Much confusion and misconception surrounds

Transcription

1 Are Safe Withdrawal Rates Still Relevant in Today s Low-Return Environment? Michael E. Kitces Partner, Director of Research Pinnacle Advisory Group Columbia, MD Safe withdrawal rates are designed to ensure sustainable spending from retirement portfolios throughout a wide range of economic scenarios. Today s economic environment of low bond yields and high cyclically adjusted price-to-earnings ratios for stocks is somewhat unprecedented, but the traditional 4.% safe withdrawal rate should remain an appropriate baseline, unless the forecast is that the fundamental engine of economic growth is damaged far worse than during the Great Depression. Much confusion and misconception surrounds the concept of safe withdrawal rates. Safe withdrawal rates can seem overly simple, as illustrated in the following example. Consider two approaches to formulating a safe spending recommendation for clients. The first is to use a traditional safe withdrawal rate of 4.%. The second is to use a Monte Carlo approach that incorporates different percentile outcomes to generate a spending plan that has a 99% chance of lasting 3 years, a 96% chance of leaving 1% of principal, and a 5% chance of quadrupling the initial wealth. Virtually no adviser would view Option 1 as a safer approach; nearly all advisers would select Option 2. Surprisingly, these two scenarios describe the same research. Whether I use Monte Carlo scenario analysis to generate a safe withdrawal rate or formulate one based on 1-plus years of historical scenarios, the conclusion is the same: The safe withdrawal rate that coincides with the 99% success rate is equivalent to the safe withdrawal rate based on extensive historical research. That rate is 4.%. In this presentation, I will review the safe withdrawal rate research and apply it to today s low-return environment to determine whether the research is still relevant in answering the two fundamental questions clients have regarding retirement: (1) How much money can I safely spend from my portfolio without worrying about the markets? and (2) How much money do I need to sustain my retirement goals? 1 This presentation comes from the Wealth Management 213 conference held in Boston on March 213 in partnership with Boston Security Analysts Society. 1 For more commentary on financial planning and other issues, visit Traditional Withdrawal Rates Whichever question a client has, the underlying goal is simply to determine a reasonable, sustainable series of cash flows given myriad risks and uncertainties. The traditional approach is to use financial planning software or a spreadsheet framework to plug in a time horizon, inflation assumptions, an asset allocation, expected asset class returns, and an initial portfolio value to determine if the plan will work. As an example, assume a 6-year-old client with a 3-year retirement time horizon, and roughly in line with 1 years worth of Morningstar/Ibbotson data, inflation averages a rate of 3.%, average stock returns are 1.%, average bond returns are 5.%. With a 6% stock/4% bond allocation, the average return is 8.%. The portfolio s initial value is $1 million. The solid line in Figure 1 illustrates how much the client can safely spend. Spending starts at about $66, a year and, with inflation, increases more than $15, a year. The dotted line is the client s portfolio value. It begins with $1 million and rises slowly in the early years because the client needs to build a cushion for the spending projected near the end of the horizon. During the second half of the period, the solid line starts trending downward, and the client reaches the end of the 3-year time horizon without running out of money. Many advisers also use a ledger chart to show annual portfolio values, growth, and spending. Again, the conclusion is the same: If there is money left at the end of the ledger, retirement works. Often, advisers prefer to plan for a small terminal account balance at the end of the time horizon, in the event that a client outlives the projected horizon, but the concept is the same. If there is not 213 CFA Institute cfapubs.org September

2 CFA Institute Conference Proceedings Quarterly Figure 1. 6/4 Retirement Portfolio over 3 Years Spending ($) 18, 16, 14, 12, EOY Balance Balance ($) 1,2, 1,, 8, 1, 8, 6, 4, 2, Spending 6, 4, 2, Age Note: EOY is end of year. Sources: Figure from Michael Kitces ( based on estimates of long-term returns from Morningstar/Ibbotson. enough money at the end, adjustments need to be made. The four factors to resolve a shortfall are an increase in savings, a lower withdrawal rate and reduced spending, a delay of the onset of retirement, or an early death. For a portfolio with an initial balance of $1 million and using the assumptions already discussed, the safe withdrawal rate is projected to be about 6.6% (or about $66, starting with a $1 million portfolio). Thus, whether advisers realize it or not, they have for many years been building into retirement recommendations a 3-year sustainable spending level of 6.6% of the initial portfolio, adjusted annually for inflation. Of course, the problem with this withdrawal rate, which is readily apparent the first time a client comes in for a review, is that annual portfolio returns have never actually been precisely 8.