Back in Action

I’m back with another edition of the Pension Couch. I produce Pension Couch articles from edited and sanitized exchanges with readers who ask me defined benefit (DB) pension questions. It’s a way for me to create posts with useful pension-related information without the additional work required to write one from scratch. In this edition, I answer a reader’s “what if” question about replacing lost pension income by taking a higher-paying non-pensionable job. As a question, it fits well with this blog’s stay-or-go Golden Albatross theme. Therefore, I believe it’s worth your time.

This article’s request came from a reader who I called Kai. He specifically asked how much he’d need to save and invest at a new non-pensionable job to replace lost annual pension income from his current pensionable job… if he decided to leave six years earlier than planned. On the face of it, that’s a straightforward question. The answer, however, required modeling his retirement savings and investment options and then determining if they could replace the potential lost pension income.

Readers ask me some form of the “replacing pension income” question a lot, which tells me two things. First, many readers have contemplated leaving their often lower-salaried pensionable jobs for higher salaried non-pensionable jobs. Second, many readers also understand these scenarios involve trade-offs connected to their pension’s ultimate value in retirement. But, as just mentioned, mathematically modeling these “what if” questions can be complicated. Fortunately, in this article, I demonstrate how to determine if replacing pension income is feasible without resorting to complex math formulas. Instead, I use a free website and free retirement planning software, which you can easily replicate, should you need to answer the same question.

Kai’s Circumstances

The following is a heavily edited summary of Kai’s circumstances and question. Gathering this data required several emails. So, for those thinking about contacting me, take note of the type of information I need when making your request.

• • Age: 44
  • Employment: State employee in the US
  • Tenure: 20 years of service (YOS)
  • Ideal retirement age: 50
  • Pension start age: 50
  • Annual pension contributions:
    • 6% of salary
    • Approximately $6,900 per year
  • Pension design issues:
    • Cost of living adjustment (COLA):
      • Requires state legislature approval
      • Typically no more than .5% to 1%, if approved at all
  • Pension options:
    • Option A: Leave current pensionable job for non-pensionable job –> work 6 more years –> retire at age 50 and collect $18,746 per year (p/y)
    • Option B: Stay at pensionable job –> work 6 more years –> retire at age 50 and collect $43,534 p/y
      • Annual difference: +$24,788 p/y for staying at the pensionable job
  • Question/Request: 
    • “If I leave now, how much would I have to save each month for the next six years to replace the difference I am losing? I know there are some unknown variables like how long am I going to live and what will be the interest rate on this savings. The only thing I know to do is run a calculator that says how much I need upfront to withdraw $25,000 for 40 years. When I do that, I come up with roughly $400,000.

My Standard Disclaimer and More

“Kai, let me provide my standard disclaimer up front for anyone who writes to me looking for advice. I’m not a financial professional. Yes, I’ve studied pension for years, both as a hobby and academically. However, whatever insight I can provide is for informational purposes only, and what you take and do with that information is up to you. Assuming I can help, I’d encourage you to obtain a second opinion before making any financial decisions. 

With that said, from my perspective, you appear on the correct path with your calculator approach. It’s a pragmatic way to solve a potentially complex math problem that at least puts you in the ballpark. Good job!

However, we could do a few things to provide a more realistic look at how much money you might need to save to replicate those extra pension payments. I would specifically recommend modeling it, including simulated stock market volatility, instead of using flat annual rates of return. That assumes you plan to invest the money rather than find an interest-bearing vehicle like a bond.”

