Are Vanguard's 10-year forecasts flawed in any way?
Are Vanguard's 10-year forecasts flawed in any way?
Hello. I'm sure most of you have heard about Vanguard's monthly reports that outline 10-year forecasts for the performance of certain asset types. In the latest issue of the report, the median return is 4% for US stocks and 7.1% for ex-US stocks. This raises the question - should we overweigh ex-US for the next 10 years (or until another one of Vanguard's forecasts shows that US stocks are expected to deliver better returns than ex-US)? Or, if one has access to an investment vehicle that is expected to return more than 4% in the next 10 years with similar or lower risk, wouldn't it make sense to dump some US stocks and invest the money in that vehicle? Of course, the answer to these question heavily depends on the validity of these forecasts. Are they statistically sound? Does it make sense to make investment decisions based on them?
- typical.investor
- Posts: 5247
- Joined: Mon Jun 11, 2018 3:17 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
Expected returns are typically quite different than realized returns. Ex-US has been expected to outperform for some years now.hithere wrote: ↑Sun May 16, 2021 2:50 am This raises the question - should we overweigh ex-US for the next 10 years (or until another one of Vanguard's forecasts shows that US stocks are expected to deliver better returns than ex-US)? Or, if one has access to an investment vehicle that is expected to return more than 4% in the next 10 years with similar or lower risk, wouldn't it make sense to dump some US stocks and invest the money in that vehicle?
Re: Are Vanguard's 10-year forecasts flawed in any way?
You pose 4 separate and good questions in the post:
should we overweigh ex-US for the next 10 years ?
if one has access to an investment vehicle that is expected to return more than 4% in the next 10 years with similar or lower risk, wouldn't it make sense to dump some US stocks and invest the money in that vehicle?
Validity of these forecasts. Are they statistically sound?
For my two cents worth re question 1 and 2 and based on that Vanguard list, I would be more worried about being too overweight in US "growth" stocks which have an expected real return of -3 percent or something per year!. The trouble is that most index funds are currently fully pregnant with US growth stocks.Does it make sense to make investment decisions based on them?
- asset_chaos
- Posts: 2628
- Joined: Tue Feb 27, 2007 5:13 pm
- Location: Melbourne
Re: Are Vanguard's 10-year forecasts flawed in any way?
While Vanguard's monte carlo simulations are likely as reasonable as anyone else's, the way they've presented the return predictions is poor because they convey a false sense of precision. A typical person would look at those tight ranges and perhaps conclude that Vanguard is guarenteeing that ex-US will outperform US (as just one example) because the bottom of the ex-US range is larger than the top of the US range. But this is misleading. The clue is in the caveat above the numbers, which says
"the figures are based on a 1.0-point range around the rounded 50th percentile of the distribution of return outcomes for equities and a 0.5-point range around the rounded 50th percentile for fixed income."
Their model runs through 10,000 scenarios where the inputs change and the model generates a distribution of 10 year returns. They report the median of this distribution with an arbitrary plus or minus 1 point range. The arbitrary plus or minus 1 point is not the real width of the distribution. They don't indicate the real width. You'd like to know that median, central prediction plus or minus one or two standard deviations, even better what range they would say is at the 95% confidence level. That's because getting the 67th percentile outcome or the 30th percentile outcome would not be surprising at all. From other data I've seen the real ranges around those 50th percentile central predictions are more like plus or minus 10% for stocks (and less for bonds).
The main point to understand about using such predictions to make portfolio changes is not that they are statistically invalid, it's that the ranges of reasonable outcomes is much larger than any of the differences between medians. Yes, ex-US is priced so that it has a reasonable probability to outperform US stocks over next 10 years, but there remains a substantial probability of the reverse to happen too. And neither outcome would invalidate the model. It's the ranges being much larger than the median differences that's telling you, in it's oblique, mathematical way, the same thing that Bogle's been telling you more directly all along: stay the course.
"the figures are based on a 1.0-point range around the rounded 50th percentile of the distribution of return outcomes for equities and a 0.5-point range around the rounded 50th percentile for fixed income."
Their model runs through 10,000 scenarios where the inputs change and the model generates a distribution of 10 year returns. They report the median of this distribution with an arbitrary plus or minus 1 point range. The arbitrary plus or minus 1 point is not the real width of the distribution. They don't indicate the real width. You'd like to know that median, central prediction plus or minus one or two standard deviations, even better what range they would say is at the 95% confidence level. That's because getting the 67th percentile outcome or the 30th percentile outcome would not be surprising at all. From other data I've seen the real ranges around those 50th percentile central predictions are more like plus or minus 10% for stocks (and less for bonds).
The main point to understand about using such predictions to make portfolio changes is not that they are statistically invalid, it's that the ranges of reasonable outcomes is much larger than any of the differences between medians. Yes, ex-US is priced so that it has a reasonable probability to outperform US stocks over next 10 years, but there remains a substantial probability of the reverse to happen too. And neither outcome would invalidate the model. It's the ranges being much larger than the median differences that's telling you, in it's oblique, mathematical way, the same thing that Bogle's been telling you more directly all along: stay the course.
Regards, |
|
Guy
Re: Are Vanguard's 10-year forecasts flawed in any way?
thanks. i wonder why they don't do +/- 1 standard deviation as you suggest.asset_chaos wrote: ↑Sun May 16, 2021 4:50 am While Vanguard's monte carlo simulations are likely as reasonable as anyone else's, the way they've presented the return predictions is poor because they convey a false sense of precision. A typical person would look at those tight ranges and perhaps conclude that Vanguard is guarenteeing that ex-US will outperform US (as just one example) because the bottom of the ex-US range is larger than the top of the US range. But this is misleading. The clue is in the caveat above the numbers, which says
"the figures are based on a 1.0-point range around the rounded 50th percentile of the distribution of return outcomes for equities and a 0.5-point range around the rounded 50th percentile for fixed income."
Their model runs through 10,000 scenarios where the inputs change and the model generates a distribution of 10 year returns. They report the median of this distribution with an arbitrary plus or minus 1 point range. The arbitrary plus or minus 1 point is not the real width of the distribution. They don't indicate the real width. You'd like to know that median, central prediction plus or minus one or two standard deviations, even better what range they would say is at the 95% confidence level. That's because getting the 67th percentile outcome or the 30th percentile outcome would not be surprising at all. From other data I've seen the real ranges around those 50th percentile central predictions are more like plus or minus 10% for stocks (and less for bonds).
