Question: What bias affects both the cost-of-living adjustment (COLA) from the Social Security Administration (SSA) and the oft-cited 4% Rule for retirement income?

Answer: Both use last year's inflation. They look back when they should be looking ahead.

Admittedly, inflation in recent months is somewhat correlated with inflation a year from now. Since 1914, the Consumer Price Index (CPI-U) in any given month has had a correlation of 0.54 with inflation 12 months ahead. Over the last 30 years, however, the correlation has declined to 0.21, a value so low that it's next-to-worthless for making predictions.

By looking backward to predict forward, both SSA and the 4% Rule are, in effect, adjusting in arrears.  SSA waits until inflation has happened, and then modifies payments as of January the following year. Similarly, the 4% Rule advises that late in the year, you should find the past year's inflation and increase your retirement spending by that amount in the year to come. In both cases, the inflationary train has left the station and reached the next stop on its journey. You can catch the next train, but you'll always be one station behind.

A better idea, clearly, would be to use an accurate predictor of next year's inflation, something better than simply assuming that the future will mimic the recent past. How might one do that?

One idea is to use the median (middle) estimate of experts whose livelihood depends on precisely forecasting future inflation. As it happens, this method is easy to implement and reasonably accurate. The Philadelphia branch of the Federal Reserve Bank conducts a survey of professional forecasters and publishes the results quarterly. Over the last 30 years, the correlation between this forecast and the year-ahead value of the broad CPI-U measure of inflation has been 0.52. As shown by the chart below, the survey tends to ignore brief divergences, up or down, and predict something like an average trend.

​Indeed, the survey correlates extremely well (0.92) with future core inflation, as measured by the CPI without its volatile food and energy components. In effect, the forecasters, although asked to predict the broadest measure of inflation, CPI-U, are actually estimating core inflation. That's good because, as argued previously in this series, core inflation is a good way to manage a retirement budget. Our Safe-Payout calculator therefore uses the survey forecast as the recommended COLA for those who set their retirement budget with the "collared inflation" method (a safe variant of the 4% Rule).

Is it possible to do even better than the survey forecast? After all, the Philly Fed is not Lake Wobegon, "where all the children are above average." In the survey, some forecasters must do better than others, but we aren't told who they are or whether their superior performance persists over time. Cherry-picking the best forecasters would be perilous, admittedly, because statistical theory says they will tend to regress, doing less well in the future than they have in the past.

With that precautionary observation in mind, take another look at the chart above. It offers an additional forecast based on an econometric model from academic research. Shown by the red line in the chart, the model's correlation with the broad CPI-U metric has been 0.83 since 1978. The model is a simplified version of the so-called "triangle model" of Robert J. Gordon, an economist at Northwestern University who developed the model and has applied it to data for a 51-year period from 1962 to 2013. 

In the simplified form shown here, the model says future inflation depends on:

• Inertia, or the tendency of core inflation to persist. While Gordon uses a weighted average of the previous five years of inflation, my analysis of data since 1978 found equally good results by simply using the most recent 3-month average of the trimmed mean PCE measure of inflation.
• Demand for employment, as measured by the gap between short-term unemployment and a "natural rate." If, say, only 3% of the labor force has been unemployed for six months or less, and the natural rate is higher than that, then jobs are so easily found and quickly filled that upward pressure on wages and prices will soon push inflation higher.
• Supply of critical goods, which can cause inflation to fall or rise if temporary supply-shocks boost or diminish the availability, and thus affect the price, of those goods. In the simplified model, I used just one value, which had statistical significance for the period 1978-2015: an average of the energy component of the CPI for the previous quarter.

As developed by Gordon, the triangle model was designed to predict inflation one quarter ahead (three months), not a full year. In fact, the simplified version did very well indeed, when predicting next quarter's CPI-U, both since 1978 (0.93) and for the more difficult period of the last 30 years (0.74). However, when predicting inflation a year ahead, over the past 30 years, the triangle model did no better than the middle-of-the-pack for professional forecasters.

The Social Security Administration (SSA) began making annual cost-of-living adjustments (COLA) in 1975. They compute the COLA value in the early fall from three recent months of data on the Consumer Price Index (CPI).

Should those three months happen to witness a large but temporary gain or drop in consumer prices, the SSA COLA would affect payments for a full year, starting in January, as if the temporary event had been long-lasting. In 1986, for example, the prices of oil and gas were plunging when the COLA was computed. In 1987, although energy prices stabilized and inflation quickly returned to its previous level, retirees were stuck for the full year with SSA benefits set at a lower level because of the earlier, temporary downdraft. Of course, it can work the other way, too. A temporary uptick in the CPI when the COLA is computed can give retirees a nice benefits for 12 sweet months the next year.

There are better alternatives. But they are not the ones most often cited by commentators, which were covered in the first post in this series. Have a look at the chart below. It shows the SSA COLAs since 1975 (green dots), the full CPI-U index (blue line), and two alternatives for computing COLAs.*

Before considering the alternatives, compare the SSA COLA to the actual CPI-U. In 1980 and 2006, the SSA COLA was well above the CPI-U level, giving a nice boost to retiree's budgets the following year. In 1986 and 2010, both the SSA COLA and the CPI-U were low, but the CPI-U bounced back the following year to a normal level of inflation. Yet retirees had to budget for the previous year's low COLA of 1% or less.

The SSA COLA and CPI-U both vary from year-to-year to a visibly greater extent than the two alternatives, shown by the yellow and red lines on the chart. What are these alternatives, and what makes them different?

