Retirement Cash Buffer Simulation Study

Disclaimer

• This analysis is for informational and educational purposes only.

• There could be bugs in our programmatic analysis. We are not liable for any actions you take as a result of anything you read below. Please conduct your own due diligence before making investment decisions.

• Any decisions influenced by the information in this analysis are made at your own risk.

Overview

This analysis is an extension of the Retirement Asset Allocation Simulation Study linked below:

Retirement Asset Allocation Simulation Study

In the Retirement Asset Allocation Simulation Study, we quantified the trade-offs between various stock and bond asset allocations during both working / saving and retirement / spending years. We observed that high stock allocations resulted in the best average returns, the most volatility, and poor worst-case performance (through the Great Depression, for example). This got us wondering if keeping some retirement assets in cash — and taking withdrawals from that cash during poor investment return years — could mitigate the downside of high stock asset allocations while preserving most of their power to produce high returns.

Here we will attempt a data driven answer for the following questions as they pertain to the retirement / spending years:

• Will keeping a portion of retirement assets in cash — and taking withdrawals from that cash during poor investment return years — result in higher retirement account values compared to a no-cash approach?

• What is the optimal amount of cash to maintain?

• If cash has been depleted during poor investment return years, how quickly should it be replenished during strong investment return years?

• Is there an optimal investment return threshold for when to make withdrawals from cash vs investments?

• Is there an optimal investment return threshold for when to replenish cash?

• What are the trade-offs between different approaches?

To answer these questions, we will use historic stock and inflation data to simulate (back-test) different asset allocations and compare outcomes.

Note: Cash is assumed to be held inside a tax advantaged account (401k, 403b, IRA, etc.) to avoid tax implications. All references to “stocks” should be interpreted to mean index funds of stocks rather than individual stocks (e.g. Google (GOOGL), Amazon (AMZN), Apple (AAPL), Deere and Co. (DE)).

Data

Two data sources were used in this analysis:

• S&P 500 Total Return (price changes plus dividends, used to represent the stock portion of asset allocations)

  • https://www.slickcharts.com/sp500/returns

• Consumer Price Index (CPI) (used to make inflation adjustments to dollar values)

  • https://data.bls.gov/timeseries/CUUR0000SA0?years_option=all_years

Below are annualized summary plots of the data sources.

Retirement / Spending Cash Buffer Simulation Details

The following assumptions were used for the cash buffer simulation:

• Retirement length of 30 years

• Starting retirement savings of $2.5M

• Annual spending of $100,000 per year (4% of retirement savings)

• One 30-year simulation will be run starting each year from 1926 to 1993

• All investment returns are reinvested

• All invested dollars will have a 100% stock / 0% bond asset allocation

• If there is cash available, all withdrawals are made from cash during poor investment return years

• If the available cash is below the target amount, it is replenished during strong investment return years

• The amount of investments that can be liquidated into cash in a single year is capped (more on this later)

• Money in the cash buffer does not earn interest

• All plot values will be inflation adjusted to 2023 dollars

The variables we’ll be trying to optimize are the cash amount (quantified in years of spending), the investment return threshold for when to make withdrawals from cash vs investments, the investment return threshold for when to replenish cash (10% total return, 15% total return, etc., to reduce complexity the investment return threshold for spending cash vs investments will be the same as the investment return threshold for replenishing cash), and the maximum amount of cash to liquidate in years that meet the investment return threshold (quantified in years of spending).

Cash Amount Optimization

We’ll start by optimizing the amount of cash. A 15% investment return threshold was used for the decision to spend cash vs investments and when to replenish cash. The maximum single year liquidation was set at 1 year of spending. The investment return threshold and maximum liquidation variables will be optimized in subsequent sections.

In the plots below, each gray line represents the simulation for a single start year. The red line is the mean value across all the simulated start years; the blue line is the median value across all the simulated start years. These plots can be confusing at first glance. If this type of plot is unfamiliar to you, it might help to just look at the best case at the end, the worst-case at the end, and the average; be sure to pay close attention to the vertical axis scale which changes from one plot to the next.