The main point to understand about using such predictions to make portfolio changes is not that they are statistically invalid, it's that the ranges of reasonable outcomes is much larger than any of the differences between medians. Yes, ex-US is priced so that it has a reasonable probability to outperform US stocks over next 10 years, but there remains a substantial probability of the reverse to happen too. And neither outcome would invalidate the model. It's the ranges being much larger than the median differences that's telling you, in it's oblique, mathematical way, the same thing that Bogle's been telling you more directly all along: stay the course.
cheers,
grok
RIP Mr. Bogle.
Re: Are Vanguard's 10-year forecasts flawed in any way?
I use Vanguard's forecast in my spreadsheet because it is far more sophisticated and rationalized than any swag I could come up with.
I do not use Vanguard's forecast for decisions. In the short term, markets go up or down with either equally probable. For the long term, I assume that equity will trend up and fixed will remain flat. Decisions on the margin, therefore, plot a course that partially participates in unexpected gains and partially shields from unexpected losses.
I do not use Vanguard's forecast for decisions. In the short term, markets go up or down with either equally probable. For the long term, I assume that equity will trend up and fixed will remain flat. Decisions on the margin, therefore, plot a course that partially participates in unexpected gains and partially shields from unexpected losses.
"Plans are useless; planning is indispensable.” (Dwight Eisenhower) |
"Man plans, God laughs" (Yiddish proverb)
- nisiprius
- Advisory Board
- Posts: 52109
- Joined: Thu Jul 26, 2007 9:33 am
- Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry
Re: Are Vanguard's 10-year forecasts flawed in any way?
You ask about "statistically sound." Vanguard is doing something that IMHO is phony and dishonest: "Please note that the figures are based on a 1.0-point range around the rounded 50th percentile of the distribution of return outcomes for equities..." Now why in the world would anyone do that? What is this magic "1.0-point range?" No, that is not statistically sound. There's no reason in the world why the uncertainty in a stock forecast should be the magic round number ±1%.
This has nothing much to do with Vanguard, particularly. Although if you ask "what is Vanguard's special strength?" it is probably not market forecasting.
I'm going to approach this from three directions.
1) if you should adjust your portfolio based on market forecasting, why doesn't Vanguard do it? I am referring specifically to a class of mutual funds called "tactical asset allocation funds" in general, and to Vanguard's LifeStrategy Funds in particular. If you read the description of the Vanguard LifeStrategy Moderate Growth Fund, it says:
The point is, it wasn't always that way. From very soon after launch in 1994 at least through until about 2012--I don't have the exact dates, but a very long time--60/40 was merely the "neutral point," and the allocation was adjusted up and down according to a quantitative model. In the 1980s and 1990s tactical asset allocation funds were very popular, almost standard for balanced funds. There was one in my employer's 401(k) plan--Fidelity Asset Manager--and probably in many others.
So the LifeStrategy funds used tactical asset allocation and adjusted holdings based on market forecasts for at least fifteen years. And then around 2012 or 2014 or so... they stopped. They froze the allocations: 80/20 for Growth, 60/40 for Moderate Growth, 40/60 for Conservative Growth, 20/80 for Income, and quit using forecasts.
Tactical asset allocations funds are almost unknown today. And the reason is simple: they didn't work. They weren't disastrous, but once they had a long track record behind them it was clear that as a class they were underperforming simple fixed allocations.
2) This is "investing advice inspired by Jack Bogle," and he wrote:
3) People want forecasts, and the demand creates a supply. They're garbage. GMO's forecasts are much more famous than Vanguard's and are often discussed in the forum, yet they have been worse than the toss of a coin. Yet people still regularly bring them up as if were sure-thing accurate. Perhaps the insights behind them are valuable. The forecasts, as information to act on, are not.
Years ago I read an absolutely marvelous book which everyone should read: The Fortune Sellers: The Big Business of Buying and Selling Predictions, by William A. Sherden. It's about predictions in general, and it's stunning. Predicting next week's weather is about the best of any predictions humans make. I know that sounds crazy, but read the book.
Examples can be multiplied of impressive economic predictions, made with great confidence, that turn out to be completely wrong. One I'm fond of is Bloomberg's poll of 68 economics in April 2014 for their forecasts of the 10-year Treasury rate. 68 out of 68, 100%, unanimity, said it would rise, the only disagreement being by how much. It fell.
Be skeptical of forecasts, even when "everybody" says so.
In GMO's case, in a sense the predictions come with a warning label that nobody reads: the full reports (apparently) show ± bars on the predictions, i.e. a range and not a number. The ranges are like "6% ± 10%." So they can never be disproven, because the final results seven years later, even if they aren't even close to being in the same order as the predictions, are still within the stated margins of error.
Vanguard also leaves themselves a complete "out:" "These probabilistic return assumptions depend on current market conditions and, as such, may change over time." When you think that through, it's just a sophisticated way of saying "I have lost three pounds over the last six weeks so I am forecasting that I will reach 170 pounds by the Fourth of July, but, that depends on my current eating habits and may change over time." It is extrapolating current conditions which not only may change, but--over ten years--almost certainly will.
On a large scale, one might say that to the extent forecasts like these have any validity, the people in positions to make big decisions use them to do things that change what happens. For example, the Federal Reserve has a stated inflation target of 2% per year, so if forecasts the Fed trusts show that we are headed for 7% per year, they are going to do everything in their power to make the forecast wrong.
I don't think so. No.
This has nothing much to do with Vanguard, particularly. Although if you ask "what is Vanguard's special strength?" it is probably not market forecasting.
I'm going to approach this from three directions.
1) if you should adjust your portfolio based on market forecasting, why doesn't Vanguard do it? I am referring specifically to a class of mutual funds called "tactical asset allocation funds" in general, and to Vanguard's LifeStrategy Funds in particular. If you read the description of the Vanguard LifeStrategy Moderate Growth Fund, it says:
So these are fixed percentages. 60/40 all the time.The fund holds 60% of its assets in stocks, a portion of which is allocated to international stocks, and 40% in bonds, a portion of which is allocated to international bonds.