%. Even if they average out to that point, returns are more volatile, and it turns out that volatility matters. Beginning with Poor Returns. Consider the alternate version of this scenario shown in Figure 2. Average equity returns are still 1% over the whole time horizon, but that includes zero during the first two years, 2% during the last two years, and 1% for all the years in between. When the portfolio starts out with two years of zero returns, instead of wealth slowly winding downward when the client is in his 8s as a result of spending, wealth begins to decline when the client is in his 6s and spending falls short by a sizable margin. The client runs out of money six years before retirement, which is a whopping 2% short of his 3-year goal. Figure 3 illustrates a scenario in which the best returns are at the beginning of the horizon. The portfolio s growth exceeds the projected spending needs enough that the spend-down does not begin until the last two or three years. This portfolio includes a sharper downturn at the very end, when the portfolio earns zero returns. Nonetheless, this client not only achieves her spending goal over the entire 3-year time horizon but also never dips into her principal. The difference between Figure 2 and Figure 3 is four years of returns and is ultimately dominated by the two initial years of returns at the beginning of the horizon. When the portfolio starts out with poor returns, as shown in Figure 3, it is not able to sustain spending even when returns improve later, whereas when the early returns are good, even a severe downturn later may not threaten the plan. Effects of Inflation. Variability in inflation can also have significant effects on the portfolio. An increase in the inflation estimate from 3.% to 3.5% can exhaust funds in 26 years instead of 3 years. Conversely, a half-point decrease in inflation can extend spending in that same portfolio to 35 years. In other words, just a 1.% swing in inflation can shift a 3-year target by 9 years. The combined effect of both the sequencing of returns and changes in inflation can be especially severe. In another example, instead of using the 8- to 1-year history of capital market returns and inflation, I use the 3-year history from 1969 to The nominal large-cap equity return was 68 September CFA Institute cfapubs.org

3 Are Safe Withdrawal Rates Still Relevant in Today s Low-Return Environment? Figure 2. 6/4 Retirement Portfolio over 3 Years with % Returns First 2 Years and 2% Last 2 Years Spending ($) 18, 16, 14, 12, EOY Balance Balance ($) 1,2, 1,, 8, 1, 8, 6, 4, 2, Spending 6, 4, 2, Age Sources: Michael Kitces ( based on estimates of long-term returns from Morningstar/Ibbotson. Figure 3. 6/4 Retirement Portfolio over 3 Years with 2% Returns First 2 Years and % Last 2 Years Spending ($) 18, 16, 14, 12, 1, 8, 6, 4, 2, EOY Balance Spending Balance ($) 1,6, 1,4, 1,2, 1,, 8, 6, 4, 2, Age Sources: Michael Kitces ( based on estimates of long-term returns from Morningstar/Ibbotson. 13.4%, and five-year Treasury bonds generated an 8.6% annualized return. This period was a fantastic time to be an investor over the entire 3-year time period: A 6/4 portfolio returned nearly 11.5% compounded annually for 3 years. Unfortunately, annual inflation averaged 5.3% over this same 3-year period. Figure 4 shows the effects of inflation on spending. The spending requirement begins at about $7, and finishes at $35,. This client spends $1 million in just the last 3 years from a portfolio that begins with $1 million and has to last 27 years before the spending requirement grows to $1 million in the last 3 years alone. The good news is that it is not very difficult to grow a portfolio with 11.5% returns compounding annually on a 6/4 portfolio. In fact, the results 213 CFA Institute cfapubs.org September

4 CFA Institute Conference Proceedings Quarterly Figure 4. 