Modeling Assumptions

Each of the examples I modeled for Kai included the following assumptions based on the information he provided me:

• • Kai’s retirement lifespan: Age 50 to 90
  • Pension start age: 50
  • Income tax rate: 22%
  • Long-term capital gains (LTCG) tax rate: 15%
  • Annual projected amount that investments need to produce: $25,000
  • Was Kai willing to spend his accumulated savings to zero to generate $25K annually? Yes

The modeled scenarios also include these same baseline assumptions when applicable:

• • Average annual inflation rate: 3% with a +/- 3% volatility potential
    • 3% is the approximate historical inflation avg. for US inflation
  • Average annual stock market returns: 10.5% with +/- 20% volatility potential
    • 10.5% is the average pre-inflation adjusted return for the S&P 500

Minimum IRA/401K penalty-free withdrawal age: 60

Example 1

My attempts to model Kai’s situation got complicated quickly, so let’s start with the basic scenario first. For example 1, I ran a simple 40-year scenario using this online calculator. I withdrew $25K annually using a spend-to-zero trajectory from a lump sum investment as it earned a straight rate of return. Open the link, or refer to screenshot 1 (above) to follow along as they display the numbers I used.

• • The 10.5% interest rate/rate of return reflects the average historical S&P 500 returns (pre-inflation).
    • I assumed he’d invest this lump sum of amassed post-tax cash in its entirety in a taxable investment account using an S&P500 index fund.
  • I set the $25K annual withdrawals to occur at the beginning of the year. So, he’d start each year of retirement with a $25K withdrawal from his accumulated nest egg.
    • If I set the withdrawal to the end of each year, there was a significant difference. You can change those settings with the “PMT made at the” radial button.
  • Finally, I entered potential lump-sum totals that allowed a spend-to-zero scenario to work over 40 payment periods (years in this case). I found that a lump sum of $258,300 invested on day one of retirement will last for 40 years using all the above assumptions.
    • If you divide that $258,300 by the 72 months separating Kai from retirement at 50, it totals $3,587.50 a month. That’s how much he’d need to save each month to make this scenario work.
  • Importantly, this scenario does not reflect inflation, so this is a gross return. Since Kai’s pension plan states a COLA is discretionary, I assumed there wasn’t one. This scenario also doesn’t account for investing and growing his monthly savings to $258,300 during those six years (or 72 months) before retirement.
    • If he wanted, Kai could turn this into an inflation-adjusted scenario by using a 7.5% rate of return to reflect the US’s average historical inflation rate of ~3%.
  • Pros for example 1:
    • It’s simple and math-based.
    • The goal appears easily obtainable.
  • Cons for example 1:
    • It assumes straight investment returns with no stock market volatility, which is unrealistic.
    • Kai would need to find a 10.5% interest rate bond to replicate this scenario in real life, but this wouldn’t be possible without courting many risks. We’re essentially talking junk bond status.
      • An immediate insurance annuity that pays $25K annually might be an alternative. Still, I’d be surprised if he found one that did so for only $258,300. I told Kai he might want to investigate this option.
  • Bottom line: While useful as a lower bound for answering how much money Kai would need to save, this method is more theoretical than practical.

Retirement Calculators and Modeling

Next, I turned to my preferred retirement planning tool, Flexible Retirement Planner (FRP). While there are numerous retirement planning software tools a person can use, I use FRP because it allows me to model different “what if?” scenarios. It’s also free. Notably, FRP uses Monte Carlo scenarios to simulate the randomness (i.e., volatility) within stock market returns from year to year. IF interested, the most comprehensive coverage I’ve read on the pros and cons of different retirement planning tools/calculators is over at the Can I Retire Yet? website. Check out their articles here to determine which one best suits your needs.

In the meantime, I assumed Kai didn’t have FRP familiarity, so I attached screenshots of my models to my email so he could follow along. I embedded those same screenshots at the beginning of each example below.

Example 2

Screenshot 2 (below) shows my simulation; use it to follow along.

For example 2, I modeled the impact of volatility, taxes, and inflation on saving and investing in a taxable investment account in the six-year run-up to your retirement. Kai started saving at age 44 and grew a nest egg big enough to allow $25K annual withdrawals at age 50. He then withdrew $25K annually from that account after he retired. The results were radically different from example 1.