The point is, it wasn't always that way. From very soon after launch in 1994 at least through until about 2012--I don't have the exact dates, but a very long time--60/40 was merely the "neutral point," and the allocation was adjusted up and down according to a quantitative model. In the 1980s and 1990s tactical asset allocation funds were very popular, almost standard for balanced funds. There was one in my employer's 401(k) plan--Fidelity Asset Manager--and probably in many others.
So the LifeStrategy funds used tactical asset allocation and adjusted holdings based on market forecasts for at least fifteen years. And then around 2012 or 2014 or so... they stopped. They froze the allocations: 80/20 for Growth, 60/40 for Moderate Growth, 40/60 for Conservative Growth, 20/80 for Income, and quit using forecasts.
Tactical asset allocations funds are almost unknown today. And the reason is simple: they didn't work. They weren't disastrous, but once they had a long track record behind them it was clear that as a class they were underperforming simple fixed allocations.
2) This is "investing advice inspired by Jack Bogle," and he wrote:
What I want to stress is that although he did talk about adjusting the ratio in accordance with your own financial profile, notice that he did not say anything about adjusting it in response to market forecasting.When All Else Fails, Fall Back on Simplicity.
There are an infinite number of strategies worse than this one: Commit, over a period of a few years, half of your assets to a stock index fund and half to a bond index fund. Ignore interim fluctuations in their net asset values. Hold your positions for as long as you live, subject only to infrequent and marginal adjustments as your circumstances change. When there are multiple solutions to a problem, choose the simplest one....
Although the stock market’s wild and wooly odyssey since I wrote them makes those words seem an eon away, I believe more than ever in that basic principle: Rely heavily on index funds, and begin with the idea of a 50/50 bond/stock ratio, adjusting the ratio in accordance with your own financial profile.
3) People want forecasts, and the demand creates a supply. They're garbage. GMO's forecasts are much more famous than Vanguard's and are often discussed in the forum, yet they have been worse than the toss of a coin. Yet people still regularly bring them up as if were sure-thing accurate. Perhaps the insights behind them are valuable. The forecasts, as information to act on, are not.
Years ago I read an absolutely marvelous book which everyone should read: The Fortune Sellers: The Big Business of Buying and Selling Predictions, by William A. Sherden. It's about predictions in general, and it's stunning. Predicting next week's weather is about the best of any predictions humans make. I know that sounds crazy, but read the book.
Examples can be multiplied of impressive economic predictions, made with great confidence, that turn out to be completely wrong. One I'm fond of is Bloomberg's poll of 68 economics in April 2014 for their forecasts of the 10-year Treasury rate. 68 out of 68, 100%, unanimity, said it would rise, the only disagreement being by how much. It fell.
Be skeptical of forecasts, even when "everybody" says so.
In GMO's case, in a sense the predictions come with a warning label that nobody reads: the full reports (apparently) show ± bars on the predictions, i.e. a range and not a number. The ranges are like "6% ± 10%." So they can never be disproven, because the final results seven years later, even if they aren't even close to being in the same order as the predictions, are still within the stated margins of error.
Vanguard also leaves themselves a complete "out:" "These probabilistic return assumptions depend on current market conditions and, as such, may change over time." When you think that through, it's just a sophisticated way of saying "I have lost three pounds over the last six weeks so I am forecasting that I will reach 170 pounds by the Fourth of July, but, that depends on my current eating habits and may change over time." It is extrapolating current conditions which not only may change, but--over ten years--almost certainly will.
On a large scale, one might say that to the extent forecasts like these have any validity, the people in positions to make big decisions use them to do things that change what happens. For example, the Federal Reserve has a stated inflation target of 2% per year, so if forecasts the Fed trusts show that we are headed for 7% per year, they are going to do everything in their power to make the forecast wrong.
Last edited by nisiprius on Sun May 16, 2021 6:54 am, edited 1 time in total.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
- RickBoglehead
- Posts: 7852
- Joined: Wed Feb 14, 2018 8:10 am
- Location: In a house
Re: Are Vanguard's 10-year forecasts flawed in any way?
My forecast is a 2% return on my taxable assets, and 6% on my retirement accounts. With 2/3rds of my assets taxable, that's an overall 3.3% overall.
Avid user of forums on variety of interests-financial, home brewing, F-150, EV, home repair, etc. Enjoy learning & passing on knowledge. It's PRINCIPAL, not PRINCIPLE. I ADVISE you to seek ADVICE.
Re: Are Vanguard's 10-year forecasts flawed in any way?
The best advice might be to regard this sort of thing as "not even wrong."*
I think nisi"s discussion is excellent. It does make one wonder what point there is in publishing something like this.
*"not even wrong" comes from this dismissive anecdote: "The phrase is generally attributed to the theoretical physicist Wolfgang Pauli, who was known for his colorful objections to incorrect or careless thinking.[2][3] Rudolf Peierls documents an instance in which "a friend showed Pauli the paper of a young physicist which he suspected was not of great value but on which he wanted Pauli's views. Pauli remarked sadly, 'It is not even wrong'."
https://en.wikipedia.org/wiki/Not_even_wrong
I think nisi"s discussion is excellent. It does make one wonder what point there is in publishing something like this.
*"not even wrong" comes from this dismissive anecdote: "The phrase is generally attributed to the theoretical physicist Wolfgang Pauli, who was known for his colorful objections to incorrect or careless thinking.[2][3] Rudolf Peierls documents an instance in which "a friend showed Pauli the paper of a young physicist which he suspected was not of great value but on which he wanted Pauli's views. Pauli remarked sadly, 'It is not even wrong'."
https://en.wikipedia.org/wiki/Not_even_wrong
-
- Posts: 6103
- Joined: Sat Mar 11, 2017 6:51 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
As others are suggesting, the equities models have so much irreducible uncertainty as to not really be actionable in my view. About all I would say is properly understood, the uncertainties around those estimates continue to support a diversified approach to equity investing.
Bond models also have a lot of uncertainty in some ways, but there are also some aspects to bond analysis that are a little more straightforward, in that you can compare bonds to at least derive market estimates of returns to term, expected inflation, credit risk, and so on.
