational to use an annuity for “protection” for the very small chance that someone will die when it will be worse if they live.

2. Taxable accounts allow tax loss harvesting much more easily and efficiently.  Annuity losses have to exceed the 2% of AGI threshold, and the taxpayer must itemize.

3. If the investor needed the money early, he or she could be vastly worse off in three potential ways:  1) surrender charges, 2) the time period was too short to favor the annuity alternative, 3) early withdrawal penalties for pre-59½ distributions.

4. Using an annuity increases the standard deviation of returns relative to the taxable alternative.  This is contrary to what is desired.  In other words, for any given time horizon there is a rate of return where annuities and the taxable alternative are equivalent.  If the investment experience has been good (i.e. better than breakeven), then the annuity will be the superior choice.  If the investment experience has been bad (i.e. below breakeven), then the taxable alternative would have been better.  In other words, when purchasing an annuity, very good returns get better, and very bad returns get worse.  This is undesirable in most cases.

5. Finally, sometimes an advisor has placed an annuity inside of an account that is already tax advantaged.  The rationale is that the client is risk-averse and wants this “protection” even with the higher costs.  The expected payoff is computed by multiplying the average percentage the account is likely to be down (when it is at a loss), times the probability of being down, times the probability of dying.  This figure would be compared to the marginal cost of the annuity vs. the alternative investment in the account.  My calculations show this to be a bad bet because the probability of dying is too low – unless the annuity owner is in his or her nineties.

Let me dilate further on #1 above.  The death benefit has a computable value that will be greatest the very first year of the annuity because the investment has a positive expected return.  Even if some losses are bigger in years after the first one, the chance of them happening goes down even faster.  Using a mortality table, we can compute what a rational investor should be willing to pay for the insurance.  Still, that isn’t the whole picture because if the annuity has increased in value, the heirs lose on the tax treatment, and the odds of being up are much higher than of being down.

So, no matter what the mortality is, even if the investor dies after the first year, the annuity “protection” is a small net loss unless future investment performance will be dramatically worse than history. And, if that assumption is valid, purchasing an annuity is not optimal because high returns are required for it to make sense if the investor lives.  Note that that is the best year!  After year one, it gets dramatically worse. This means the “protection” is on average worth much less than zero because of the adverse tax treatment.

If the annuity is purchased within an already tax-advantaged account, we can ignore the second part of the analysis above and simply look at the benefit vs. how much the annuity costs (the incremental cost over an alternative mutual fund).  The downside protection is only worth the probability of being down, times the average magnitude, times the probability of death.  There are no annuities inexpensive enough to make sense in a tax-advantaged account, unless life expectancy is less than about 5 years.

Despite what our foregoing analysis, if an investor wants/needs to hold very tax-inefficient investments, and does not have sufficient “room” in tax-advantaged (e.g. retirement) accounts, and does not need the income generated, annuities can be the correct solution to put a tax-efficient “wrapper” around those inherently inefficient vehicles. (Tax inefficiency is a function of both the tax rate and the amount recognized each year.  Treasury bonds in the late 1970’s or early 1980’s for example were very inefficient holdings.)

Finally, annuity salespeople will claim that people don’t buy (i.e. they aren’t sold) the “plain vanilla” annuities I have used in my examples above.  That they are actually buying some sort of market guarantees or protection.  It is hard to debunk this because the products are changed constantly so it is like playing whack-a-mole, but I would simply point out that it is impossible for insurance companies to offer returns without the commensurate risk. They can do this with other types of insurance only because the risks are uncorrelated (all the houses don’t burn down at the same time).  In the market all the investments do tend to decline simultaneously so there is no advantage to risk pooling.

 

Tax loss harvesting is the sale of securities in a taxable account that have declined in value since their purchase in order to recognize the loss for tax purposes.  Most people (including professional financial advisors) do this at the end of the year determining whether the transaction is worthwhile by comparing the immediate tax savings to the transaction cost.  This is not optimal two different ways.

First, positions should be evaluated throughout the year for opportunities to save on taxes; there is no reason to wait until the end of the year.  Because investments tend to go up more than down and there is little serial correlation in the market, taking losses when they are economically meaningful is prudent.

