Safe withdrawal rates for Australian retirees

David Blanchett/Anthony Serhan/Peter Gee | 02 Feb 2016

Page 1 of 1

David Blanchett is Morningstar's US-based head of retirement research, Anthony Serhan is Morningstar's managing director of research strategy Asia Pacific, and Peter Gee is a Morningstar fund analyst.

 

There is a growing body of literature on safe withdrawal rates for retirees. Most of this research has reached conclusions based on historical returns, primarily historical returns in the United States.

While it's impossible to predict the future, the market environment today for Australians is unique, and therefore it makes more sense to base withdrawal rates off expectations versus history.

In this paper we briefly explore safe withdrawal rates from the perspective of historical returns, both international and domestic, but more importantly provide some context of safe withdrawal rates given our return expectations. There are four primary findings from this research.

First, Australians may have unrealistic future return expectations given how well the markets have performed historically.

Second, while safe withdrawal rates today are similar to historical averages, they are lower and may be significantly lower when incorporating improvements in mortality and the impact of fees.

Third, current minimum withdrawal rates for account-based pensions in Australia may lead to investors depleting retirement assets too soon.

Finally, a balanced portfolio is likely the best allocation for Australia retirees. Overall, while these findings are less optimistic than past research on the topic of safe withdrawal rates, they are nevertheless an important starting place for retirees and financial advisors today.

 

Safe initial withdrawal rates

Initial research by Bengen (1994), among others, suggests an initial safe withdrawal rate from a portfolio is 4 per cent of the assets, where the initial withdrawal amount would subsequently be increased annually by inflation and assumed to last for 30 years (which is the assumed duration of retirement).

This finding led to the creation of the "4 per cent Rule," a concept that is often incorrectly applied:

• The "4 per cent" value only applies to the first year of retirement, whereby subsequent withdrawals are assumed to be based on that original amount, increased by inflation.

• Additionally, the assumed retirement period is 30 years, which may be too short or too long based on the unique attributes of that retiree household.

• The idea of "safe" was measured by the likelihood that you would still have money left after 30 years. The withdrawal rate generated from such analysis does not reflect the expected or median return on assets, but rather returns at the lower end of possible outcomes.

Exhibit 1 provides some insight as to how the "4 per cent Rule" and other withdrawal rate heuristics have largely been determined. The returns for the analysis come from the Dimson, Marsh and Staunton dataset and the assumed portfolio is 50 per cent US shares and 50 per cent US bonds.

The analysis in Exhibit 1 demonstrates the highest possible initial withdrawal rate based on an assumed individual retiring in the US where retirement lasts 30 years, but the starting point of that 30-year period varies by time and reflects the investment returns and inflation patterns of each 30-year period.

Early research on safe initial withdrawal rates relied heavily on historical returns, especially historical US returns--this can partially be attributed to data availability as well as the fact many early researchers in retirement were based in the US.

 

Exhibit 1 Initial sustainable withdrawal rate % -- Where the 4% rule comes from


chart


Source: Morningstar

 

The shaded area in the chart is the lowest safe withdrawal rate over the entire period. Therefore, 4 per cent was selected as the safe initial withdrawal rate. There are a number of problems extrapolating these results to Australian retirees today.

First, the analysis did not include fees at all. There is a definite cost to investing that needs to be considered when estimating withdrawal rates.

Second, the analysis assumes retirement lasts 30 years, while in reality the expected duration of retirement (and respective modeling period) should vary by retiree.

Third, this problem ignores the experience of retirees in other countries. Just because a 4 per cent initial withdrawal has been safe in the US does not mean it would have been safe in Italy (Exhibit 3 demonstrates that it hasn't).

Finally, it assumes past returns are a reasonable basis for retirees today. While the past provides some window into the future, the markets today are in a different place than historical long-term averages, and this needs to be taken into account when advising a retiree on a safe initial withdrawal rate.

 

Historical returns: An international perspective

Return assumptions are a significant driver when estimating a safe initial withdrawal rate, likely second behind the length of retirement in overall relative importance. While historical US returns provide some context as to how safe a variety of withdrawal rates would have been for Americans, it does not create the appropriate historical context for Australians.