• Core is a measure that's been computed for decades as an alternative to the full CPI-U. It uses the same data, but leaves out Food and Energy. If domestic gas, home heating oil, or imported food should rise of fall temporarily, the full CPI-U will follow along, but the "core" inflation index won't. Wait a minute! Don't retirees have gas stoves, drive cars or take buses that use gasoline, and buy foods that may have volatile prices? Of course they do. That's why the CPI-E index ("E" is for "elderly") includes Food and Energy, in portions calibrated to the spending habits of older households. But because it includes them, the CPI-E has the same flaw evident in the chart for the SSA COLA. Any index the includes Energy is prone to the boom-and-bust cycle of the gas and oil business. Possibly, some Foods have the same failing, especially if they are imported and subject to foreign currency rates. Omitting Food and Energy from an inflation index simply removes troublesome volatility.
• Trimmed is another measure derived from the CPI-U. Like the Core index, it omits certain items from the calculation. Of the 40-plus items in the CPI-U, the "trimmed" index omits the 3 or 4 that showed the highest inflation in a given month, and the 3 or 4 that had the lowest inflation. If gasoline prices surged or collapsed in a given month, they are omitted. The same goes for utility bills or medical services or food away from home or any other component of CPI-U. The method is agnostic, however, about what to drop. Unlike the core inflation index, which always drops the same items, the trimmed index zeroes in on the current offenders, whatever they may be.

Both the core index and the trimmed index are available from the Federal Reserve Bank, going back 30 years or more. More recently, some have advocated a related index, the median CPI, which simply finds the component of the CPI that's right in the middle in a given month, at a level of inflation that exceeds half the components and is itself exceeded by the other half. 

To see how the core and trimmed indexes are better, consider the statistics in the chart below. It shows that over the 33 years since 1983, when both indexes were available, they had long-run averages similar to those of CPI-U and SSA COLA. Thus, neither the core index nor the trimmed index would have given a retiree more or less over the long run, than would have been generated by the full CPI-U or the standard SSA COLA. If the trimmed or core index gave more in a particular year, they tended to give correspondingly less in another year. The compound rate of the core index was virtually identical to the SSA COLA, while the trimmed index was about 0.1% higher. The median or middle rate over the 33-year span was virtually the same for the trimmed rate as for the SSA COLA.

Look next at the standard deviation, a measure of volatility. The trimmed and core indexes had less volatility than the SSA COLA and the CPI-U. This is statistical evidence of the pattern visible in the historical chart, where the path over time was smoother for the trimmed and core indexes.

The smoothness of the two alternative measures is important for another reason. It makes the trend of cost-of-living adjustments more predictable for retirees. This year's COLA is more like next year's inflation. In the historical chart, the 0% SSA COLAs in 2009, 2010, and 2015 would have been small but non-zero values, had the trimmed or core index been used. Conversely, the spike to nearly 6% for the SSA COLA in 2008 would have been a more normal value between 2% and 4% with the other two indexes.

The final chart, shown above, provides statistical evidence that the two indexes were more predictable. Both the CPI-U and SSA COLA were utterly unable to predict next year's inflation. Their correlation was virtually zero, both with their own value in the following year, and with the future value of the full CPI-U. In contrast, the core and trimmed indexes in a given year were highly correlated with their own value the following year (r > 0.7). They were even somewhat correlated with next year's CPI-U (r > 0.3).

As noted in the first post in this series, it doesn't help retirees to ask SSA to compute an index that simply weights all the normal components differently, whether in a way that matches the spending patterns of older individuals (CPI-E) or by a method that focuses on supposedly necessary but volatile expenses. What really matters to retirees is to have a method that offers reassurance on questions like these:

• Will I be able to live within my budget next year?
• Will my COLA this year be like it was last year?
• Over my retirement years, will my SSA benefits truly keep up with inflation?

Because an index that removes volatile components is both predictable in the near-term and aligned with inflation in the long-run, it is superior in addressing these concerns. This is the direction SSA should take in revising their COLA calculations.

* For CPI-U and Core inflation, the plotted value is a compounded annual rate of change. First, for a given month, the non-seasonally-adjusted value is divided by the same value a year ago. This value is then compounded for the months of July, August, and September, to approximate the three-month method used for SSA COLA. For Trimmed inflation, the plotted value is compounded in September over the preceding 12-month period, using the seasonally adjusted monthly values of the 16% Trimmed Mean CPI as calculated by the Federal Reserve Bank of Cleveland. In effect, all three measures are doubly smoothed. The CPI-U and Core measures first smooth out seasonal effects by comparing the non-adjusted, current month to a year ago, then smooth again by compounding over three months. The Trimmed value first smooths out seasonality by averaging the seasonal adjustments on each expense category within a given month, then smooths over the full year by compounding 12 months of rates. Although the methods are somewhat different, the Core and Trimmed values end up having very similar volatility.

Data sources: FRED for CPI-U, Core CPI-U Less Food and Energy, and 16% Trimmed Mean CPI. Social Security Administration for its annual COLA values.

This post was part of a series on building a portfolio of mutual funds or Exchange Traded Funds. Much of the series has been moved to articles on the Portfolios menu:

• Basic Portfolios: Some simple ideas that work remarkably well.
• Diversify: Effective ways to diversify with bonds and stocks, globally.
• Factor Investing: Specialized portfolios that employ techniques from academic research.
• Respond to Inflation: Realistic options for inflation-protection, near-term and long-term. (Not yet moved, this topic is covered below, in this post.

On the topic of inflation, Andrew Ang's fine textbook, Asset Management, offers many good insights, to which I am much indebted (see his chapter 11). I've superimposed my own analysis, however, and sometimes relied on additional sources.

It's helpful, I think, to focus on inflation with three sets of questions:

• Near-term: At a given time and for the next year or two, how should an investor's portfolio adapt to inflation or, equally important, to deflation? This question matters if, and only if, you plan to spend some of your investments in the next year or two.
• Long-term: What investments, if held a long time, will best protect against inflation? Do the same investments protect against deflation? And how long is long enough?
• Correlates: Both near-term and long-term, what investments (if any) are correlated with inflation (or deflation)? If one applies diversification criteria, do any of the popular candidates, such as TIPS, commodities, and real estate, truly add value?

