Monday, June 30, 2014

Defending Systematic Withdrawals

Jonathan Guyton and Michael Kitces are both leading advocates of systematic withdrawal strategies for retirement, as Wade Pfau pointed out in a recent post. Pfau observes that "it is somewhat uncommon to hear rigorous intellectual defenses of systematic withdrawals." Most academics prefer other strategies with greater downside protection.

The video is informative and I suggest you watch it regardless of how you feel about "safe withdrawal rates." Guyton and Kitces are two very bright people, but I am not a fan of systematic withdrawals except for households who have a lot of savings.

When I say "a lot", I'm referring to those over-savers who need to spend perhaps 3% or less of their savings every year in retirement. They probably have adequate risk capacity to manage the downside of SW.

As I watched this interview, I realized that I am perhaps not as far from Michael's position as I had thought. If you're wealthy enough, I have no problem with SW, and that is the target market for most financial planners. I focus my attention more toward retirees who are not over-savers, for whom I believe SW is often a very risky approach.

Risk tolerance, or how much risk lets you sleep at night, is different than risk capacity, or how much risk you can afford to take. (Here's a cute analogy to explain the difference.) Risk capacity is the limiting factor when choosing a retirement strategy and the more wealth you have relative to your annual spending, the greater your risk capacity.

Kitces notes in the video that in many cases the overall portfolio constructed with a floor-and-upside strategy (or "discretionary - non-discretionary strategy", as it is called in the video) is practically identical to what would be required for a SW strategy. This is technically accurate, but omits important points.

First, a major purpose of floor-and-upside and time-segmentation strategies is to make retirement finances more transparent and manageable. We could, using the same logic, organize our household budget into just two categories: "income" and "stuff I need to pay for". That wouldn't provide much transparency into our budget, however, so we create subcategories for food, rent, transportation and the like.

In this same way, floor-and-upside and time-segmentation strategies create subcategories, or "buckets", to help us visualize where the money will come from to pay our living expenses in future years, or whether a shortfall will jeopardize groceries and housing, or just the cable bill.

In other words, these strategies help us manage our retirement plan.

Time-segmentation attempts to minimize the risk of needing to sell assets when they are cheap and to provide a safe source for near-term spending. Floor-and-upside tries to ensure that in the worst case we can still pay for non-discretionary expenses. SW lumps them all into one large "total return" bucket, much like the "stuff I need to pay for" category.

The second important but unmentioned point is that SW recommends a spending rate that some might consider safe or sustainable. Life annuities also determine a spending rate based on the payout rate of the annuity we purchase. The SW rate is probabilistic while the annuity payout is contractual.

Time-segmentation and floor-and-upside strategies, to the contrary, do not inherently calculate a spending rate. SW rules of thumb and consumption-smoothing are two ways one could establish such a rate for these strategies.

Lastly, these strategies rebalance asset classes differently. The asset allocations may start out being quite similar, but they will most likely drift apart over time.

These are important differences   the strategies provide different spending, different risks and different asset allocations over time. The initial portfolio allocations may be "practically identical", but the strategies are not.

I find comfort in the observation that the strategies frequently have similar initial portfolio allocations because I would have a hard time explaining why they wouldn't.

One of them will perform best for you over your lifetime. Unfortunately, we can't predict which one that will be. That's the bet you have to make.

Kitces also mentions in the interview his personal quandary, which he has stated before, that a retiree who expects a wonderful retirement might consider it a failure if that retirement only turns out to be OK. This is the tradeoff one makes when forgoing the possible upside standard of living with the SW strategy in order to mitigate the risk of going broke in old age with a floor-and-upside strategy.

Personally, I don't understand the quandary. If I were offered the opportunity to take the worst-case scenario off the table (dying broke) by giving up the possibility of more trips to Europe and a second home, I'd jump at it.

But, that's a decision retirees with a lot of wealth will get to make for themselves.


  1. I do think there are important distinctions btwn 'SW' and 'safety first' methodologies. However, I don't think any of the three points made in this blog get at the heart of the matter.

    First, the argument that SW is so simple that it can be characterized has "income" and "stuff I pay for" is not really an accurate characterization. SW strategies all contain a budgeting component, and any intelligent discussion I've read includes detailed budget categories which are almost always segregated into essential and discretionary. That's how we come up with the budget in the first place. We don't suddenly become unconscious of spending patterns just because they're not in buckets.

    Second, I don't think it's accurate to infer that a spending rate is not fundamental to time segmentation and flooring strategies. It has to be, otherwise we wouldn't know what the budget (or bucket or floor) size needs to be. Saying that time segmentation and flooring strategies don't "...inherently calculate a spending rate" seems like a distinction without a difference.

    Third, AA btwn SW and Time/Flooring strategies certainly does diverge over time, with purer Time Segmentation strategies resulting in increasing equity allocations as we age; personally, not where I want to be. If a graduated (periodic bucket reallocation) approach is used, it seems to make Kitces point.

    This is a great subject that would be better discussed, I think, from a different viewpoint, which is actually a point you make in the blog: Unless you have BIG MONEY (>=33xAnnual expenses), give up on upside and obtain 'guaranteed' income. So, I'd suggest a more useful discussion would be to start with that position, then discuss what 'ranges' (Annual Spending/NW) are most appropriate for SW and 'safety first' (Time Segmentation, Flooring, etc.) strategies.