6/4 Retirement Portfolio over 3 Years in High-Inflation Environment Spending ($) 4, 35, 3, 25, 2, 15, 1, 5, EOY Balance Spending Balance ($) 1,8, 1,6, 1,4, 1,2, 1,, 8, 6, 4, 2, Age Sources: Michael Kitces ( based on estimates of long-term returns from Morningstar/Ibbotson. reveal that the client can safely spend 7.4% of the portfolio, or $74,38, as a starting level. But Figure 5 shows the client s retirement scenario when the sequencing of returns and inflation is taken into account: The client is broke in Year 12. In other words, although the results went all 3 years with the average return, this client ran out of money given the actual sequence of results that produced that average return (and sadly, it was just as the greatest bull market of the century began). Safe Withdrawal Rates Thus, ultimately, the challenge of safe withdrawal rates is to determine how low the spending rate has to be such that, if an unfavorable sequence occurs, the portfolio will still make it to the end in other words, how much of a safety margin is necessary, as a reduction in spending, to protect against unfavorable sequences. Determining the answer, in turn, requires analyzing not just average returns but the worst-case sequences and scenarios found in history. Figure 6 shows the initial safe withdrawal rates that would have worked for a 6/4 portfolio rebalanced annually for any particular 3-year time period going back to the 187s, given the actual sequence of stock, bond, and inflation results that occurred through the time period. Thus, the first bar on the left shows that for a retirement that ran from 1871 to 19, a withdrawal rate of $95, in Year 1, or 9.5% of the starting account balance, would have worked with spending subsequently adjusted for actual inflation each year and the portfolio s growth reflecting the actual sequence of stock and bond returns that occurred. But the focal point of the chart is not what worked in 1871, nor the increase in the spending rate, nor even that the average spending rate was about 6.5%. Instead, the focal point is the spending rate that can be sustained during the most adverse scenarios. In other words, the point of safe withdrawal rates is simply to target spending at the worst-case scenario that could be found in history. If returns for a current retiree are comparable to one of those Great Depression style eras, the funds should last for 3 years; if the results are anything more favorable, the client will simply finish with more money or may be able to safely increase spending in the future when it turns out a terrible sequence is not occurring. And what was that worst-case scenario safe withdrawal rate historically? It has been about % (depending on the exact data set used), which means there is a safety margin of about 2% (i.e., retirees must withdraw about 2% less than the average rate of 6.5% to protect against bad sequences). This difference is not trivial; it amounts to one-third of a client s annual spending, but that is the impact of managing against the risk. And notably, this has little relationship to average returns; in some scenarios, like the 1969 retiree, the average returns may be quite good, but the withdrawal rate must still be low to protect against the bad sequence. In fact, ultimately there are a few key bars (3-year retirement scenarios) from Figure 6 that drive the 4% safe withdrawal rate. Those four low 7 September CFA Institute cfapubs.org

5 Are Safe Withdrawal Rates Still Relevant in Today s Low-Return Environment? Figure 5. 6/4 Retirement Portfolio over 3 Years Including Return Sequencing and Inflation Spending ($) 4, Balance ($) 1,2, 35, 3, EOY Balance 1,, 25, Spending 8, 2, 6, 15, 1, 5, 4, 2, Age Note: Returns and inflation are from 1969 to Sources: Michael Kitces ( based on estimates of long-term returns from Morningstar/Ibbotson. Figure 6. Safe Initial Withdrawal Rates by Starting Year with a 6/4 Portfolio, Initial Withdrawal Rate (%) Starting Year Source: Michael Kitces, The Impact of Market Valuation on Safe Withdrawal Rates, The Kitces Report (March 28): bars on the chart are 197 (which was the onset of the credit crisis and crash of 197), 1929, 1937, and The crash of 197 is not as well known as the adverse events of 1929 and 1937 (and the later crash of 1973), but they all were similar in that they typically involved a market crash, and more than 15 years of overall stagnant markets, before markets finally made new highs. In addition, the safe withdrawal rate research found that the optimal portfolio allocation to stocks is about 6% (ultimately varying between 4% and 7%, depending on the exact study). These balanced 213 CFA Institute cfapubs.org September

6 CFA Institute Conference Proceedings Quarterly allocations work best because when a portfolio is too bond heavy, it is ravaged by inflationary scenarios, and when it is too stock heavy, it is ravaged by bear markets. Safe Withdrawal in Challenging Times. So, how bad are the most challenging times, the low points in the historical analysis? Table 1 encompasses all four of these challenging historical environments. Comparing the average of the bad periods with the overall 14-year average reveals some fairly consistent trends: below-average stock returns of more than 1 bps and lower-than-average bond returns of almost 1 bps, both of which lead to lower rebalanced returns. This illustration assumes annual rebalancing, which generates a small amount of rebalancing alpha, but the rebalanced portfolio during the bad times still lags the rebalanced portfolio by about 6 bps during the 14-year average. Inflation during these periods was above the overall average as well, so real returns are compressed even further. Thus, during bad periods, stocks averaged 6.