• • Results:
    • With the assumptions set, I plugged numbers into the “Taxable Annual Savings” box to determine which amount produced a spend to zero chart like the one on the bottom half of screenshot 2.
    • $75,100 saved and invested annually ($6258 p/m) between 44 and 50 produced the effect.
    • The median portfolio value by age 50 was $513,399. This amassed total is a product of Kai’s modeled savings and investment returns from age 44 through 49.
    • Since this was a spend-to-zero scenario, withdrawal rates (i.e., the percentage of Kai’s portfolio that an annual $25K withdrawal represented) ranged from 4.9% to 87.1%.
    • Saving $75.1 K a year produced a 50/50 chance that Kai’s amassed nest egg at age 50 would last long enough to sustain $25K annual withdrawals to 90. That’s an extremely risky proposition for retirement planning purposes.
    • The squiggly black line that runs across the chart’s purple bars and spikes at age 64 represents the chance of Kai running out of money by that year of retirement.
      • Essentially, this is the worst 10% of the random outcomes produced by the Monte Carlo simulation.
      • In other words, if the sequence of return risk (SRR) doesn’t work in Kai’s favor, and those 50/50 odds break against him at almost every turn, the earliest age at which he might exhaust funds is 64.
      • That isn’t good if Kai lives to 90.
    • Pros :
      • This is a far more realistic spend-to-zero model than example 1 because it injects the uncertainty of investment volatility.
      • It also depicts the headwinds that taxes and inflation create.
      • Thus, the risk is better quantified.
    • Cons:
      • The model doesn’t take full advantage of all tax-saving methods.
      • The 50/50 risk involved in the model is worrying.
    • Bottom line: A risky but valuable model. Ideally, Kai would want to save more to buy down that 50/50 risk to something more reasonable.

Example 3

Screenshot 3 (below) shows my simulation; use it and follow along.

The only thing I changed for Example 3 was to buy down the risk for Kai by saving more. I wanted to find an annual savings amount that got Kai to a statistically safe probability of success (i.e., 90% or more). This was a theoretical exercise more than anything else.

• • Results:
    • Saving $156,000 annually from age 44 to 49 produced a 90% probability of success.
    • Withdrawal rates as a percentage of portfolio value only varied from 0.8% to 2.4%, meaning it’s no longer a spend-to-zero scenario. Rather, it’s a highly conservative safe withdrawal rate (SWR) scenario.
      • The median projected ending portfolio value was over $3 million, well above the ~$1 million portfolio value that the model started with at age 50.
    • As a result, Kai could essentially ignore the squiggly black line in this scenario because even the worst 10% of the model’s outcomes  (i.e., those calculations which modeled the worst potential sequence of returns risk (SRR)) never produced an average yearly portfolio value of $0 before age 90.
  • Pros:
    • Risk-free scenario
  • Cons:
    • Too conservative
  • Bottom line: Interesting as a mental anchoring point, but probably unrealistic for Kai’s practical use since he was not looking to bequeath this money to someone else.

Example 4

Screenshot 4 (below) shows my simulation; use it and follow along.

Example 4 mirrors example 2 but splits Kai’s savings in the six years preceding retirement. I split savings between maxing out Kai and his spouse’s 401K at a combined $41K annually and placing the remainder of the savings in a taxable investment account. Once again, I shot for a spend-to-zero scenario. Interestingly, by doing so, it raised the required combined annual savings amount compared to example 2.

• • Results:
    • This model required a combined total annual savings of $76,200 ($6350 p/m) to meet the spend-to-zero requirement.
      • $41,000 annually into Kai and his spouse’s combined 401Ks
      • $35,200 annually into a taxable investment account
    • The projected portfolio value at retirement was $530,653.
    • This model does not account for any matching funds employer(s) might provide for 401K contributions.
    • Like example 2, example 4 carries a 50/50 chance of running out of money before age 90, with the unluckiest 10% of calculations (i.e., the black squiggly line) showing exhaustion of funds as early as age 64.
      • In other words, this is still an extremely risky proposition for retirement planning purposes.
    • Withdrawal rates as a percentage of portfolio value range from 4.6% to 47.2% of portfolio value, as expected in a spend-to-zero scenario.
  • Pros:
    • Once again, this example models the risk caused by investment return and inflation volatility and the impact of taxes more realistically.
    • Since I can’t model employer match, this example may be cheaper in real life if Kai’s employer(s) provides matching 401K funds.
  • Cons:
    • More expensive than example 2, probably due to the 22% tax bracket for Kai’s 401K withdrawals
    • Still risky
  • Bottom line: More accurate from a tax perspective, but slightly more expensive than example 2 if Kai and his spouse don’t get employer matching funds for the 401Ks. If Kai and his spouse get matching funds, this would probably be slightly less expensive than example 2.

