Key ages and smooth withdrawals


For the benefit of others, I'll share an approach I just devised that seems to work well for me. Instead of assuming smooth withdrawal over my lifetime with an expected age of 100, I'm setting my smooth withdrawal end date to the year I'm 71. That's the year after I start withdrawing SS.

This works for me because my wife and I have retirement accounts of approximately equal value, she's a few years younger than me, and will continue working 3 years after I retire. In the current scenario, that's early 2018 for me and early 2021 for her.

If I set the last year of withdrawals at age 100, the value of the smooth withdrawals is far too low to provide for our needs in the years between my retirement, and my starting SS. An important fact is that we're very constrained as to regular assets, but I think that's typical of a lot of people.

Anyway, by setting my last year as 71, all of my retirement accounts are exhausted at that point but the higher withdrawal rate supports us until SS kicks in. My wife will start drawing SS at retirement at 62, at which point I also get my spouse benefit. We'll draft her retirement accounts smoothly over her lifespan.

So, my point is, if you have a spouse/partner with retirement accounts you can rely on, consider shortening the smooth withdrawal period for one or the other. It made my plan much easier to manage. Without that I had to use special withdrawals for the first 7 years and, as others have described, had to play whack-a-mole to avoid hitting 0 on real assets.

I hope others find this useful.



dan royer's picture

It's very helpful. Interestingly, that's our situation too. My wife has a defined pension (I do not) but I have more 403b assets. So we're compressing her 401K into about a ten-year period so that it smooths that near-term constraint and gets us to age 70 when the bonus to SS kicks in.

I don't think it's unusual that you have the same constraint. Many families do. The advantages of putting money into tax-deferred savings leads most middle-class folks to put the preponderance (or all) of their savings into those assets.

It's clear that's what we should do, but ESP doesn't handle the lack of "regular" assets very elegantly. Considering the number customers in that situation, I suggest that management direct some effort into improving it, in that respect.

In my opinion, attempts should be made to optimize retirement withdrawals with a goal of smooth standard of living, not so much smooth withdrawals. Consider earned income, pensions, real estate, special receipts, etc., all fixed income streams as entered by the user. Use regular assets as the first source of a buffer, as it does now. But if ESP can't keep SOL level with that, it should adjust retirement withdrawals to do so.

Those retirement withdrawals would be governed by the key ages, as they are now. That's the control the user could exert. Maybe an option button in the settings to "allow variable retirement withdrawals". But the goal would be to eliminate the Whack-A-Mole that's required in regular-asset-constrained economies. Those cannot be considered anything except normal.

That said, compressing the consumption of one or the other of the partners' retirement accounts into a shorter period is an option, currently. (I suggest you include that in a video, or tip sheet.) An important caveat is to be careful to not leave yourself without a savings cushion, should economic events not take the direction you anticipate. Single users without recourse to a partner's assets would be well advised to not do that.

Thanks Chris,

There are a bunch of tweaks like this that can be made. In the optimization challenge, for one spouse I set the age of first withdrawal to 1 year before EOL. This skewed the spouse's tax-sheltered withdrawals to match RMDs so they were no longer smooth. In turn, for this profile, it helped reduce taxes and increase consumption. This is exactly the sort of thing I was hoping to get out of the optimization challenge. Thanks for sharing.

Re: most people putting the bulk of their financial assets into tax sheltered accounts, this is one reason for younger ESPlanner users to test Roth, regular asset growth, etc. to try for some level of balance. That will make more options available near retirement.


Brian, I tried your RMD approach. I guess it could work if I had more regular assets, but I don't, and it doesn't, for me. If I did have more regular assets, I probably wouldn't be compelled to play with retirement withdrawal rates in the first place.

What does benefit me is to play around with my age of last withdrawal. Oddly, in one of my profiles where I work an additional year, I hit my peak SOL by stopping withdrawals only 2 years before EOL. Changing that to 1 year earlier or 1 later drops the SOL.

Thanks for your insights.

With respect to whether to put money in pre-tax or post-tax retirement accounts, it seems like ESP users looking for smooth withdrawals might favor one over the other, as suggested by the results. But that limits optimization of smooth SOL by requiring smooth withdrawals. I'd rather allow variable withdrawals, if it improved SOL. Unfortunately, ESP doesn't support that. If it did, the result of which type of retirement account you put your investments in might differ.

Hi Chris,

If I may, it appears you are drawing some conclusions without enough info. On the old forum, there were some old posts which explained why regular asset withdrawals are varied to smooth consumption while trying to keep tax-sheltered asset withdrawals relatively constant (per profile and RMDs). Essentially, the computational complexity of ESPlanner's algorithms led to very long compute cycles with all variables free. This was sufficiently long that ESPlanner may not be a viable product. Meaning running the reports would take a really long time instead of the few seconds they do now.

To cope, tax-sheltered asset withdrawals were typically kept smooth or nearly so. Now in the past several years, various enhancements have been added, such as those we've been discussing, that do allow some form of varying tax-sheltered asset withdrawals (e.g. special withdrawals, key ages early/late/other combos, etc.). It is up to the user to experiment to see which combination results in better results, hence my interest in optimization.

You are on the right track digging in like you are, experimenting, using the forum so you'll get more out of the software than most.

Specifically regarding investing in tax-sheltered/regular assets, try experimenting with several new (test) profiles/families with varying levels of assets and asset types. You'll see that having some balance in assets (and types) does give the user more options and can boost (optimize) consumption with a larger range of levers and outcomes. Of course, each one is different and constrained in one way or another. But you'll learn a lot about what optimization tweaks work best in your specific case.


I'd be okay with overnight processing. That could be an option with 37 layers of "are you really sure you want to do that?" dialog boxes, default = no. Doing nothing, it operates as it does now; check the box and it frees the program to optimize retirement withdrawals, too. I'd probably enable that rarely after I'd gotten close to what I thought was my ideal plan, then let ESP tweak it.

Of course, I have no idea how difficult that would be, programmatically.

We use cookies to deliver the best user experience and improve our site.