In Retirement, Your State Tax Bill Could be Zero

In Retirement, Your State Tax Bill Could be Zero

If you search for information on state personal income taxes, you’ll likely find a list that focuses on the highest marginal rate in each state, with states like California, Hawaii and Oregon topping the list of high-tax states. But for retirees, focusing on tax brackets alone (especially the top tax bracket!) can be downright misleading. A range of tax policies related to age and income type can make even a supposedly “high tax” state like California into a low- or zero-tax state for many retirees. So, when it comes to evaluating states for their tax-friendliness to retirees, we need to look deeper.

“Zero-Tax” States

Seven states have zero personal income tax at the state level. Another two states tax only certain types of investment income (like interest and dividends) but not other income (like wages and pensions). Clearly, all else being equal, these states are the most tax-friendly to all taxpayers, retirees included.

  • States with no personal income tax: Alaska, Florida, Nevada, South Dakota, Texas, Washington, Wyoming
  • States where only certain investment income is taxed: New Hampshire, Tennessee

But, for retirees, the list of states that offer the chances of zero state income taxation (or at least taxation that is far below what you would pay in your working years) is much longer. Depending on the make-up of your retirement income picture, you might be looking at low or no state income tax in retirement even if you don’t live in the above nine tax havens.

For example, 29 states and the District of Columbia exclude all Social Security income from taxation. Three states on this list – Hawaii, Illinois, and Pennsylvania – also exclude pension and qualified retirement plan income. Fifteen states include some level of capped exclusions for pension and other retirement income, from a low of $4030 per spouse (Montana) to a high of $65,000 per spouse (Georgia). Twenty-four states also offer additional deductions, exemptions or credits to people age 65+, leading to even lower taxes for many retirees. 1

California as a Low-Tax State?

In order to explore what the state tax landscape looks like for retirees, we estimated the 2019 state personal income tax burden of the following two households in all 50 states and the District of Columbia. 2

Household A

For Household A, the results of this test may be surprising. For this family, in addition to the nine no-tax or investment-income-tax-only states, 14 states had an estimated $0 state tax bill. On this list are California, Hawaii and New York, three of the states commonly thought of as high-tax states.

Additional States with a $0 tax bill for Household A

In another eleven locations (AZ, CT, DC, IA, ID, ND, NE, NJ, OH, OK, VT), this family’s effective rate would have been below 1%. (And in Oklahoma they would have paid just $33, or 0.04% in taxes.)

The states where this family would have paid the highest tax were Utah (3.6% effective rate) and West Virginia (3.1% effective rate) – neither of which is commonly highlighted as a high-tax state.

Household B

Because of its higher total income and its investment income, the number of zero-state-tax states for Household B is smaller: we estimate zero state taxes in the seven zero-tax states and Georgia. But this family’s tax burden in many states is still lower than one might expect. Fifteen states had effective rates below 2%. On this list, we once again find California and New York.

States with a less-than-2% effective tax rate for Household B

Retiree-Friendly States

The tax burden in any given state will depend on a family’s total tax picture. But these two example households show that some states with high top marginal rates can actually be quite tax-friendly to retirees. Some states, like Georgia, Hawaii, Illinois and Pennsylvania, often masquerade as zero-tax states for some retirees. Typically, the more a household depends on Social Security and qualified retirement plans and pensions, the higher the likelihood that they will come close to paying no state income tax.

More Information

Decisions about where to live in retirement go far beyond financial considerations. But it is important to have the right facts when evaluating how tax-friendly your state of residence in retirement might be. To help us move beyond a simplistic and misleading focus on top marginal tax brackets and into the true possible state tax experiences of retirees, Income Lab has released a useful Guide to State Personal Income Tax for Retirees that provides commentary and information on all 50 states and the District of Columbia.

1 In some states, these capped retirement income exclusions and age-based deductions, exemption and credits may be subject to income-based phase-outs.

2 We assumed that Social Security and Pension/Qualified Plan Income was evenly divided between spouses.

A Stock Indicator that can Improve Retirement Income Decisions

A Stock Indicator that can Improve Retirement Income Decisions

In times of economic turmoil, it’s helpful to know that there is a long-term stock market indicator that can help us make better retirement income decisions. Numerous economic factors can inform retirement income decisions, but none has been discussed more than the cyclically-adjusted price-earnings ratio (CAPE), a long-term measure of how much investors are willing to pay for corporate earnings.1 This ratio plummeted during the recent stock market drawdown. But historical evidence shows that as CAPE goes down the percentage that a retiree has been able to withdraw from his or her portfolio has gone up. Let’s review this evidence and see what it may mean for today’s retirement investors.

