## Tuesday, October 20, 2020

### Variable Percentage Withdrawal: Garbage In, Garbage Out

The concept of Variable Percentage Withdrawal (VPW) for retirement spending is simple enough: you look up your age in a table that shows what percentage of your portfolio you can spend during the year.  The tricky part is calculating the percentages in the table.  Fortunately, a group of Bogleheads did the work for us.  Unfortunately, the assumptions built into their calculations make little sense.

If we knew our future portfolio returns and knew how long we’ll live, then calculating portfolio withdrawals would be as simple as calculating mortgage payments.  For example, if your returns will beat inflation by exactly 3% each year, and your \$500,000 portfolio has to last 40 more years, the PMT function in a spreadsheet tells us that you can spend \$21,000 per year (rising with inflation).

Instead of expressing the withdrawals in dollars, we could say to withdraw 4.2% of the portfolio in the first year.  If the remaining \$479,000 in your portfolio really does earn 3% above inflation in the first year, then the next year’s inflation-adjusted \$21,000 withdrawal would be 4.26% of your portfolio.  Working this way, we can build a table of withdrawal percentages each year.

Of course, market returns aren’t predictable.  Inevitably, your return will be something other than 3% above inflation.  You’ll have to decide whether to stick to the inflation-adjusted \$21,000 or use the withdrawal percentages.  If you choose the percentages, then you have to be prepared for the possibility of having to cut spending.  If markets crash during your first year of retirement, and your portfolio drops 25%, your second year of spending will be only \$15,300 (plus inflation), a painful cut.

A big advantage of using the percentages is that you can’t fully deplete your portfolio early.  If instead you just blindly spend \$21,000 rising with inflation each year, disappointing market returns could cause you to run out of money early.

Choosing Withdrawal Percentages

One candidate for a set of retirement withdrawal percentages is the RRIF mandatory withdrawals.  These RRIF withdrawal percentages were designed to give payments that rise with inflation as long as your portfolio returns are 3% over inflation.

Unfortunately, the RRIF percentages would have a 65-year old spending only \$20,000 out of a \$500,000 portfolio.  Some retirees chafe at being forced to make RRIF withdrawals, but when it comes to the most we can safely spend in a year, most retirees want higher percentages.

A group of Bogleheads calculated portfolio withdrawal percentages for portfolios with different mixes of stocks and bonds.  Most people will just use the percentages they calculated, but they do provide a spreadsheet (with 16 tabs!) that shows how they came up with the percentages.

It turns out that they just assume a particular portfolio return and choose percentages that give annual retirement spending that rises exactly with inflation.  You may wonder why this takes such a large spreadsheet.  Most of the spreadsheet is for simulating their retirement plan using historical market returns.

The main assumptions behind the VPW tables are that you’ll live to 100, stocks will beat inflation by 5%, and bonds will beat inflation by 1.9%.  These figures are average global returns from 1900 to 2018 taken from the 2019 Credit Suisse Global Investment Returns Yearbook.

So, as long as future stock and bond returns match historical averages, you’d be fine following the VPW percentages.  Of course, about half the time, returns were below these averages.  So, if you could jump randomly into the past to start your retirement, the odds that you’d face spending cuts over time is high.

For anyone with the misfortune to jump back to 1966, portfolio spending would have dropped by half over the first 14 years of retirement.  More likely, this retiree wouldn’t have cut spending this much and would have seriously depleted the portfolio while markets were down.

The VPW percentages have no safety margin except for your presumed ability to spend far less if it becomes necessary.

Looking to the Future

But we don’t get to leap into the past to start our retirements.  We have to plan based on unknown future market returns.  How likely are returns in the next few decades to look like the average returns from the past?

30-year bonds in Canada pay less than 1.2%.  For them to beat inflation by 1.9%, we’d need to have 30 years of 0.7% deflation.  That’s not impossible, but I wouldn’t count on it.  It seems crazy to expect bonds to deliver 1.9% annual real returns in the coming decades.  Bond returns may get back to historical averages at some point, but retirees can’t expect much for some time.

Expecting 5% annual real returns from stocks may be sensible enough, as long as you have a high capacity for reducing retirement spending if it becomes necessary.  If your ability to reduce retirement spending is more limited, you need a safety margin in your assumed stock returns.  For my own retirement spending plans, I use inflation+4% for stocks (minus investment costs) and inflation+0% for bonds.