Example 5

Screenshot 5 (below) shows my simulation; use it and follow along.

For giggles, I modeled the same scenario as example 4 but used $41K in Roth 401K contributions instead. I didn’t know if Kai and his spouse had access to Roth 401Ks. Still, I was curious about the differences between saving in pre-tax and post-tax retirement vehicles. Surprisingly, this proved the cheapest of all the FRP modeled examples.

• • Results:
    • This model required a combined total annual savings of $68,500 ($5,708 p/m) to meet the spend-to-zero requirement.
      • $41,000 annually into combined Roth 401Ks
      • $27,500 annually into a taxable investment account
    • Projected portfolio value at retirement $475,705
    • Like example 2, example 5 carries a 50/50 chance of running out of money before age 90, with the unluckiest 10% of calculations (i.e., the black squiggly line) showing exhaustion of funds as early as age 62.
      • This is the riskiest proposition of all the FRP simulations! Probably because it has the smallest portfolio value at retirement.
      • It relies more on investment returns to sustain those annual $25K withdrawals, further exposing it to SRR.
    • Withdrawal rates as a percentage of portfolio value ranged from 5.3% to 58.5% of portfolio value, as expected in a spend-to-zero scenario.
  • Pros:
    • Cheapest of the FRP modeled examples
  • Cons:
    • Riskiest of the FRP modeled examples
  • Bottom line: Still risky when compared to modeled volatility. However, tax-free growth inside a Roth vehicle makes a difference!

GM’s Assessment

Here’s what I wrote to Kai after I ran all the scenarios:

“Although Example 1 is the cheapest, it’s not practical unless you find a bond with a 10.5% fixed interest rate. Unless you have access to two Roth 401Ks, then Example 5 (the second cheapest) probably isn’t workable either. Although, whatever your solution, it does make an argument for maxing out your Roth vehicles first. Example 5 also courts a severe sequence of returns risk (SRR), as do all the FRP modeled spend-to-zero scenarios (i.e., examples 2, 4, and 5). That leaves examples 2 or 4 as the most workable, the decider for which should probably be whether or not there’s a 401K match from your employers.

Either way, unless you’re OK with a 50/50 chance of running out of money before age 90, you will need to save more than those models show. I’m not suggesting you need to save anything near what example 3 modeled. However, from a planning perspective, saving northwards of roughly $76.2K annually ($6350 monthly) seems necessary for any of this to work. That’s a big step up from the 6% ($6900) taken from your paycheck annually for your pension!

The only alternative I could think of would be pricing an insurance annuity to start at age 50 that paid out $25K annually. Given the opaque insurance annuity markets, I can’t tell you what price you’d pay. You’d also need to see what inflation would do to that since it’s my understanding most insurance annuities don’t account for inflation. Much of the academic literature I read found that insurance annuities are not nearly as efficient as large pension funds’ annuities. In other words, your dollars stretch a lot further through your pension fund than purchasing an insurance annuity as an individual. I mentioned this in Pension Series Part 25 (Pension Design).  

I’ll turn it over to you to process all of this at this point, and hit me up with any follow-up questions you might have.”

Kai’s Initial Response

“Hi, Grumpus. After taking the time to review, here is my takeaway from your last email.