An important conclusion will be that sensible protection against inflation and deflation is possible, but by different methods for the near-term than for the long-term. Furthermore, the popular candidates may not be the best or the only methods you should consider.

Here's an interesting test. You come to me in December, asking "What will inflation be next year?" I caution that nobody knows the perfect answer, but one indicator is quite good and amazingly simple. "It has correlated 0.64 with next year's inflation for the past 100 years," I proclaim. Then I show you data for the past century, using the chart below. Can you guess what my indicator is?

No, it's not the price of oil or gold. It's not the return on inflation-indexed bonds, or the median price of homes, or any exotic financial product. And it's a lot simpler than the economic model favored by this year's hot economist. It's last year's inflation. Yup. Next year's inflation is pretty well predicted by inflation over the most recent 12 months. That's because inflation (or lack of it) tends to persist. *

Despite the dire tone of some articles and commentary, inflation is not like a thief in the night who steals your heirloom jewelry, sets your home ablaze, and swiftly vanishes before dawn, leaving your wealth devastated. He's more like the unwelcome guest, your financially reckless cousin, who emails you one day and shows up on your couch the next, feasts from your pantry, maxes all your credit accounts, totals your car, and requires strenuous measures to be evicted, leaving you poor, thin, weary, and forgiveably ungracious after a prolonged ordeal.

The point is that you'll know when inflation is becoming a problem. One of the more obvious clues will be that interest rates are rising. That clue, as it happens, also defines your best defense.

T-Bills
As Ang and others before him have pointed out, the asset most strongly correlated with inflation and the best single hedge against it is a 3-month Treasury Bill. As inflation rises or falls, the interest rate on T-Bills tends to follow. Because T-Bills can be redeemed and reinvested at the newest rate every three months, the response to inflation, although lagging somewhat, is timely. 

Even so, it's not a perfect hedge, mainly because policies of the Federal Reserve sometimes (not always!) hold T-Bill rates below the inflation rate. That happened for nearly two decades during the Great Depression and World War II, and again, more recently, for six years and still counting since the financial panic of 2008-2009.

These limitations notwithstanding, a T-Bill offers good protection against inflation for expenses you know you will have to pay within the next year or so. Here's a concrete example:

• A hypothetical saver needs $1000 per month to pay expenses that are not covered by her other sources of income, such as work, Social Security, or an annuity.
• In late December, she has $3000 in a bank account (earning next to nothing) to cover her expenses through March. She withdraws $12,000 from her investment or retirement account, and buys a 3-month T-Bill. (She can do so easily and with no fees whatsoever at www.treasurydirect.gov.)
• In April, she redeems the T-Bill and puts enough in her bank account to cover the next three months' expenses. If her expenses have risen (or fallen) a bit in the past three months, she sensibly increases (or decreases) the amount set aside in her bank account, thus maintaining a smooth standard of living. (To simulate these adjustments, I exactly matched them to the change in consumer prices, up or down, for the previous quarter.) The remainder gets reinvested in a new 3-month T-Bill.
• She does the same in July and October.
• At the end of December, she looks at the amount she will get by redeeming the most recent T-Bill, to see if it will cover the first three months of the coming year. She wants it to cover her real expenses, which may have risen above $1000 per month because of inflation over the past year.

Had this strategy been followed every year from 1934 to 2014, the results would have been as shown in the chart below. The value depicted in the chart is the surplus (or deficit) over the amount needed for the first quarter of the coming year, adjusted for inflation.

Before 1952, when interest rates were restrained by the federal government, the results were weak. On average, the amount available after a year was nearly 9% less than the amount needed for the next three months. These were desperate times, financially, and it is not clear that any one-year savings plan could have done better.

From 1952 through 2014, the results were good. On average, there was a real 1% surplus, and surpluses occurred much more often than deficits. On the assumption that future federal policies may resemble those enforced since the early 1950's, the T-Bill strategy looks attractive.

Short-Term Treasuries
What if you were willing to take a bit of a chance on longer-term bonds? For example, how about a short-term Treasury fund whose bonds have average maturities around 2 years? One might expect such a fund to respond to inflation less quickly than T-Bills, but to pay a higher rate of interest and to be penalized somewhat less when federal policies suppress the interest on T-Bills. The next chart shows the results from using 2-year Treasuries instead of 3-month T-Bills.

Clearly, there is much more variation from year to year, which is not good for budgeting. On the other hand, the average surplus for the period 1952-2014 was higher, around 2%. In several years, the surplus was a full month's expenses. Here's an idea: What if the surpluses were carried over, to offset subsequent deficits?

A variant of that idea turns out to be highly effective, perhaps the best strategy of all. It works like this:

• In the first year, save two years' expenses ($24,000 for our hypothetical saver) instead of one ($12,000).
• Use a fund of 2-year Treasuries instead of T-Bills.
• Don't spend any surplus. Don't try to replenish any deficit. Simply add next year's expenses at the end of each year, adjusted according to the past year's inflation, exactly as in the T-Bill strategy.

The extra deposit at the outset serves as a cushion, creating, in effect, a reserve fund.

This strategy accumulates unspent surpluses over time, even in difficult periods. The chart above illustrates how the strategy would have worked in three very different 26-year periods. The first period, 1934-1959, saw a toxic mix of occasional high inflation, recessions with steep deflation, and persistent suppression of Treasury rates. During this time, the strategy lost some of its reserves but never went to ground. Bear in mind that the values in the chart are real dollars (inflation adjusted). Thus, the $3000 of buying power that remained on reserve at the end of 1959 would have seemed much larger in nominal dollars. The second period, 1961-1986, spanned multiple recessions and successive years of high inflation. Yet it maintained good reserves throughout. The final period, 1988-2013, benefited from a bull market in Treasuries and ended with enough surplus to cover more than two years of inflation-adjusted expenses.