%, bonds averaged 1.%, and rebalanced 6/4 portfolios averaged just 4.5% on a real-return basis. The equity risk premium in these environments is only 5.% across the 3-year time periods. These differences in return have a huge impact over 3 years. Even over very long-term periods, returns do not necessarily average out. These scenarios were also sequenced in the worst possible way; not only were returns below average for 3 years, but poor returns at the beginning of the time horizon had to be survived before getting to the decent years at the end. And unfortunately, these portfolios never fully caught up during the entire 3-year time horizons. It is notable that our safe withdrawal rate discussions are based on 3-year returns, but research shows that 3-year returns do not matter as much as 15-year returns, as shown in Figure 7. If the first half of the retirement horizon is marked by poor real returns, even fantastic returns in the second half are inconsequential because growth compounded on a zero balance is zero if the portfolio has already been depleted. Conversely, when the retirement period begins with fantastic returns, a bear market Table 1. 3-Year Nominal Returns in Various Historical Environments 3-Year Nominal Return Starting in 197 Starting in 1929 Starting in 1937 Starting in 1966 Average of Bad Periods Overall 14-Year Average Stocks 7.77% 8.19% 1.12% 1.23% 9.8% 1.35% Bonds Rebalanced Inflation Source: Michael Kitces, What Returns Are Safe Withdrawal Rates REALLY Based Upon? The Kitces Report (August 212): www. kitces.com. Based on data from Schiller ( Figure 7. annualized Real Returns of 6/4 Portfolio for 15 Years vs. 3- Year Safe Withdrawal Rate Annualized Real Return and Safe Withdrawal Rate (%) Year Safe Withdrawal Rate Year Annualized Real Return Starting Year Source: Michael Kitces, The Impact of Market Valuation on Safe Withdrawal Rates, The Kitces Report (March 28): 72 September CFA Institute cfapubs.org

7 Are Safe Withdrawal Rates Still Relevant in Today s Low-Return Environment? at the end of the horizon has little impact. In fact, the correlation between early returns and the safe withdrawal rate is an incredible.9. Ultimately, this 15-year phenomenon explains a lot about the worst historical scenarios. For the 1966 retiree, the first 15 years of real returns on stocks was.1% and on bonds was 1.1%. The rebalanced portfolio had a real return of.6% for 15 years. The 1937 retiree would have had better equity returns at 3.5%, but real bond returns were 3.1% annualized for 15 years as bonds went from a deflationary period to a period of ultra-low interest rates and, finally, to a period of rising interest rates. The rebalanced portfolio would have had a real return of 1.2% for the 15-year time horizon. This example illustrates what kinds of scenarios determine the 4.% safe withdrawal rate a real return of 1.% on stocks for 15 years, negative real returns on bonds for 15 years, and a real return of less than 1.% on a rebalanced portfolio for 15 years. Eventually, the second half of these horizons yields much better returns, but until then, spending must be low enough so that when the good returns finally show up, the portfolio has retained enough of a balance to capitalize on them. Safe Withdrawal and Market Valuation. One of the reasons 4.% has been determined as a safe withdrawal rate is that the timing of bad returns is presumed to be unpredictable, but that presumption is not necessarily true. Figure 8 shows the relationship between safe withdrawal rates and the Shiller cyclically adjusted P/E (CAPE). The dotted line is the safe withdrawal rate for any particular 3-year time period. The solid line represents the CAPE. The chart shows a near-perfect inverse correlation. Thirty-year withdrawal rates are remarkably well predicted by the ten-year trailing P/E on the date the retirement horizon begins. The reason the CAPE works so well is that it is a fairly good predictor of 15-year returns (which, as I noted, are themselves a strong predictor of 3-year withdrawal rates). This fact is important because these withdrawal rates can be determined in advance, based on already-then-known trailing data. In other words, Figure 8 illustrates that the onset of bad periods and the need for 4% safe withdrawal rates is not entirely random. The four worst time periods for retirement 197, 1929, 1937, and 1966 correlate perfectly with the four peaks of the solid line. Returns and sequencing matter, but what matters most is fundamental valuation. Fundamental valuation is not a short-term predictor, but over the long term, it does indicate that a high valuation at the starting point creates quite a headwind for sustainable withdrawals. A Safe Withdrawal Rate Today. Today s environment of historically low interest rates combined with very high CAPEs is somewhat unprecedented. Usually, when interest rates are low, equity valuations have already fallen and are very low. Poor economic news drives down equity prices, interest rates fall in response, and then equities rally as the economy recovers. Because the current environment is unprecedented rates have fallen but valuations are still high there is much uncertainty as Figure 8. Starting P/E vs. Safe Withdrawal Rate over a 3-Year Period Starting CAPE 35 Safe Withdrawal Rate (%) Safe Withdrawal Rate Starting CAPE Starting Year Source: Michael Kitces, The Impact of Market Valuation on Safe Withdrawal Rates, The Kitces Report (March 28): CFA Institute cfapubs.org September