But I don't really pay attention to that, because I have a very narrow focus on what role contracts for future payment (a category including bonds and some other stuff) play in our financial plan, and I am not really interested in forecasts for the stuff I don't have a plan to use.
Bond models also have a lot of uncertainty in some ways, but there are also some aspects to bond analysis that are a little more straightforward, in that you can compare bonds to at least derive market estimates of returns to term, expected inflation, credit risk, and so on.
But I don't really pay attention to that, because I have a very narrow focus on what role contracts for future payment (a category including bonds and some other stuff) play in our financial plan, and I am not really interested in forecasts for the stuff I don't have a plan to use.
Re: Are Vanguard's 10-year forecasts flawed in any way?
it would be nice if vanguard would publish a retrospective look at how their past forecasts have performed.dbr wrote: ↑Sun May 16, 2021 8:54 am The best advice might be to regard this sort of thing as "not even wrong."*
I think nisi"s discussion is excellent. It does make one wonder what point there is in publishing something like this.
*"not even wrong" comes from this dismissive anecdote: "The phrase is generally attributed to the theoretical physicist Wolfgang Pauli, who was known for his colorful objections to incorrect or careless thinking.[2][3] Rudolf Peierls documents an instance in which "a friend showed Pauli the paper of a young physicist which he suspected was not of great value but on which he wanted Pauli's views. Pauli remarked sadly, 'It is not even wrong'."
https://en.wikipedia.org/wiki/Not_even_wrong
Vanguard overlords- are you listening?
RIP Mr. Bogle.
Re: Are Vanguard's 10-year forecasts flawed in any way?
I'd recommend listening to the Bogleheads Podcast interview with Joe Davis, the head economist at Vanguard.
https://www.bogleheads.org/blog/portfolio/joe-davis/
In his interview he addresses a lot of questions about Vanguard's forecast and why that product even exists. Short answer: He does not recommend individual investors use his forecast for tactical asset allocation decisions. His primary clients are Vanguard fund managers and other institutional investors Vanguard sells their forecast to.
This is another case where we all should really understand that Vanguard is a complex business not a charity or non-profit dedicated to doing what's right for any and all individual investors. Vanguard has businesses and interests that are not relevant to a long term individual investor. They have posts and press releases that are not relevant to many if not most individual investors.
https://www.bogleheads.org/blog/portfolio/joe-davis/
In his interview he addresses a lot of questions about Vanguard's forecast and why that product even exists. Short answer: He does not recommend individual investors use his forecast for tactical asset allocation decisions. His primary clients are Vanguard fund managers and other institutional investors Vanguard sells their forecast to.
This is another case where we all should really understand that Vanguard is a complex business not a charity or non-profit dedicated to doing what's right for any and all individual investors. Vanguard has businesses and interests that are not relevant to a long term individual investor. They have posts and press releases that are not relevant to many if not most individual investors.
Re: Are Vanguard's 10-year forecasts flawed in any way?
Everybody’s forecasts are flawed. Nobody can accurately predict the future. And that includes Vanguard. (If they could, why would they be such heavy indexers).
Somebody, maybe Buffet, pointed out that forecasts tell you a lot about the forecaster, and nothing about the future.
I’d love to see a write up on their previous forecast that covered the most recent 3 or 4 years. These boards are full of crazy smart people, and the collective “we” can’t accurately predict 10 months into the future......
Don’t do anything based on a forecast.
Somebody, maybe Buffet, pointed out that forecasts tell you a lot about the forecaster, and nothing about the future.
I’d love to see a write up on their previous forecast that covered the most recent 3 or 4 years. These boards are full of crazy smart people, and the collective “we” can’t accurately predict 10 months into the future......
Don’t do anything based on a forecast.
Re: Are Vanguard's 10-year forecasts flawed in any way?
Isn't reacting to these market forecasts just an attempt to market time? Part of the Boglehead philosophy is "Never try to time the market".
Buy - Hold - Rebalance! No mention of anticipating a market change.
Buy - Hold - Rebalance! No mention of anticipating a market change.
-
- Posts: 2390
- Joined: Mon Dec 17, 2018 5:49 pm
Re: Are Vanguard's 10-year forecasts flawed in any way?
I make no changes in my portfolio based on forecasts. I save as much as I can, keep enough cash for a comfortable margin of safety for most reasonable scenarios and the rest goes into a low cost equity index fund. None of what I do requires forecasts.hithere wrote: ↑Sun May 16, 2021 2:50 am Hello. I'm sure most of you have heard about Vanguard's monthly reports that outline 10-year forecasts for the performance of certain asset types. In the latest issue of the report, the median return is 4% for US stocks and 7.1% for ex-US stocks. This raises the question - should we overweigh ex-US for the next 10 years (or until another one of Vanguard's forecasts shows that US stocks are expected to deliver better returns than ex-US)? Or, if one has access to an investment vehicle that is expected to return more than 4% in the next 10 years with similar or lower risk, wouldn't it make sense to dump some US stocks and invest the money in that vehicle? Of course, the answer to these question heavily depends on the validity of these forecasts. Are they statistically sound? Does it make sense to make investment decisions based on them?
“You only find out who is swimming naked when the tide goes out.“ — Warren Buffett
- asset_chaos
- Posts: 2628
- Joined: Tue Feb 27, 2007 5:13 pm
- Location: Melbourne
Re: Are Vanguard's 10-year forecasts flawed in any way?
As the article is on Vanguard's advisors site, I speculate charitably that they "dumb it down" to make it easier for advisors to communicate to clients a simple, clear message. Uncharitably I might speculate that showing the full distribution would expose the whole exercise as not worth the candle.grok87 wrote: ↑Sun May 16, 2021 4:56 amthanks. i wonder why they don't do +/- 1 standard deviation as you suggest.asset_chaos wrote: ↑Sun May 16, 2021 4:50 am While Vanguard's monte carlo simulations are likely as reasonable as anyone else's, the way they've presented the return predictions is poor because they convey a false sense of precision. A typical person would look at those tight ranges and perhaps conclude that Vanguard is guarenteeing that ex-US will outperform US (as just one example) because the bottom of the ex-US range is larger than the top of the US range. But this is misleading. The clue is in the caveat above the numbers, which says
"the figures are based on a 1.0-point range around the rounded 50th percentile of the distribution of return outcomes for equities and a 0.5-point range around the rounded 50th percentile for fixed income."