If we recalculate the VPW tables with these new assumptions, the annual withdrawal percentages drop by nearly a full percentage point.  This may not sound like much, but let’s look at an example of a 65-year old spending from a \$500,000 portfolio invested 50% in stocks and 50% bonds.  The “official” VPW tables would have this retiree spending \$25,000, but only \$21,000 by my calculations.  It’s not hard to see who most retirees would rather believe.

Pre-Retirees

This disagreement over reasonable assumptions makes a big difference for pre-retirees deciding how much money they need to retire.  For the 65-year old in the earlier example wanting to spend \$25,000 per year, the Bogleheads say to save \$500,000, and I say nearly \$600,000.

Clearly I could never make it as a financial advisor.  I’d be worried about protecting people from future spending cuts, and more “optimistic” advisors would scoop up all my clients by telling them what they want to hear.

Conclusion

Even when smart people develop good retirement spending tools, the results are only as good as the baked-in assumptions.  We can’t count on the high bond returns of the past, and it makes sense to have some safety margin in expected stock returns.  As with so many other calculators, if you input garbage assumptions, the results you get out will be garbage as well.

1. I'm not very good at predicting the future, but I assume Blackrock can do a better job than me. Their numbers are pretty close to yours. https://www.blackrock.com/institutions/en-us/insights/charts/capital-market-assumptions

1. Anonymous: Based on Blackrock's numbers, I may be a little optimistic in assuming bond returns will match inflation. However, for now instead of actual bonds I use EQ savings accounts paying 1.5% (nominal), which isn't too far off inflation.

2. When I change the inflation to a different %, the amount to be withdrawn does not change. The calculations seem to be based on the nominal return of 5% for stocks and 1.9% for bonds. I do agree the bonds return can be reduced to 1.5% for now but why would the stock return be reduced to 4%?

1. Hi Kevin,

I'm not sure where you're changing the inflation figure, but having the withdrawal percentages not change when you change the inflation assumption is a sign that the returns are real values (not nominal).

The Credit Suisse document is clear in stating that 5% for stocks and 1.9% for bonds are real (inflation-adjusted) figures.

If the 5% and 1.9% figures really were nominal, then VPW would be designed to target constant nominal withdrawals over the full retirement. This would be a plan for eating cat food later in life.

3. Hi Michael,

Inflation is on line 144 of the list tab. If I change the stock% to 4% and fixed to zero the I do get much lower withdrawal amount based on my original portfolio size. That means working many more years :( thank you for the post

1. Hi Kevin,

OK, I found the inflation figure at B144 of the Lists tab in the VPW Accumulation and Retirement Worksheet. It's only used to discount the value of a fixed (non-indexed) pension.

I'm sorry to be the bearer of bad tidings, but it's better to know now that you need to work longer, especially if you're like me and are very unlikely to find work in a decade or two that pays anything close to the salary I got while young. I'm glad I worked an extra couple of years instead of having to work for a decade or more when I'm older.

4. Thank you Michael. Do you use VPW using lower growth or do you use other withdrawal strategies? Is so do you have any article/spreadsheet I can look at.

1. Hi Kevin,

I do use VPW with lower growth assumptions, although nobody called it VPW back when I worked on it. However, I've been thinking about a version that assumes less ability to reduce spending, perhaps Bengen's 4% rule back-testing where you set a lower limit on how low your spending can go. I haven't yet figured out how to combine the idea of back-testing with more modern return assumptions (mainly that bonds return nothing over inflation).

2. Below are some relevant posts.

https://www.michaeljamesonmoney.com/2019/04/my-bucket-strategy-for-retirement.html

https://www.michaeljamesonmoney.com/2014/02/cushioned-retirement-investing.html

https://www.michaeljamesonmoney.com/2013/10/a-retirement-income-strategy.html

https://www.michaeljamesonmoney.com/2013/10/a-retirement-income-strategy-revisited.html

5. Hi Michael,
The reduced spending is what I have struggled as well. My math skills are not as great as yours so for now I am thinking of saving approx 5 years of cash flow vs VPW in case there is a 50% market crash. For simplicity if VPW allows for withdrawal of \$50K per year and \$35K as max spending in case of a crash, I would save 5X\$15K=\$75 in cash outside of the VPW portfolio. I know it may not mathematically make sense but I am not sure how else to do it.

