Consumption Smoothing with Social Security
I am testing a profile where retirement account withdrawals start at 60 and social security starts at 70. ESPlanner results show a much smaller consumption rate for the years before social security kicks in because the retirement account withdrawals are constant before and after social security. How do I get the program to smooth consumption by spending more of the retirement account before social security kicks in and less after? I know I can force some manual smoothing by lying to ESPlanner and telling it I will be annuitizing part of the retirement accounts for a fixed ten year period. Surely this sort of workaround is not required to get a consumption smoothing program to do something this basic. Please help me understand what am I missing?
RSS
This has been discussed elsewhere and you're not missing anything. We don't allow anything but smooth withdrawals from beginning to end. That call was made long before I came on board. I've been arguing in favor of being able to make explicit withdrawals from retirement accounts on a year by year basis but haven't been able to gain a lot of traction although the last time we discussed this I think I got some agreement.
Best,
Dick Munroe
I talked with Larry about this the other night. He suggested playing with the date of last withdrawal, pulling it in towards today. The program then takes more money out of retirement each year, socking into savings in the years before social security kicks in and then taking it out as needed. With a little trial and error I came up with a nice smooth consumption. The magic number for last withdrawal for me was 78. The problem is that, due to increased taxes, the average annual consumption level is much lower than without the smoothing, and even lower than doing smoothing by taking a portion as an annuity.
Of course, you realize that this workaround is not simulating the reality of what I want to do - take just enough extra out of the 401k each year before social security kicks in so that yearly consumption is level. I guess your idea of allowing explicit withdrawals from retirement accounts on a year by year basis would work, but ideally, ESPlanner would make the calculations for you.
I also tried another approach - I took part of my current 401k balance and told ESPlanner it was my wife's. Then by playing with the mix of how much was my $ versus hers, and adjusting her first/last withdrawal dates I was able to get some interesting results. They were not quite as smooth, probably due to spousal social security benefits, there was still retirement funds being stashed in savings and the process was messy (I haven't fully conquered it), but it was the highest average annual consumption I had seen. Of course, I can only play with that work-around because my wife does not have her own 401k.
This particular smoothing task (401k and social security) is a very basic concept that is entirely too difficult to simulate with ESPlanner. The program limitation of "smooth withdrawals from beginning to end" is keeping it from delivering the real goal of "smooth consumption from beginning to end". ESPlanner should be looking at other sources of income such as social security, pensions, special receipts, etc. and adjusting withdrawals automatically to smooth consumption. "Consumption smoothing synthesizes all of your life resources and calculates a time path of withdrawals from financial resources that will sustain a smooth living standard." ("Spend" pg 253.) I don't need a $200 program to calculate "smooth withdrawals from beginning to end", I can do that with a 4% rule of dumb. I hope my arguments help you with your traction issue.
Thanks.
The automatic adjustment of draws from the retirement accounts isn't ever going to be possible for technical reasons.
Essentially it boils down to the way the algorithms work. In order to do this we would have to introduce the retirement accounts and the draws as state variables. For every additional state variable the computational load goes up dramatically (it's at least geometric, so you at least double the computational load for each new state variable). 10 years ago, ESPlanner took 15 minutes or so to run. Without fundamental changes in the algorithms, the 30 seconds and under times for the Main reports and the under 2 minute times for a full Monte Carlo simulation are solely due to Moore's law in action. Computers have caught up with number of computations necessary to deliver the reports ESPlanner needs to generate in a reasonable time.
So more state variables, more memory, more computes, more expensive computers to run ESPlanner on, and more time necessary to generate reports (computers can only get so much faster).
I can envision a time when we have a large enough customer base so that we can afford to bring up a web version of ESPlanner on truly outrageous iron. ESPlanner could have substantially more state variables and a number of other algorithms that we haven't been able to use because we simply can't get them to run fast enough. But that's a discussion for another day.
What we can reasonably do is to allow you some finer control over the draws. Not automatic, but it doesn't break the bank from a computational perspective by continuing to make the draws exogenous. We've been kicking this around for a while and it is a feature I want to see but I haven't been able to get general agreement (although I'm getting closer).
Best,
Dick Munroe
That is too bad. It would seem on the surface that a two stage answer for retirement withdrawals - one before and one after social security - would not be that difficult or computationally intensive, but you are the expert. For the record, I'd trade a couple more minutes of run time for a more useful answer.