Cash Amount Comparison Plots

Next, we will compare the mean and median performance between the different cash amounts.

Now that we’ve compared the range of outcomes and average performance for each cash amount, let’s look at how the worst-case scenarios compare. For this simulation, the worst time to retire without a cash buffer was in 1929 (at the beginning of the Great Depression).

We’ll also look at how the best-case scenarios compare. For this simulation, the best time to retire without a cash buffer was in 1942.

Investment Return Threshold Optimization

In the section above, we saw that a cash amount of about 4 years of spending had the best performance in worst-case scenarios and didn’t significantly hinder performance in average or best-case scenarios. Using a 4 years of spending cash amount, we’ll now optimize the investment total return percentage (i.e. investment return threshold) which will dictate whether to spend cash vs investments and when to replenish cash. We’ll leave the maximum single year liquidation at 1 year of spending and optimize that variable in the next section.

In the plots below, each gray line represents the simulation for a single start year. The red line is the mean value across all the simulated start years; the blue line is the median value across all the simulated start years.

Investment Return Threshold Comparison Plots

Next, we will compare the average performance between the different investment return thresholds.

Now that we’ve compared the range of outcomes and average performance for each investment return threshold, let’s look at how the worst-case scenarios compare. For this simulation, the worst time to retire without a cash buffer was in 1929 (at the beginning of the Great Depression).

We’ll also look at how the best-case scenarios compare. For this simulation, the best time to retire without a cash buffer was in 1942.

Maximum Liquidation Optimization

In the section above, we saw that a cash amount of about 4 years of spending and a minimum investment return threshold of around 20% had the best performance in worst-case scenarios, and didn’t significantly hinder performance in average or best-case scenarios. Using a 4 years of spending amount and 20% minimum investment return threshold, we’ll now optimize the maximum single year liquidation. Like the cash amount, the liquidation cap will be quantified in years of spending. For example, in this simulation where we’re spending $100,000 per year, a 0.5 years of spending liqidation cap would equate to $50,000; a 1 year spending liquidation cap would equate to $100,000.

In the plots below, each gray line represents the simulation for a single start year. The red line is the mean value across all the simulated start years; the blue line is the median value across all the simulated start years.

Maximum Liquidation Comparison Plots

Next, we will compare the mean and median performance between the different maximum liquidation amounts.

Note: Values at and above 2 Years of Spending all produce the same line.

Note: Values at and above 2 Years of Spending all produce the same line.

Now that we’ve compared the range of outcomes and average performance for each maximum liquidation amount, let’s look at how the worst-case scenarios compare. For this simulation, the worst time to retire without a cash buffer was in 1929 (at the beginning of the Great Depression).

We’ll also look at how the best-case scenarios compare. For this simulation, the best time to retire wihtout a cash buffer was in 1942.

Optimal Cash Buffer vs No Cash Buffer

To recap, we’ll compare the performance of an optimal cash buffer (4 years of spending in cash, 20% minimum investment return threshold, 2 years of spending maximum single year investment liquidation) to no cash buffer.

Final Values

Optimal Cash Buffer vs No Cash Buffer Comparison Plots

Next, we will compare the mean and median performance between the optimal cash buffer and no cash buffer simulations.

Now that we’ve compared the range of outcomes and average performance for the optimal cash buffer and no cash buffer simulations, let’s look at how the worst-case scenarios compare. For this simulation, the worst time to retire was in 1929 (at the beginning of the Great Depression).

We’ll also look at how the best-case scenarios compare. For this simulation, the best time to retire without a cash buffer was in 1942.

Conclusions

The following conclusions should be interpreted within the confines of the simulation assumptions described in the Retirement / Spending Cash Buffer Simulation Details section.

• There are substantial advantages to maintaining a cash buffer in worst-case investment performance scenarios.