Their model runs through 10,000 scenarios where the inputs change and the model generates a distribution of 10 year returns. They report the median of this distribution with an arbitrary plus or minus 1 point range. The arbitrary plus or minus 1 point is not the real width of the distribution. They don't indicate the real width. You'd like to know that median, central prediction plus or minus one or two standard deviations, even better what range they would say is at the 95% confidence level. That's because getting the 67th percentile outcome or the 30th percentile outcome would not be surprising at all. From other data I've seen the real ranges around those 50th percentile central predictions are more like plus or minus 10% for stocks (and less for bonds).
The main point to understand about using such predictions to make portfolio changes is not that they are statistically invalid, it's that the ranges of reasonable outcomes is much larger than any of the differences between medians. Yes, ex-US is priced so that it has a reasonable probability to outperform US stocks over next 10 years, but there remains a substantial probability of the reverse to happen too. And neither outcome would invalidate the model. It's the ranges being much larger than the median differences that's telling you, in it's oblique, mathematical way, the same thing that Bogle's been telling you more directly all along: stay the course.
cheers,
grok
Regards, |
|
Guy
- nisiprius
- Advisory Board
- Posts: 52109
- Joined: Thu Jul 26, 2007 9:33 am
- Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry
Re: Are Vanguard's 10-year forecasts flawed in any way?
The more I stare at those ±1% ranges for stocks, ±0.5% for bonds, the more steamed up I get.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
-
- Posts: 163
- Joined: Sun May 09, 2021 6:10 pm
Re: Are Vanguard's 10-year forecasts flawed in any way?
Here is a link to a report showing the methodology to arrive at predicted annual returns for the decade ending 2030 of 4.7% for US equities and 8.1% for international. https://personal.vanguard.com/pdf/ISGUS ... online.pdfhithere wrote: ↑Sun May 16, 2021 2:50 am Hello. I'm sure most of you have heard about Vanguard's monthly reports that outline 10-year forecasts for the performance of certain asset types. In the latest issue of the report, the median return is 4% for US stocks and 7.1% for ex-US stocks. This raises the question - should we overweigh ex-US for the next 10 years (or until another one of Vanguard's forecasts shows that US stocks are expected to deliver better returns than ex-US)? Or, if one has access to an investment vehicle that is expected to return more than 4% in the next 10 years with similar or lower risk, wouldn't it make sense to dump some US stocks and invest the money in that vehicle? Of course, the answer to these question heavily depends on the validity of these forecasts. Are they statistically sound? Does it make sense to make investment decisions based on them?
Is the model flawed? No, I don’t think it is. But it’s merely a model based on probabilities. Think of a weather forecast that gives you an 80% chance of thunderstorms. Sometimes it’s right. Sometimes it’s not.
As far as whether you should take action in response to the predictive model—that depends. Are you attempting to be tactical or are you strategic. If you’re strategic, meaning you’ve set up a portfolio that is diversified to capture the market returns calibrated to your risk tolerance through fixed income allocation, then no, you probably shouldn’t. Ask yourself this question- is your portfolio ready for whatever may happen?
That being said, I believe you can make adjustments to your portfolio, but it absolutely must be very well planned and strategic. I personally would not attempt tactical moves but that’s a personal choice.
Re: Are Vanguard's 10-year forecasts flawed in any way?
Of course they are flawed. Any market moves made in response to these types of forecasts are doomed to (random chance of) failure. However, they do have some value in terms of long term planning. Plug some numbers into your portfolio and your savings rate to see if you are on-track, woefully short or headed to riches. Will it be accurate? Not really. But it will be accurate enough if you periodically revisit where you stand using the same methods.
-
- Posts: 6201
- Joined: Mon Apr 11, 2011 12:28 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
I don't see why. It's just a forecast.
20% VOO | 20% VXUS | 20% AVUV | 20% AVDV | 20% AVES
-
- Posts: 1772
- Joined: Sun May 13, 2018 3:41 pm
Re: Are Vanguard's 10-year forecasts flawed in any way?
Take these forecasts with a grain of salt. Nobody knows nothing. However I do like to read, listen, or watch videos of Joe Davis from Vanguard who does have a role in putting out these forecasts. I can't speak for him, but I would wager that even he would say that these forecasts are not actionable for investors that have a diversified portfolio.
Re: Are Vanguard's 10-year forecasts flawed in any way?
These could be excellent forecasts, the best that can be made with the available data. And still not good/wise to use for changing investing plans. I guess it depends on what "flawed" means to you and what use you contemplate making of them.
-
- Posts: 1244
- Joined: Thu Dec 29, 2016 3:59 pm
Re: Are Vanguard's 10-year forecasts flawed in any way?
I never understand why they always seem to publish forecasts based on data that’s 4 months old. It shouldn’t be hard to do a quick adjustment based on current market prices. Instead they force me to whip out a calculator or spreadsheet and do the adjustment manually.
In some cases it wouldn’t matter, but the market is up quite a bit YTD. And this happened last year too, with them publishing a rosy forecast in August, which was based on March 30th asset prices.
In some cases it wouldn’t matter, but the market is up quite a bit YTD. And this happened last year too, with them publishing a rosy forecast in August, which was based on March 30th asset prices.
Re: Are Vanguard's 10-year forecasts flawed in any way?
As a 100% US investor, I've been sleeping well at night, relying on Vanguard's track record of being wrong. I hope they can maintain it. But I've noticed I don't get the aggressive pop - up messages to add international anymore when I open their app. That worries me a little.
Last edited by wesgreen on Sun May 16, 2021 9:33 pm, edited 1 time in total.
Re: Are Vanguard's 10-year forecasts flawed in any way?
They are flawed, yes, in that they are unbearably conservative. They fail to read in historical market data which is 10+% historical S&P averages. It increased clicks and causes angry threads on bogleheads, so I guess it’s “mission accomplished.”
No, it’s never different this time.
-TheDDC
No, it’s never different this time.
-TheDDC
Rules to wealth building: 75-80% VTSAX piled high and deep, 20-25% VTIAX, 0% given away to banks.
- asset_chaos
- Posts: 2628
- Joined: Tue Feb 27, 2007 5:13 pm
- Location: Melbourne
Re: Are Vanguard's 10-year forecasts flawed in any way?