CAPE’s Historical Relationship to Retirement Income

Retirement is a long-term event. CAPE is a long-term indicator. So, it’s not surprising that CAPE has a lot to say about retirement income. The box plot in Figure 1 shows the historical distribution of available 20-year portfolio withdrawal percentages,2 grouped by initial CAPE value, since 1881. (The “boxes” show the middle 50% of data. The “whiskers” show the lowest and highest 25%.)

Figure 1

CAPE levels (divided into historical quintiles): Low – CAPE between 4.6 and 11.2; Below Average – 11.2 to 14.8; Average – 14.8 to 17.8; Above Average – 17.8 to 22.2; High – 22.2 to 44.8.

The high-level pattern is clear. When CAPE was low, available withdrawal percentages were high. When CAPE was high, much lower withdrawal rates were possible. The same is true for 30-year withdrawal periods.4

CAPE’s Historical Relationship to Retirement Nest Eggs

This pattern might tempt us to think that high CAPEs are bad for retirement income. But the story isn’t quite that simple. When CAPE is high, portfolio balances tend to be high as well. And we don’t spend percentages – we spend dollars. The average available withdrawal rate in high-CAPE periods was 43% lower than in low-CAPE times. But those high-CAPE periods were at the tail end of some of the best stock markets in history. Figure 2 shows how much money a hypothetical retiree would have had at retirement (in 2020 dollars) if, prior to retirement, they had saved $1000/month for 30 years in a 60/40 portfolio.

Figure 2

The lesson is that high CAPEs are also high points for retirement nest eggs. The best time in history to retire wasn’t a point when the highest withdrawal rate was possible, but a time when a healthy withdrawal rate could have been applied to a large portfolio.

CAPE and Retirement Income Risk

Figure 1 holds a few more lessons for retirement income planning. For example, it suggests that the predictive power of CAPE is highest when CAPE is high. The range of withdrawal percentages available during periods of high CAPE is the smallest of any CAPE regime. This is true for both the total range (from the minimum to the maximum, between the ends of the “whiskers”) and for the middle 50% (the box, called the interquartile range).

Importantly, the lower tail of the withdrawal distribution (the lower whisker, which is the range from the lowest available withdrawal rate to the 25th percentile) and the range from the median to the 25th percentile (the bottom half of the box) are also smallest in high-CAPE periods. This means that the risk of severely overestimating available retirement withdrawals has been lowest when CAPE was highest. Overestimating sustainable withdrawal rates is the primary risk that can lead to unwanted downward income adjustments in retirement, so it is comforting to know that this risk is lowest when CAPE is highest.

High-CAPE periods are on the other extreme. The range between the lowest and highest withdrawal rates in low-CAPE periods is massive. Thankfully, rates were also substantially higher than they were in other CAPE regimes: the lowest withdrawal rate in history during a low-CAPE period was roughly equal to the average rate in high-CAPE periods, and two-thirds of the available withdrawal range in low-CAPE times lies above the highest high-CAPE withdrawal rate. So, high CAPE means high withdrawals rates but also higher risk of overestimating available withdrawals.

What’s CAPE Been Up to Lately?

Now that we know a bit about historical patterns, let’s look at how CAPE has behaved recently. Since 1881, average (median) CAPE is 16.3. But in the last 30 years CAPE has been consistently higher, with an average of 26.0. The February-March 2020 stock market drawdown brought about the largest drop in CAPE since the Great Recession.

Figure 3

Viewed across 140 years of history, the March 2020 low point is still an “above average” CAPE. However, we hadn’t seen CAPE this low since 2012, during the long recovery from the Great Recession. And this reduced CAPE value was just above the lowest levels we saw in the worst of the protracted Tech Bubble crash (2000-2003).5

The economic repercussions of the COVID-19 pandemic are still developing, so only with hindsight will we know whether this was the low point for CAPE in this market cycle. But we do know that in the past, as CAPE goes down, the pain retirees may have felt in their investment portfolio has been eased somewhat by an increase in the withdrawal percentage they could afford to take from those portfolios. We also know that, no matter the CAPE value, CAPE can help us make better retirement income decisions.

