Asset Allocation

Interesting tidbit in the Fidelity report provided by @wallace471:

Developed Markets (ex-U.S.)
• We expect most non-U.S developed countries, including Australia, Canada, the UK, France,
Germany, and Japan, to lag the real GDP growth of the U.S. through 2042, mainly due
to birth rates and other demographic trends. This is expected to keep earnings growth
subdued relative to the U.S. and emerging markets.
• Return estimates for developed-equity markets outside the U.S. are 3.3% in real terms over
the next 20 years, topping U.S. stocks (at 3.0%). We expect a diminished return for developed non-
U.S. markets versus the long-term historical average, and slightly higher volatility compared
with the U.S., based on a higher concentration of more-cyclical sectors.

Fidelity's take on developed markets is not as rosie 🌺 as others.

@pizzaman I'm not sure I'd focus on the interquartile range as much as I would the difference between the dotted line (predicted rate from 10 yrs earlier) and the solid line (actual observed rate for that year). From 2011 to 2020 the distance between the two seems equal to or slightly less than a point. I'm not defending Fidelity's projection one way or another, but that post-10 year difference seems reasonably small to me (they actually converge at a few points too). Now if those lines had a much wider distance between them, and/or the up/down trend was off or contrary to actual rates, I'd have less confidence in it.

@nc-cpl I guess I don't understand Figure II-2. Isn't a 10 year ROR prediction a single number, say the US stock market will grow at an average 3.5% over 10 years? Did Vanguard make a prediction for each year over a 10 year span??? What am I missing?

@pizzaman Below is the explanation provided below the chart. In plain speak, each point is comparing the forecast 10 years ago to the actual performance over that 10 years. It is amazing that they are that close. The interquartile range is important to look at since when probabilistic "estimates" are made about any variable, it is important to know the range of uncertainty in those estimates. In the chart, they are 50% confident the actual estimate will be between the upper and lower estimate. There is a 25% chance that the actual return will be higher than the estimate and a 25% chance it will be lower than the estimate. Based on the chart, I would feel confident that the model they are using is doing a reasonable job.

The reason that these estimates of future returns are done on an annual basis is because market conditions change constantly. Starting yields for bonds is a great predictor of future 10 year returns. Stock forecasts will change based on yield, growth in yield and valuations which change over time. The first 10 years of retirement is the most critical and (in my opinion) that is where I would use estimates based on market conditions at the time and update them every year as conditions change. Historical returns don't know what the market conditions are at the time you run your model. There may be some of the years starting conditions that are close but who knows which ones.

The chart shows the actual 10-year annualized return of a 60/40 stock/bond portfolio compared with the VCMM forecast for the same portfolio made 10 years earlier. For example, the 2011 data point at the beginning of the chart shows the actual return for the 10-year period 2001–2011 (solid line) compared with the 10-year return forecast made in 2001 (dotted line). After 2022, the dotted line is extended to show how our forecasts made between 2013 and 2022 (ending between 2023 and 2032) are evolving. The interquartile range represents the area between the 25th and 75th percentile of the return distribution. The portfolio is 36% U.S. stocks, 24% international stocks, 28% U.S. bonds, and 12% international bonds. See the Appendix section titled “Indexes for VCMM simulations” for further details on asset classes.

@golich428 So a question for anyone monitoring this thread...are you using Vanguards data as your ROR input for this year? Splitting the difference between two sources? Averaging across all of them? Curious where people are landing and planting their stake in the ground since we need to get this into the tool to see how it plays out for the rest of the year (and further).

In looking for an analysis of the accuracy of firm RoR forecasts, I came upon John Rekenthaler's column in Morningstar who has from time to time provided his view of long term RoR forecasts.

Among these (there are several artcles), the following may be relevant in this ongoing discussion...

In "Can Warren Buffet Forecast the Stock Market?"(11/5/2020), Rekenthaler examines Buffet's prescription ("rule of thumb") for predicting future stock market returns:

Stock total returns = GDP growth + inflation + dividend yield

https://www.morningstar.com/articles/1009443/can-warren-buffett-forecast-the-stock-market

He looks at various applications of the formula, including rolling time periods, up to 30 years.