The problem is not the reporting of the median of the distribution as the forecast, the problem is the arbitrary and artificial ranges they imposed around the median. At best these artificial ranges are misleading because they give a veneer of precision to the forecast that is unsupported by the full output of the simulation.
I don't want to put words into others' mouths, but I suspect nisiprius is "steamed" because he expects a much higher quality of data honesty out of Vanguard. I wouldn't let one of my students publish data with an invalid analysis, and I'm too disappointed that someone at Vanguard has.
Regards, |
|
Guy
Re: Are Vanguard's 10-year forecasts flawed in any way?
If you use them for what they are (a prediction) and you keep in mind predictions are just that, predictions and not guarantees, then they are fine. Should you modify your asset allocation based on these predictions or take any action at all? That is a different question. I personally believe it's simpler to pick an asset allocation and stick with it. Ignore the noise, predictions, talking heads, etc. and just keep it simple. Since you specifically mention Equity Ex-US, I am a fan of keeping the 20 - 30% in Ex-US. It's fallen out of favor the past decade due to underperformance but that is exactly why you diversify. At some point you'd suspect US to underperform in relation to Ex-US which is where most predictions (not just Vanguard) are today. But when or by how much or will it even happen? Who knows. No one does for sure. But, if you are diversified, you should be able to come out in a good spot no matter what happens.
Bottom line, the forecasts are what they are. Don't look at them as anything other than a prediction. It's probably best to look at predictions from multiple organizations and see where there are consistencies. That might at least give you a bit more confidence in the predictions but again, they are just predictions. Not a whole lot different than pre-game sports predictions. Sometimes they are right, sometimes they are wrong.
Bottom line, the forecasts are what they are. Don't look at them as anything other than a prediction. It's probably best to look at predictions from multiple organizations and see where there are consistencies. That might at least give you a bit more confidence in the predictions but again, they are just predictions. Not a whole lot different than pre-game sports predictions. Sometimes they are right, sometimes they are wrong.
Re: Are Vanguard's 10-year forecasts flawed in any way?
No, Vanguard's predictions are not "statistically sound." By definition nobody knows the future, and nobody knows what the financial markets will do in the future. Anyone can make predictions but predictions are just educated guesses, nothing more. Do not confuse predictions and guesses with "statistical" validity. Statistics refers to analyzing past numbers, not future guesses.
Does it make sense to make investment decisions based on them? To some extent, sure, I think it is reasonable to conclude that stocks will return more than bonds with greater risk and I think it's reasonable to make investment decisions based on that conclusion, but much beyond that, no. And while it is "reasonable" to do so, do not fool yourself into thinking there are any guarantees.
Does it make sense to make investment decisions based on them? To some extent, sure, I think it is reasonable to conclude that stocks will return more than bonds with greater risk and I think it's reasonable to make investment decisions based on that conclusion, but much beyond that, no. And while it is "reasonable" to do so, do not fool yourself into thinking there are any guarantees.
Re: Are Vanguard's 10-year forecasts flawed in any way?
I think of statistics as using a sample to understand an underlying population. For example, ask 1,000 random people a question and use the results to estimate how the entire group to which they belong would answer the question, or pick items off an assembly line and test them to judge the quality of the entire output of that line.mptfan wrote: ↑Mon May 17, 2021 6:42 am No, Vanguard's predictions are not "statistically sound." By definition nobody knows the future, and nobody knows what the financial markets will do in the future. Anyone can make predictions but predictions are just educated guesses, nothing more. Do not confuse predictions and guesses with "statistical" validity. Statistics refers to analyzing past numbers, not future guesses.
Using statistics to predict stock prices implicitly assumes past and future stock prices represent a single unchanging population and that we have enough reliable data to make meaningful statements about the entire population. I don't see any good reasons that this would be true.
-
- Posts: 6103
- Joined: Sat Mar 11, 2017 6:51 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
Yeah, forward-looking models are more properly understood as probabilistic. And although that is a species of quantified uncertainty, it is not the same species as something like sampling error.Seasonal wrote: ↑Mon May 17, 2021 7:27 amI think of statistics as using a sample to understand an underlying population. For example, ask 1,000 random people a question and use the results to estimate how the entire group to which they belong would answer the question, or pick items off an assembly line and test them to judge the quality of the entire output of that line.mptfan wrote: ↑Mon May 17, 2021 6:42 am No, Vanguard's predictions are not "statistically sound." By definition nobody knows the future, and nobody knows what the financial markets will do in the future. Anyone can make predictions but predictions are just educated guesses, nothing more. Do not confuse predictions and guesses with "statistical" validity. Statistics refers to analyzing past numbers, not future guesses.
Using statistics to predict stock prices implicitly assumes past and future stock prices represent a single unchanging population and that we have enough reliable data to make meaningful statements about the entire population. I don't see any good reasons that this would be true.
And I don't think this is mere semantics. A lot of people I encounter seem to look at backtesting and such as solving the predictive modeling problem, as long as they can get a big enough sample. But that's not a valid assumption, that the past is a representative sample of the future. Instead, we can maybe learn some things about causation and such from investigating the past, which can help us build better predictive models. But doing that sort of work is way more complicated, and subject to many more uncertainties, than I think some people have internalized.
Re: Are Vanguard's 10-year forecasts flawed in any way?
It's not clear to me that this is necessarily a problem.
They could've not given a range. But then the chance of the projection being right would be about zero, no matter the number.
They could've given a much wider range. But then the chance of the projection being right would approach 100%, and then they'd be criticized for not having projected anything.
So I interpret the report as meaning "the model we used says this outcome appears to be more likely than other outcomes".
I'm not seeing where they're entrenching themselves and saying "we declare that this will be the outcome, and if not, something's terribly wrong".
I think it's just a projection.
Strategic Macro Senior (top 1%, 2019 Bogleheads Contest)
-
- Posts: 15289
- Joined: Fri Apr 10, 2015 12:29 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
Monte carlo simulations to forecast 10-yr equity returns are quite flawed. The similulation must use some probability distribution for 10-year returns, but we don't know the distribution of 10-yr equity returns. This makes it a bit like trying to forecast the weather in Chicago using historical climate data for Los Angeles.