The program has so much potential, but if it cannot compute a smoother consumption for a basic social security and retirement account scenario, then I really think the book oversold itself and the program under delivered.
Best of luck in your ongoing endeavors with this.
Dan
Well, this is something that we'll have to agree to disagree on. As I said, there are a number of things that we can do to make it possible for you to manage draws yourself but we won't compromise the commercial viability by slowing things down dramatically.
Even given this issue, the results you're getting are far superior to those of any other planning tool available. Yes, there are limitations. This is one of them and exists for sound technical reasons. We can, as I said, address some of these issues by providing finer control over retirement accounts, but we will always be making these decisions within the larger context of our business and unless we get a lot more push from the customer base this particular issue will not be addressed "quickly" as much as I would like to see this dealt with. I've been working on this issue for two+ years now but there are only a small handful of users (countable on the fingers of both hands), including the FPs, who feel this is anywhere close to being a show stopper so I don't have a lot of leverage.
So if you want to see some of the finer control features implemented, you'll have to drum up some user support and communicate with Larry to get the decision made to actually commit resources.
Best,
Dick Munroe
No disagreement on speed - it needs to be fast. I just meant for me, if it had to be a little slower to give a good answer, *I* could live with it.
I have no idea how large your install base is, or their makeup in terms of background or expertise. Based on your comment, your current audience is obviously happy with the program. So I must be out of touch.
It seems to me you have created a fine tool for those clever enough to recognize the limitations of the ESPlanner results and who are also smart/patient enough to navigate the workarounds and understand even their weaknesses. If that, and the folks who don't know any better, is your target audience, then you are golden.
However, I had figured Larry was trying to reach a broader audience with his book and I assumed I was raising legitimate concerns for that broader audience. I feel that audience will be disappointed with a consumption smoothing program that cannot even smooth a simple 401k and social security scenario.
I'm sure you know that even if as you said, ESPlanner is far superior, there is always room to raise the bar and always someone wanting to steal your lunch. Being a programmer myself, I know that sometimes "sound technical reasons" is code for "I haven't quite figured it out yet". With no one pushing it will not get figured out. (Apologies, not trying to lecture.)
Unfortunately, I have no idea how to drum up user support. I haven't even drawn a lurker into this thread. :)
Dan
I for one would love to be able to input how much to withdraw from my IRA every year either as a percentage or $ amount. I can't figure out a way to do this in the current program. I can only do it for the current year. It seems that the increase in run time for this approach would be minimal. There would need to be a subroutine that determines if the Required minimum distribution rules have been met but that subroutine must already exist.
Just my opinion
Thanks
James
As long as the draws are exogeneous (outside our control or on autopilot) then you're right, there is no additional run time cost incurred. Once you make the draws part of the smoothing process, all bets are off.
We continue to consider allowing explicit draws from retirement accounts as a new feature. The more folks (and thus far, there really aren't that many demanding this) that want it, the more likely it is to happen.
Best,
Dick Munroe
So it would take more processing time to automatically smooth to “x” after age 70 and to “x minus social security” before 70?
As long as we have to smooth things, automatically, yes. The problem is the desire to smooth the draws dynamically (rather than exogenously which is what happens now) which interact with the taxes which interact with the regular assets which interact with the survivor states which interact with the life insurance which interact with ...
The interactions are what increase the computational load and we're pushing the state of the art at the moment in terms of getting things done. It's not as simple as subtracting social security.
Best,
Dick Munroe
I have been handling the SS vs. Roth vs. 401k withdrawal optimization separately on a set of spreadsheets that use the ESP consumption and spending as inputs- goal is no income tax on anything after retirement. Having just updated to 2.14, it seems there may be a way to hack this using the 529 function since it is driven by the withdrawal schedule you input and then smooths the contributions over time. Have not yet tried messing with this and haven't seen what the tax implications would be, but it might be a place to start for the power user who wants this functionality (although it would be semi-automatic at best).
Regards,
Bill
The 529 contributions aren't smoothed either. Like the draws from the retirement accounts, they too are exogenous and are held constant in real terms from "now" until the year by which you say the expenditure is to be fully funded. Your economy is smoothed around the expenditures necessary to fund the 529 plan.
Best,
Dick Munroe
Oh well, thought I was on to something.
