Group Think: Strategies for ESPlanner
I’m looking for some collective wisdom on how to best tackle “predicting the future.”
First off, I know that predicting the future cannot be done; but what can be done is to create a collection of scenarios to see different potential paths. So let me construct one strategy and let other forum members comment and improve upon it.
The Issue: While everyone’s personal situations vary, we collectively face a future of uncertainty at the macro level. There is much talk about future tax rates or inflation rates changing; but what is the range of impact to our future personal economic plans (as laid out by ESPlanner)?
The Goal: Create a strategy for modeling how sensitive our personal plans are to larger economic “winds of change.”
One Possible Plan: Even though I have a relatively fast computer, I do find that the calcs with the Monte Carlo feature turned on takes a considerable amount of time. So I am suggesting the following:
1) Start with Monte Carlo turn off. Do the macroeconomic modeling first. Then come back and fine tune the model with the Monte Carlo feature on.
2) Next, set up a “matrix of change.” This is where I’m looking for some collective feedback: what combinations should we be changing. The number of variables quickly grows exponentially so come intelligent pairing of change agents would be optimal.
The items that would appear to be pertinent are:
• Inflation
• Tax rate Changes (Federal, Payroll, State)
• Social Security Changes
• Housing market variations
• Nominal Rates of Returns
(Since I'm recommending turning off Monte Carlo and then follow up with it back on when fine tuning, do I even care about changing the investment nominal rates of returns: meaning that this should be excluded in our sensitivity study?)
Question #1: What is missing?
Next: can someone provide thoughts of what combinations make sense?
Also, do we even think about adding in the dimension of future change (ESP lets us "pick the year" of changes in taxes, social security. My mind already aches with too many variables!)
A follow up post will provide a template to drop in your ideas. But note that just this post alone will show how many variables exist in my limited view of the world. Ideas and strategies on optimum pairs or combinations would be greatly appreciated!
RSS
Matrix for changes (a template) - Put in your recommendations starting with the low and high ends of the range, and the recommended increments to run through the scenarios.
• Inflation
Bottom of Range:
Top of Range:
Increment Steps:
• Tax Rate Changes - Federal
Bottom of Range:
Top of Range:
Increment Steps:
• Tax Rate Changes - Payroll
Bottom of Range:
Top of Range:
Increment Steps:
• Tax Rate Changes - State
Bottom of Range:
Top of Range:
Increment Steps:
• Social Security Changes
Bottom of Range:
Top of Range:
Increment Steps:
• Housing market variations
Bottom of Range:
Top of Range:
Increment Steps:
Does anyone have any clever tricks for capturing the essence of change?
While I could carefully archive each 100+ PDF report, I’m looking for a few key indicators to capture in a table (Excel) that summarizes the difference with each economic scenario. Goal is to capture the minimum number of “results†so as to compare across multiple scenarios.
Here are my straw recommendations:
* Living standard per adult (at 10/15 year intervals into the future)
* Total income (at 10/15 year intervals)
* Net worth (at 10/15 year intervals)
Should I care about anything else? (Hate to run through a 10 or 30 scenario study and then find that I should have recorded other items of change.)
At one point I ran a series of what I think of as "sensitivity" experiments. I altered a single variable at a time to see how sensitive my plan was to changes in that variable. For some of the variables that the plan turned out to be sensitive to, I ran a number of scenarios, altering that variable across a range (e.g., start withdrawing from IRA in year A; start withdrawing in year A+1, etc.). I some cases, ten years was the appropriate step, in other cases one year. In some cases (e.g. future tax increases) the plan proved to be far less sensitive than I would have guessed, and so I didn't even bother altering across a range of values - one experiment was enough to tell me what I wanted to know.
All of this was done, as you've suggested, with Monte Carlo turned off. The point was not to get real results, but to find out what variables had the greatest effect, to narrow down what to try altering as I tried to come up with a real plan.
Your plan will probably be sensitive to different variable than mine was, but some of the things I tried were:
rate of return on real estate
when I chose to downsize
when I chose to retire
when I chose to take Soc Sec
when I chose to begin ret withdrawals
when I chose to end ret withdrawals
effects of different rate of inflation
how much my house will sell for
Based on the results of these runs, I created a new "baseline" profile, and used it to run a series of Monte Carlo sensitivity scenarios to see how different asset allocation strategies performed.
Now that I've found an asset allocation strategy, I'm re-checking some of the original sensitivities with my new asset allocation strategy.
A lot of work, but worth it if this planning allows me to retire early.