1 Traditional P/E ratios divide price by recently reported or projected earnings. But these earnings numbers are volatile, and the resulting measure isn’t very good at helping us develop expectations about future returns. CAPE, on the other hand, uses long-term (usually 10-year) average inflation-adjusted earnings and is much more useful for developing long-term expectations.

2 These withdrawal levels estimate how much of an initial portfolio balance a retiree could have withdrawn annually from a 60/40 stock/bond portfolio, with adjustments for inflation thereafter, for 20 years. Research suggests that a static approach to retirement income produces poor outcomes, but this approach is useful for comparing possible retirement income levels across historical periods. Past performance does not guarantee future results. Results based on index returns. The returns of actual investments may be higher or lower than index returns. Results would differ based on different investment mixes and retirement income plans.

3 There are many potential confounding factors when trying to interpret CAPE data. For example, accounting rules for earnings calculations have changed over time, meaning the denominator for the CAPE ratio is not consistent at all points of history. Changes in dividend/share buyback behavior and the supply of and demand for stocks may also have affected the expected numerator of the ratio over time.

4 It is common in retirement research to focus on 30-year retirement periods. We focus on 20-year periods in Figure 1 in order to increase the number of periods in our data set that represent high CAPE values. (Many of these high-CAPE periods have occurred recently, making them out-of-sample when looking at 30-year retirement periods.)

5 As of this writing (April 17, 2020) CAPE had recovered to 27.4, just above the 30-year average, but stock markets remained volatile.

How Do Black Swan Events Affect Retirement Income?

How Do Black Swan Events Affect Retirement Income?

The term “Black Swan” has become a shorthand in finance for a large, sudden market drop.1 The dramatic stock market drawdown and economic events of the last month (like the 17 million new claims for unemployment in the last three weeks) have led many to apply the term to the COVID-19 pandemic. Many retirees are worried about whether this Black Swan will require massive lifestyle changes. For businesses that operate with a high amount of leverage and large short-term liabilities, these events can be an existential threat. But evidence from previous Black Swans shows that, historically, they have never had a catastrophic effect on retirement income.

Notable Black Swan Events in History

For present purposes, we’ll define a “Black Swan” market event as a US stock market decline of 30% more in a month or less.2 Since 1928, there have been five such events.3

  • Great Depression: -42% (Oct-Nov 1929) and -32% (Sep-Oct 1931)
  • Crash of 1987: -31% (Oct 1987)
  • Great Recession: -30% (Sep-Oct 2008)
  • COVID-19 pandemic: -33% (Feb-Mar 2020)

Retirement is Not Like Running a Hedge Fund

Many people are understandably worried about Black Swan events’ impact on their investments. But the immediate financial impact of a Black Swan depends very much on the nature of an investor’s liabilities. Fast, unexpected events with massive market impact have an immediate impact on those with large short-term liabilities. For example, many hedge funds are highly leveraged: they borrow money in order to invest it or use financial instruments with built-in leverage, like futures and derivatives. As a result, a hedge fund’s immediate liabilities can balloon in a Black Swan. Because investment banks reconcile (or “mark to market”) many positions daily, a Black Swan event can have a significant impact on a hedge fund’s ability to stay in business. In fact, events with much smaller drawdowns than those highlighted above, like the Asian Financial Crisis of 1997, have caused hedge funds to fold.

But retirees are nothing like hedge funds. The average retiree does not invest heavily in using margin (leverage). Additionally, basic retirement liabilities – namely, monthly living expenses – are each relatively small and are due over a long timeframe.

Additionally, though retirement investment accounts will almost certainly suffer during a Black Swan event, many retirees derive a significant amount of their income from sources like Social Security or similar pension, which is not directly affected by Black Swan markets. That means that a portion of the retirement income that covers core living costs is immune to these market disruptions.

Furthermore, living costs are unlikely to increase significantly for retirees during a Black Swan event. If anything, economic shocks can cause short-term liabilities to go down temporarily as people instinctively pull back on discretionary spending. (This is particularly clear in the COVID-19 pandemic, where stay-at-home orders and social distancing have reduced expenses related to travel, eating out, etc.) As a result, the short-term, extreme impact of a Black Swan event is simply not as large for retirees as it is for some other investors.

Long-Term Effects of Short-Term Swans

Given the large and sudden nature of Black Swans, it is tempting to think that the worst timing for retirement would be immediately before such an event. Imagine this: you begin retirement near the top of a bull market. You’ve planned your spending based on a nest egg that has grown with that bull market. A few months later, the bottom falls out of the market. Sounds catastrophic, right?