I also have a separate 529 set of spreadsheets to calculate the contributions and state tax benefits in each year. Since I am liquidity constrained in most years until my kid is out of college, I manually adjust the yearly 529 contribution to meet my overall target, maximize consumption, and minimize regular asset savings and then input into ESP as special expenditures. Not sure if the ESPlanner 529 will offer added functionality, but that is a discussion for another thread.
Back to this topic, as noted elsewhere, changing the amounts and withdrawal dates for your and your partner's retirement accounts allows a two-stage retirement account withdrawal rate at the expense of disregarding your actual retirement assets. You can back into what would be required in current assets to maintain a certain standard of living in the future. Not sure if that would help with your SS question or not.
Regards,
Bill
Dick,
I like to add my hat to the list of customers seriously looking for a feature to optimize the retirement account withdrawals. I noted this earlier this summer when I first started using the product. The product is great, but this one point is really throwing me off for planning purposes. Appreciating the computation issues of doing this automatically I could work with a manual override. I'd greatly appreciate a change that it would adjust the retirement account balances +/- and calculate taxes consistent with the adjustment. In my case I will draw extra for 4-6 years before I collect SS and will also reduce my withdrawals after selling a cottage much later in retirement. All that goes to say that I'd like to be able to make adjustments in any year. The good news is that I can see there is upside for my consumption - I just haven't figured out how much.
If there is anything in particular that I can do to identify the extent of the customer needs/requirements to justify the upgrade I would be willing to assist
Thanks
Jeff Wortman
Looks like y'all are going to get [at least some of] your wish. We are beta testing and will release in this update cycle (2009-11-01) a mechanism by which you can specify year by year withdrawals from your retirement accounts. We're finalizing the user interface now and have tested the alpha versions successfully. Excluding the tax issues associated with drawing from your retirement accounts BEFORE your year of first withdrawal (which I believe we will be allowing but not calculating the tax consequences of) you specify a withdrawal, that withdrawal gets made (possibly modified by the MDRs) in that year and the remaining balance of the account gets withdrawn smoothly from then on. If you make another special withdrawal later on, that withdrawal gets made and the account balance gets withdrawn smoothly from then on. The assumption the code makes is that there are no further special withdrawals and that withdrawals are to be made smoothly once a special withdrawal has been made.
As per earlier discussions the process of figuring out the amount of special withdrawals is not and will not be made automatic, but we feel that we've covered most of the bases for those folks who want to deal "differently" with their retirement accounts.
Best,
Dick Munroe
Thanks Dick - I appreciate the effort to address this and will put it to good use
Jeff
This is a bit late, but FWIW, the new special withdrawals feature meets my needs nicely enough. Thank you.
You're most welcome.
Best,
Dick Munroe
Ditto. I increased my consumption by almost $1K per year by frontloading my IRA withdrawals - drawing out more before SS kicks in. Interesting that this is the exact opposite of some conventional advice, which is to delay withdrawals on tax defered accounts.
Well, as usual the "advice" doesn't factor in all the issues. The big problem is the taxation of the SSA benefits. Consider the case where you're drawing both RIB and from your regular IRA at the same time. You're paying taxes on both income streams and you've lost the benefit of any delayed retirement credit you might have accrued by delaying taking benefits. If you can delay taking your RIB by drawing money from your taxable IRA you win both ways, paying taxes on only one stream while increasing your income by 8%/year (more or less) by delaying your RIB.
It should be a big win (not that I've actually run a case) but a little algebra will tell you that your RIB doesn't have to be very big ($12,500) to get and extra 1K/year in income for every every year you delay taking your RIB. Even adding in reasonable taxes on the extra income, it should still be a pretty big win if you can delay a couple of years.
The caveat here would be that your RIB is really an annuity and won't go down. If that assumption doesn't hold all bets are off.
Best,
Dick
Dick: Regarding your 9 Oct 2010 post--What is RIB?
Sorry, Social Security Administration for Retirement Insurance Benefit, i.e., what most people think of as their SS income.
Best,
Dick Munroe
Hi, You can also try converting your regular IRA to a Roth by entering special withdrawals from the regular IRA and entering contributions to your Roth in the same year. The reason this is relevant to the Social Security benefit taxation issue that Dick raised is that Roth withdrawals aren't included in your Adjusted Gross Income.
If your AGI is high enough in years you are taking your Social Security benefits, first 50 percent and then 85 percent of your benefits will be subject to federal income taxation.
best, Larry