How Should You Plan for Your Spending to Change Throughout Retirement?

It’s challenging enough to figure out how much you’ll want to spend at the start of retirement.  Even more challenging is deciding how your spending will change as you age.  These choices make a big difference in how much money you’ll need to retire.  They also shape the spending options you’ll have available throughout retirement.  Here I explore the good and bad parts of common wisdom on retirement spending to arrive at my own spending plan for retirement.  Spoiler alert: the “go-go, slow-go, no-go” narrative is good marketing, but it has cracks.

Two extremes

Some people focus on the early part of their retirement.  They want as much money as possible available early on while they’re still young enough to enjoy it.  They seem to think of their older selves as a different person who they care less about than their current selves.

Others focus on their older selves and worry about running out of money at some point.  These people usually spend far less than their portfolios allow, and they tend to be resistant to spreadsheet evidence that they’d be fine spending more.  Some make frugality part of their value system, and others are genuinely fearful.

A rational retirement spending plan is somewhere between these two extremes.  But where?

The default

Before retirement spending research over the past decade or so, the default was to assume that retiree spending would rise with inflation each year.  In real (inflation-adjusted) terms, we assumed that retiree consumption would be flat over time.

This doesn’t mean that consumption would be flat in the transition from working to retirement, though.  Many expenses go away in the typical retirement.  Average retirees pay less income tax, have paid off their mortgages, spend less on children, and no longer have many work-related expenses like commuting and clothing.  On the other hand, retirees often spend more on hobbies.  Some retirees are exceptions, but retirement experts say typical retirees need 45-70% of their working income to have the same standard of living.  But after retirement starts, we used to assume flat consumption over the years.

It’s tempting to think that having retirees’ spending rising with inflation would have them matching the spending increases of their younger neighbours.  However, this isn’t true.  Human progress causes our consumption to rise faster than inflation over the long term.  Compared to a century ago, workers are far more efficient today, and they have a wide array of products and services available that people in the 1920s never dreamed of.  Progress will continue, and with each passing decade, more amazing products will become available.

If you want to fully participate in our progressing economy, you would need to plan for annual retirement spending increases of about inflation+1%.  It may be rational to decide you won’t need the latest iPhone or whatever amazing new product that will come along, but it’s important to realize that planning for flat consumption is already a compromise.  If you were keeping up with your neighbours at the start of retirement, you would be falling behind a decade or so later.

Go-go, slow-go, no-go

The idea that we should plan to spend less each year through most of retirement has some of the best marketing around.  In his book, The Prosperous Retirement, Michael Stein referred to three general phases of retirement:

• Go-go years: From 60-65 to 70-75.  High activity and spending.
• Slow-go years: From 70-75 to 80-85.  Activity and spending decline.
• No-go years: From 80-85 on.  Minimal activity with healthcare and long-term care costs.

This framework is easy to embrace for anyone who is still a long way from the slow-go age.  We’ve all seen old-timers who seem unable to do much, and more importantly, they seem very different from us.  However, if you ask someone in their early 70s if they’re into their slow-go years, don’t expect a polite response.

Already, most descriptions of the three phases have the go-go years ending at 75 instead of 70-75.  With so many baby boomers now in their 70s, it’s not surprising that they don’t like to see themselves as slow-go.

Setting these self-image issues aside, are these older boomers spending less than they did in their 60s?  If they are spending less, some will be doing so by choice and some by necessity because they have limited savings.  How significant is this group who overspent early?  Do you really want to model your own retirement in part on this overspending group?

In the end this vivid narrative paints a compelling picture of someone (but not you!) slowing down and eventually stopping altogether, but it doesn’t prove anything about how you should plan your retirement.

The research

One of the early papers researching retirement spending patterns is David Blanchett’s 2014 paper Exploring the Retirement Consumption Puzzle.  This paper along with many subsequent papers have established without a doubt that the average retiree’s inflation-adjusted spending declines in early retirement and increases late in retirement as health care and long-term care costs rise.

That seems to settle it, right?  We should follow the research and plan for declining consumption through early retirement, and possibly plan for health spending and long-term care costs late in retirement.  But there’s a disconnect.  We know what average retirees do, but is this what they should have done?

The average Canadian smokes about two cigarettes per day.  Does this mean we should all plan to smoke two cigarettes each day?  Of course not.  This average is brought up by the minority of Canadians who smoke.  If we take the smokers, whose behaviour we don’t want to emulate, out of the data, the average drops to zero.  In reality, the best plan is to not smoke at all.