  2. Hoot, I agree with much of what you say, but while budgeting can and must be added to an SW strategy (and as you point out, any intelligent discussion would), budgeting is not used to calculate a sustainable withdrawal rate. For TS and Floor strategies, spending can be determined by budgeting, consumption smoothing, current annuity rate limits or other methods. Those strategies don't inherently supply a spending rate; SW does and it is the same spending rate regardless of portfolio value, budget, or proportion of non-discretionary expenses. The “sustainable” rate is determined from historic portfolio returns solely by your portfolio allocation and expected (remaining) years in retirement.

    I agree that a spending rate is a fundamental requirement of all strategies. My point is that TS and Floor strategies don't tell you how to calculate it.

    SW strategies may have budgeting added to them, but discretionary and non-discretionary expenses are lumped into a single bucket in the sense that they are both subject to the same risk and return (that of the portfolio). The point of Floor strategies is to invest non-discretionary capital more safely. I know Michael Kitces believes that SW strategies have a floor, but I don’t. At least not in the way that most of us define floors. So, I’m comfortable with that characterization.

    I'm not suggesting we become unconscious of spending patterns when they are not in buckets. I'm suggesting that it is easier to remain aware of them if they are. We could be conscious of our spending without a budget, it's just harder.

    I'm not sure that I want increasing equity allocation as I age, either, although a recent paper by Kitces and Pfau recommends that. A TS strategy should be revised as our view of the future changes and I would deal with it in that way. My portfolio wouldn't become more aggressive. If I thought I had less than 15 years to live, I wouldn't invest any money that I needed in stocks. That bucket would disappear. If I were 90, the bond bucket would also disappear.

    On the other hand, that is another point determined by wealth. If I had a lot more money than I needed at age 90, I would gladly become more aggressive with that capital as I aged. Which is a great segue to your last point.

    I agree that how much wealth you need for the various strategies would be an interesting discussion and I hope someone picks up on it. I try to focus my blog on helping people who are not wealthy (their spending need is WAY beyond 3%) and I think “safety first” is always the way for them to go.

    You make interesting points and I truly appreciate you taking the time to write a thoughtful comment that will help other readers think through the issues!

    1. Dirk, thanks for reiterating your commitment to serving the less wealthy reader (me). Most of the places I go to get financial insights are obviously for the really well-off; and of no help therefore to my modest circumstances. I find your posts very commonsensical.

  3. Dirk, very good post, and brings up a question for me. Over the years I've struggled with budgeting. Early on it was essentially burying my head in the sand. I didn't have enough money, lived paycheck-to-paycheck, and just tried to make ends meet. Not proud of it but that's the way it was. Later in life when I did make more, we got religion about debt and savings, and wiped out the former and concentrated on the latter - thank goodness. And I learned to do things like painstakingly recording every expenditure to develop a much better feeling for where it all went.

    I wonder how many others are like this, and then to my question: what do they do when retired and the paychecks stop? Suddenly they no longer have the automatic throttle on spending other than their own will power which may be hard to manage. Do you have postings on this?

    What we do is to add enough money into our checking account on the first of each month to cover normal expenses. I expect by the end of the month that that amount will approach zero. If it stays positive, good. And if it goes negative it's because of irregular and/or unforeseen expenses such as vacation, bigger house maintenance expense, etc. But overall I have a view of the annual amount/budget. Pretty simple and I think it works for us, especially since we have good savings. But even if not (perhaps even especially) then putting what one has to spend into checking each month is a convenient way to replicate a paycheck-to-paycheck approach. Thoughts?

  4. Wow. Budgeting isn't an area where I have a lot of advice to pass along. It is, as you say, difficult and ultimately requires self control.

    You could, of course, buy a life annuity and let the insurance company dole it out. I say that only partly in jest, since I believe there are many people for whom this is not an unreasonable approach. A spendthrift or an adult with a serious substance dependency, for example, would be a good candidate.

    Here are a few thoughts, though. Budget a total expenditure for the year based on your retirement plan, then allocate it across months. Not every expense is monthly.

    I prefer to limit check writing because credit cards and debit cards provide a way to download excellent spending records. With checks, as with cash, you have to enter the payee and "for" fields manually.

    I use to download spending information from all of my accounts: banks, brokerages, credit cards, even the mortgage company. Mint even does its best to fill in the category, though you will have to review and edit those.

    I find Mint reports inadequate, but some people find them usable. I download everything from into Excel, but I know spreadsheets very well. (Hint: learn pivot tables.)

    Mint is terrible with Vanguard and similar brokerage/banking arrangements. Vanguard sends so many transaction records you don't need that finding expenditures is a needle in a haystack problem. Schwab Bank works very, very well.

    Online banking helps, too. Don't pay for it. There are too many good, free options.

    I think the key is knowing how much you have already spent versus budget BEFORE you spend more.

    Good luck!

    1. There are some of us who're left-brained and others who're right-brained, nothing is completely one way or the other, and I think budgeting is one of those things that some find straightforward (like myself) and others not so. My assumption in writing my comment is that many have trouble with it and so when they reach retirement (either when chosen or otherwise) that they could use an easy approach to managing their cash and spending. Sure, if all could use or other, there'd be no problem.

    2. I don't really know anyone (including me) for whom budgeting comes easily.