- abuss368
- Posts: 27850
- Joined: Mon Aug 03, 2009 2:33 pm
- Location: Where the water is warm, the drinks are cold, and I don't know the names of the players!
- Contact:
Re: Are Vanguard's 10-year forecasts flawed in any way?
Hard to say as it is a forecast.
Will international have a day in the sun? No doubt at some point as all asset classes eventually do.
Honestly I stopped reviewing forecast years ago as it will make no difference to our strategy.
Tony
Will international have a day in the sun? No doubt at some point as all asset classes eventually do.
Honestly I stopped reviewing forecast years ago as it will make no difference to our strategy.
Tony
John C. Bogle: “Simplicity is the master key to financial success."
-
- Posts: 1914
- Joined: Sun Jul 26, 2020 2:29 pm
Re: Are Vanguard's 10-year forecasts flawed in any way?
"...the penalties for incorrect predictions are negligible...." Daniel Kahnemantypical.investor wrote: ↑Sun May 16, 2021 2:59 amExpected returns are typically quite different than realized returns. Ex-US has been expected to outperform for some years now.hithere wrote: ↑Sun May 16, 2021 2:50 am This raises the question - should we overweigh ex-US for the next 10 years (or until another one of Vanguard's forecasts shows that US stocks are expected to deliver better returns than ex-US)? Or, if one has access to an investment vehicle that is expected to return more than 4% in the next 10 years with similar or lower risk, wouldn't it make sense to dump some US stocks and invest the money in that vehicle?
"The only function of economic forecasting is to make astrology look respectable." John Kenneth Galbraith
Retirement is best when you have a lot to live on, and a lot to live for. * None of what I post is investment advice.* |
FIRE'd July 2023
-
- Posts: 6103
- Joined: Sat Mar 11, 2017 6:51 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
So the standard way to do this is to report exactly what you are doing in terms of what percentage of modeled outcomes fall within your stated ranges. Like, you can state explicitly your intervals contain 80% of modeled outcomes, or 90%, or 95%, or 99%. And then you set the intervals to include that percentage of modeled outcomes.HanSolo wrote: ↑Tue May 18, 2021 12:56 amIt's not clear to me that this is necessarily a problem.
They could've not given a range. But then the chance of the projection being right would be about zero, no matter the number.
They could've given a much wider range. But then the chance of the projection being right would approach 100%, and then they'd be criticized for not having projected anything.
So I interpret the report as meaning "the model we used says this outcome appears to be more likely than other outcomes".
I'm not seeing where they're entrenching themselves and saying "we declare that this will be the outcome, and if not, something's terribly wrong".
I think it's just a projection.
I think it is fair, therefore, to object to setting the intervals first, and then not actually reporting what percentage of modeled outcomes fall within the intervals. That is at best simply not a meaningful reporting, and arguably can be misleading in the context of the standard practices.
Re: Are Vanguard's 10-year forecasts flawed in any way?
What's your definition of overweighing ex-US?
By many BH standards, I'm overweight ex-US, at about 41-42% ex-US stocks.
But I'm just holding global market weight.
Global stocks, IG/HY bonds, gold & digital assets at market weights 75% / 19% / 6% || LMP: TIPS ladder
- Peculiar_Investor
- Site Admin
- Posts: 2432
- Joined: Thu Oct 20, 2011 12:23 am
- Location: Calgary, AB 🇨🇦
- Contact:
Re: Are Vanguard's 10-year forecasts flawed in any way?
You forgot “It's tough to make predictions, especially about the future.” ... Yogi Berra and/or Niels BohrAlwaysLearningMore wrote: ↑Tue May 18, 2021 8:09 am "...the penalties for incorrect predictions are negligible...." Daniel Kahneman
"The only function of economic forecasting is to make astrology look respectable." John Kenneth Galbraith
Normal people… believe that if it ain’t broke, don’t fix it. Engineers believe that if it ain’t broke, it doesn’t have enough features yet. – Scott Adams
- BrooklynInvest
- Posts: 1162
- Joined: Sun Jul 28, 2013 9:23 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
Almost all asset managers do this. They can be useful inputs to advisors' asset allocation models, but that's it. It wasn't that long ago that FINRA wouldn't approve the things for retail investors because they're very easy to misinterpret.
Have no idea whether Vanguard's return assumptions are any better than any other firms... and don't care Sometimes the studies make for interesting reads.
Have no idea whether Vanguard's return assumptions are any better than any other firms... and don't care Sometimes the studies make for interesting reads.
Re: Are Vanguard's 10-year forecasts flawed in any way?
Yes, Vanguards forecasts are hugely flawed. Vanguard has no ability to predict the future better than anyone else. Don't let Vanguard's predictions sway your asset allocation, stay the course!
Re: Are Vanguard's 10-year forecasts flawed in any way?
If not just Monte Carlo simulations that are flawed if we don't know the distribution of 10-year future equity returns.Northern Flicker wrote: ↑Tue May 18, 2021 1:57 am Monte carlo simulations to forecast 10-yr equity returns are quite flawed. The similulation must use some probability distribution for 10-year returns, but we don't know the distribution of 10-yr equity returns. This makes it a bit like trying to forecast the weather in Chicago using historical climate data for Los Angeles.
-
- Posts: 6103
- Joined: Sat Mar 11, 2017 6:51 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
And just to clarify something, the model doesn't start with a probability distribution over returns. If it already had that, it would be done.Seasonal wrote: ↑Tue May 18, 2021 10:07 amIf not just Monte Carlo simulations that are flawed if we don't know the distribution of 10-year future equity returns.Northern Flicker wrote: ↑Tue May 18, 2021 1:57 am Monte carlo simulations to forecast 10-yr equity returns are quite flawed. The similulation must use some probability distribution for 10-year returns, but we don't know the distribution of 10-yr equity returns. This makes it a bit like trying to forecast the weather in Chicago using historical climate data for Los Angeles.
Instead, the model performs the simulations using a dynamic model incorporating macroeconomic and financial risk factors for each region like yield curves, inflations expectations, various other leading economic indicators, and other historic risk factors by asset type. The dynamic statistical relationship between these risk factors and asset returns is based on historic data.