To see how a hypothetical retiree in this position would have fared historically, let’s imagine households with $1 million (in 2020 dollars) in retirement savings, invested 60% in stocks and 40% in bonds, who began retirement six months before each of the Black Swan markets above. Though in real-life retirees’ income should probably change over time based on longevity, economic conditions, and so on, we’ll simplify here and have each hypothetical household spend according to the well-known (though very flawed) “4% rule”. They’ll spend 4% of their initial retirement balance each year, adjusted for inflation.

Below we see how this nest egg fared over time. In the first three historical Black Swan markets, our households would have made it 30+ years into retirement without running out of money. In only one case (1929) would the household have lived through most of this lengthy retirement with less than they had started with. The 1929 case was particularly bad since this household would have had the one-two punch of 1929 and 1931.

We haven’t yet lived through 30 years since the 2008 financial crisis, but so far, this hypothetical household is doing okay. Because the 2008 Black Swan was followed by a longer, deeper market drawdown, it took a while for this household’s balance to recover. But it’s back at a healthy inflation-adjusted level 12 years later. (It’s worth noting that one thing that can cause Black Swan markets to have a lasting impact on retirement is panic selling.)

I wouldn’t wish the returns from these scenarios on anyone, but it’s clear that past Black Swans did not have what it took to destroy retirements. That’s because retirements play out over many years or many decades: plenty of time for fast-moving, brief Black Swans to look small in the rearview mirror.

1 In his books Fooled by Randomness (2001) and The Black Swan: The impact of the highly improbable (2007), Nassim Taleb defines a “Black Swan” as an event that lies outside the realm of regular expectations (like seeing a black swan if all you’ve ever seen were white swans) and carries extreme impact, but that we concoct explanations for after the fact, making it seem explainable and predictable in retrospect.

2 Based on the daily closing price of the S&P 500 Index. Past performance does not guarantee future results.

3 Many events that do not meet this (admittedly arbitrary and restrictive) definition have been called “Black Swans” elsewhere. I have no issues with the broader application of this term. I’ve chosen a narrower definition here in order to focus on the largest, quickest market drawdowns.

What do Unemployment Claims Mean for Retirement Income?

What do Unemployment Claims Mean for Retirement Income?

In the last two weeks, the U.S. Department of Labor has released the highest weekly initial unemployment claims numbers on record. Almost 10 million Americans filed for unemployment over a two-week period at the end of March. To put these numbers in perspective, the previous weekly high was 695,000 in September of 1982. The Great Recession saw a peak in new claims of 639,000 in April of 2009. In other words, the size and pace of new unemployment due to the COVID-19 pandemic is staggering, and we may be only at the beginning of this unemployment spike.1

Initial Unemployment Claims

  • March 21, 2020             3.3 million
  • March 28, 2020             6.6 million
  • April 13, 2009               639,000
  • Sept 27, 1982                695,000

Those in retirement may not have just lost employment, but It is natural to ask what these sorts of gloomy economic headlines might mean for retirement income planning. It turns out that, historically, initial unemployment claims do hold information that is useful for retirees who support their retirement income with portfolio withdrawals. Though data is somewhat limited (initial claims data is only available since 1967), the four-week rolling average of initial claims has between a 61% and 83% correlation to available withdrawals for retirement income, depending on the amount of time that income must last.2

For example, available 20-year withdrawal rates peaked at around 14% just two months before the peak in initial claims in 1982. Compared to a median available withdrawal rate of 8.8% over the period shown (and a low of 4.8%), we know in retrospect that 1982 retirees had access for a short while to withdrawal rates that were 59% higher than usual.

A similar matching of peaks was seen in the financial crisis. (Since only 11 years have passed since 2009, we use available 10-year withdrawals here.)

Of course, correlation is not the same as causation. These high correlations are probably best explained by the tendency of initial unemployment claims to peak (almost by definition!) just as things start to get better economically. These unemployment claim peaks can be close to points when securities prices reach a low point. And we know that available withdrawal rates are highest just after these market low points. Historically, sequences of returns following market bottoms are very advantageous for retirees.

Though the current environment seems bleak, it’s important to remember that the same was true at previous points when unemployment spiked. But with hindsight, we can see that things were darkest before the dawn. For example, after 1982, future investment returns would have more than made up for the market losses that had been suffered around the time of that spike in unemployment claims, and retirees would have been rewarded for staying the course. The same is true for the Great Recession.

