Friday, September 2, 2011

Dan Ariely on Spending in Retirement

Noted behavioral economist Dan Ariely has a new blogpost making a few controversial points.  Here I would like to focus on his criticism of the traditional replacement rate rules of thumb (which are somewhere in the neighborhood of 70%).  Those rules of thumb are indeed just rules of thumb and everyone will have a unique situation, but these rules were derived by looking at household survey data and investigating how much people spend on average before and after retiring. He said that in conducting surveys about what people would like to spend after they retire, he adds up the costs of these activities and finds that their ideal replacement rate is more in the neighborhood of 135%.

I’m confused, as this seems to be the end of the story.  I could imagine if he asked people about what they would like to be spending currently, he might also find that they would like to be spending much more as well.

Economics is all about “maximizing happiness” as he rightly notes, but this must be done in light of one’s constraints.  And it usually involves “consumption smoothing,” which means that people can get the most happiness if they are able to spend about the same amount from year to year rather than letting this fluctuate (as his approach would do, someone is going to have to save an awful lot to have a 135% post-retirement replacement rate, which means they will have a much lower standard of living prior to retirement).

So we can’t just suggest that financial planners are giving bad advice because his surveys show that people would like to spend much more after retirement.

I have a new paper that will be in the upcoming October 2011 Journal of Financial Planning, “Getting on Track for a Sustainable Retirement: A Reality Check on Savings and Work.” That link is to a rough draft version.

Using the approach outlined in my paper, we can see what still awaits someone hoping for a 135% replacement rate when they retire. This approach is about how much one needs to replace from their personal savings; Social Security and other income sources would be added on top. So let me use “115%” as the target replacement rate from savings, which would provide somewhere in the neighborhood of 135% with Social Security too.

And let’s consider the case of a 40-year old. This 40-year old already has accumulated some wealth, and the following table shows her feasible retirement age for different scenarios regarding her current wealth accumulation, future savings rates, future stock allocations, and desired replacement rates.  More about specific assumptions are provided in the paper.

What we can see from this table is that unless she has already accumulated quite a lot of wealth (as represented by multiples of her final salary), she is going to be rather old before reaching the point where she can feel comfortable retiring with such a high replacement rate.  And this assumes she can keep the same job and same salary for so long. Life is indeed about tradeoffs, such as in the bottom part of the table where we see that if she currently has saved 4 times her salary, she can potentially retiree at age 69 with a 115% replacement rate, but she can retire at age 58 with a 50% replacement rate. Now she needs to decide whether to delay retirement in order to spend more or not.  

I think this is a real tradeoff that Prof. Ariely must consider, and it seems to me that a good financial planner should be able to help someone evaluate these tradeoffs to come to the best happiness-maximizing compromise solution.  


Trying to Reach a Wealth Target in a Volatile World

A rather interesting discussion is taking place in the "Redefining Risk" thread at Bogleheads. Many different aspects are being considered now, and here I'm interested in the issue really gaining momentum on page 5 about what sort of issues to build into the analysis when trying to guide people toward their wealth accumulation goals.  John Norstad wrote an especially interesting and funny comment here.  In moving through your career, you have all kinds of options to make adjustments to your strategy as you compare where you stand and where you want to be at retirement, such as making changes to your savings rate, your asset allocation, your retirement age, your wealth accumulation goals.

In my recent paper, Getting on Track for a Sustainable Retirement: A Reality Check on Savings and Work, which will be included in the upcoming October 2010 Journal of Financial Planning, one of my major points was to emphasize just how difficult it is to know whether you are on tracking to meeting a particular wealth accumulation goal at your retirement date.  That is why I encouraged a different approach toward knowing whether you are on track, which is explained in that paper.

Since then, I have been playing around with some simulations that sort of anticipated John's comments.  These matters are only half finished, but I thought I could share a few figures now.  

This analysis is based on someone with a constant real salary saving for their retirement.  I use Monte Carlo simulations based on Ibbotson Associates's SBBI data for 1926-2010, for the S&P 500 and intermediate-term government bonds.  For 5 different asset allocations, the portfolio arithmetic returns and standard deviations are as follows:

% stocks   expected real return   standard deviation
  0          2.52                  6.84
  20         3.75                  7.09
  40         4.99                  9.43
  60         6.23                  12.76
  80         7.46                  16.49
  100        8.70                  20.39