Second, the calculation of tax savings and costs is more complicated than it first appears.  Consider first the standard analysis:

• Assume the original investment was $10,000.
• The investment has declined in value to $9,000.
• The investor is in the 15% tax bracket (i.e. this will offset other long-term capital gains).
• An equally attractive alternative investment that is not a wash sale violation is available.
• The transaction costs are $20 for each trade ($40 total – one sell and one buy).  Transaction costs should include not only the explicit commissions but also bid/ask spread costs and market impact costs.

This appears to be the proverbial “no brainer” – spend $40 to save $150 (a $1,000 loss times 15% tax savings).

In fact, given the information above, whether this is a prudent strategy or not is undeterminable.  Three key pieces of information are missing:

• How long does the investor expect to keep the “old” investment in their portfolio if they don’t harvest the loss?
• What will the tax bracket be at that time for that transaction?
• What is the hurdle rate for the return on this strategy?

Let’s assume if the investor doesn’t sell to take the loss it would be sold in 5 years anyway for some reason (perhaps we estimate this by simply knowing the investor has a 20% portfolio turnover rate).  Further, assume the tax bracket at that time will continue to be 15%.  The future transaction costs aren’t relevant since it will be sold at that point anyway – it doesn’t matter which investment it actually is.

At this point, it is simply a time value of money problem.  With tax loss harvesting, the investor saves $150 in taxes now less the $40 for transaction costs for a net of $110.  However, in 5 years the cost basis is $1,000 lower than it would have been had the old investment been kept.  Thus, the tax bill at that time is $150 higher than it would have been.  Essentially the $110 now cost $150 in five years.

Since the rate of return that will grow $110 to $150 in 5 years is 6.40%, this selling is only advisable if the current tax savings can be invested to earn more than 6.40%.  Alternatively, it can be viewed as an opportunity to borrow at 6.40% for 5 years.  That might or might not be attractive depending on the situation.

To recap, tax loss harvesting is more attractive to the extent:

• The loss to be harvested is large.
• The transaction costs are low.
• The alternative investment is attractive.
• Future tax rates will be low.
• The investment will probably be held for a long time.

Conversely, tax loss harvesting is unattractive if:

• The loss to be harvested is small.
• The transaction costs are high.
• The alternative investment is unappealing (such as keeping cash for 30 days to avoid the wash sale rules and then repurchasing the original investment).
• Future tax rates will be high.
• The investment will be sold shortly anyway for unrelated reasons.

Tax gain harvesting is potentially advantageous when a taxpayer finds themselves facing a lower rate in a particular year than they are likely to face in the near future.  The optimal strategy is frequently to harvest those gains and pay the taxes early, but at the lower rate.  For investments that will be held for the long term, but not too long, it may be optimal to sell and lock in the gain at the lower rate and then immediately repurchase the investment.  This would mean that only future gains will be taxed at the higher rate.

To prevent taxpayers from selling investments at a loss to get the tax deduction and then immediately repurchasing the investment (or doing something roughly equivalent economically) there are what are known as “wash sale rules” that disallow taking the deduction until the investment is “really” sold.  These rules do not apply to recognizing capital gains.

Many investors may have large carryforward losses.  The current tax code only allows the recognition of $3,000 per year of capital losses against ordinary income.  The remainder is carried forward to future years where it may be used against subsequent recognized gains.  For taxpayers in this situation, recognizing gains early will not be advantageous as they are not actually being taxed at the long-term gain rate, but rather simply using up some of the accumulated losses.

Assume a stock was bought for $50,000 and is now worth $150,000 and would incur taxes of $15,000 (a 15% rate) if sold this year and that the same sale in the future would incur taxes of $20,000 (a 20% rate).  If it is anticipated that the investment would be sold in ten years but instead is sold now and repurchased, the investor pays $15,000 today to avoid taxes of $20,000 in ten years.  That is a rate of return of just 2.92% (annualized).  It would probably be better to invest the $15,000 instead of paying taxes and earn enough to pay the $20,000 later with money left over.  (This simplistic analysis ignores transaction costs from the “extra” trades and taxes on the alternative investment, so the return is slightly overstated.)  Conversely, if it is anticipated that the investment would be sold in two years rather than ten, the rate of return is 15.47% (annualized).  That is compelling.  To recap, capital gain harvesting is more attractive to the extent:

• There are substantial embedded capital gains on investments in a taxable account
• The investor does not have offsetting recognized losses or loss carryforwards
• The investment will be sold anyway in a relatively short period of time.