Carrying this thinking over to retirement spending, we need to know how many retirees overspent early in retirement and now regret it.  You don’t want to emulate these people.  If we could remove these people from the data, the average spending from the remaining retirees might give a better picture of what you should do.  In addition, we might want to remove retirees from the data if they badly underspent.

The retirement spending smile

The Blanchett paper refers to a “retirement spending smile” that is widely misunderstood.  If we draw a chart of average retiree spending over time, it starts high, falls for a decade or two, and then rises again at the end of life.  People refer to this chart shape as a smile.  However, in Blanchett’s 2014 paper, the smile actually referred to a chart of changes in retiree spending.

So, Blanchett observed that retiree spending changes little in early retirement, then starts to decline and this decline grows in mid-retirement, then the decline slows or even reverses to spending increases late in life.

Here is a chart of Blanchett’s annual spending change data:

Notice that the points don’t really look much like a smile.  The measure of how well a curve fits some data is called R-squared.  Blanchett reports that his spending smile curve has about a 33% R-squared match with the data.  This is a rather weak match, and is a sign that he didn’t have enough data.  Another sign of too little data is the big changes over a short time.  There is no obvious reason why the spending drop should be so much more at 80 than it was at 78.  

What is important but unclear is how much of this data comes from overspenders and underspenders who you don’t want to emulate.  Blanchett considers the question of whether retirees spend less “by choice or by need,” and admits that “it is impossible to entirely disentangle this effect.”  To explore this question he divides the retiree spending data into four groups based on whether their spending is high or low and whether their net worth is high or low.  He then studied each group separately.

Unfortunately, his limited data set forced him to restrict this subgroup analysis to retirees between 65 and 75.  Notice that according to the current version of the go-go, slow-go, no-go narrative, all these retirees are still enjoying their go-go years.  Despite this, Blanchett found a spending smile in two of the four groups.  When spending and net worth were both low or both high, there was a spending smile, although the R-squared measurement was somewhat weak in each case: 61% and 53%.  The low spending high net worth group just kept spending more each year, and the high spending low net worth group just kept spending a lot less each year.

It’s not clear that any of this analysis of the age 65-75 cohort helps much in deciding how you should design your own retirement.  Let’s call the groups with spending and net worth either both high or both low the rich and poor groups.  You’re unlikely to want to model your retirement on the other two groups.  If we take Blanchett’s spending changes for the rich and poor groups and turn them into consumption amounts each year, we get the following chart:

The poorer retirees don’t change their consumption much at all.  What little movement we see over time could be just randomness in the data.  The richer retirees show a pronounced drop in consumption.  However, this drop starts right from the beginning of the so-called go-go years.  What does this mean?

The most powerful part of the go-go, slow-go, no-go narrative is the mental image of an old codger in a rocking chair not able to do much.  But this image doesn’t help us make sense of a more than 20% drop in spending in richer retirees’ latter 60s.

Maybe some richer retirees make big purchases like a cottage that skew the data.  Certainly there are some who make some carefully planned bigger purchases early, and some just overspend early and are forced to reign in their consumption.  It’s hard to know how many of these retirees you’d want to exclude from the data because you wouldn’t want to model their mistakes.

This research leaves me with more questions than answers.  The same is true for other retirement spending studies I’ve read.  It’s hard to figure out which data to exclude due to the retiree making mistakes.  

Some study authors try to argue that retiree overspending is not a factor in their data based on some analysis they perform.  This makes no sense.  The idea that overspenders don’t exist at all is obviously wrong.  What we really need is a study designed to capture data on whether retirees underspend or overspend so that we can exclude their data.  This would allow us to build a model of retiree spending that would be worth emulating.

Conclusion

We have no definitive answers given the evidence available.  At one extreme you could plan for rising inflation-adjusted spending by about 1% each year to allow you to fully participate in future human progress that improves goods and services.  At the other extreme you could plan for declining consumption in line with several academic studies, and just ignore the fact that some retirees aren’t worth emulating because of their mistakes.

An apparent advantage of the declining consumption plan is that you could retire sooner on less savings, or if you save the same amount, you could spend more in early retirement.  Both of these options are very appealing for those who don’t like their jobs or who haven’t saved enough.  The downside is the possibility of forced spending reductions and ultimately running out of money.  I’ve seen this up close in my extended family; it’s quite sad and was expensive for me in one case.

My choice is a compromise between these two extremes.  I plan for flat inflation-adjusted spending, and try to be prepared to cut spending if the stock market seriously underperforms.