8 CFA Institute Conference Proceedings Quarterly to its effect on safe withdrawal rates. Shorter-term Treasury Inflation-Protected Securities currently have negative real returns. Those negative real returns might continue, and equities and other assets might experience below-average real returns in the future as well. There are no similar historical scenarios to model the current environment against. Bond yields today are similar to what they were during the average of the bad periods; the 15-year TIPS yield is slightly negative at.15% historically, which is close to where it is now. Notably, that means to breach the 4.% withdrawal rule would require not just poor bond returns, but worse than just negative real returns, and a rebound in rates would remove much of this risk. In addition, it is notable that the worst safe withdrawal rates require not only negative real returns on bonds but also very poor 15-year stock returns a real return on equities that is less than 1% for the next 15 years to merely match the historical scenarios. Accordingly, if the 4.% rule seems too risky in today s environment, the math reveals that doing worse would require the S&P 5 Index to be priced at only about 1,675 in 228. In other words, to truly breach the 4.% withdrawal rule would require the S&P 5 to be lower in 228 than it was in To be fair, on a global basis, those extreme scenarios have happened. Wade Pfau applied the safe withdrawal rate approach on a global scale. 2 He examined 1 years of data over different rolling periods to determine the lowest withdrawal rates for various countries. Canada, Sweden, Denmark, the United Kingdom, Australia, and Switzerland have safe withdrawal rates that range from 3.6% to 4.4%. But for other developed nations, the safe withdrawal rates are much worse. The safe withdrawal rate in Japan is.4%, and in Germany, it is 1.1%. In France, it is 1.2%, in Belgium 1.5%, and in Italy 1.6%. There is a common theme among these countries with very low safe withdrawal rates: They have all been on the losing side of a global war (or at least had the war fought on their domestic soil, with severe damage to their workforce and productive capacity). In such an environment, it is perhaps not a surprise that those who have retired are going to have a bad retirement. But the situation also suggests that short of preparing for World War III, the safe withdrawal rate may be more stable than it is often given credit to be. 2 Wade Pfau, An International Perspective on Safe Withdrawal Rates: The Demise of the 4 Percent Rule? Journal of Financial Planning, vol. 23, no. 12 (December 21): Today s Environment There is some debate about whether today s lowreturn environment will truly turn out to be below some of the thresholds I have discussed in a historical context. Certainly, it is hard to deny that the environment today is at least close to some of these thresholds. Nonetheless, that situation is not necessarily fatal to a retirement plan. If I ran a Monte Carlo analysis for a retiree who was five years into his retirement beginning in 197, 1929, 1937, or 1966, there would be many scenarios in which the risk of failure appears to be increasing, but in the end, none in which the retirement plan actually fails because ultimately in all these environments, fundamentals took hold. Market returns can be miserable for many years, but at some point, stocks or bonds again yield better returns. Viewed another way, the cyclicality of asset prices and market returns may paint a dismal picture at the low points, but cyclicality also helps get retirees back on track. It is also notable that there are many other assumptions built into the withdrawal rates that are being challenged in current research. One of the major assumptions is that clients will continue their spending, adjusted for inflation, every year for life. But in truth, most 88-year-old clients are not quite as active as most 61-year-old clients. Fortunately, declining spending in the later years becomes a buffer, which means in practice that some spending adjustment valves that are not modeled in research provide for additional management through bad environments in the real world. There are other reasons to be supportive of safe withdrawal rates even in today s environment. One is that the safe withdrawal research is based on remarkably undiversified portfolios relative to today s standards. It is based on a simple portfolio investing 6% in S&P 5 stocks and 4% in intermediate government bonds, with annual rebalancing and no changes in the investment strategy for 3 years. Increasing diversification leads to higher withdrawal rates, particularly because diversification helps hedge against some of the really disastrous scenarios. The safe withdrawal rate in 1966 was low because of the highinflation environment. If the 1966 portfolio had held some assets for inflation protection, such as real estate and commodities, the withdrawal rate would have increased from 4.% to 5.% or 6.%. There are a number of different levers and ways that withdrawal rates can be adjusted in today s environment to manage difficult environments. 74 September CFA Institute cfapubs.org

