Tuesday, April 17, 2012

Variable Withdrawals in Retirement

To get up to speed on using variable withdrawals in retirement, I’ve been studying “Variable Withdrawals in Retirement” from Bob’s Financial Website, among other resources. His article is the best overview I’ve seen so far.

The classic 4% withdrawal rate rule is about constant inflation-adjusted withdrawal amounts. The benefit of this approach is that it provides a smooth and predictable income stream for as long as wealth remains. But this method also has disadvantages. Wealth can run out. Also, it is probably not a realistic explanation for how people spend in retirement, since people probably don’t ignore market fluctuations. They may increase their spending when markets go up, and they are probably not going to play the implied game of chicken by continuing to spend the same amount each year as their wealth plummets toward zero.

So what happened? We simply withdrew what we needed and kept an eye on our portfolio balance. Most years our balance went up and we spent the money on vacations, luxuries and charity. When our balance went down we tightened our belt and economized.

This is what most people do and it works.

At the opposite end of the spectrum is a rule to spend a constant percentage of the remaining portfolio each year. This allows for spending to increase when wealth grows. As well, though spending amounts may drop to uncomfortably low levels, this spending rule prevents wealth from ever running out. The main disadvantages of using this rule are that the spending path is unpredictable and fluctuate widely over time.

What many of the other methods described at Bob’s Financial Site seek to do is to obtain some of the advantages of the constant percentage rule, while also placing bounds to help smooth spending. In other words, seek a compromise between constant inflation-adjusted withdrawal amounts and constant percentages of wealth. Strategies fitting this bill include William Bengen’s Floor and Ceiling, Jonathan Guyton’s Decision Rules with Guardrails, Bob Clyatt’s Rational Investing, and Robert Carlson’s Endowment Formula.

One exception to this trend is Paul Merriman’s Flexible Withdrawal Rate. He moves in the opposite direction from smoothing consumption by having you actually increase your withdrawal rate when wealth increases. I don’t think this makes much sense because you should be saving some of that windfall for a rainy day. People get diminishing returns from spending.

The other rules Bob discusses which I haven’t mentioned yet include Peter Ponzo’s Sensible Withdrawals (actually Taylor's strategy from above sounds related to this), Ben Stein and Phil DeMuth’s Five Year Plan, and some contributions by Cut-Throat and LBill from Bogleheads. Actually, I think some components from these approaches will ultimately lead to an optimal spending rule, but none of these approaches is quite there yet. From Peter Ponzo we get the idea of having a baseline spending floor, and from Cut-Throat and LBill we get the idea of calibrating withdrawal percentages to remaining life expectancy. I will come back to these issues again someday.

But for now, I would like to show four different Monte Carlo simulated outcomes when using a 5% withdrawal rate and a 50% allocation to large-cap stocks and a 50% allocation to intermediate-term government bonds. The three strategies I compare are constant inflation-adjusted withdrawal amounts, constant percentage of remaining wealth, and a version of Robert Carlson’s Endowment Formula in which the spending amount combines 50% of the 5% constant inflation-adjusted withdrawal amount and 50% of the constant 5% percentage of remaining wealth. This allows for some response to market conditions and income will still fluctuate, but not by as much as with the constant percentage rule. Also, wealth can still run out, though not as soon as with the constant inflation-adjusted withdrawal rule. 

These simulations show spending amounts for an initial wealth of 100. This can be scaled, i.e. spending of 5 with wealth of 100 is the same as spending of $50,000 with wealth of $1,000,000, for instance.

As you look at these spending paths, I have a question for you. How do you decide which spending path is more desirable? If you were forced to pick one of these three strategies, which would you choose?  Why? For the sake of argument, suppose that below are the only four possible outcomes from market returns and there is a 25% chance for experiencing each of these four simulated paths in your retirement. Which strategy do you pick?

First, here is a good outcome in which markets boom after about 6 years into retirement and wealth then grows rapidly. While constant inflation-adjusted withdrawals stay fixed at the same real amount, the constant percentage rule leads spending to grow quite a bit, and the endowment rule shares a lot of those gains. Spending is highest with the constant percentage rule, and therefore wealth is the lowest with this rule.

