My friend Jay sent me an article years ago that argued the best withdrawal strategy for retiree was “Bonds first.” You set a constant dollar withdrawal amount for spending and an initial mix of stocks and bonds. You sell bonds for your spending until you’ve depleted them; that then leaves you with a 100% stock portfolio that you’ll then sell for your spending. This differs from the normal or “Usual” strategy that one can evaluate in a Retirement Withdrawal Calculator (RWC) like FIRECalc: all RWCs assume that you rebalance back to your target mix of stocks vs. bonds at the start of a year. The purpose of this post is to examine “Bonds first” to see if that tactic gives the same safety – the same number of years of full withdrawals from a portfolio in the worst-case sequence of returns – as the “Usual” process of rebalancing back to your design mix of stocks and bonds. Conclusion: it does.

== The attraction of Bonds first ==

“Bonds first” recognizes that stocks generally outperform bonds. By selling bonds first from your portfolio, you’re allowing stocks to compound in value. When you start selling stocks only, you’ll have greater portfolio value than if you were selling them year after year. For example, stock returns have far outdistanced bonds for the three years of the six year since the start of our plan at the first of 2015. The math that gets me back to my design mix of stocks vs. bonds for the start of the next year told me to solely sell stocks for our spending. And then it told me to sell __more__ stocks to buy bonds to get to my design mix. That means I’ve foregone stock returns and have a lower total portfolio value now than if I had previously sold “Bonds first”.

The disadvantage of “Bonds first” is that it is very hard to get in your head that you should ever be 100% in stocks when you are retired and withdrawing from your portfolio for your spending. I find it IMPOSSIBLE. I would go crazy if I were 100% in stocks no matter how much I had. I hate the thought of having no bonds to sell in a year when stocks crater. It’s worse than that: I can’t imagine how I would feel if I was 100% stocks and withdrawing, say, 5% from my portfolio each year and saw my portfolio decline to LESS THAN HALF its value in real spending power in 24 months. That happened in 1973-1974! That happened in 36 months in 2000, 2001, and 2002!

== We use an RWC to judge ==

We use an RWC to judge the safety of our decisions for our financial retirement plan. An RWC shows how a portfolio fares over time for sequences of returns that the RWC constructs. For a set of inputs, an RWC tells you the number of years you can count on for a full withdrawal for your spending. The fewest number of years will be result of the worst case or Most-Horrible sequence of returns that the RWC constructs.

The actual Most-Horrible sequence of returns since the 1870s starts in 1969. I described that sequence here and here. It’s really Most Horrible. The worst seven-year return period for stocks starts in 1969. The worst seven-year return period for bonds starts a few years later. The poor returns in that sequence, coupled with your on-going withdrawals, shrink a portfolio to its tipping point, less than half its initial value. It spirals to depletion even when returns turn to be well above average.

The “Usual” RWC uses a fixed mix of stocks and bonds. The RWC rebalances a portfolio back to its design mix at the start of every year. You can’t use the usual RWC to examine a different approach like “Bonds first”. The only way to do that is to build a more detailed spreadsheet that allows you to pick what you want to sell for your spending each year and start the next year with a revised mix.

I built a basic spreadsheet I described and link to in this post. I altered it for this post to track stocks and bonds separately. I can input “Bonds first” and calculate portfolio value year-by-year.

== The test ==

I want to compare “Bonds first” to “Usual”. Does “Bonds first” match “Usual” in the number of years of full withdrawals for spending?

I input the following into my spreadsheets: initial $1,000 portfolio value; mix of 80% stocks and 20% bonds; total investing cost of 0.10%; and an annual withdrawal amount for spending of $44 in constant spending power. All returns are inflation-adjusted.

== The two options ==

Option 1: This is the base case, the “Usual” way all RWC’s work. I assume a portfolio is rebalanced to its design mix after the annual $44 withdrawal for spending for the upcoming year. You see that spreadsheet here.

Option 2: I use the expanded spreadsheet, and I withdraw “Bonds first.” The first $44 withdrawal comes from the starting balance of $200 of bonds (20% of the $1,000). In the example, I consume the bonds in about 3¼ years. You see that spreadsheet here.

== The comparison ==

I display how many years a portfolio provides a full withdrawal for spending for the two options:

“Bonds first” for all practical purposes has the same safety as “Usual”. The portfolios have almost the same amount at the end of the 20th years, basically just enough for a full withdrawal for the 21st year, but they don’t have enough for the 22nd year.

I don’t display the results from other inputs to the spreadsheet, but I’d reach the same conclusions over a very wide range of spending rates, mix of stock vs. bonds, and investing cost. I find no case where “Bonds first” is significantly different in terms of safety than the “Usual”.

**Conclusion**: This post examined if you should consider a tactic of “Bonds first” as an approach to deciding what to sell from your portfolio for your annual spending. Is that approach as safe as the “Usual” approach of rebalancing back to your design mix every year? I judged the effect on a portfolio using the Most-Horrible sequence of returns for stocks and bonds in history. I find found that “Bonds first” and “Usual” result in the same number of years of full withdrawals from a portfolio in the worst-case sequence of returns.