In Figure 2 we recreate the analysis in Figure 1, but instead of using historical US returns (as we do in Exhibit 1) we use historical returns for an Australian investor. We also include a portfolio fee of 1.00 per cent.

 

Exhibit 2 Initial sustainable withdrawal rate % -- The "4% Rule"… a (historical) Australian perspective


chart


Source: Morningstar

 

Had early withdrawal rate research been based on the analysis in Exhibit 2, early research would not have suggested that a 4 per cent initial withdrawal rate is safe, rather it would be closer to 2.5 per cent.

A 2.5 per cent initial withdrawal rate implies a retiree needs 40 times the desired retirement income goal (1/2.5 per cent=40). This is significantly more wealth than is inferred from the 4 per cent rule, which is only 25 times the desired retirement income goal (1/4 per cent=25). Clearly, the assumed returns have a significant impact on the analysis.

To provide an even greater perspective this analysis is done for each of the 20 countries in the Dimson, Marsh, and Staunton dataset, with the results included in Exhibit 3.

The results in Exhibit 3 include the initial withdrawal for varying target probabilities of success, where retirement is assumed to last 30 years, the portfolio is invested domestically in 50 per cent shares and 50 per cent bonds, and the annual portfolio fee is 1.0 per cent of assets.

 

Exhibit 3 Safe initial withdrawal rates at various target success rates by country


chart


Source: Morningstar

 

The true safe withdrawal rate varies significantly by country and target success rate. For example, using the historical returns in Japan a 95 per cent target success rate would yield an initial withdrawal rate of 0.2 per cent versus 3.0 per cent for Australia.

Interestingly, the highest initial withdrawal rates across the 20 countries have been based on US returns. This suggests safe withdrawal rates may be overly optimistic on a global perspective. For example, using the US returns and targeting a 90 per cent success rate yields an initial safe withdrawal rate of 3.54 per cent.

This is the highest initial withdrawal among the 20 countries and is considerably higher than the 20-country average, which is 2.29 per cent.

Historical returns obviously play an important role when determining safe withdrawal rates. Exhibit 4 includes the historical inflation-adjusted (real) returns and risk for shares and bonds by country over the entire test period (from 1900 to 2014) as well as the returns for a 50/50 portfolio.

 

Exhibit 4 Historical inflation-adjusted returns and risk by country: 1900-2014


table

Source: Morningstar

 

Australians have clearly been very fortunate in terms of historical returns. For example, the real return of shares (bonds) over the period in Australia was the third (ninth) highest among the 20 countries.

The relative risk of shares (bonds) has also been relatively low, with the second (eleventh) lowest level of volatility. The balanced portfolio in Australia had the third-highest return and the fourth-lowest risk.

Over this same period, inflation in Australia ran at 3.81 per cent per annum, which would make the nominal return on Australian shares closer to -13 per cent (when inflation is added to the 8.94 per cent shown in Exhibit 4).

Overall, these returns indicate that Australians have been relatively fortunate from a historical returns perspective. This may be problematic when modeling retirement if investors have unrealistic expectations about future returns.

While it's certainly possible Australian capital markets may continue to outpace their international peers, it's important to get a better understanding of the risks and expectations facing Australian investors today before reaching any conclusions.

 

Investing and retiring in Australia today

In Australia today, there are three primary types of retirement income stream products available: account-based pensions (investment product with structured drawdown), annuity (guaranteed product), or a combination of both.

Account based pensions dominate the income streams market in Australia. These products are essentially a managed investment with minimum annual drawdowns required by legislation. There are a number of reasons behind the popularity of account-based pensions:

1. Control of capital--account holder has discretion over the drawdown rate above legislated minimum.

2. Estate planning advantages--upon holder's death the balance may be transferred to the estate or the pension payments can continue to be paid to a dependent beneficiary.

