Does ‘bucketing’ achieve its goal?

The practice of ‘bucketing’ has increased in popularity since the Global Financial Crisis as a risk minimisation strategy for clients moving into the retirement phase.

Bucketing involves putting aside (into a separate ‘bucket’) a pre-determined amount of defensive assets (cash and fixed interest) to allow pension payment drawdowns to continue for many years after a significant market downturn, hopefully avoiding the need to sell down growth assets at deflated prices.

How effective is this as a strategy to minimise retirement risk? Increasing the exposure of a portfolio to cash and fixed interest reduces the expected volatility of the portfolio’s returns, so bucketing reduces risk in that context. But coinciding with the reduced volatility is a reduction in the portfolio’s expected return rate also.

More importantly, is return volatility the most important risk metric for retirees? There are numerous possible risk metrics for retirees. One risk is that the desired income stream ceases due to complete loss of capital (referred to in this article as Bust Scenario risk). 

In this article we use mathematical modelling to test the probability that particular outcomes could occur. In the model 2,000 scenarios have been used to assess the likelihood that certain outcomes are more or less probable. As such, the outcomes and probabilities should not be used as any form of prediction of a particular outcome for any specific investor. 

In chart one there are 599 Bust Scenarios (30 per cent) for a balanced portfolio (see Appendix for details) being drawn down at five per cent per annum of the starting capital ($100), indexed with inflation, over 35 years from the 2,000 economic scenarios generated using a stochastic model (see Appendix for more details).

The average return of all scenarios over 35 years is 6.9 per cent per annum, with volatility of 11.5 per cent.

Chart 1: 2,000 scenarios of balanced portfolio with 5% drawdowns

Source: Macquarie

‘Bucketing’ 30 per cent of the portfolio into cash has a number of impacts.

Firstly, the average return of the entire portfolio (70 per cent balanced and 30 per cent cash) over 35 years reduces to 6.4 per cent per annum, meaning less capital is expected to remain at the end of the term. However, as expected, the return volatility also reduces to 8.1 per cent. 

Chart 2: 2,000 scenarios of balanced + 30% cash portfolio with 5% drawdowns

Source: Macquarie

Perhaps more significantly, the number of Bust Scenarios has increased to 908. Note that in chart two the first Bust Scenario does not occur until the 14th year, whereas in chart one the first Bust Scenario occurs in the 11th year.

Although one risk metric (volatility of returns) has decreased by 30 per cent (from 11.5 per cent to 8.1 per cent), another risk metric (the chance of the income stream not lasting the 35-year term) has increased by 50 per cent, from 30 per cent to 45 per cent.

The impact of different term requirements and cash bucket sizes

The analysis above assumes the time frame required for the income stream is 35 years and 30 per cent of the starting portfolio capital has been placed in a cash bucket.

Table one shows the impact of different income stream term requirements and reduced levels of cash in the bucket.

Table 1: return and risk for 5% income stream

Income stream term (years)	Metric	Portfolio composition
		Balanced 100%	Balanced 90% Cash 10%	Balanced 80% Cash 20%	Balanced 70% Cash 30%
15	Average return per annum	6.9%	6.8%	6.4%	6.4%
	Volatility	13.1%	11.9%	9.4%	9.4%
	Crash Scenarios	1%	0%	0%	0%
20	Average return per annum	7.0%	6.9%	6.5%	6.5%
	Volatility	12.2%	11.0%	8.6%	8.6%
	Crash Scenarios	3%	3%	2%	2%
25	Average return per annum	7.0%	6.9%	6.5%	6.5%
	Volatility	11.9%	10.7%	8.4%	8.4%
	Crash Scenarios	10%	10%	10%	10%
30	Average return per annum	7.0%	6.8%	6.4%	6.5%
	Volatility	11.7%	10.5%	8.3%	8.2%
	Crash Scenarios	20%	20%	22%	24%
35	Average return per annum	6.9%	6.8%	6.4%	6.4%
	Volatility	11.5	10.4%	8.2%	8.1%
	Crash Scenarios	30%	33%	41%	45%
40	Average return per annum	6.8%	6.8%	6.4%	6.4%
	Volatility	11.4%	10.4%	8.1%	8.0%
	Crash Scenarios	41%	46%	61%	66%

Source: Macquarie

Increasing the cash exposure reduces the Bust Scenario risk with a 15-year term, but increases it if the term requirement is 30, 35 or 40 years. Over a 25-year term, bucketing neither reduces the Bust Scenario risk, nor increases it.

The impact of increasing exposure to growth assets

If it’s accepted that reducing the overall portfolio exposure to growth assets may increase the risk of a Bust Scenario when the required term exceeds a certain point (for example, 25 years in the balanced portfolio example above), then a subsequent question is what of the converse position? Does increasing the overall exposure to growth assets impact on the Bust Scenario risk? 

Chart three shows that at a five per cent initial drawdown level, a Growth portfolio (80 per cent growth assets, 20 per cent defensive assets) results in a low risk (23 per cent) of failure (ie Bust Scenarios) over 35 years, and a high average return per annum (7.4 per cent). 

Chart 3: 10,626 portfolios by Bust Scenario probability and average return – 5% drawdown

Source: Macquarie

These results are however, very sensitive to drawdown level assumptions. As drawdown levels reduce to very low levels, there’s increasingly less difference in the Bust Scenario risk between the various portfolios (conservative, balanced and growth), so the focus moves to the average return rate.

Conclusions

The concept of ‘bucketing’ is intended to reduce risk for retirees by avoiding the realisation of growth assets at deflated prices during a market downturn. However, the increased exposure to defensive assets that results from bucketing reduces a portfolio’s average return rate, and while it also reduces the volatility of returns (a common risk metric), bucketing can actually increase the Bust Scenario risk in some circumstances.

Appendix - stochastic model details

The Mercer Capital Market simulator

• A stochastic forward-looking model
• Generates 2000 scenarios of yearly serially correlated and mean-reverting asset class returns and inflation
• Applies non-normal return distributions
• For illustrative purposes the analysis by Macquarie is based on Mercer’s long-term “equilibrium” assumptions. Mercer provides both “equilibrium” and “market aware” assumptions 

Income stream scenarios

• Starting income levels as percentage of initial balance: five per cent
• Income level indexed to inflation
• 40-year cash flow projection inclusive of terminal value (if any)
• Portfolio return = Internal rate of return i.e. money-weighted return
• Returns assumed to be net of fees

Chart three

Each dot represents one of 10,626 different possible portfolios using five per cent asset allocation ‘buckets’:

• Cash
• Australian fixed interest (AFI)
• Australian Equities (AE)
• Developed market equities (DME)
• Emerging market equities (EME)

Portfolio	Drawdown level – 5%
	Return (% pa)	Risk per annum
Cash	4.0	4.6
Australian Fixed Interest	4.2	4.4
Australian Equities	7.1	27.9
Developed Market Equities	6.7	26.2
Emerging Market Equities	6.8	44.1

Conservative portfolio: 40 per cent growth assets, 60 per cent defensive assets

Balanced portfolio: 60 per cent growth assets, 40 per cent defensive assets

Growth portfolio: 80 per cent growth assets, 20 per cent defensive assets 

David Barrett is head of technical services at Macquarie.