I actually created my own VPW spreadsheet and I did it for several reasons.
1. The original spreadsheet only comprehended a total stock market and total bond market portfolio in the backtesting section. That's not my portfolio.
2. The original spreadsheet uses a fixed expected real return for stocks and bonds based on a Credit Suisse paper from a couple of years ago. It's based on long term (mid 1800's) worldwide returns.
3. I want to be able to treat all future cash flows as my "portfolio". In simple terms when I retire, I will have my savings and some point after I retire, I will have SS. Others may have a pension, etc. Anyway, rather than just withdrawing from my portfolio the percentage dictated, I would withdraw that percentage + the equivalent amount of SS I will eventually get once I do start taking SS. Once I start taking SS, I stop the additional withdrawal.
4. The author of VPW basically reinvented the PMT function already available in excel. The calculations are a lot simpler if you just use PMT.
OK, your actual question was what do I use for estimated future returns. The answer is varied. And some may use an average of any of the answers below.
First, remember that the estimated future returns that you plug into must be real future returns, not nominal future returns. And on whatever day you decide to take your withdrawal, find the data below and use that in the calculation.
Choices for the stock portion of your portfolio
1. 1/CAPE (aka 1/PE10). Not perfect but it is decent. You can always find this at:
Shiller PE Ratio or on Professor Shiller's website. For example, today CAPE sits at 32.45 So 1/CAPE=3.08% Note, CAPE is derived by professor Shiller for the S&P 500 index. Probably pretty close if you use a total stock market fund. But if you deviate much from those two, you may need to adjust this number upwards or downwards. Or you can just use it and if your choice of stock fund has higher returns, than that, then the dollar amount withdrawn will automatically reflect it, eventually. This is probably the easiest metric to use as it's always kept up to date.
2. After that, there are a number of websites that make their predictions for the future. Just be aware that they differ in whether they give future returns in nominal terms or real terms. If they give it in nominal terms, you'll need a future inflation guess to go along with it. If you do that, then make sure you do the calculation as a geometric, not arithmetic product. For example. If future nominal returns are 5% and your inflation guess is 2%, then future real is (1+5%)*(1-2%)-1 = 2.9%
List of websites with predictions (there are more if you google)
1. Robeco:
https://www.robeco.com/en/themes/expected-returns/
2. Blackrock:
https://www.blackrock.com/institutions/en-us/insights/portfolio-design/capital-market-assumptions
3. Research Affiliates:
https://interactive.researchaffilia...currency=USD&model=ER&scale=LINEAR&terms=REAL
There are many others. Generally, they are all in reasonable agreement with each other for the most part. I know at least one person who averages 1/CAPE with the data from Research affiliates for example.
What about bonds?
1. The same websites above sometimes also give future bond returns.
2. You can also get a good idea by going to Morningstar and pull up your bond fund. The SEC yield is a decent predictor of nominal future returns for your bond fund. For example, the 30 day SEC yield for VBTLX is currently 3.12% according to Morningstar.
But you still need to take this number and convert it to a real return. So, you need a future inflation predictor to convert the nominal predicted return to a real predicted return. Again, some of the websites above provide an answer. Some alternatives to that:
1. University of Michigan has a survey with consumer expected inflation rates. It's available here:
https://fred.stlouisfed.org/series/MICH In May of 2018, the data shows 2.8%
2. You can also look at the difference in yields between nominal bonds and TIPs of the same duration. In the ideal world that difference would be the expected inflation rate. Fortunately FRED has already done the calculation for you and its available here for 10 year expected inflation
https://fred.stlouisfed.org/series/T10YIE As of July 6th, it's 2.12% Or you can look at the 5 year numbers which aren't much different and can be found here:
https://fred.stlouisfed.org/series/T5YIFR
Now just weight the stock and bond expected returns as your AA. For example, if you're 60/40 then 0.6* stock expected + 0.4 * bond expected and plug that into the PMT equation each year to calculate your %withdrawal.
That's about it. Don't get too hung up on absolute accuracy here. VPW/PMT is somewhat adaptable. If your future actual returns are higher than your expected returns, then the dollar amount in your portfolio will be higher. If the dollar amount is higher, then the withdrawal will also be higher. And vice versa.
In my own backtesting, I have found that this smoothing works a little better for a stock-heavy portfolio than for portfolios with a higher percentage of bonds. This makes intuitive sense because a higher dosage of bonds will already reduce your return variability by itself.
You can also add what's called "short term smoothing" to the withdrawals in order to limit how much you want to allow the withdrawal to change year to year. This is only limited by your imagination, but note that if you oversmooth, then you may deplete your portfolio faster resulting in lower withdrawals later on.
Hope this helps...