9 Are Safe Withdrawal Rates Still Relevant in Today s Low-Return Environment? For advisers who believe the markets are going to breach the kinds of 15-year real returns I discussed earlier, where it does not matter how good the returns are in the second half of the horizon because the portfolio may be depleted in the first half, spending could be reduced to 3.5% or 3.% to ensure there is enough money to take advantage of good returns when they finally occur. Alternatively, in this case, an adviser could simply build a laddered bond portfolio of TIPS or buy an inflation-adjusted annuity to yield approximately % in real spending levels and skip investing in growth assets altogether. But it is worth bearing in mind that if clients have that negative an outlook on the world, they should begin to worry about the stability of insurance companies, the financial system, and government bonds as well. Unfortunately, black swan tail events can even affect investments that have guarantees. But for most advisers, the best course of action is to monitor the portfolio regularly and moderate clients spending levels so that if a horrible sequence of returns does occur, clients will still have assets left when the good returns show up. Depending on how the retirement is going, spending can be moderated or increased as needed. Fortunately, monitoring retirement plans on an ongoing basis and making adjustments as needed is the most common approach among practitioners, but that does not invalidate the importance of setting a proper initial spending level to start the retiree down the path. Conclusion As I conclude, the one insight I want to emphasize is that safe withdrawal rates have nothing to do with average returns; they are derived from worst-case scenarios. Safe withdrawal rates are determined according to sequencing risk and the timing and volatility of returns. Most of the worst-case scenarios arise from environments of near-zero real returns on stocks and bonds for the first 15 years of retirement. These kinds of scary environments are the ones that dictate 4.% withdrawal rates, not the kinds of environments that necessarily breach the 4.% rule. Depending on the investment outlook, an adviser can manage accordingly by monitoring clients regularly, by actually reducing spending, or in extreme cases, by buying guaranteed investments if the world experiences another catastrophic black swan event. Ultimately, active monitoring and risk management are key. Many of the worst-case scenarios are caused by events that today can be hedged and diversified against; the 4.% rule is based on a portfolio of good, old-fashioned 6% large-cap/4% intermediate-cap government bonds. I hope advisers recognize that the bar that has been set for the safe withdrawal rate is remarkably low and should defend client portfolios against the current environment. Although it is important to recognize low-return environments as they occur, it is equally important not to let recent market returns have so much impact that they distort the outlook for the future. This article qualifies for.5 CE credit. 213 CFA Institute cfapubs.org September

10 CFA Institute Conference Proceedings Quarterly Question and Answer Session Michael E. Kitces Question: Have you done this type of analysis with a declining equity glide path or with asset allocations other than 6/4? Kitces: A declining equity glide path actually leads to worse withdrawal rates. Mechanistically reducing equity exposure by 1% or 2% a year leads to scenarios where, if there is a bear market for the first half of retirement, there are not enough risk assets to benefit from a bull market when the recovery finally arrives. I did an extension on the safe withdrawal rate research a couple of years ago and found that when risk exposure was increased or decreased based on P/Es, the results were better than when risk exposure was adjusted by mechanistic rules. Even if such tactical, valuation-based moves would be a bit early in overweighting and underweighting equities, the client would still end up with materially better results by reducing the volatility. Even without an increase in returns, sequencing is improved and sustainable income is noticeably higher. The bottom line is that there is some value in managing risky asset exposure, but mechanistic paths are a poor way to do it. Only when done on a more dynamic basis does such an approach lead to higher sustainable income. Question: How useful are historical capital market assumptions given the high levels of government debt in developed countries and the prospects for hyperinflation? Kitces: The capital market assumptions I discussed earlier that generated the 4.