  • In the worst-case investment performance scenario, the difference was about $10.0M after a 30-year retirement. Specifically, if you start retirement / spending without a cash buffer in 1929, you run out of money (-$1.5M if you allow the simulation to keep running); if you start retirement / spending with an optimal cash buffer in 1929, you end up with $8.5M.

• There are modest advantages to maintaining a cash buffer in average investment performance scenarios.

  • The difference was about $1.6M after a 30-year retirement. Specifically, if you don’t use a cash buffer, on average you end up with $13.7M. If you do use a cash buffer, on average you end up with $15.3M.

• There are modest disadvantages to maintaining a cash buffer in best-case investment performance scenarios.

  • In the best-case investment performance scenario, the difference was about $1.8M after a 30-year retirement. Specifically, if you start retirement / spending without a cash buffer in 1942, you end up with $34.2M; if you start retirement / spending with an optimal cash buffer in 1942, you end up with $32.4M.

• A cash buffer is more effective at minimizing downside and maximizing upside when compared with conservative asset allocations (high bond holdings, for more on this see the Retirement Asset Allocation Simulation Study linked in the Overview section).

• A cash buffer will not significantly reduce the volatility associated with an aggressive asset allocation (high stock holdings).

• The optimal parameters for a cash buffer retirement spending approach include:

  • 4 years of spending cash target

  • 20% minimum investment return threshold for deciding when to take withdrawals from cash vs investments and when to replenish cash

  • 2 years of spending maximum single year investment liquidation

• Everything else held constant, the year you retire has a huge influence on your retirement savings.

  • In the optimal cash buffer scenario, the difference was $26.8M after a 30-year retirement. Specifically, if you start retirement / spending in 1964, you end up with $5.6M; if you start retirement / spending in 1942, you end up with $32.4M.

  • In the no cash buffer scenario, the difference was $34.2M after a 30-year retirement. Specifically, if you start retirement / spending in 1929, you run out of money; if you start retirement / spending in 1942, you end up with $34.2M.

Implementation

Most people are used to running their finances with a pay day every two weeks, twice per month, or once per month. Since we only have one data point for each year in our S&P500 Total Return and Consumer Price Index (CPI) (used to make inflation adjustments to dollar values), we unfortunately can’t simulate retirement withdrawals at a frequency similar to common pay day intervals. Below is an outline of how you might implement the cash buffer system with annual withdrawals during retirement; presumably the same approach would work with more frequent withdrawals, but we don’t have the data granularity required to prove that.

• 1 - Enter retirement with a cash buffer held inside a tax advantaged account (401k, 403b, IRA, etc.) to avoid tax implications. In the Pre-Retirement Cash Buffer Prep Simulation Study linked below, we investigated the optimal way to build this cash buffer before retiring. In short, the differences between approaches we could come up with were trivial.

Pre-Retirement Cash Buffer Prep Simulation Study

• 2 - Near the end of each calendar year, check the total return percentage for your investments (dividends plus price change).

• 3 - Use your total return value from step 2 to decide whether to spend your cash buffer or investments for the following year.

  • 3.1 - If the total return is above your investment return threshold, proceed with steps 3.1.1 and 3.1.2, otherwise jump to 3.2.

    • 3.1.1 - Liquidate a year’s worth of spending from your investments and deposit the proceeds into your bank account.

    • 3.1.2 - If your cash buffer is below the target amount, liquidate up to two years’ worth of spending into your cash buffer.

  • 3.2 - If the total return is below your investment return threshold, proceed with steps 3.2.1, 3.2.2, or 3.2.3.

    • 3.2.1 - If there is cash in your cash buffer, move one year’s worth of spending into your bank account.

    • 3.2.2 - If there isn’t any cash in your cash buffer, liquidate a year’s worth of spending from your investments and deposit the proceeds into your bank account. Note: this is very unlikely to occur if you’re following the plan designed by this analysis.

    • 3.2.3 - If there is some cash in your cash buffer, but not enough for an entire year, move the cash you have into your bank account and liquidate whatever additional assets are required to satisfy a year’s worth of spending. Note: this is also very unlikely to occur if you’re following the plan designed by this analysis.