Flexible Retirement Planner Support Forum

Hi Jim,

I've setup a scenario where I run it at Below/Moderate/Above Average styles. They'll come back with a success probability of 99.8,98.9,95.8%. I run those without the back to work turned on.

I then turn on back to work and rerun for the 3 investment types, and the probability of success barely changes (99.9,99.7,98.0%), but I have a back to work probability of 70-90%?

I've tried both flexible and stable spending rules. I'm not sure I understand why I would have such a high probability of back to work, when I only had a small number of failures.

I have the back to work set at 85% of portfolio for 2 years. I would think this should put me at 100% success?

Any insights?

Thanks

Generally with this sort of stochastic modeling exercise, there's isn't a meaningful difference between a result with a 98% probability of success and one with a 100% probability of success. There's a built-in imprecision to tools like this and this is just the nature of this type of modeling approach. The built-in imprecision comes from guaranteed errors in estimating plan inputs along with necessary simplifying assumptions in the model like treating returns as if they were normally distributed. So the caveat is just to not read too much into small differences in the results.

That said, my best off-the-cuff guess for the reason you're seeing little or no impact from the back-to-work feature would be because you're only allowing 2 years of back to work. It'd be interesting to see how high you'd need to set the back to work years to knock off those last 10-20 failures that you're seeing (and get the prob up to 100).

Also, it's possible that the 2 year back-to-work limit is having an impact, but in more subtle ways. It'd be interesting to see what impact enabling back to work is having (if any) on things like the 10% and 90% portfolio value bands, the median ending portfolio value, and the average spending shortfall. It may be that the influence of enabling back to work shows up in these outputs rather than in the probability of success.

Thanks ! I'll play around with the variables as you suggest