% safe withdrawal rate already incorporate environments like 1.% real returns on rebalanced portfolios from now until the late 22s to early 23s, assuming that there is some growth during the second half of the horizon. To breach the 4.% rule in the United States would require an extreme scenario, such as experiencing 15 2 years of zero real returns on stocks or bonds and failing to experience growth even after that. The countries that have safe withdrawal rates below 4.% are countries where the economic capacity was destroyed by war or some other catastrophic event. It takes extreme scenarios to breach the 4.% withdrawal rates. Certainly, those scenarios can happen, but frankly, there are few places to hide from hyperinflation and financial system meltdowns if those are the futures managers want to paint for clients. Question: Have you done this type of analysis in a severely deflationary environment, such as the one Japan has experienced? Kitces: The existing safe withdrawal rate research has modeled a U.S. domestic case around how 4.% withdrawal rates would have fared through the deflation scenario in the Great Depression. A 4.% withdrawal rate worked through the Great Depression because government bonds generated enormous real returns during the early part of the retirement horizon while equities were getting slaughtered; rebalancing annually helped to mitigate the effects of negative equity returns. We are in the process of analyzing outcomes of a Japan-like scenario since the late 198s, but we do not have a full dataset at the moment (since it has not been 3 years from the peak yet). Notably, one insight we do see from the deflation scenarios in particular and this insight pertains to asset allocation research overall is that deflation hedges in a portfolio are necessary. Thus, government bonds, despite the poor outlook on their returns going forward, arguably still deserve a place in the retirement portfolio because they are still one of the best deflation hedges. They will clearly not fare well in a more inflationary environment, but in the case of an inflationary scenario, a bear market in bonds can be mitigated by holding such real assets as commodities and real estate in a diversified portfolio. In fact, because of this deflation (or even disinflation) hedging, those who retired in the year 2 and lacked significant exposure to government bonds are experiencing substantial declines in spending rates today. Retirees who had hefty exposures to government bonds are doing fine. They have been getting poor bond yields, but if inflation trends downward and turns negative, then government bonds will be one of the few assets that will continue to provide key real returns through the difficult time period. Question: Have you incorporated detailed forecasts of future tax implications and tax rates into your analysis? Kitces: There is a lot of research on how taxes affect withdrawal rates. In the research I published 76 September CFA Institute cfapubs.org

11 in 21, the tax drag generally amounts to between.25% and.75% of the withdrawal rate, so the 4.% rate falls to somewhere between 3.75% and 3.25%, depending on the tax rate assumptions. The tax drag is not quite as bad as some may think because the sequencing of taxes is favorable; taxes are paid in bull markets and not in bear markets. In fact, losses in bear markets can result in tax deductions or even a tax refund. In other words, taxes are a drag, but they also have a natural self-mitigation effect. A high tax rate of 3% or 4% will translate into only 15 2% of the withdrawal rate because the sequencing effects are favorable. Question: Have you evaluated incorporating structured products that can help smooth volatility, such as principal guaranteed notes or buffer notes; puts or calls around portfolios; and other financial products that have been discussed with respect to longevity planning, such as annuities? Kitces: In terms of options, derivatives, and structured products for hedging strategies, I Q&A: Kitces am optimistic about their potential to materially improve safe withdrawal rates. The behavior of these products through high-inflation environments and deflationary environments needs to be researched because most of these products have not existed throughout historical highinflation and negative-inflation environments (and are difficult to model using hypotheticals). But from a theoretical framework, they should fare well, and I do believe there is a lot of potential in that space. 213 CFA Institute cfapubs.org September