Again, my purpose for working on these simulations was to demonstrate how hard it is to know if you are on track for reaching a wealth accumulation goal.  In this case, the goal is to be able to replace 50% of your final salary using fixed real withdrawals from your portfolio and using the 4% "safe withdrawal rate".  This means you need 12.5x your final wealth.
Now, this first figure shows the case for someone who plans to work for 30 years.  Assuming they will contribute to their savings at the end of the year for 30 years, based on the expected fixed geometric returns they use an excel spreadsheet to calculate the savings rate they need to use that would provide them with 12.5x final salary at their retirement date. What this figure figure shows is, for each different asset allocation, the distribution for the number of years it takes to actually reach their wealth accumulation goal.  Because stocks have higher returns and higher volatility, a lower savings rate is required on average, but you can also see that the distribution for the career length is also wider.  For something that you "expect" to take 30 years to happen, it could actually take anywhere from 15 to 60+ years.  Planning to reach a goal is harder than it looks!

That is assuming that the person doesn't make any adjustments during their career.  So far, I've gotten around to considering two other adjustments.  The first is to adjust the savings rate depending on your progress toward your goals, and the second is to adjust your asset allocation depending on your progress toward your goals.  This work is still unfinished and there is more I want to do, but the following is what I have so far. It's not complete, but what this is meant to show is how easy it is to be "tricked" into thinking you are or are not on track, compared to what actually ends up happening. As well, it shows how someone who is trying to carefully target in on their goal, just how hard it can be even with a willingness to make large annual adjustments to their savings rates and asset allocation.

Let's consider someone working over a 30-year career and trying to aim in on the 12.5x wealth accumulation goal at their retirement date.  They are going to make an earnest effort at doing this.  Let's just consider the case for now of someone starting with a 60/40 asset allocation.  Here is what he does:

1. For the first five years of his career, he does save more than required to help get him off to a good start (this "more" is the 75% percentile of savings rates needed to get to the goal from some separate simulations I haven't specifically discussed).

2. After that, the savings rate he uses is what he gets from using the expected returns for his portfolio and comparing his current wealth accumulation to the final goal.  If he is ahead of schedule (the blue line is above the red line in the top part of these figures - though that red line is specifically for 60/40 and his asset allocation may diverge from that), he can save less.  If he is behind schedule, he saves more.  His savings rate is allowed to vary between 0 and 40%.
3. Regarding asset allocation, I have built in a way for him to lower his stock allocation from the starting level, but not to increase it. The rule is, if he finds that his required savings rate is below zero, he starts checking the savings rate needed with lower stock allocations until he finds another positive savings rate.  Then he uses new asset allocation.  This can happen down to 0% stocks.  Again, I have not yet built in a way for him to increase the allocation after that, but I need to do that too.  Also, about locking in a goal with all bonds, he really should buy some bonds that will precisely get him to his goal.  But so far I still only have him investing in a bond mutual fund that still has some volatility.  This can cause him to potentially fall behind schedule again from a point where he should be able to lock in a goal.

The point again, of showing this, is to show how difficult it still is to reach the goal, despite allowing for very volatile savings rates.  While the volatility of final wealth is not so much now, just consider, still, how difficult in would be to use these different savings rates each year that you really cannot plan for in advance.

Despite doing all this extra work, the distribution of wealth outcomes is still rather wide, compared to the goal of a 12.5x wealth accumulation.  Here is about this distribution from 1,000 simulations:
Summary Statistics for Terminal Wealth
Mean:              12.438
Minimum             6.695
5th Percentile      9.039
1st Quartile       11.463
Median             12.498
3rd Quartile       13.512
95th Percentile    15.656
Maximum             21.61
Std. Deviation      1.902

Here are a few sample illustrations of this from the Monte Carlo simulations.

This first one shows perhaps about as good as one can possibly expect to do. This person was fortunate to spend most of his career ahead of schedule.  He was able to lower his stock allocation to make it easier to lock in his goal.  He did still have somewhat volatile savings rates, as he could start lowering them between years 5 and 15, but then got some bad luck that pushed him off track compared to where he needed to be (the red line is the expected wealth path for 60/40, and after switching to 0/100 he needs to be higher at any given moment.  In the end though, he did make it and achieve his goal.  Not everyone is so lucky.

Consider this guy.  He was slightly ahead of schedule for more than 20 years.  Never enough to lower his stock allocation, but his savings rates could drop quite a bit and were close to zero for a while.  Then, sometime after year 20 some bad market returns arrived which pushed him off track, forced him to increase his savings rate to 40%, and still left him quite far below his goal of 12.5 after 30 years.  He seems to be at about 7x his salary. 

Here is another case in which the goal was reached, but savings rates were pretty volatile in the mean time.
And just for fun, here is the case of the person with the lowest final wealth accumulation of anyone:

And the person with the highest final wealth accumulation of anyone: