I am happy to announce that these concepts are now available as a way to help build a financial plan in inStream, a new comprehensive web-based financial planning software available for financial planners and advisors. I serve as their chief financial planning scientist.
The basic thesis is that it is very difficult in practice to do retirement planning in the traditional way:
- Figure out a budget for how much to spend from financial assets in retirement
- Decide on a sustainable withdraw rate based on what has worked in U.S. history
- Figure out "The Number," which is the wealth accumulation target for retirement that allows the safe withdrawal rate to support the desired expenditures
- Figure out how much to save to get to the wealth target by the retirement date.
As for Step 4, which Michael Kitces explained earlier at his blog in what ended up as the cover story for the December 2013 Journal of Financial Planning, the problem is that it relies too much on wealth continuing to grow at a given compounded rate of return even as the retirement date approaches. Wealth in stock and bond funds generally doesn't grow at a fixed rate of return, and Michael suggested that this means that one's actual retirement date will be uncertain as one has no idea when they will actually be able to meet their wealth target. With "safe savings rates," I went in another direction to suggest that retirement will be feasible at the planned retirement date as long as one saves consistently from their salary in a way that gives a sufficient high probability that they will be able to meet their spending goals regardless of the wealth accumulation at retirement and the withdrawal rate actually needed to meet the spending goals.
As well, the idea that one needs to save $1 million dollars (or whatever) can be very daunting and abstract. With safe savings rates, the output is much clearer and can actually be acted upon: you need to save x% of your salary for your retirement plan to be sufficiently sustainable.
With "safe savings rates," there are just two steps:
- Same as above
- Figure out what savings rate to use to have a sufficiently high probability of being able to meet one's retirement spending goals.
Let me provide a basic example to illustrate the idea using inStream.
Our individual is 37 years old. He hopes to retire at age 66 and makes plans to live through 100. His goal is to spend an inflation-adjusted $65,000 per year in retirement. He has some flexibililty which manifests through a willingness not to take inflation adjustments for his spending in years following a negative market return for his portfolio. This decision about distribution rules leads to the following:
He currently has $50,000 in his Roth IRA and plans to contribute the maximum allowable through retirement. He also has $25,000 in taxable assets. He bases his plan on the expectation that he will earn $90,000 per year in inflation-adjusted terms through age 65. His plan calls for an 80/20 stock-bond allocation until age 50, then a 50/50 allocation until age 65, and then a 40/60 allocation from ages 66-100. Portfolio management fees are 0.2%.
The question is: How much of his salary should he be saving in order to have a 90% chance that his wealth will not be depleted by age 100?
The answer is... drumroll...
16%. That's the savings rate through age 65 that will give him a 90% chance of sustaining his plan through age 100. The above image also shows more details about the distribution of his spending. Remember that the portfolio freeze rule is turned on, so spending does not adjust for inflation after years with negative returns and in the 10th percentile of the distribution, his spending fell from $65k to $58k because of this.
If I turn off the portfolio freeze rule, then we are back to the classic inflation-adjusted spending amount. In this case, the safe savings rate increases to 18%:
And so now, it is now possible to apply safe savings rates in real world practice. It's no longer just a program I've written on my computer. As the title of that second article implied, I hope this can help give people a reality check about the feasibility of their retirement plans.
This is great stuff Wade, congrats. Very happy to see that innovations in planning research is working its way into financial planning software.
ReplyDeleteThanks!
Deleteyes, congratulations Wade. I am sure that is going to be an extremely helpful tool for retirement planning.
ReplyDeleteI see the software is for advisors /planners. Do you envision that one day there may be a safe savings rate calculator ( a truncated version of the full planning software) that could be accessed by "do-it -yourselfers"? Something without a $1,200pa price? I suspect it would be greatly valued by many: the go to resource for savings considerations.
Derek, Thanks.
DeleteYes, I'd like for such a thing to happen, though I don't have any immediate plans for how to go about doing this. You definitely have a good a good idea!
It could be "InStream Basic" along the lines of laurence kotlikoff has ESPlanner Basic that I am sure promotes interest in the paying service:
Deleteso maybe a marketing idea for InStream
Your blog is very nice. Every best article need best writing skills..great
ReplyDeleteA very nice integration of retirement savings and drawdown strategies!