1. 1. Example 1 requires $43,000 saved a year. This scenario only works if I were to be fortunate and get a consistent 10.5% return and leaves no room for error, such as bad returns in the first couple of years of retirement. This is the extreme of the aggressive side.
   2. Example 2 requires $75,000 saved a year. More risk as it is only a 50% chance of success.
   3. Example 3 requires $156,000 saved a year. This is virtually a lock that, with this much saved, I shouldn’t run out of money after 40 years. It is the extreme end of the conservative side.
   4. Example 4 requires $76,000 a year, and it is similar to example 2 but splits the savings between pre-tax retirement and a brokerage account. This also has close to a 50% chance of success.
   5. Example 5 requires $68,500 a year. It’s like example 4, except all Roth 401k up to the max and then the rest in brokerage. This also has close to a 50% chance of success.

So the overall takeaway is that I could make $43k more in another job, and it “could” even out, but it is highly unlikely. If I save around $70k, I have a 50% chance, and it’s closer to 100% if I save $156k a year. I think this makes my decision pretty easy, at least for the near future, that I should stay [at my pensionable job].

If my job ever gets to where I dislike it, it will be about more than money. However, if I stay, my next big decision will be at 50. At that point, I could leave and take the $43k or stay for four years more and get $63k, which is my full pension. That will be a decision I will have to wait and make then.

Again, I thank you so much for this valuable information. I left you a donation [https://www.buymeacoffee.com/GoldenAlbatross] which isn’t enough to pay you what you’re worth, but I wanted to give you something. Please let me know if there is anything I can do for you.”

My Follow-Up: The Importance of Modeling Volatility

“Kai, I agree that your situation seems reasonably clear from the money/math perspective. Staying till 50 provides the highest monetary reward with the lowest amount of risk. That assumes your pension plan is well-funded, which we didn’t discuss. However, that should be easy enough for you to research and incorporate into your decision-making.

Considering how significant modeling investment return volatility proved in the differences between example 1 and examples 2-5 reminded me of my DIY COLA article. Specifically, it reminded me of an exchange in the comments in which a reader questioned the figure Big ERN came up with for the DIY COLA. However, the commenter used a straight return calculation ala example 1 [above]. Once I took the commenter’s numbers and pumped them through the volatility machine that is FRP, I got something much more akin to Big ERN’s original calculations.

My lesson from that episode was that while straight return calculations serve a purpose, they can often mislead. I’m not saying that a financial scenario modeled using Monte Carlo simulation doesn’t have drawbacks. Still, it represents the real world better than an assumed straight return scenario.”

Retirement Spending Need vs. Pension Amounts

“Switching gears, I hope circumstances don’t dictate that you consider leaving earlier than 50. As you stated, if they do, it will be about more than just the size of the pension at that point, which is a great way to think about it. After all, you can’t put a price tag on everything. However, I hope that my analysis provides you with the insight needed to engage in the last six years at your job with a clear goal in mind. It’s much better than the alternative of grinding it out with no clear end, which, speaking from experience, can be a drag.

Regarding your next decision at 50, I would emphasize determining how much money you need for annual spending in retirement first. Then decide how much you want your pension versus withdrawals from investments to cover. Your Gap Number is the uncovered spending gap between your fixed income and annual projected retirement spending. Ultimately, the size of the pension is only important when compared to one’s needs and desires. A pensioner’s Gap Number tends to reflect those needs because it’s derived from a retirement spending plan, which reflects what a pensioner perceives they will value in retirement.

Finally, a big “thank you” for the donation. To date, that’s the largest one-off donation a reader has made, and it will help keep the website’s lights on!”

Kai Gets the Final Word

“As far as deciding on whether to leave before 50, at this point it is not too hard of a decision. There are things I like and don’t like about my job, but overall I like it enough to stay. Hopefully that will remain the case for the next 6 years. Thank you again for all the great info and let me know if I can do anything for you.”

Like what you read? Here are three ways you can support the blog: 

1) Buy the website a beer on its Buy Me a Coffee page. All beers “bought” go straight to keeping the site running ad-free.

2) Purchase The Golden Albatross: How to Determine If Your Pension Is Worth It from ChooseFI publishing. All royalties from book sales also go directly towards keeping the site up and running.

3) Re-post links to my articles on your social media feeds, and message your favorite personal finance podcast or radio show to suggest they cover my content.