Short-Term TIPS
Originally, they were called Treasury Inflation-Protected Securities, or TIPS. But they never truly guaranteed protection from inflation, and the Treasury Department now calls them Treasury Inflation Indexed Notes or TIINs, if their maturity is five years. The new name is apt because in recent times, their value as an inflation hedge has been about as good as, well, TIIN.

Surprised? Consider this example. In October 2010, you could have bought TIPS from the Treasury Department that would mature four years and eight months later, in April 2015. Those TIPS would have paid a measly interest rate of 0.5%, which would have seemed reasonable because T-Bills paid even less. At the time, the big institutions that bid at Treasury auctions expected inflation to start rising. So they bid up the price of the TIPS, thereby forcing buyers to pay, on average, $1055 for a bond that could be redeemed in nearly five years for $1000 plus inflation. They were counting on inflation to recoup the extra $55 and then some. As it all worked out, the cumulative inflation between October 2010 and April 2015 was 8.2% (equivalent to 1.7% per year). The buyer was therefore paid about $1082 in April 2015, plus those flimsy interest payments twice a year. For the initial investment of $1055, the buyer got a real (inflation adjusted) total return of about 1% per year. The five-year bet on inflation paid off, in this case.

But what if the original buyers' expectations were wrong? It can happen. In July 2014, no one anticipated that inflation for the next 12 months would be virtually zero, but that's what it was (0.2%). If you had bid on TIPS in early 2014 expecting a 1% real return, you would be holding a loss after the first year. Because mutual funds and ETFs buy, sell, and redeem TIPS continuously, they are vulnerable to such miscalculations. They can lose money when their inflation bets are too high. 

For some evidence, consider the chart below. It shows the real (inflation-adjusted) returns since late 2009 for two funds: STPZ, the PIMCO ETF for short-term TIPS; and VSGBX, a Vanguard fund for short-term bonds from the Treasury Department and other federal agencies. The PIMCO fund holds TIPS exclusively; the Vanguard fund holds no TIPS at all. Over the six-year period, both funds virtually matched inflation, with a terminal value near $1.00 for every dollar originally invested. But the TIPS fund was more volatile, rising or falling as its inflation bets came out winners or losers.

It would be nice to know if the long-term trend will resemble the short history shown in this chart. Alas, six years is the longest history available for a short-term TIPS fund as of September 2015. Pending additional evidence, a rational investor could use a short-term TIPS fund, instead of short-term Treasuries, but the outcome might not be any better, and the ride may be less smooth. An educated guess is that short-term TIPS may do better when inflation is rising, while short-term Treasuries might excel when deflation and recession are lurking.

Academic studies and historical data imply that near-term protection against inflation (and deflation) can be secured by either of these strategies:

• Hold a year's worth of anticipated expenses in 3-month T-Bills, redeeming and re-investing every three months. Virtually as good, use a savings or money-market fund if it is safe, very liquid, and pays interest at or above the rate of T-Bills.
• Or, hold two years of anticipated expenses in a fund of short-term U.S. government bonds. Withdraw expenses for a year, and replenish with anticipated expenses for the next year, keeping any surplus on reserve. A fund of short-term TIPS may be reasonable, but other U.S. government bonds are likely to offer more consistent performance.

In the next and final post in this series, we'll look at long-run inflation and the best strategies to mitigate its effects on your portfolio.

* In the chart, an outlier that goes against the trend is the solitary point in the lower right quadrant. It's for 1921 when, after World War I brought inflation that exceeded 15% annually, the economy sank sharply into recession. Within the year, high inflation switched to extreme deflation. It's the only such whipsaw in the past century. In fact, it's an exception that proves the rule because the preceding inflationary trend persisted for four years, and the subsequent deflation lasted for two, followed by several years of low inflation and the booming economy of the Roaring Twenties.

Disclaimer: Historical data cannot guarantee future results. Although a mixture of bonds and stocks may be safer than investing exclusively in one class of assets, diversification cannot guarantee a positive return. Losses are always possible with any investment strategy. Nothing here is intended as an endorsement, offer, or solicitation for any particular investment, security, or type of insurance.

An annuity that gives you guaranteed payments, for the rest of your life, may seem attractive. But it likely comes with fees and restrictions that may be hard to decipher. Furthermore, the net benefit to you may be less than you think, and more expensive than other options. Still, for some, certain annuities make sense. The key questions are:

• What's an annuity, and do you need one?
• How can you find the best annuities from the best companies?
• What annuities should be avoided, because better alternatives are available?

For answers, click here to read our article on annuities, which is located on the Retirement tab. Or, for detailed examples based on our research, continue reading the post below.

Here are some examples of Single-Premium Immediate Annuities (SPIAs) and other options. They are based on actual quotes and real products, for a hypothetical married couple in their late 60's, seeking to invest $100,000 from an existing retirement account, with 100% benefits to be paid as long as either spouse lives. Although the quotes and products are genuine, I've made the companies anonymous because they may not operate in the state where you live, your age and other variables may differ, and the companies may impose contractual stipulations that depart from the quotes and website-data available to me. These examples should not be construed as recommendations to buy any particular security or insurance product. Rather, they are intended to illustrate how you might examine and evaluate annuity options applicable in your circumstances.

