Monday, April 27, 2015

Retirement Spending Assumptions and Net Worth

In my last post, Spending Typically Declines as We Age, I reviewed the results of research by David Blanchett (PDF) and Sudipto Banerjee (PDF) that shows expenditures in retirement typically decline as we age. Most retirement spending strategies assume, as I noted in that post, that future real spending will remain constant throughout retirement.

The amount that we can safely spend from retirement savings in the current year depends heavily on the assumptions we make about future spending trends. If our future spending needs will decline, spending rules that assume constant real spending will be unnecessarily conservative and, of course, if future spending will increase, those spending rules will recommend spending that may not be sustainable.

What assumptions should our retirement plan make about future spending? The two papers I reference offer some clues.

First, Banerjee reports that spending in retirement increased for only 16% of the households in the data he studied, while it declined for 66% of them. It is significantly more likely that your expenditures will decline as you age, but they might not, so our retirement plans should also consider worst case outcomes.

We could assume a worst case, that expenditures will increase perhaps 1% per year on average, but that would significantly increase the predicted cost of retirement. If we assume we will live 30 years or more, that our market returns will be quite conservative by historical standards and that our expenses will grow in retirement, we will quickly realize that hardly anyone could afford that retirement. Making lots of conservative assumptions doesn't make a very good plan.

Blanchett offers additional insights by segmenting the data based on the level of annual spending relative to net worth. He creates four categories of consumption: low spenders with high net worth, low spenders with low net worth, high spenders with high net worth and high spenders with low net worth. By figuring out into which group you best fit, you may be able to narrow the field of spending assumptions for your plan.

The dividing line for high and low spenders was $30,000 per year and the hurdle for high net worth was $400,000 in Blanchett's study. These are the median values for his data sample, not for the population of retirees. In other words, more than $30,000 of annual spending made households in the data sample high spenders relative to other households in the sample, but that amount wouldn't make you a high spender relative to all other retirees in the U.S. today. The breakpoints for the larger population of retirees would likely be much higher. Blanchett is showing that expenditures in retirement depend on the relationship between annual spending and net worth; he is not claiming that these are the dividing lines for all retirees.


Blanchett notes that two of these groups, Low Spending, Low Net Worth and High Spending, High Net Worth retirees, consume efficiently (green), while the other two groups consume too much (red) or too little (yellow).

Following is a diagram from Blanchett's paper plotting these four segments. Please note the very important point, as I explained in Spending Typically Declines as We Age, that these graphs show annual rate of change in spending and not annual spending, itself. With the exception of Low Spending, High Net Worth households (the red squares on Panel B) almost all of the annual changes in expenditures are negative, meaning spending declines throughout retirement for the other three groups. (Double-click the chart for a larger image.)


Notice that the following graphs of annual spending are quite different than Blanchett's graphs of annual spending change above. Because some readers have mistaken the Blanchett "smile" rate-of-annual-change graphs for annual expenditures graphs, I provide both in the examples below.

In fact, Blanchett's paper shows this in his Figure 7, though most of that paper addresses annual spending change and not annual real dollar spending. My graphs will show typical real dollar annual spending that is derived from the rate-of-change graph to its right. I place the spending function on the left because I believe that information will be more meaningful to most of my readers, and I switched Panels A and B in Blanchett's Figure 7 for consistency – the charts on the left always show annual spending.


Let me be very clear about this. The Blanchett smile curves, like Panel A above, show how quickly typical spending changes each year. The spending curves, like Panel B above, show how much spending changes in real dollars and, in most cases, spending goes steadily downward throughout retirement, like the curves in Panel B.

Now, let's look at Blanchett's four spending/net worth classifications.

Low Spending, Low Net Worth households likely spend a large portion of their budget on non-discretionary expenses with little opportunity for reducing expenses later in retirement. We usually give up some discretionary items later in retirement, like extensive travel and sports, and these households have fewer of those to eliminate, so expenditures don't decline a lot with age.


High Spending, High Net Worth households also consume efficiently and will likely see greater declines in spending than Low Spending, Low Net Worth households, because they will have more discretionary spending to eliminate as they age.


Low Spending, High Net Worth households appear to have the highest probability of increased expenditures throughout retirement, but they can afford it. They are underspending. A likely cause for such an increase in expenditures, in fact, is recognition over time that they have the resources to spend more.

This graph also demonstrates the key point that increasing expenditures don't necessarily mean that retirement is getting more expensive and decreasing expenditures don't mean it is getting less expensive. They mean that retirees are spending more or less. Expenditures, the subject of this analysis, are not the same as expenses. Sometimes expenditures change because retirees have to spend less and sometimes it is because they can spend more.


High Spending, Low Net Worth households also consume inefficiently and they are likely to recognize as they age that their level of spending is unsustainable. This realization will eventually lead to declining expenditures later in retirement, as can be seen in the following chart.


The next chart combines all four annual spending curves for comparison. The two curves in the middle result from efficient consumption. Inefficient consumption produces the two extremes.


Your personal ratio of annual spending to net worth should suggest whether your own spending is more likely to rise, fall, or remain somewhat constant throughout retirement.

To summarize this information, Banerjee tells us that two-thirds of retirees will experience declining expenditures as they age and only 16% will see increased spending. Blanchett tells us that spending will likely increase for Low Spending, High Net Worth households as they realize they are able to spend more as they age, as it will likely decrease for High Spending, Low Net Worth households as they realize they are running out of savings. Among households that consume efficiently, those with High Spending and High Net Worth are more likely to spend less later in retirement because they will have more discretionary expenses to "age out of" than will Low Spending, Low Net Worth households.

Some expenditures change for reasons that have little to do with how much annual spending a retiree targets or how wealthy they are. Some changes are the result of aging. Health care expenses tend to increase as we get older but we also become less active and other expenses decline. I recently read that international air travel declines among septuagenarians and domestic air travel declines among octogenarians. I suspect the sales of bungee-jumping and rock-climbing gear decline in those market segments, as well. I spend less on hair care expenses.

These two studies deal with typical retiree spending patterns and assume that expenditures will follow some trend, rising, constant or declining, throughout retirement. They don't, however, deal with the most likely scenario for an individual household, irregular net spending.

Both income and expenses in retirement are likely to vary significantly over time. Income will vary, for instance, as Social Security benefits ramp up for retired couples and more income needs to be withdrawn from savings early in retirement. There may also be large planned expenses later in retirement, like college for a child or grandchild. Net spending is the important consideration, the difference between annual income and annual expenses. Irregular net spending from savings might look like the red bars in the following chart:


These irregular net spending years, whether they are caused by changing income or changing expenses, must be considered when calculating a sustainable amount to spend in the current year. Like steadily rising or steadily declining expenditures, spending rules assume flat future spending and don't accommodate irregular net spending very well.

It is helpful to know that expenses typically decline in real terms throughout retirement, but yours may not. You need to plan for expected spending declines but be prepared for a worse case. Like portfolio returns, total expenditures in retirement are unpredictable.

So, most retirement income strategies assume constant spending throughout retirement and most retirement expenditure studies show that constant spending isn't the norm. What's a retiree supposed to do with that?

In my next post, Spending Rules That Fit the Patterns of Retirement, and Some That Don't, I'll explore spending strategies in light of future spending expectations.

5 comments:

  1. Dirk, thanks for summarizing the two articles. I like where this is going. Good information, and can't wait for next blog post. Brad

    ReplyDelete
  2. In Blanchett's 2x2 segmentation by consumption and net worth, I find the $400K threshold for "high net worth" disconcerting, given that according to Blanchett, net worth in his study includes the net present value of future Social Security payments as well as pension income. With a modest SS income stream alone (say, $20K/year starting at age 65), it would be easy for many new retirees to meet this high net worth criterion, but in reality they may have few assets to tap for irregular or emergency expenses and would not consider themselves "well off" by any means.

    ReplyDelete
    Replies
    1. I agree, but I think you're reading too much into these numbers. To quote Blanchett, "The approximate median consumption in our sample is $30,000 per year and the approximate median net worth is $400,000."

      David isn't claiming that these amounts are representative of the larger population of retirees, only for his data sample. So, you can read into this that if you had more than the median spending and more than the median net worth of the sample you were, relatively speaking, a High Spending, High Net Worth retiree. You cannot read into it that any retiree in America today with more than $400,000 is High Net Worth, for example, or that more than $30,000 spending makes you a High Spender.

      If David had spending data for the entire population of retirees, you can bet the median spending would be much larger than $30,000 and median net worth would be higher, too.

      You have to work with the data that is available. He is simply showing that expenditures in retirement are a function of the relationship between your spending and your net worth.

      Delete
  3. When doing my planning, I use the notion of "mandatory" expenses as a baseline. I define these to be the expenditures necessary to sleep in a comfortable house, be able to eat decent three meals a day, and have the mobility needed in my environment to run basic errands. I enumerate those expenses in today's dollars, then draw a curve representing 3% inflation out to the end of my planning window. Then, when I plot my incomes, the difference between the two is what I can use for 'discretionary' expenses, those that I don't need to incur every month to live.

    I've given some thought about how such a categorization would behave in the terms you describe; essentially, I think it would be replacing one type of mandatory expense, say, housing, for another, such as assisted living. But, I think it would take data that hasn't yet been collected in that manner to substantiate. Your thoughts on the concept would be most welcome...

    ReplyDelete
    Replies
    1. That's a good way to do it. It's what economists (and I) recommend.

      I think a major point of the Blanchett paper is that expenditures in retirement are highly dependent on discretionary versus non-discretionary expenses. People with lots of discretionary expenses are likely to see expenditures decline sharply as they age, while those with mostly non-discretionary expenses won't be able to reduce spending as much. So, simply put, those two categories behave in these terms: discretionary expenses can decline a lot as we age but non-discretionary expenses won't.

      You are correct that housing is non-discretionary. You might reduce the expense somewhat by moving somewhere cheaper, but unless you move back in with the kids, housing costs aren't going away and may not decline much ,at all.

      Does that answer your question?

      Delete