Adding New Use-Defined Assets in Monte Carlo
I'm just starting to play around with user-defined assets.
Why is there no “relative risk” for new funds that I’ve added? Does this matter? Will this affect the way Monte Carlo results are calculated?
Why are the historical returns for some assets shown in real dollars (e.g., cash, large cap stocks, and intermediate government bonds) while others are shown in nominal dollars (e.g., DFA funds)?
Since economics-based planning only allows for one future ROR change, is it OK to use Monte Carlo planning mode to model multiple specific future ROR changes? For example: Let’s say I want to see what would happen if I earned 7% over each of the next five years (2015-2019), then 5% over the next three years (2020-2022), then a 20% bear market correction in 2023, followed by a 6% return thereafter. Could I just set up new assets with “fake” historical returns of 7%, 5%, -20%, and 6%, assign these assets to a portfolio, and then apply the portfolios to each year? Would something like this work on the basic and detailed reports? Or am I missing something in the way the math works in Monte Carlo? I would ignore the Monte Carlo trajectory, distribution, and range reports if I was using this approach.
Mon, 06/08/2015 - 23:16
Taking these in order...
- Why no "relative risk" as an input for UDAs? This is because the mean return and relative risk are now calculated by the program. If you input the series of returns for your asset, you will see those values appear in the window on the Build Portfolios tab.
- Why real returns for some assets and nominal for others? I'm 99% sure it's just the way these were input by the ESPlanner team. For Cash, Large Cap Stocks, REITs, Commodities, etc. they use values from Ibbotson reports as the reference and use the longest time frame available with ten years minimum. For DFA, I'm assuming they just found the series of returns for DFA funds online (which are normally nominal returns) and used the new MC copy/paste functionality to input these.
- For your last question, you'd have to experiment with this. There could be MC issues with some profiles as earlier versions used to be somewhat sensitive to "bogus" UDA values, but that's just a guess. As a test, I did exactly what you said for your series of returns in a test profile. The Basic and Detailed reports seem to show exactly what you'd expect for the sequence of returns. I'd have to play around with this more to feel entirely comfortable with this, but it does seem possible to implement multiple future RoRs with this approach and you even could split these up with different RoRs for regular assets and for each spouse's retirement assets, if you wanted to go into that level of detail. The MC reports (in my test profile) showed very little variation from the lowest to highest results, as you would expect, and don't appear to be too useful in this scenario. I only used 10 years of returns for each RoR level, aggressive spending behavior, and standard precision. This could be a nice way to model a large decline at different points in time. It's not as severe as Upside Investing and much more flexible. Something to think about and play around with more. Good idea. Thanks.