• SPIA-3%. This immediate income annuity raises its payments by 3% each year. The first year's payment is $4544, a 4.54% rate on the $100,000 deposit. With a COMDEX rank of 96, the insurer is among the top 30.
• SPIA-CPI. Also an immediate income annuity, this one increases its payments by the same percentage as the previous year's increase in the Consumer Price Index. Its first year payment is $4349, or 4.34%. The insurance company's COMDEX rank is 90, which puts it solidly in the top 100.
• Rider. This variable annuity is a partnership between a well established, highly regarded investment firm and an insurance company with a COMDEX rank of 92 (the top 60). Underlying the annuity is a mutual fund invested 60% in domestic and international stocks, and 40% in domestic and international bonds. The mutual fund, if purchased by itself, would have fees under 0.2%. Purchased as a variable annuity, it has additional fees of 0.29%. A guaranteed income rider adds 1.2% more. The total fees are pro-rated as 0.127% deducted automatically every month. In year one, for a $100,000 purchase, the annuity pays you $4500 (4.5%). Net, every month that year, the insurer deducts $501.67 from your account, paying $375.00 to you and $126.67 to itself. These values may change, depending on the performance of the mutual fund. It's possible for the annuity fees to increase while the payments to you remain flat. On the other hand, a legal rider to the annuity contract guarantees that payments to you will never decrease, and stipulates a formula by which they will increase if the underlying investments rise above their value at the time of purchase.
• Self-CPI. This option is not an annuity. It's a direct investment in the same mutual fund that underlies the Rider option. But the fees are lower, under $16 monthly, because none of the annuity fees apply, and the fund is held in an IRA account with no employer-related fees. It's a self-managed plan, where the investor uses the "collared inflation" method described here and implemented in our retirement income calculator. This method starts with an initial payment of $4020 (4.02%) because of the age of the hypothetical couple. It increases the payment each year by the smaller of (a) 6.7% or (b) last year's change in the Consumer Price Index. The limit of 6.7% is a "collar" which, historically, has been necessary to protect against depleting the fund in the worst case of 40+ years in retirement with inflation initially at extreme levels.

I was able to obtain market data since 1972 for the domestic stock, international stock, and domestic bond indices underlying the Rider and Self-CPI options (but not for the international bond index, which was added to the mutual fund only very recently). This data was used to simulate what would have happened if an investor had begun each of the options above under three scenarios:

• From January 1972 to December 2014, a 43-year retirement with wide variation in markets and inflation.
• A 21-year retirement starting in 1973, when stocks crashed then rose, and inflation started very high.
• A 21-year retirement starting in 1994, when stocks and bonds both rose and fell, but inflation was low.

The simulations are necessarily conjectures. The annuities being simulated were not available to be purchased during these time-periods, and had they been, their prices, interest rates, and contractual terms might have been different from what is sold today. Thus, the analyses reported below must be viewed as suppositions about what might happen if today's annuity products were to experience future markets for stocks, bonds, and interest rates resembling markets of the past.

First, the good news. All four options provided payments every year for 43 years. Even better, the Self-CPI option had a residual value of $209,650, adjusted for inflation, in 2014. Yes, you read that right. With the money left in 2014, the investor's heirs would have had twice the buying power of the couple's original $100,00 investment in 1972. The annuities, as annuities, had no residual value. Instead, any profits went to the insurers.

However, because 43 years is a long time, because inflation was high in the 1970's, and because the period included multiple large swings in the values of stocks and bonds, the four options paid out very different amounts from year to year. The chart below shows how they differed, over time. It's important to note that in the chart, all values are inflation-adjusted to show constant buying power, pegged to January 1972.

Viewed in terms of buying power, the SPIA-CPI option was the clear winner. When adjusting for last year's inflation, it sometimes lost a bit of buying power if current-year inflation was high. But over the long term, it guaranteed constant income, in real, inflation-adjusted dollars. 

The SPIA-3% option underestimated the high inflation of the 1970's and never caught up. Had this annuity been purchased with a 4% annual increase instead of 3%, the initial payment would have been a bit smaller, but the decline during high inflation would have been more limited. Eventually the initial buying power would have been nearly restored, as the compound rate of inflation for the entire 43-year period was 4.14%.

The Self-CPI option, because of its 6.7% cap on annual increases, fell behind inflation in the 1970's, but leveled out from 1980 onward. In effect, this method made a trade-off. It kept a large bequest as a result of cautious withdrawals.

The Rider option, when adjusted for inflation, fell short on its promise that payments would never decrease. Yes, in nominal dollars, the same amount was paid every year throughout the 1970's, but within a decade, inflation cut the buying power of those dollars in half. When huge bull markets in stocks and bonds began in 1982, the Rider option swiftly recovered, only to lose ground again after the dot-com bubble burst in 2000. Of all four options, this one had the least consistent payouts, adjusted for inflation.

The foregoing analysis comes with a big caveat. It's for a very long retirement in a unique historical period. Your experience in retirement will almost certainly be different. Consider, then, two other examples, both 21 years in length, one from January 1973 to December 1993; the other, from January 1994 to December 2014.

During these two periods, the underlying mutual fund of the Rider and Self-CPI options had compound, inflation-adjusted returns a bit higher than 5% per year. In this respect, the stock and bond markets were, overall, about average during both periods, despite some dramatic swings along the way.

At 21 years, both periods were also average in another sense. They fit the joint life-expectancy of a couple in their late 60's.

A key difference, however, is that in the first period, 1973-1993, inflation was abnormally high (6.05%), while in the second, 1994-2014, inflation was much lower (2.30%). Because of inflation, the SPIA-CPI option did the best job of maintaining payments at constant buying power in the 1973-1993 period; the other methods lost about 25%, falling close to $3000, inflation-adjusted. In the low-inflation period from 1994 to 2014, however, all the options preserved buying power, and the SPIA-3% and Rider payments actually bettered inflation. The chart below displays the average payments, after adjusting for inflation.

As shown in the next chart, only SPIA-CPI generated annual increases that kept pace with the 6% rate of inflation in the first period, while all the options matched or beat the tame 2.3% rate of inflation in the second period.

Finally, as with the 43-year analysis, the Self-CPI option was unique in preserving a bequest. Adjusted for inflation to reflect constant buying power, Self-CPI preserved principal approximately equal to the initial investment, in both 21-year periods: $90,756 at the end of 1972-1993, and $144,631 at the end of 1994-2014. The three annuities, by design, left nothing.

The examples presented here are for selected options and time periods. As such, they cannot be taken as guarantees of future performance or as definitive recommendations of certain products. Your results may differ, and will, as with any investment, entail a risk of loss. With those important cautions in mind, consider the following key insights from these examples:

• If one's goal were to guarantee constant buying power, SPIA-CPI would have been the best of the options examined above. A SPIA with a 4% or 5% annual increase might have performed comparably for periods of high inflation, but would have been more than necessary under normal levels of inflation.
• If one's goals were to preserve constant buying power and also leave a bequest, Self-CPI would have been the best of the options studied here. However, it would require active management by the investor (or by a paid adviser), which sets it apart from the annuity options.

Do you see your Social Security benefits as a retirement account? Your view, perhaps, is that you made deposits into a personal account with payroll taxes, and, when the time comes, you will withdraw from that account. Actually, Social Security was devised not as a government-sponsored retirement account, but as a form of insurance for retirees, the disabled and their children, and widows and widowers. Payroll taxes are premiums that are paid into a pooled fund, and payments to retirees and other qualified individuals are benefits to insure them against lost income. 

Viewed as insurance, Social Security benefits should be postponed to the latest possible age, because doing so maximizes the amount of the benefit. By waiting longer to collect your benefit, you get more insurance against loss of income later in life. You might need it if you live longer than expected or have large unplanned expenses. One excellent source of information is the website, Maximize My Social Security. For a modest fee, it provides a sophisticated online calculator to compute your best option for collecting benefits.

As detailed in other posts in this series, it is difficult to use stock and bond investments to construct a retirement portfolio that will last a lifetime and reliably protect against inflation. Social Security benefits address this difficulty because they are both adjusted for inflation and paid until you depart. There's a flip-side, of course. You can't bequeath your Social Security benefits to an heir (although a spouse, former spouse, or child may be entitled to benefits of their own, because of their relationship to you). Your stock and bond investments, however, can be bequeathed to a charity or a person, if any assets remain in your estate. Because of their complimentary nature, you need both Social Security and some investments. The following list spells out the reasons, based on the payout risks likely to be of concern to a retiree:

Inflation: Will it erodes your income?

• Social Security: Guaranteed protection.
• Investments: Some protection, depending on your investment strategy, but no guarantee. Read this post for details.

Payout Volatility: Can you rely on sufficient income every year?

• Social Security: Guaranteed delivery of the same inflation-adjusted amount, every month.
• Investments: Likelihood of higher payouts some years, lower ones other years.

Longevity: Will your income sources last as long as you do?

• Social Security: Guaranteed to last your lifetime.
• Investments: May last your lifetime if you adopt a good strategy, but failure is possible.

Emergency: What if you are struck by a costly event or illness?

• Social Security: No help here. Your benefits won't increase if you suddenly need more.
• Investments: If you are wise with your strategy, you can set aside some retirement savings as a reserve or contingency fund.

Bequest: Will you be able to leave an inheritance to your heirs or charity?

• Social Security: No help here. You have guaranteed benefits to you, not to your heirs. There's no account balance that would go to your estate.
• Investments: Again, it depends on your strategy. It's possible, as other posts in this series explain, to follow a plan that has a high probability of leaving something for your estate, provided you avoid emergency payouts.

If you have both Social Security benefits and retirement investments, do you also need an annuity? For many retirees, an annuity of any kind may add little value. 

To see why, consider a typical retiree. Her actual Social Security benefits might cover 40% or more of her pre-retirement income (see, for example, a classic study by Munnell and Soto). This figure would rise if she maximized her Social Security benefits, because the actual practice, as noted in the reported studies, is that most retirees minimize their annual benefits by taking them too soon. As it happens, 46% is about how much of a typical household income is paid for housing and food expenses, according to the most recent survey by the U.S. Department of Labor. Your circumstances may differ. Still, to a good approximation, Social Security benefits may suffice to pay for your most critical expenses.

Thus, your income from investments, although it may vary from year to year, may only need to cover about 60% of your budget. If your investment income were to fall by, say, 30% in a given year, that's really 30% of 60%, or 18% of your total income. The key question is whether your budget can sustain a drop of 18% or so. If you have a reserve fund or can reduce some optional expenses or have ample investments, the answer may be yes. If not, you may want to consider annuitizing some (not all) of your investments, thereby raising your guaranteed income to a higher percentage than the 40% or so that Social Security might cover. 

Should you chose to buy an annuity, it would be sensible to get one that resembles Social Security benefits. As explained in a related article, the best annuity is fixed (not a variable annuity); it pays for your lifetime (not for a limited period); and it's inflation-adjusted.

While your individual circumstances must guide your decision, you might, as a rule of thumb, consider your options very carefully before locking up more than two-thirds of your budget in the total of Social Security plus guaranteed annuities. Certainly, if your retirement investments are more than adequate to cover your income needs beyond what Social Security insures, then converting those investments to annuities would have little value. Other posts in this series provide some guidance on making such decisions.

Disclaimer​: Although a mixture of bonds, stocks, and guaranteed benefits may be safer than investing exclusively in one class of assets, diversification cannot guarantee a positive return. Losses are always possible with any investment strategy. Nothing here is intended as an endorsement, offer, or solicitation for any particular investment, security, or type of insurance.