Typically, statistics that tell a tale of economic woe hold within them a message of hope for retirement income. Though spikes in unemployment claims usually coincide with periods when nest eggs shrink, they have never in the past required retirees to reduce their income in proportion with their losses. If the past is any guide, these peaks are also when the withdrawal rates available to retirees are highest. These countervailing forces can provide significant cushion to income in tough times.

1 In future Lab Notes we’ll track the unemployment rate as well, which lags behind new claims statistics.

2 P-values for the null hypothesis that there is no relationship between available withdrawal rates and initial claims are vanishingly small. Available withdrawal rates are based on gross returns from a generic 60% stock/40% bond portfolio with monthly withdrawals adjusted for inflation. Results would vary for other withdrawal plans and other portfolios. Past performance does not guarantee future results.

It’s Time to Retire Static Planning

It’s Time to Retire Static Planning

The analytics behind today’s retirement plans focus on how things might work out if a given set of choices (investment allocations, portfolio withdrawals, cash flows, etc.) are carried on into the future without change. But this static approach to planning is a crude and ineffective way to meet the two major goals most retirees share:

(a) Maximize standard of living in retirement
(b) Leave a legacy / maintain a financial safety net

The proof of static planning’s failure can be found in plain sight in one of the charts most familiar to today’s financial planners: a graph of possible outcomes if a static plan is followed.

possible outcomes of a  static plan

Source: Inflation-adjusted (real) portfolio balances for a 1000-scenario Monte Carlo simulation of a $43,000/year inflation-adjusted income taken from a $1 million 60/40 stock/bond portfolio.1

The range of possible outcomes that could result from following a static plan is huge, ranging from the stratospheric (over 50% of scenarios ended with over $2 million, with a maximum of $12.8 million left over) to the dismal (running out of money). Let’s be clear: both outcomes are a type of failure. In the outcomes that ran out of money this plan clearly failed both the standard of living and legacy goals. But in scenarios that produced an outsized legacy, these retirees skipped income that could have improved their standard of living, thus failing to meet goal (a).

A plan that has this range of outcomes is simply a bad plan. To follow this static plan would be like setting out on a journey to the moon knowing that you might end up burning up in the sun or landing on Neptune. Luckily, just as with a lunar landing, a retirement journey allows for course corrections along the way. And those course corrections allow retirees a much higher likelihood of meeting or even exceeding their goals. But in order to implement intelligent course corrections, we need to abandon static planning. Dynamic plans would not only include a set of behaviors for today, but consideration of how to adjust as time goes on and circumstances change.

To demonstrate the value of dynamic planning, we simulated the retirement experiences of couples who began their retirement at every monthly point in the last 130 years. These retirees began with a relatively conservative income level and reevaluated their entire situation every month as they moved forward in time. This included looking at changing longevity expectations, portfolio balances, purchasing power, and the risk of their income level. Each couple got an income increase if risk went down significantly and a pay cut if risk went up significantly. Other than these risk-based changes and adjustments for inflation, they kept income steady.2

In this study, 91% of simulated retirees experienced more income than they had originally planned for. On average, these retirees received 166% of their originally expected income over 30 years. At the high end, the retirees from July of 1982 ended up with 294% of what they had planned for.

But not all scenarios were positive: 9% of scenarios received less total income than originally planned. On average, though, this shortfall was just 5.4% (94.6% of expected income). On the low end, November 1965 retirees received 88.5% for their originally expected income.3

30- year income experience with dynamic plan
To summarize, retirees who began their retirements relatively conservatively and adjusted over time were far more likely to have a positive than a negative overall retirement income experience. And those who had positive experiences gained far more than was lost by those who had negative experiences, both on average and in the best/worst cases.

In all cases, overall retirement experiences showed a reasonable balance between the goals of maximizing standard of living and maintaining a legacy or safety buffer. (No test scenario resulted in portfolio failure.) Different dynamic plans would have different outcomes, but the overall patterns are clear: dynamic plans improve retirement income and the ability to meet goals (a) and (b).

It’s time to abandon unrealistic, static planning and embrace a more realistic, dynamic approach.

