Handling a Fixed Annuity

This chapter addresses the issue of how to handle withdrawals from a portfolio when a portion of your required income stream is supplied by a fixed annuity. I did not see a way to analyze this with just the data available from the Jarrett and Trinity studies, so I tried to find the portfolio size required to cover the impact of inflation (over the specified 30 year period). In the first year the entire annuity would be spent and no withdrawal would be made from the portfolio. In the second year the annuity would be spent and the required increase due to inflation would be taken from the portfolio. This was repeated each year and a portfolio was considered ‘successful’ if the portfolio lasted through the 30^th year. I also used the same range of asset allocations, the same rebalancing assumptions, etc. as stated previously. The size of the initial portfolio was varied between 4x annuity amount and 10x the annuity amount in increments of 1x.

The results of this analysis are presented in three tables:

1) The percent of the 54 cases (retirement starting years from 1926 through 1979 inclusive) that left a positive balance in the portfolio at the end of 30 years (Click Here).

2) The median ending portfolio value (the number where half the ending values were greater and half were less). Median seems more meaningful to me than average in this particular case (Click Here).

3) The third table is the same as the second table but with the final value expressed as a percentage of the initial portfolio value.(Click Here)

The results were quite interesting and are different from the results of the Fixed withdrawal methodology referenced in the previous studies in these two ways (among others):

Very aggressive equity allocations optimize both the end portfolio value (true for the Fixed Withdrawal cases) and the worst case results (not true for the fixed withdrawal case). The Fixed Withdrawal analysis optimized the worst case scenario at 40% equities (probably closer to 50% but I didn't bother with that level of analytical granularity). The bridge between these two differing sets of results is the fact that a fixed annuity has some of the same characteristics as a bond.

The cost to get from an 80% probability to 100% probability of portfolio success (based on historical data) is much higher in the fixed annuity case compared with the case of no annuity. I played around with the numbers and found that the portfolio size required to generate a 100% success rate divided by the portfolio size required to generate an 80% success rate was around 1.4 for the case of fixed, inflation-adjusted withdrawals from a portfolio (ala' Trinity/Jarrett). For the case of inflation protection of a fixed annuity that ratio is closer to 2.0. Those last few portfolio failures were truly tough to make go away in the case of a fixed annuity and the stagflation of the 1970’s/early 80’s. My personal reading of the Jarrett/Trinity data led me to planning assumptions revolving around the 100% success data points. I may reconsider this for the fixed annuity case.

A suggested methodology for handling the case where part of your income requirements are supplied by a fixed annuity is (for calculation purposes) to separate your assets into two baskets:

A basket to provide inflation protection for the fixed annuity

A basket to provide the income required in excess of the fixed annuity.

The hypothetical case of a fixed annuity of $25,000 and a requirement for an inflation adjusted  retirement income target of $50,000, would require the following asset base assuming the factors of 9 (inflation protection of the fixed annuity) and .044 are chosen:

$25,000 x 9 + (50,000-25,000)/.044 = $793,000 (gulp)

There is nothing necessarily 'mathematically correct' about the handling this problem as 'two baskets of money'. But it seems to me to be a reasonable approach and should be a good upper bound. For those of you who are mathematically oriented, this assumption is closer to correct when the conditions that negatively affect the Fixed Annuity case and negatively affect the Fixed Withdrawal case are well correlated. This tends to be the true (and unfortunately also generates the highest required portfolio values).

Readers with retirement horizons longer than 30 years, will need to make minor adjustments (something that is partially addressed in the Withdrawals Forever chapter).