This post is the fifth and last in a series about investing to protect against changes, up or down, in consumer prices. Previous posts argued that historically:

• Stocks have offered the best protection against inflation over the very long term (18+ years), 
• In the short term (0-5 years), investing exclusively in one type of bond fund may be ill-advised. Instead, a mix of short-term corporate and treasury bonds plus short-term TIPS may be prudent.
• Over a holding periods from 1 to 20 years, intermediate-term Treasuries are paradoxical. They offer better average returns but a greater risk of lost purchasing power, compared to short-term Treasuries.
  
• Adding a small allocation to stocks may create a portfolio well suited for 1-year to 5-year investments, having compound returns like intermediate-term bonds and price-protection like very short-term bonds.

In this post, I investigate how increasing your portfolio's allocation to stocks may generate sensible strategies for longer-term investing of 5 to 20 years.

The portfolios in this analysis are based on the data described in a previous post. There I considered a mix of 10% U.S. stocks, 20% intermediate-term Treasuries, and 70% short-term TIPs and bonds, which historically has had compound returns like intermediate-term Treasuries and price-protection like diversified short-term bonds. As an investment strategy, this mix looked attractive for short time-horizons, perhaps 1-5 years. As a point of comparison for longer periods, I flipped the percentages of stocks and short-term bonds, creating this portfolio:

• 70% U.S. stocks
• 20% 10-year Treasuries
• 10% 2-year Treasuries

The chart below shows how, from 1920 to 2014, this portfolio compared to one invested entirely in stocks and to the portfolios reported earlier in this series. All the portfolios were rebalanced quarterly, if more than 3% out-of-balance, using methods explained here.

If you had invested in 1920, totally ignored all the exciting and disturbing events that transpired for the next 94 years, and cashed out in mid-2014, putting all your money on stocks would have been the best strategy. If, on the other hand, you were not able to anesthetize yourself against churning markets and had less than 94 years of patience, would the portfolio with 70% in stocks have been better? In the chart above, the answer is not obvious. Yes, there was somewhat less churn, but it was still a bumpy ride.

Suppose, instead, that your time-horizon is 17 years. Perhaps you are investing a lump-sum toward the future college education of a newly born grandchild. Or you are investing this year's 401(k) contribution, with a plan to purchase a life-time fixed annuity when you retire 17 years from now. (As a retirement strategy, this may be a sensible plan in some limited cases, if your overall savings are low.)

If you had invested for 17 years starting in 1928, your results would have been as shown in the next chart, which has the same set of portfolios. This particular period had deflation in the early 1930's, inflation during the 1940's, and a severe economic downturn during the Great Depression. Voila! In this instance, the portfolio that had 70% invested in stocks was the best of the lot.

Of course, this one example proves nothing. To see why, consider a second example, the 17-year period starting in 1966. Here a very sharp recession occurred, but no deflation. Instead, inflation was high and persistent. In this case, the bond-heavy portfolios all did better than the stock-heavy portfolios. However, the stock-portfolios kept pace with inflation overall, and a 70% allocation to stocks did a bit better than 100% in stocks.

The flaws in the analysis above are twofold. We should ...

• Consider all time periods, not just the two that started in 1928 and 1966.
• Examine stock-allocations that increase from around 10% for short holding periods to 70% or more for long ones.

Using the same method of analysis reported earlier in this series, I examined all holding periods of 1, 3, 6, 9, 12, 15, 18, and 21 years duration, since 1920. I did so for the portfolios shown above, with a new twist. Instead of a portfolio always invested 70% in stocks, I created one that grew the stock-percentage as the holding-time lengthened:

• For a 1-year period, only 10% went to stocks. 
• For each additional year, the stock-percentage grew by 5%, until it reached 80% at 15 years.
  
• After 15 years, the stock-percentage remained at 80%.
• The allocation to short-term bonds decreased by the same amount that the stock-percentage grew. Thus it fell from 70% for a 1-year holding to 0% for 15 years.
• The allocation to intermediate-term bonds remained at 20% throughout.

As shown in the next chart, this new portfolio rarely lost purchasing power. By this metric, the portfolio with a growing percentage of stocks did better than both 100% in stocks and 100% in intermediate Treasuries. It also did nearly as well as the portfolios that were stacked with short-term bonds.

Furthermore, when losses did occur for the new portfolio (under 10% of the time in holding periods from 6 to 15 years long), the amount lost tended to be small. As the next chart shows, the losses amounted to about a half of one percent in purchasing power, on average.

Importantly, the compound return of the portfolio with increasing stock-allocations was strong even after adjusting for inflation. The final chart shows how the inflation-adjusted return grew as the stock-allocation increased.

Considering all the posts in this series, historical data implies that an investor can get price-protection (against losses to inflation and deflation) and also get good compound returns, by crafting a portfolio in which:

• The allocation to stocks grows with the length of the holding period. This provides inflation-protection long-term.
  
• A modest amount is allocated to intermediate Treasuries for all holding periods. This guards against deflation and recessions.
• The remainder is allocated to a diversified mix of short-term TIPs and bonds. This assures some stability of returns, particularly for near-term spending.
• The portfolio is rebalanced approximately quarterly, if it is more than 3% out of balance.

One might use slightly different percentages than illustrated above, and still get good results. Furthermore, with the funds and ETFs available now, it would make sense to invest the stock portion in both U.S. and international stocks, and to diversify the short-term portion across Treasuries, 0-5 year TIPS, and corporate bonds.

Historical data cannot guarantee future results. Although a mixture of bonds and stocks may be safer than investing exclusively in one class of assets, diversification cannot guarantee a positive return. Losses are always possible with any investment strategy. Nothing here is intended as an endorsement, offer, or solicitation for any particular investment, security, or type of insurance.

This post is the fourth in a series about investing to protect against changes, up or down, in consumer prices. Previous posts argued that historically:

• Stocks have offered the best protection against inflation over the very long term (18+ years), 
• In the short term (0-5 years), investing exclusively in one type of bond fund may be ill-advised. Instead, a mix of short-term corporate and treasury bonds plus short-term TIPS may be prudent.
• Over any holding period, as short as 1 year or as long as 20, intermediate-term Treasuries are paradoxical. They offer better average returns but a greater risk of lost purchasing power, compared to short-term Treasuries.
  