ReplyDeleteIt seems to me that the more years (or decades) between the present and the time a person plans to retire, the more uncertainty there would be in specifying a realistic Initial Distribution amount. For instance, with how much confidence can a 37-year-old state that $65K (present value) is a good starting target? Over the ensuing decades, there could be significant changes in how Social Security and Medicare benefits are computed. There could be unexpected health shocks, changes to marital status, etc. etc. And during the 28 years prior to age 65, the individual could advance along the hedonic treadmill to a point where that $65K seems very insufficient. While it is crucial to specify a starting Initial Distribution, it will need to be revisited and perhaps adjusted at least several times prior to retirement. The challenge, however, is that any upward revisions to the Initial Distribution rate that are made within a decade or so of retirement will potentially require uncomfortably large additions to the savings rate; it may even be impossible to catch up without delaying retirement.
Thank you. I agree with you about the difficulty of building a long-term plan. One would want to do it conservatively and perhaps save a bit extra as a buffer. Because the problem you indicate, that someone changes their retirement goals near the end of their career to seek a more expensive retirement, would be a problem for any efforts at building a retirement plan.
DeleteOne of the aspects of inStream that will be great for research purposes is that we can start to track individual behavior over time, and as the data comes in, we might start to figure out of what you've suggested has been a problem for a lot of people. This could lead to the development of guidelines to help people figure out what sort of long-term budgeting assumptions are appropriate. It will be the CRSP database for personal finance. But it will take time to collect this data.
There are two problems I see:
ReplyDelete1. Assuming he is withdrawing $65,000 (present value) at age 66: If this initial $65,000 withdrawal is for essential expenses, then foregoing the COLA increases for any reason (in this case, when portfolio return is negative in the previous calendar year) can be devastating for the retiree.
2. With inflation freeze turned off, when using my aftcast calculator, which is based on actual market history, I get different results: He needs to save $34,000/year (indexed to CPI), or about 38% of annual earnings.
3. Using total historical returns with a 0.2% management fees is unrealistic. Most people pay a lot more than that in total portfolio and management fees, one way or another.
4. In your calculator, you seem to have included the historical dividends. Historically, dividends were about double of what they have been since 1990's. It is misleading to use them going forward. By the same token, ditto for the bond yields, which might (or might not) remain low for years to come.
5. Here in Canada, this person would have been allowed to save (either on a tax-free or tax-deductable basis), about 34% of his income (18% in tax sheltered retirement account, 6% tax-free account and 10% in government pension benefits, which is a lot closer to the figure I have found with my aftcasting.
Jim Otar
Jim, Thank you for your comments. I admire your work. Let me respond:
Delete1) I agree if those are all essential expenses. I wrote, "His goal is to spend an inflation-adjusted $65,000 per year in retirement. He has some flexibililty which manifests through a willingness not to take inflation adjustments for his spending in years following a negative market return for his portfolio." With this, I'm trying to get at the idea that there are some tradeoffs between essential and discretionary which are being weighed. The idea is the more you are willing to cut if things go bad, the higher initial spending rate (or lower savings rate) you can use.
2. You've had me worried that there was something wrong with the program, but after trying a few alternatives, I do not think there is a problem with my calculation. Just to be safe, I ran a similar example through my original safe savings rate program that I wrote for my research article which used aftcasting to get at the results. So it should be quite similar to what you've done. I didn't do the variable asset allocation. I used this simplified example: work for 28 years, retire for 35 years, current assets as a percent of salary = 75,000/90,000, retirement income replacement rate = 65,000 / 90,000 and the portfolio freeze rule is turned off. Salary is constant in real terms. Fees are 0.2%. If the asset allocation is a fixed 60% stocks, then the worst case scenario in history since 1871 is a savings rate of 23.5%. It was for a 1918 retiree. I'm quite confident about this backtested number. 38% otherwise seems way to high. This retirement plan isn't that ambitious. Are you sure that you don't have any sort of wealth accumulation target and safe withdrawal rate built in? That would raise the number.
3. I'm implicitly assuming a Vanguard customer in index funds, but your point is well taken. To make the example more applicable, I should assume higher fees.
4. I'm with you on this. Basing our forward looking projections on historical averages is a terrible idea for the reasons you suggest. There will still be refinements made to the default capital market expectations, and users are free to set their own assumptions.
In this example (above): Is part of this $65,000 income (in current dollars) paid by the social security? Or, is all of it coming out of this portfolio with starting total value of $75,000 at age 37? Regards, Jim Otar
DeleteJim,
DeleteThere is no Social Security in this example.
Wade