1 Assumptions based on the 30-year average gross monthly returns and standard deviations of S&P 500 Total Return Index and SBBI Intermediate Term Government Bond Index through January 2020. Arithmetic average monthly returns: 0.66% (stock), 0.23% (bond). Standard deviations: 3.43% (stock), 1.23% (bond). Correlation: -0.28.

2 Specifically, each couple began retirement with income that had a 90% chance of being sustainable through their plan (a 10% risk level). Couples increased real income if the risk of their income reached 0% and decreased real income if the chances of maintaining their income level went down to 25%. Risk was measured as the chances that a given income amount would survive the full remaining plan length, adjusted for then-current longevity. Couples adjusted their income only when nominal income changes were 5% or more.

3 We also ran this study using different initial retirement income levels, both higher and lower than the baseline case reported here. As expected, starting retirement with a lower income reduced both total income received and the chances of a pay cut during retirement. A higher initial income results in higher total average income and a higher chance of a pay cut sometime during retirement.

Sequence of Returns Risk Has Been Greatly Exaggerated

Sequence of Returns Risk Has Been Greatly Exaggerated

When trying to decide how much someone can afford to spend in retirement, we often worry about poor investment returns early in retirement. Poor returns early in retirement can permanently limit retirement income. Researchers call this Sequence of Returns Risk.

Historically, the difference between good and bad return sequences appears huge. For example, if you retired in December 1968, just before a rough period of high inflation and poor market performance, you could have taken 4% of your initial portfolio in the first year, with later withdrawals adjusted for inflation, for 30 years.1 In contrast, if you retired in August 1982, at the beginning of a roaring bull market, the same portfolio would have supported a withdrawal rate of about 11.6% – nearly three times as much. This stark difference is usually taken to be evidence for the magnitude of Sequence of Returns Risk.

But these numbers are wrong – or at least grossly misleading. We spend dollars, not percentages. A 1982 retiree would have been able to withdraw three times the dollars available to the 1968 retiree only if both retirees had the same amount of money at retirement. But, assuming identical savings behavior prior to retirement, that is impossible.

Let’s walk through a more “reality-based” example.

Assume that both people saved $1000/month and retired after 40 years. In this case, the 1968 retiree could have spent about $6600/month. The 1982 retiree could have spent $7800/month – an advantage of 18%. That’s not even close to the 300% we expected. Where did all the extra income go?

The answer is that these two retirees had very different investment experiences while they were saving, and so had very different amounts of money when they retired. Both saved a total of $480,000 over 40 years, but the 1982 retiree had about $800,000 at retirement, compared to the 1968 retiree’s roughly $2 million. The extra dollars for the 1968 retiree erase a large portion of the expected income gap.

withdrawal rate examples
This pattern in the table above applies through history: assuming this saving behavior, the correlation between balance at retirement and percentage withdrawal rate is -0.8, very close to perfectly negative.2
correlation between balance at retirement and percentage withdrawal rate

This pattern shows us that, in the past, those who experienced great returns in their saving years and had a large nest egg at retirement were able to spend less, as a percentage of their retirement portfolios, than others. Conversely, those who experienced worse returns and had smaller retirement nest eggs than others tended to be able to spend more in percentage terms.

The examples of 1968 and 1982 demonstrate that this effect can go a long way toward cancelling out what might otherwise have been dramatic differences. Sequence of returns risk is real, and we ignore it at our peril. But for those who save and invest for retirement over time, this risk is smaller than you may have heard.3 We have market cycles to thank (or blame) for this pattern: historically, what goes up tends to go down. But thankfully, what goes down has also gone up again.

This is the source of the well-known “4% rule”. All examples here assume a 60/40 stock/bond portfolio, are stated in approximate real (inflation-adjusted) dollar terms and reflect gross index returns of S&P 500 stock index and the Ibbotson SBBI Intermediate Term Government Bond Index and the DMS US Bond TR Index. Indices are not available for direct investment. Actual investments may have provided better or worse results. Past performance is not a guarantee of future results.

With this saving behavior, the worst time to retire was August 1918 and the best time was December 1954.


withdrawal rates

That’s still a substantial difference in real dollars, but nothing close to the 300% statistic we started with. Notice that the percent income available in the best time is lower (6.2%) than that found in the worst time (7.5%). This is the final proof that percentage withdrawal rate, on its own, doesn’t tell us much about retirement income. 

3 For this example, the coefficient of variation (CV, calculated as standard deviation / mean) of percentage withdrawal rate is 27% while the CV of dollars of real income is 18%.