Since I have often ripped on constant spending as an unnecessary and foolishly high risk idea[1], let me try to explain my chart to set up the post...
A. Lifestyle Volatility. Most researchers and bloggers will explain that as one goes from a constant spend assumption to a rules based or percent of portfolio world [or the other way; more on that later] that lifestyle volatility will go up. Many retirement papers and models choose center-A because of its modeling tractability and because of (an outsized?) confidence either that more retirees than not fetishize constant lifestyle or that the principles of academic economic utility always make sense.
B. Retirement Risk. Most researchers and bloggers will be clear that moving to the right will more likely than not reduce risk from sequence of returns and, if one believes in ruin risk...then that too. This I believe.
C. Model "Virtue." This is sketchy and an opinion but I'm thinking there is some kind of break point that balances lifestyle vol and risk after which you are probably shooting yourself in the foot in terms of either self denial or choppy lifestyle as one goes to the right. Going to the left is another story.
D. Common Range. This interval, when someone writing on the topic is mostly honest, is the most common framing of the range of spending choices. But they always forget that there is a dark region to the left of constant spending. This is where one -- by design, happenstance, malice, self-destructive urges of self or others, or maybe just plain ignorance -- spends more when markets crash and less when they are stable or rising. To wit...
E. This is approximately where I was in 2008 through 2010. I won't say why.
MY WEAK DEFENSE OF CONSTANT SPEND ASSUMPTIONS
Let's try this on for size. I don't expect anyone but me to buy this. What reasons could I have to defend the constant spend assumption when so many smart people beat it down as unrealistic or inappropriate? Here are some attempts:
1. There are things worse than a constant spend that real people sometimes do in real life.
2. It is sometimes forgotten that model fidelity to the last-mile of planning and precise real world considerations is not a monolith. That kind of focus is not for everyone. So let's break the modeling task into two separate worlds, one of investigative or exploratory work where ambiguity is likely high and speed and simplicity is more helpful than not and another world of of confirmatory analysis or decision making where degrees of ambiguity have been resolved or waived away and ignored. I often work in the former zone. Many paid advisors and academics are working in the latter. The latter is a world with marginal tax rates, fee schedules, non-normal distributions, specific portfolio design, auto-correlation and mean-regression, account sequence drawdown strategies and so forth. That's good stuff but hard to model and sometimes distracting to the first world. My own theory is that the broad-brush shapes of core retirement processes are more important than fine detail in that first world. Also I don't get paid for work in the second.
3. It is, in programming terms, quite simple and respectful of my time.
4. It is a reference, a guidepost. If one knows in one's mind's eye the general effects of movements away from constant spend assumptions then it is a reasonable base case or scenario against which one can and should compare strategies. This is maybe at least a little more salient when looking at it in the context drawn above. It stands there right in the middle as a kind of an anchor. By all means add rules and adaptive systems but show me a reference point first.
5. Research (proof by assertion here...) shows both a retiree utility preference for constant lifestyle as well as adaptive behaviors both over the lifecycle (maybe Blanchett's smile shaped spending) and in response to financial stress (who, really, willingly spends themselves to zero in service of short-sighted lifestyle goals). There is no real reason to hyper focus on constant spending but also no reason to totally abandon it in investigative work.
No comments:
Post a Comment