1. Hi Kevin,

Despite the fact that my approach looks like a bucket strategy, I actually think of my portfolio as a single entity, cash and all. I use the 5 years of spending as a guide to how much to have in fixed income (bonds+GICs+cash). I prefer to lower my return expectations rather than have even more cash on the sidelines "just in case."

Some people like to think of their main portfolio (with some stock/bond mix) and have some just-in-case cash on the side. This certainly adds safety, but it also reduces the stock allocation. As I said, I prefer to just lower my future return expectations.

6. Excellent I will read the posts. Thank you.

7. The following exchange contained broken comment author links, so I replaced it with this single long comment to remove the links.

----- BHCh:

Always hard to develop a spreadsheet that would work for everyone.

The key question with VPW: can you reduce your withdrawals by X% if the portfolio drops? In many cases the answer will be “yes”.

People with small portfolios have DB income (eg state pension) dominating retirement income so taking less out of their investments isn’t as much of an impact as it seems.

People with large portfolios should be able to withstand cuts.

And those caught in the middle need a different approach. For example they could have a “buffer” cash pot which gets filled when the market outperforms growth assumptions and used when the market underperforms.

----- Michael James:

Hi BHCh,

The problem isn't the spreadsheet so much as the two numbers for the assumed real rate of return on stocks and bonds.

I agree that it's important to be able to reduce spending if portfolio returns disappoint. And yes, a base of CPP+OAS or the American equivalent helps with absorbing a portfolio withdrawal reduction. However, I don't see the sense in counting on bond returns that almost certainly can't happen.

People can differ on how much margin they need. I can even see some retirees choosing to use VPW with historical average stock returns. But it doesn't make sense to use bond returns that won't happen. We need to start with sensible return estimates and reduce them with what ever margin we feel is necessary. But this margin shouldn't be negative.

----- BHCh:

Don’t know if the real returns on bond funds are predictable. Eg What happens if inflation goes up but then BoC starts buying old bonds at inflated prices? Whatever number we assume, its always uncertain.

----- Michael James:

The future isn't predictable. But make up any bond scenario you like. The only way you'll get to 1.9% real return is if bond rates go negative and then keep going further negative to ridiculous levels like -10%.

----- BHCh:

I agree that logically prospects for bond returns are very limited but not sure what happens with bond funds if the government decides to throw money at the economy by overpaying for bonds when the inflation is rising.

And if the bonds are returning nothing for the next 50 years, perhaps we need to start looking at other sources of fixed income. Like farmland...

----- Michael James:

It's possible that interest rates could rise in the coming years, but the problem if you own bonds now is that you've locked in your interest rate. If you own 30-year bonds today and long-term rates rise, the value of your bond drops. So new investors in 30-year bonds would get higher rates, but you'd have to either stick with your crappy rate for 30 years or sell to lock in a loss and then buy a bond at a better rate. Either way, your overall bond returns stay poor for a very long time.

----- Michael James:

By the way, BHCh, I've noticed that your Blogger profile at
https://www.blogger.com/profile/08178467606233596591

isn't publicly available. This causes web crawlers to declare that my blog is full of broken links. Believe it or not, I haven't found a way to stop Blogger from displaying your profile's URL when it displays the posts that you've commented on (short of just deleting your comments, which I don't want to do). If I could just never display any profile URLs for any of my commenters, I'd do it, but I haven't found a way.

If making your profile public (possibly after removing parts you deem sensitive) would cause you other troubles, then don't worry about it and I'll keep looking for a technical solution. Otherwise, please consider making it public.

8. The following exchange contained broken comment author links, so I replaced it with this single comment to remove the links.

----- BHCh:

The other question: what is the maximum drawdown and by how much will I have to cut expenditure? Thats where bonds come in handy. And if you can’t reduce expenditure by half (or whatever it is) then yes, you need a safety margn.

----- Michael James:

I didn't look through all the backtests, but the biggest spending reduction I saw was about 50% (and it lasted for many years). People will differ on how much safety margin they need, but this margin shouldn't be negative.