3. Allows selection of risk/return profile of the assets held.

4. Product is simple and transparent.

5. Product fits easily within traditional advice models.

6. Less costly for providers than guaranteed income streams--no need to hedge or hold capital against retirement risk.

Annuities make up a very small percentage of the retirement income streams market in Australia. The main reasons behind the lack of demand include:

1. Retirees' preference to access capital and a desire to leave a bequest.

2. Individuals underestimate their life expectancy.

3. Perception that annuities are costly and do not deliver value for money.

4. Notion that the Age Pension provides longevity protection.

5. Only a limited number of providers issuing the products.

6. The low interest rate environment has hindered the attractiveness of the products.

The legislated minimum payment amount for account-based pensions is set out under Schedule 7 of the Superannuation Industry Supervision Act and came into effect on 1 July 2007 under the Government's "Simplified Superannuation" reforms.

The minimum payment amounts are worked out simply as the pension account balance multiplied by the percentage factor. The percentage factor is tabled below and is based on the beneficiary's age on 1 July in the financial year in which the payment is made.

 

Exhibit 5 Minimum annual payment for account based pensions


table


Source: www.moneysmart.gov.au

 

Historically, drawdown relief has been provided--the percentage factors have been reduced by the government during times of poor investment market performance in a bid to avoid drawing down already savaged retirement saving balances.

For the 2008, 2009 and 2010 financial years, the percentage factor was reduced by half (for example, 2.5 per cent for a 65 year old) and over the 2011 and 2012 financial years, beneficiaries were only required to drawdown 75 per cent (3.75 per cent for a 65 year old) of the legislated minimum percentage factors.

The logic behind the legislated annual minimum income payments is to ensure that the account-based pension is being used to provide an income stream in retirement. The principle is that the funds are withdrawn from the concessionally taxed superannuation environment over time and not preserved for the accountholder's beneficiaries--the superannuation system should not be used as an estate planning tool.

It is important to note that the legislated minimum payment factors were set based on average investment returns and average life expectancies at various ages. Also, this paper does not consider other financial planning strategies such as transition-to-retirement options or how account-based pensions interact with social security payments such as the aged pension.

 

Australian share-market characteristics

The Australian share market only represents approximately 2 per cent of global markets. It follows that investment opportunities in Australia are restricted when compared to global markets.

Looking at the different weights to various industries in domestic and global share markets (Exhibit 6), it can be clearly seen that the Australian market is highly concentrated to financials (inclusive of real estate) and materials.

The defensive sectors--utilities, energy, telecommunication services, health care and consumer staples--are not well represented domestically. Adding defensive stocks reduces cyclical bias and has the benefit of cushioning returns in down markets.

There is a clear need to focus on investment risk as the population ages. Diversification is a key tool to reduce the impact of wealth shocks.

 

Exhibit 6 Sector weights: Australia versus world


chart


Source: Morningstar. Data as of 31 October 2015.

 

It is also worth noting that in addition to sector concentration, the Australian share market also exhibits issuer or security-specific concentration. The top 10 stocks made up approximately 48.9 per cent of the S&P/ASX 300 index as of 31 October 2015.

Importantly, many investors have come to recognise the importance of overseas diversification in equity portfolios and the exposure has been gradually increasing. Today, the typical superannuation fund has an approximately equal split between Australian and international equities.

A home country bias in equity exposure is something seen in many markets around the world and while there are arguments for this to exist to some extent (for example, the matching of assets and liabilities), the reduction of this bias in Australia is a positive step towards managing risk.

 

Longevity risk

The other important risk when considering safe withdrawal rates is longevity risk. The Australian Bureau of Statistics (ABS) life tables are a commonly used measure of life expectancy for Australians.

Exhibit 7 includes information about how life expectancies for a newborn have changed since 1890, increasing by 32.67 years for males and 33.41 for females. By 2011, life expectancy was 79.9 years and 84.3 years for a boy and girl, respectively.

 

Exhibit 7 Life expectancy for a newborn (years): 1881–2012


chart


Source: ABS

 

The ABS tables are based on the period life expectancy methodology--it is the average number of years a person will live if the age-specific mortality rates at that point in time were to be applied for the rest of the person's life.

The reality is that mortality rates are more than likely to improve in the future due to advances in technology and medicine, so the period tables are likely to underestimate the number of years someone could expect to live.