Rekenthaler revisits this approach and compares it to 2021 firm predictions in the article "The Long-Term Forecast for US Stock Returns" (6/17/2021):

https://www.morningstar.com/articles/1043396/the-long-term-forecast-for-us-stock-returns

He finds that the "Buffet formula would appear to remain valid" with the expected long-term stock return of 6.37% After accounting for stock buy-backs, Rekenthaler suggests that the stock RoR to assume for predictions could be 7.5%.

In Exhibit 5, he then compares these "Buffet" models to the firm 2021 predictions for 10-15 year stock returns. The firms' 2021 (mean) predictions are all lower - note that the possible band of returns by each firm are not presented there.

Note that his colleague, Benz, summarizes these predictions annually. The 2021 summary is here:

https://www.morningstar.com/articles/1018261/experts-forecast-stock-and-bond-returns-2021-edition

@wallace471 Bob - Curious, given all this, whats your final takeaway and decision as far as what your putting into PRC?

Pizza Man's final answer on ROR for S&P 500.

Determine the Compound Annual Growth Rate or the CAGR for the S&P 500 from 1928 to 2022. Then do the same for 1938 to 2022, repeat until 2018 to 2022, for 10 time periods. Take the CAGR of each of the 10 time periods and average them together. This resultant number puts more weight on later (more recent) time periods. Include inflation and dividend reinvestment, and you get a real rate of return of 6.60%. This is the number I will input into PRC for stock market growth. This has just as much validity as any other hair brained idea from the big boys 😋. Patent pending and trade mark pending. So, if you use my number you will owe me royalties 🤑 🤑 🤑

http://moneychimp.com/features/market_cagr.htm

@pizzaman Well we'll all be eager to compare your 6.6% alongside the hair-brained pundits next Jan. LOL!

Great idea! If my analysis is found to be correct a year from now, I will make sure EVERYBODY knows it 🍻, proclaim my advanced financial acumen 😎, and maybe even start a newsletter, for a fee of course 🤑. If, in the very unlikely event it does not come true, I will tell no one, and no one except maybe for @nc-cpl, will even remember that I made a "prediction". So it's a win win for me 😜.

@pizzaman What about fixed-income, general and healthcare inflation numbers? I'm thinking I'll just eyeball all the recent reports, toss out the low and high outliers and average the remaining to get at a reasonable projection.

BTW - I hope you're right at 6.6%

Historical general inflation is factored into my 6.6% stock ROR. Otherwise the long term inflation rate for the US (not counting the last 6 months) is I think 3.16%, I use 4%. Healthcare inflation I have at plus 2%. Bonds are trickier. Because my asset allocation is time dependent, we will live off of my regular IRA first so it is invested in short term T-bills and notes which are around 4.7% - 4.5% now, and our remaining asset allocation (Roth IRA's and my wife's regular IRA is something like 90% stocks), so long term bond ROR does not factor into our long term forecasts very much.

@pizzaman Maybe I missed something but if your ROR os 6.6 and your inflation is factored in at 4, are you saying your nominal is 10.6%?

Yes, my 6.6% is real ROR. In PRC you are to input real rate of return for stocks, bonds and money market (cash equivalents). The 4% general inflation is used by PRC for other factors such as social security as well. You think 6.6% is a little high even if it takes into account historic inflation and ROR that is skewed toward recent times? 🤔

Using anything resembling historical averages may turn out to be accurate, who knows? Average stock returns may make you really rich, but no one said your retirement was destined to have average returns, so is that number actually what you should plan around? Remember that withdrawing 4% of the initial portfolio + inflation adjustments has run out of money in a couple of instances for 30 year retirements and that only needs something like a 1.1% real CAGR to last.

To make it more personal, ask yourself about the difference in regret of leaving money in your estate but your spouse and family secure vs. living it up but running low on funds and leaving a long lived spouse in a financial mess.

Historical data is about as good as it gets, at least to show the huge variation in outcomes that are possible. You have it at your fingertips in Pralana, just go to Run Analysis and click Historical Results, then click Run Analysis. That will show you graphically the deciles of how your assumptions stack up against historical data. Then give yourself a reality check by selecting 1965 and click View Historic Sequence and get year by year details of how a seemingly average starting date went really poorly.