I’m having a hard time understanding consumption smoothing, particularly for negative amounts. I have a conservative scenario >95% Monte Carlo simulation) and would like to better understand how much additional money I can spend while working and in retirement. My current scenario is showing a negative amount during working years and a positive amount during retirement. What is a negative expense? I’m having a hard time grasping this. Feel free to suggest a reading resource if appropriate.
@deftonezzz Hi, this is Charlie Stone with Pralana. You ask a very good question. Stuart and I spent about half hour tonight on a Zoom call discussing and investigating your question and concluded:
- There is a bug in Pralana Online (but not PRC Gold). The pre-retirement Consumption Smoothed Spending (CSS) should be $0 (in your case) and not a negative number.
- Prior to retirement, the CSS amount is reduced by the amount of your Employment payroll contributions to savings. The rationale for this is that during the working years, it is more important for you to contribute those $ to your savings than on extra spending.
- The bug in Online is that the math should be MAX(0, CSS - PAYROLL CONTRIBUTIONS) before retirement.
- We also concluded that the User Manual is pretty vague about this. It says:
- "Pralana will generally calculate a smooth spending level throughout the entire modeling timeframe; however, if the age at which you and your spouse will cease full-time employment (i.e., the retirement date specified on the Build > Scenario Assumptions > Retirement & Life Expectancy page) is still in the future, Pralana will strive to minimize withdrawals from tax-deferred and Roth accounts up to that point. In that case, you will probably observe one spending level prior to full retirement and another spending level thereafter."
- So we will also make this more clear in the user manual.
Thank you for bringing this to our attention. A fix will be in the next Pralana Online release, likely on Sat or Sun, Aug 2 or 3. It will be mentioned in the release notes.
Charlie
thank you!
@deftonezzz Well, I made that fix to not allow consumption smoothed spending to be less than $0 and now I need to revisit it again.
The next release will allow you to specify your own targets for the consumption smoothing algorithm:
- Deterministic: you can specify a legacy amount (e.g. $1,000,000) instead of the default $0.
- Monte Carlo and Historical: you can specify a target success rate other than 90% (you can enter 50% to 98%, currently).
Here is the problem: Suppose your final effective savings is $2,000,000 and you want to find the consumption smoothed spending that will allow you to leave $3,000,000 to your heirs. In this case, the consumption smoothed spending needs to be negative....indicating if you want to reach your goal, you need to spend LESS, not more.
So in some cases (but not your specific example), CS Spending may be negative.
The next release will allow you to specify your own targets for the consumption smoothing algorithm:
- Deterministic: you can specify a legacy amount (e.g. $1,000,000) instead of the default $0.
Compare to Actuarial?
The next release will allow you to specify your own targets for the consumption smoothing algorithm:
- Deterministic: you can specify a legacy amount (e.g. $1,000,000) instead of the default $0.
Compare to Actuarial?
I like the proposed changes, seem like low hanging fruit with nice value for the user. I know I don't like being tied to 90% success with the consumption smoothing.
Consumption Smoothing can be thought of as a variation of Actuarial, though it is arrived at through iteration rather than calculation. Both spending methods aim to take the portfolio down to zero (after leaving aside legacy) by maximizing consumption.
There are some differences: Pralana's version of actuarial evenly doles out the yearly allotment from the whole portfolio each year, and then takes essential expenses from that. (It is like bogleheads VPW in this regard.) Whereas Consumption Smoothing evens out the spending left over after taking into account the income and essential expenses going forward (the equivalent of net present value of future expenses and income). So it's focused on creating steady discretionary spending, whereas actuarial is more like creating your own even "salary" that you then have to use to pay your expenses out of (and with discretionary spending varying year to year depending on essential expenses).
(There are of course other forms of actuarial than Pralana's version that end up being just like consumption smoothing. And there are versions of consumption smoothing (like the Maxi-fi program) that focus on smoothing out discretionary spending rather than total spending.)
I proposed an enhancement awhile back that would allow users to fine-tune the actuarial with a growth factor. It would allow you to fine tune the slope, either up or down, in order to tailor to e.g. you want more at the beginning when you're health or you want more at the end to be safe. The same proposal could in theory happen with Consumption Smoothing: you could take out $n more in the first ten or twenty years and then (by default) consumption smoothing take out a bit more in later years.
The lifecycle investing and withdrawal methodology is like a rabbit hole!