The cohort life expectancy method takes into account assumptions of improvements in mortality rates over a person's lifetime. In other words, instead of being based on the mortality rates for all ages in a given year, the cohort life expectancy approach takes the age specific mortality rate year by year for the particular year in which the person would be that age.

Projections based on the cohort life expectancy method are highlighted in the 2015 Intergenerational Report and detailed in the below table. A baby boy born in 2015 is expected to live 91.5 years and a baby girl 93.6 years.

Clearly, these projections paint an increasing risk to retirement funding that is not necessarily fully understood by Australians today.

 

Exhibit 8 Australia's projected life expectancy (years)


chart


Source: 2015 Intergenerational Report

 

Return expectations

It is impossible to predict the future with absolute certainty. Therefore, investors must determine what return assumptions to use in a model.

While using historical returns is sometimes viewed as a simpler path than attempting to forecast returns, we believe using forecasted returns is the best approach since it incorporates today's market valuations.

Morningstar makes the following assumptions about the expected long-term behaviour of key asset classes including cash, domestic and international fixed interest, domestic and international property, and domestic and international shares.

Historical asset class performance is not simply used to generate the capital market assumptions used as inputs to the modelling process. We use a supply-side, building-block approach to forecast equity returns.

First introduced by Diermeier, Ibbotson and Siegel (1984), and later adapted to stocks by Ibbotson and Chen (2003), the supply-side model is based on the idea that equity returns can be decomposed into underlying economic and corporate fundamentals.

Fixed-interest returns are derived using a similar approach based on expectations for cash rates, inflation and credit spreads. Assumptions refer to the long term, that is, multiple decades.

It is therefore possible that fluctuations in markets may move outcomes away from the base-case strategic assumptions during the short to medium term. These concepts are displayed visually in Exhibit 9.

 

Exhibit 9 Building blocks for equity and fixed income returns


chart


Source: Morningstar

 

Long-term returns for most capital markets are generally forecast to be lower than long-term history. This is particularly the case within equities, where above-average valuations in many markets have diminished future return expectations. Interest-generating assets are also being affected by lower prevailing market yields.

Morningstar uses several valuation models to estimate the fair value of equity asset classes and assumes reversion to fair value over a 20-year period. Our research suggests that a combination of multiple valuation measures has a significantly better predictive power than any single valuation model.

Specifically, our valuation models rely on several forward-looking measures of normalised earnings such as profit margins, return on book-equity and inflation-adjusted average earnings over the business cycle.

Other equity buildings blocks incorporate earnings growth, total yield (dividends and buybacks) and inflation. Exhibit 10 includes information about our projected returns for a variety of asset classes.

 

Exhibit 10 Return and risk assumptions for various Australian investments


table


Source: Morningstar. *Includes return from imputation credits. # Arithmetic returns are used in the return projections. Return forecasts will show returns closer to the arithmetic mean over shorter periods, but will converge to the Geometric mean over the very long term. The returns shown are before fees, taxes and inflation

 

Safe withdrawal rates: A forward-looking perspective

In the previous sections we provided a review of some of the risks for Australian retirees on making withdrawal rate decisions based on US historical equity and bond returns.

In this section, we are going to run some additional forecasts to determine safe withdrawal rates for retirees using the expected returns from Exhibit 10.

The portfolios used in these examples are built on a more diversified combination of Australian and international shares, bonds and cash than those used in the previous simulations.

Exhibit 11 includes information on the probability of success for various initial withdrawal rates based on different equity allocations, under the assumption that retirement lasts 30 years.

 

Exhibit 11 Success rates for various initial withdrawal rates and portfolios (30 year retirement period)


chart


Source: Morningstar

 

As shown in Exhibit 11, and as you might expect, the greater the certainty you want about the withdrawal rate (higher probability of success) the less equities required in the portfolio.

Once you are at the 70 per cent probability level, more equities produces higher withdrawal rates although most of this benefit is achieved at the 50 per cent equity level.

Exhibit 12 includes information on the probability of success for various initial withdrawal rates based on different retirement periods, under the assumption that the portfolio is invested in 50 per cent shares and 50 per cent bonds.

