How to secure your retirement with 2 ETFs?

Most of us invest in order to be retire comfortably.

As an investor, my personal strategy is to pick specific stocks that pays me regularly. Once this payout reaches a point where it can sustain my lifestyle, I can officially retire. I share more about it here, but that’s not the focus of today’s article.

I like picking stocks and managing my investment portfolio. But not everyone wants to actively invest during retirement for two main reasons:

• Risk of losing your retirement monies
• You want to enjoy your life and spend time on other more important things

Hence today, we’re going to explore the feasibility of a simple, no-brainer retirement portfolio consisting just 2 ETFs.

Why two ETFs?

ETFs are exchange traded funds that tracks an underlying index passively. With ETFs, investors do not need to pick individual stocks (which means you don’t need to lose hair or sleep over “which stock to buy?”, “can buy now?” type of questions).

Ideally, ETFs will provide you with market index returns.

In this article, I examine the feasibility of a retirement portfolio consisting of 50% of global stock ETF (represented by Vanguard Total World Stock Index Fund ETF or “VT”) and 50% of global bonds ETF (represented by Vanguard Total World Bond ETF or “BNDW”).

The reason we chose these ETFs was that they are both diversified globally and do not have a bias towards growth or value. The two ETFs are also relatively cheap with expense ratios much lower compared to other ETFs (less than 0.1%).

How much do you need to sustain your lifestyle during retirement?

As with any retirement-type thought experiment, it’s important to know your goals.

In this exercise, I’m using the frequently recommended safe rate of withdrawal – 4%. Assuming you retire at age 55 with $1,000,000, that works out to be $40,000 per year.

And assuming an average lifespan of 85, this exercise will attend to cover at least 40 years of retirement. (This works out to a grand total of $1,600,000.)

Good news, thanks to the ability to write programming scripts and reuse the computer programs written by other people, we no longer need to wait for 40 years to figure out how feasible our retirement plans are.

One way to assess the feasibility of a retirement plan is to use a computer to randomly generate returns of a portfolio over 1,000 lifetimes and see whether in these alternate universes, how many retirement plans succeed, and how many would fail.

Can you fund your retirement with a 2-ETF retirement portfolio?

I ran several simulations using Python, and I think it’s looking rather good.

After running the simulations 1,000 times with randomly generated returns and inflation numbers over 40 years, with the assumption that $40,000 is withdrawn annually (inflation adjusted), this is how it looks:

The graph above simply shows that of a 1000 such portfolios, 921 would successfully retire (92.1%).

So, can you secure your retirement with 2 ETFs?

In short, yes. You have a 92.1% of success.

The better news, the above was ran on the assumption that you’re looking to rely solely on your portfolio to retire. Singaporeans have access to schemes like the CPF Life that could help increase your odds of success.

With the CPF Life Escalating life plan, we can only start withdrawals at 65. Assuming that CPF Life produces an inflation-adjusted $5,000 a year, this reduces the timeline of the simulation to 30 years and expenses to $35,000 a year. I redid the simulations and got this:

Great news, the success rate improves to 98.2%.

Generally speaking, it can be extremely challenging to reduce the failure probability to zero. The retiree is better off finding better ways to increase the lifetime payouts of CPF Life or to prepare to have two years of living expenses to fend off the probability of a recession happening within two years of retirement.

How did I simulate this thought experiment?

For the curious, here’re more details of how I ran my thought experiment.

If this does not interest you, skip to directly to the conclusion below for my thoughts on how you can use this information!

1. Obtain the statistical properties of the portfolio

The first step is to assess the statistical properties of the portfolio. I extracted price data from Yahoo Finance to calculate the four statistical properties of the blended portfolio consisting of these two ETFs from 2008 to the present day. This timing captures two recessions and is appropriate for this current period as world economies slowly recover from the COVID-19 pandemic.

• Return – This is how much on average you can earn if you invest the portfolio. The higher the return, the more profitable the portfolio is.
• Standard Deviation – This measures how volatile the portfolio is. The higher standard deviation, the risker the portfolio is.
• Skew – This measures how much returns are biased. A positive number means that there is a larger number of returns lie above the mean. A negative number means that a larger number of annual returns lie below the mean.
• Kurtosis – This measures how fat-tailed the returns can be. A positive kurtosis means that a large number of returns are extreme. A negative kurtosis means that more returns lie closer to the mean.

A portfolio that is equally-weighted between VT and BNDW has the following characteristics:

Over the past 12 years, the portfolio had returns close to 10% with a volatility of about 18.75%.

You may want to note that this portfolio may tend to generate devastating losses during difficult times due to the negative skew and high kurtosis.

2. Model your expenditure during retirement

In this scenario, we will assume a starting net worth of $1,000,000 at age 55.

As mentioned above, in our example, the retiree is assumed to have an ordinary lifespan of 85 and will need $40,000 a year, but this amount has to be moderated by inflation.

To err on the safe side, this simulation should cover a longer duration of about 40 years.

The exciting thing about inflation is that it also exhibits non-normal behaviour. By extracting government data, inflation averages about 2% a year but it has a standard deviation of 2.21%, negatively skewed at -1.21% and has a kurtosis of 1.88. The retiree has to expect spikes of expenses in years of high inflation or even some deflation during his golden years.

3. Simulate your retirement

Computer programs like Python have the statmodels programming library that can generate a random signal that is similar to the portfolio built provided that you can tell the program what return, standard deviation, skew and kurtosis is. We can also do the same to project the inflation experienced by the retiree.

All that remains is to run the simulation 1,000 times with randomly generated returns and inflation numbers and see how different retirement portfolios perform.

We look at the first scenario where we run this over 40 years and withdraw $40,000 every year adjusted by inflation. This is the same graph shown above:

The program says that, out of 1,000 universes, 92.1% of the scenarios succeeded but 7.9% of the time, the retiree ends up with nothing before the forty years is up. 92% success rate is not too bad for a retirement plan.

4. Retirement beyond your investment portfolio

Beyond the investment portfolio consisting of ETFs, I also considered the use of CPF Life to improve the odds of success.

In this case, the retire notes that he can postpone retirement by 10 years by working until he is 65. After 65, we can start withdrawing from CPF Life Escalating life plan. As mentioned previously, I assumed that CPF life produces an inflation-adjusted $5,000 a year, reducing the timeline of the simulation to 30 years and expenses to $35,000 a year.

Running the same simulation over 30 years and reducing the withdrawal rate to 3.5% of $1,000,000, we get the following. This is the same graph shown above:

The number of successful outcomes improves to 98.2%.

Conclusion – what can we learn from this simulation?

Generally speaking, it can be extremely challenging to reduce the failure probability to zero.

The retiree is better off finding better ways to increase the lifetime payouts of CPF Life or to prepare to have two years of living expenses to fend off the probability of a recession happening within two years of retirement.

There are several essential takeaways from this exercise:

• A simple two ETF portfolio that covers global stocks and bonds can form a decent basis for a retirement portfolio.
• Fat-tails in investment returns are a constant bane to retiree portfolios and the main reason why 100% fool-proof retirement portfolios do not exist. 
• If you wish to retire with more than 95% chance of success successfully, your inflation-adjusted expenses should be well-within 3.5% of starting portfolio size to have a decent chance of surviving the next 30 years.
• For a decent and comfortable retirement, Singaporeans should learn to manage their CPF Life well and max out monthly annuity payouts after age 65.

A last bit of good news: a retirement web app for Early Retirement Masterclass alumni to simulate retirement portfolios is currently being built with most of the program logic already finished. I expect this tool to be launched before January 2021.