Second, here is an okay outcome. The constant percentage rule allows spending to fluctuate at levels sometimes more and sometimes less than the constant real amounts. 

Third, here is a bad outcome. With market declines and collapsing wealth, the constant percentage rule leads to much lower spending that the constant real amount rule... until year 23 when failure occurs. There is nothing more to spend because more was spent in the earlier years. Neither of the other rules run out over the 30-year horizon, though spending amounts have fallen by more than half in some years (in real terms) for the constant percentage rule.

Finally, here is a very bad outcome. With constant real spending, wealth is all but gone in year 16. The endowment rule holds on with the last of the wealth taken out in year 21 and nothing left by year 22. The constant percentage rule can never run out, but by the 15th year of retirement this rule is only supporting spending levels that are about 20% of the initial spending amount.


  1. Thanks for all the great graphs and comparisons. For us the key factor is how much of our spending is discretionary. As over 50% of our spending is totally discretionary (mostly travel) we would opt for the constant % method. We understand and expect large fluctuations in our spending but appreciate the simplicity of the constant % method.

  2. Thanks for sharing your wonderful research with us.

    We currently use the 4% of remaining wealth rule & I would choose it given the scenarios above.

    We retired early at 48/45, so we were looking at the possibility of 40+ years in retirement. We decided that we would work in order to handle any short-fall. At retirement, we felt that we had enough padding in our budget that we could handle a sharp decline in our budget & that's exactly what happened in 2009. We worked for 9 months in 2010 to give the portfolio a chance to repair itself (in hindsight, it was more psychological than financial). We also moved from NJ to Denver last year (moving out of NJ was a contingency plan we had at ER) which has lowered our cost of living a little.

    Our withdrawals from the portfolio will drop sharply when we start taking SS, so we may rethink this withdrawal method as we get closer to that date.

    We find ourselves almost 4 years into this journey with our portfolio only marginally higher (in nominal terms) than it was when we began. Our budget is back to its initial nominal value, but we find that we are very happy & live well while spending less than that amount.

    (If you are interested, you can find more at early-retirement.org by looking for posts by Walkinwood)

  3. Thank you both for sharing.

    early-retirement.org is a great resource right up there with Bogleheads.

    I'm beginning to see more of the appeal of the constant percentage approach, especially when combined with an income floor existing outside of the financial portfolio. And it is only marginally more difficult to work with in programming simulations. You can't set the lifetime consumption path at the start, you just have to update the spending amount period by period.

  4. Have you ever looked at Henry Heblers analyzenow.com website? He demonstrates some life expectancy based withdrawal plans.

    1. I read Bud's article "Withdrawal Strategies in Difficult Times." Yes, I think that his Methods 1 and 2 are what Cut-Throat and LBill have in mind. Good stuff.

  5. Wade, I like your idea of applying constant-percentage-of-wealth combined with an income floor. For the income floor piece, I believe the most basic (and non-discretionary) expenses are likely to grow with inflation and would not be a good match for constant-percentage-of-wealth--likely a bettter match with an inflation-adjsuted SPIA.

    Above the floor, perhaps lifestyle considerations enter in. The individual who wants the same amount of discretionary spending every year might use a SPIA. But for someone who doesn't mind taking some investment risk in the hopes of having more for discretionary spending, the constant-percentage-of-wealth might make sense.

    Another wrinkle--for someone who is comfortable with variable spending, but wants longevity protection for that spending, an immediate variable annuity might be a good product. I've been ignoring this product because there are virtually no sales, but it may be worth a look for fitting into the retirement planning picture.

    1. Thank you Joe. Now that I did that last post about annuities, which I wrote in the same code for this variable withdrawals work, I am now ready to test this possibility. And my preliminary analysis suggests it is worthwhile to go with increasing percentage withdrawals based on remaining life expectancy, rather thank constant percentage withdrawals. Well, that is if you don't care much about the annuity side.