The chart illustrates the large impact the retirement period can have on the initial withdrawal rate, in particular the jump from 20 to 30 years. At the 70 per cent probability level, increasing the retirement period from 20 to 30 years reduces the withdrawal rate by approximately 26 per cent (5.7 per cent to 4.2 per cent).

Similarly, the probability of success for a 4 per cent initial withdrawal rate is 99 per cent for a 20-year period, 78 per cent for a 30-year retirement, and 46 per cent for a 40-year retirement.

 

Exhibit 12 Success rates for various initial withdrawal rates and retirement periods (50% shares & 50% bonds)


chart


Source: Morningstar

 

Exhibit 13 has been provided to include additional information about specific appropriate withdrawal rates for different portfolio allocations, retirement periods, and target success levels.

 

Exhibit 13 Withdrawal rates by portfolios... Time period + target success rate


table


Source: Morningstar

 

The initial savings required to fund retirement can be estimated by taking 1 divided by the target initial withdrawal rate. For example, if 4 per cent is the assumed safe withdrawal rate the initial savings required to fund the retirement income goal would be 25 times that income need (1/4 per cent=25x).

The initial withdrawal rates in Figure 13 differ significantly, whereby longer retirement periods, higher probabilities of success, and more conservative portfolios tend to yield lower initial withdrawal rates.

The impact of the probability of success is definitely notable. For example, assuming a 50 per cent share and 50 per cent bond portfolio, and a 30-year retirement period, a 99 per cent probability of success yields an initial withdrawal rate of 2.8 per cent (1/2.8 per cent=36x multiple) versus an initial withdrawal rate of 4.6 per cent (1/4.6 per cent=22x multiple) for a 50 per cent probability of success, which are significantly different levels of required savings.

Overall, the results in Figure 13 suggest the actual amount of required savings to fund retirement is a very personalised and complex decision where a financial advisor has the potential to add significant value.

 

Implications

How should retirees and financial advisers use this research? First, the assumed retirement period should vary by client. Recognise that a 30-year time horizon is ideal for a hypothetical 65-year old retiree who dies at age 95. But based on the ABS life tables remaining life expectancy at age 65 is less than 30 years (approximately 23 years), so many will die with money unspent.

That represents a failure from being too conservative. Our simulations with retirement lasting over 30 years resulted in some relatively low safe initial withdrawal rates; however, it may be possible to hedge this longevity risk through annuitisation (the pooling of longevity risk).

Most retirees will also not need to spend the same amount every year. For couples, the longer-lived member won't spend as much as a single-person household. Retirees generally decrease spending as they experience physical and mental limitations throughout retirement.

In addition, most retirees are willing to cut spending a little when markets don't do as well as they had hoped. Incorporating variability into spending can increase the safe initial withdrawal rate significantly.

The probability of success is only one way to measure outcomes for a retiree. It fails to show the magnitude of the failures early in retirement, and it doesn't consider the security of retirees who live well beyond the 30-year timeframe.

By failing to consider the magnitude of failure, portfolio risk is increased, leaving retirees vulnerable to adverse market events, particularly those early in retirement.

Lastly, when comparing Exhibit 5: Minimum Annual Payment for Account Based Pensions to the minimum withdrawal rates generated in these scenarios it is clear that the current required minimums are set well above our projected safe withdrawal rates.

The way these two rates operate is different after the first year: the minimum per cent rate is applied to the current balance each year and increases over time in line with the government schedule; while the safe per cent rate is applied at the start and sets the annual dollar withdrawal amount which is then increased by inflation each subsequent year.

The current minimum withdrawal rates are increasing the probability that assets will be depleted much earlier in retirement and result either in pension payments that do not keep track with inflation; or the complete extinguishment of retirement assets before death.

This modelling has not addressed the question of adequacy levels but it does show the direct relationship between assumed withdrawal rates and the amount of capital required at the outset.

Any attempt to increase withdrawal rates to ensure assets are used for retirement incomes also brings with it the increased probability of longevity risk (outliving your assets) and greater reliance on government funded social security.