In this post and another one to be published soon, I add some stocks to the mix, to see whether doing so might resolve the paradoxical properties of bond-only strategies and provide a sensible strategy for intermediate-term investing of 5 to 20 years.

The core question for today's post is whether a portfolio that contains a small allocation to stocks plus a good mix of bonds can meet two objectives:

1. The rate of return is at least as good as the historical rate for 10-year U.S. Treasuries. That's about 2.3% to 2.6% above the rate of inflation.
2. The risk of lost buying power is similar to 2-year U.S. Treasuries. It's consistently very, very low, even for holding periods as short as one year,

One reason to expect that stock investments would help a portfolio achieve these objectives is the experience of 2011-2014 in U.S. markets. During this period, inflation was very low, and short-term interest rates were even lower. Consequently, Treasury bonds and TIPS actually fell in value when adjusted for inflation, while stocks (and to a lesser extent, corporate bonds) rose. Another reason is that during a period of protracted inflation such as the 1940's, stock-prices might rise enough to compensate for losses in bonds.

As additional background on how a diversified portfolio can help investors manage the contradictory risks of inflation, deflation, and interest-rate polices, I recommend a recent article from Vanguard's chief economist and the manager of their TIPS funds. It's well worth reading.

To cover a long historical record that included inflation, deflation, and economic growth, stagnation, and recession, I analyzed data for the period from January 1920 through June 2014. TIPS did not exist for most of this time, nor did I have adequate data on corporate bonds or non-U.S. investments for the entire period. The analysis was therefore limited to three classes of investments:

• U.S. large-company stocks, as represented by the S&P 500 with dividends reinvested (from Shiller).
• U.S. 10-year Treasury bonds, with interest reinvested (also from Shiller).
• U.S. 2-year Treasury bonds, as reported by the U.S. Treasury Department since 1976, and extrapolated backwards to 1920 by a statistical model that I developed which had virtually perfect accuracy (explaining 99% of the variance). The model regressed interest rates on 2-year treasuries against the interest rates of 10-year treasuries and 3-month T-bills.

For all three class of investments, I took two additional steps:

• The monthly rate of return was adjusted for that month's change in the Consumer Price Index.
• The portfolio was rebalanced quarterly, if and only if the average deviation from target allocations was 3% or more. This strategy was used because of findings I've previously reported for various rebalancing strategies. (See overall recommendations here and specifics for the quarterly strategy here.)

After looking at numerous combinations, I settled on the following as an excellent portfolio, within the constraints of the data available for the analysis:

• 10% U.S. stocks
• 20% 10-year Treasuries
• 70% 2-year Treasuries

A portfolio like this may be advantageous for investments that are possibly to be spent soon, such as an emergency or reserve fund, or a safety-net for unplanned health care costs, or the portion of a retirement portfolio set aside to cover living expenses for the next 1-5 years. In such cases, the goal is to achieve a modest return above inflation, if possible, while minimizing any risk of lost purchasing power. 

Below is a chart that shows historical results since 1920 for the 10-20-70 portfolio, and for portfolios invested entirely in 10-year or 2-year Treasuries.

The charts shows that the mixed portfolio, over the long term, had virtually the same compound rate of return as 10-year Treasuries, handily surpassing 2-year Treasuries and thereby satisfying objective #1. In addition, the chart suggests that the variation from year-to-year was lower for the mixed portfolio than for 10-year Treasuries, which would satisfy objective #2. This was particularly true from 1930 to 1950, when 10-year Treasuries rose strongly during the initial years of the Great Depression, then rapidly lost buying power because of the inflation that accompanied World War II.

For a closer look at objective #2, consider the next chart. It shows the percent of periods when a portfolio lost ground to inflation, for periods ranging in length from 1 year to 21 years. This chart confirms that for the mixed portfolio, the risk of lost buying power was always less than for 10-year Treasuries, and was nearly as low as 2-year Treasuries.

Furthermore, as the next chart shows, when losses did occur, they were very small for the mixed portfolio. In the rare occasions when losses occurred, the cumulative loss in buying power, over any holding period, was virtually zero for both 2-year Treasuries and the mixed portfolio. In contrast, for 10-year Treasuries, during the more frequent periods when losses occurred, the cumulative loss was, on average, about 1% to 3% in buying power.

Historically, for U.S. investments since 1920, a portfolio invested 10% in stocks, 20% in 10-year Treasuries, and 70% in 2-year Treasuries would have satisfied the twin objectives of having returns similar to intermediate bonds while providing robust protection against changes, up or down, in consumer prices. It may be a good portfolio for near-term objectives such as a reserve fund. However, the 10-20-70 mix would not be prudent for inflation-protection in long-term investments such as retirement plans or college savings more than 5-10 years in the future. That will be the subject of the next post in this series.

Note, too, that if one were to implement this portfolio with vehicles available today, it would make sense to have the stock portion invested in a fund that includes both U.S. and international stocks, and to have the 70% in short-term bonds diversified across Treasuries, TIPS, and corporate bonds. And the 10-20-70 percentages are not absolute. Modifying them by 5% or 10% would have generated similar results, historically. If your investment firms offers an all-in-one fund close to the 10-20-70 targets, and the fees are under 0.4%, I would personally find the all-in-one fund more attractive than having to manage a set of funds or ETFs exactly matching the targets.

Historical data cannot guarantee future results. Although a mixture of bonds and stocks may be safer than investing exclusively in one class of assets, diversification cannot guarantee a positive return. Losses are always possible with any investment strategy. Nothing here is intended as an endorsement, offer, or solicitation for any particular investment or security.