I was introduced to PRC in February, 2020. I spent a lot of time learning the tool and entering all my data. I copied the results into the Excel workbook I use for current year budget and net worth. Each month, I tracked how close my actuals were to the end of year projections. All good. Now, another year has passed and there is a new version of PRC. I also have new actuals for the beginning of year values in all of my accounts, along with updated values for salary, social security estimates, etc. Here's my dilemma. Last year (2020) was a banner year for the stock market. If I use those values to create future projections, I can't imagine meeting those new targets by the end of the 2021. Should I instead use the previously projected (fixed rate) values as the starting point? It would take advantage of PRC2021's updates, but keep the same baseline, as it were. On the other hand, what makes 2019 end of year values the right benchmark - it was just my first year with PRC. If the stock market drops in 2021, then I would start 2022 with much lower actuals and things could be back to where they started.
What do other people do each new year - try to maintain some baseline with the latest PRC and use that for comparisons with actuals, or start from scratch and model anew?
I'm hoping that actual users will add their own practices here but I'll give my two cents worth as the designer. I think the right approach (and what I do myself) is to update the model each January with updated year-end account balances (actuals) and the latest information you have on assumptions like inflation, rates of return (ROR) for each of your asset classes, income, expenses, etc. ROR assumptions should always be long-term expectations. Updating the account balances annually will account for the good years (like 2020) and the bad years as they become actuals, but the tool is fundamentally trying to model the long term so it would be a mistake to use any one year's actual returns as a long-term assumption.
I started using this in 2017 and have settled on the practice of using end of year balances as well. I also go through and update all my ROR assumptions, asset allocation etc. in January. Since it is an annual model and not a monthly one, I do not update my balances throughout the year when I use the tool to explore some decision.
For ROR assumptions, I do not use historical. I review forward looking ROR assumptions from various sources such as Vanguard, Research Affiliates, and Money For the Rest of Us, taking into consideration market conditions at the EOY. Some of these guys use a building block approach for equities based on dividend yield, earnings growth and valuations. For bonds, they are based more on current yield to maturity not historical. I create a low, base and high set of ROR's for each asset class and use them.
The optimization algorithm uses strictly the tax-bracket-restricted method and runs though all combinations of percentage-of-accounts-to-convert and marginal tax brackets, saves the total net savings at the end of the modeling period for each of these cases, and then compares the best case (i.e., the one with the largest net savings at the end) to the baseline case (i.e., one in which no Roth conversions are done). If the baseline case is better than the best optimized Roth conversion it generates the message about "no conversions" being the best option.
If you think PRC is not producing correct results then I encourage you to submit a Gold problem report along with an export data file so it can be investigated.
Thank you both for your thoughts. You've convinced me to model each year anew. I do keep a snapshot of my budget spreadsheet at each year end, so I can always go back to look at that year's forward projections if I want to see how it changes from year to year.
Stuart, I would like more of your thoughts on expected RORs. Currently most prognosticators have very low projected RORs for the next 10 years mostly because of high current valuations of equities and low interest rates. But for a 30 year plan (I am in retirement now), many suggest using RORs closer to historical. Is there a way to model one set of (lower) RORs for 10 years, and then another set (higher) for the subsequent 20 years?
I update PRC once a year using year end balances. I prefer to plan based on the assets, liabilities, estimated income and expenses that I have. I track year end balances by account in a spreadsheet over time in the form of an annual net worth statement. I do the same for annual sources of income by the line items on my tax return. I do not track progress against any plan.
I use PRC to better understanding the range of possible future outcomes based on different spending scenarios. I am retired and I track a range of safe spending levels using different spending models. In terms of potential future investment gains I consider the lower range of outcomes to carry more weight when the markets are richly valued like they are now.
Lloyd, regarding your question about a way of using lower RORs for the next 10 years and then higher rates thereafter, there is a way to do this as long as you don't have more than 5 asset classes. As I think you know, you can specify up to 10 asset classes and corresponding RORs on the Financial Assets > Asset Classes page, and these returns are fixed for the entire modeling period. To accomplish your objective, you could replicate your asset classes using a different set of RORs and then use the Financial Assets > Asset Allocation page to specify that the first set of assets is used for Period 1 (say, the next 10 years) and the second set of assets is used for Period 2 (say, years 11 and beyond). Does that make sense?
Yes, that does and I can get by with 5 asset classes for this exercise. I'll try this approach and let you know if I have any further questions. Thanks for your help.
Lloyd
Stuart, I would like more of your thoughts on expected RORs. Currently most prognosticators have very low projected RORs for the next 10 years mostly because of high current valuations of equities and low interest rates. But for a 30 year plan (I am in retirement now), many suggest using RORs closer to historical. Is there a way to model one set of (lower) RORs for 10 years, and then another set (higher) for the subsequent 20 years?
Hi Lloyd, my two cents (you get what you pay for) is to not overly complicate it like this. More moving parts means more complexity, more things to maintain, more ways for things to go wrong.
Remember that Pralana is a long-term projection tool. I'd use the projected long-term rates, understanding that the standard deviation built into each class will take care of the ups and downs. You can adjust the SD if you like.
I tend to be conservative for the clients I use Pralana to model for. It leads to more good surprises and happier endings. But like Greg Golich said, it's good to be educated and informed. Watch those fantastic outlook webinars from trusted sources like Vanguard. I like to watch the Schiller CAPE metric in order to make short-term decisions such as adjusting the withdrawal rate in my year-end review and year-forward planning in December, from perhaps the "standard" 4% to something like 3.5 when things are trending poorly, or 4.5 in good times. Bengen has recently said it should be the "4.5% rule" I believe, but again I like to be conservative.