Conclusions

This paper provides a relatively comprehensive overview of safe withdrawal rates for retirees, based on both historical returns as well as forward-looking returns.

Overall, these findings suggest that financial advisers and retirees in Australia should use lower initial safe withdrawal rates than noted in prior research-- the lower end of the range now starts towards 2.5 per cent and not the previous 4.0 per cent.

The generous capital market returns of the prior century that bolstered a comfortable and long-lasting retirement portfolio may give 21st-century clients a false sense of security.

The paper also highlights the way probability of success can be used to understand potential outcomes. While expected returns are a midpoint operating at the 50 per cent probability of success, our definition of "safe withdrawal" has been calculated in the range of 70 per cent to 99 per cent success.

Helping retirees understand the certainty of retirement incomes in this context is an important step to better meeting expectations.

While this analysis provides a useful framework to consider the question of retirement spending it also highlights the importance of understanding the specific needs and preferences of a retiree in framing investment objectives.

To watch a video relating to the aforementioned analysis, please click here.

 

References

Australian Bureau of Statistics 2014. "Australian Historical Population Statistics," cat. no. 3105.0.65.001.

Bengen, William P. 1994. "Determining Withdrawal Rates Using Historical Data." Journal of Financial Planning, vol. 7: 171–180.

Department of Treasury (Australia), 2015. "2015 Intergenerational report: Australia in 2055"

Diermeier, Jeffrey J., Ibbotson, Roger G., Siegel, Laurance B. 1984. "The Supply for Capital Market Returns." Financial Analysts Journal, vol. 40, no. 2: 2-8.

Ibbotson, Roger G., Chen, Peng. 2003. "Long-Run Stock Returns: Participating in the Real Economy."
Financial Analysts Journal, vol. 59, no. 1: 88-98.

The authors would like to thank James Foot for his assistance with the capital market assumptions and explanation of the building blocks approach.

 

© 2016 Morningstar, Inc. All rights reserved. Neither Morningstar, its affiliates, nor the content providers guarantee the data or content contained herein to be accurate, complete or timely nor will they have any liability for its use or distribution. This information is to be used for personal, non-commercial purposes only. No reproduction is permitted without the prior written content of Morningstar. Any general advice or 'class service' have been prepared by Morningstar Australasia Pty Ltd (ABN: 95 090 665 544, AFSL: 240892), or its Authorised Representatives, and/or Morningstar Research Ltd, subsidiaries of Morningstar, Inc, without reference to your objectives, financial situation or needs. Please refer to our Financial Services Guide (FSG) for more information at www.morningstar.com.au/s/fsg.pdf. Our publications, ratings and products should be viewed as an additional investment resource, not as your sole source of information. Past performance does not necessarily indicate a financial product's future performance. To obtain advice tailored to your situation, contact a licensed financial adviser. Some material is copyright and published under licence from ASX Operations Pty Ltd ACN 004 523 782 ("ASXO"). The article is current as at date of publication.

This report appeared on www.morningstar.com.au 2021 Morningstar Australasia Pty Limited

© 2021 Morningstar, Inc. All rights reserved. Neither Morningstar, its affiliates, nor the content providers guarantee the data or content contained herein to be accurate, complete or timely nor will they have any liability for its use or distribution. This information is to be used for personal, non-commercial purposes only. No reproduction is permitted without the prior written content of Morningstar. Any general advice or 'class service' have been prepared by Morningstar Australasia Pty Ltd (ABN: 95 090 665 544, AFSL: 240892), or its Authorised Representatives, and/or Morningstar Research Ltd, subsidiaries of Morningstar, Inc, without reference to your objectives, financial situation or needs. Please refer to our Financial Services Guide (FSG) for more information at www.morningstar.com.au/s/fsg.pdf. Our publications, ratings and products should be viewed as an additional investment resource, not as your sole source of information. Past performance does not necessarily indicate a financial product's future performance. To obtain advice tailored to your situation, contact a licensed financial adviser. Some material is copyright and published under licence from ASX Operations Pty Ltd ACN 004 523 782 ("ASXO"). The article is current as at date of publication.