It then runs the Monte Carlo simulation using current local market conditions, and then also applies a currency overlay. In doing so, the model assumes a long-term convergence on what they call their global capital markets equilibrium, which represents an economic theory of how global capital markets should operate over the very long term, under normal conditions. But the simulations are designed to allow for many different paths to long-term convergence.
If that all sounds pretty dicey, it is! But they are quite right that the idea of having an economic theory of what markets should do over the long term given normal conditions, and applying that to a historically-derived model of risk factors but with a generous allowance for different paths, is better than simply trying to form expectations based on averages derived from some limited set of historical data.
Better in the sense of less bad, but by no means guaranteed to be right!
Re: Are Vanguard's 10-year forecasts flawed in any way?
Do bookmakers have the ability to predict the future on sports bets?
Do you ignore the odds entirely?
Or factor it into your decision making process?
It can be helpful to know if something is a long shot bet vs a medium risk bet when doing planning.
Global stocks, IG/HY bonds, gold & digital assets at market weights 75% / 19% / 6% || LMP: TIPS ladder
-
- Posts: 6103
- Joined: Sat Mar 11, 2017 6:51 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
Putting fair odds on upcoming games is way easier than predicting relative returns over 10 years.
If you look at Vanguard's output in terms of the big picture, a lot is basically standard stuff. Stocks are riskier than bonds and have higher expected return. There are small and value premiums, not fully explained by volatility estimates. Corporate bonds have higher expected returns than Treasuries. Treasuries are currently running behind expected inflation--which you can also learning by looking at the reported daily Treasury real yield curve.
To the extent there is something worth even commenting on, it is just that ex-U.S. equities have a higher expected return than U.S. equities, mostly due to the assumption of some valuation convergence after a run up in relative U.S. valuations. But, you really didn't need a fancy model for that conclusion--either you buy that valuations are likely to converge, don't buy it, or don't know either way.
The problem is if they reported the actual underlying confidence in these estimates, it is probably much more shaded toward "don't know either way" than "likely to converge".
But yes, they do think it is at least marginally more likely to converge than not, assuming nothing abnormal happens. But that's basically baked into the way their model works.
Re: Are Vanguard's 10-year forecasts flawed in any way?
Does anyone have Vanguard's ( or anyone's else) 10-year prediction from 2010?
Looking back, how accurate were their predictions then?
Looking back, how accurate were their predictions then?
Re: Are Vanguard's 10-year forecasts flawed in any way?
This is part of the issue that rankles people, I think.NiceUnparticularMan wrote: ↑Tue May 18, 2021 12:15 pm To the extent there is something worth even commenting on, it is just that ex-U.S. equities have a higher expected return than U.S. equities, mostly due to the assumption of some valuation convergence after a run up in relative U.S. valuations. But, you really didn't need a fancy model for that conclusion--either you buy that valuations are likely to converge, don't buy it, or don't know either way.
It's one of the issues that comes up on US vs ex-US threads.
For the last several years, almost all of the long term forecasts, whether from Vanguard, BlackRock, JP Morgan, etc, have been forecasting lower 10 year returns for US vs ex-US due to high US valuations.
When that doesn't materialize in Year 1, 2, or 3, people go "see --- these guys know nothing!"
Of course this ignores that fact that:
a) There are a range of probabilistic outcomes the studies forecast. The median is just the median.
b) It hasn't been 10 years yet
I think some of the rejection of the forecasts is based in statistical methodology critiques, but I think part of the rejection is that these forecasts anticipate lower future returns that many just don't want to look at.
I'm willing to bet if these forecasts predicted 10 more years of amazing US stock and bond returns, many people would be shouting how awesome they are.
Global stocks, IG/HY bonds, gold & digital assets at market weights 75% / 19% / 6% || LMP: TIPS ladder
-
- Posts: 6103
- Joined: Sat Mar 11, 2017 6:51 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
Yeah, confirmation bias is an incredibly powerful force, in general and not least in financial discussions.watchnerd wrote: ↑Tue May 18, 2021 12:43 pmThis is part of the issue that rankles people, I think.NiceUnparticularMan wrote: ↑Tue May 18, 2021 12:15 pm To the extent there is something worth even commenting on, it is just that ex-U.S. equities have a higher expected return than U.S. equities, mostly due to the assumption of some valuation convergence after a run up in relative U.S. valuations. But, you really didn't need a fancy model for that conclusion--either you buy that valuations are likely to converge, don't buy it, or don't know either way.
It's one of the issues that comes up on US vs ex-US threads.
For the last several years, almost all of the long term forecasts, whether from Vanguard, BlackRock, JP Morgan, etc, have been forecasting lower 10 year returns for US vs ex-US due to high valuations.
When that doesn't materialize in Year 1, 2, or 3, people go "see --- these guys know nothing!"
Of course this ignores that fact that:
a) There are a range of probabilistic outcomes the studies forecast. The median is just the median.
b) It hasn't been 10 years yet
I think some of the rejection of the forecasts is based in statistical methodology critiques, but I think part of the rejection is that these forecasts anticipate lower future returns that many just don't want to look at.
I'm willing to bet if these forecasts predicted 10 more years of amazing US stock and bond returns, many people would be shouting how awesome they are.
I do wish a few more people at least noted the historical fact that most of the recent overperformance of U.S. had been driven by valuation change, followed by changes in exchange rates, and that only a small portion is attributable to relative earnings growth--which in fact was mostly offset by dividend yield (which I believe means if you were reinvesting, you barely got any earnings growth advantage at all).
And in fact, Vanguard is continuing to forecast better earnings growth in the U.S. (American exceptionalism!). They are just forecasting that will now be more than outweighed by reversals on valuation, and a bit by dividend yields doing more to outweigh earnings growth.
So, stories about the U.S. business environment being better for earnings growth, and such, are already baked into the forecasts. It is the rest of the story that Vanguard is expecting to reverse, and that part of the story does seem to be neglected by some.
-
- Posts: 15289
- Joined: Fri Apr 10, 2015 12:29 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
The Monte Carlo still needs to use a probability distribution for each variable in the model, and putting it all together in the model implicitly derives an implied probability distribution for what you are trying to forecast with the model.NiceUnparticularMan wrote: ↑Tue May 18, 2021 10:59 amAnd just to clarify something, the model doesn't start with a probability distribution over returns. If it already had that, it would be done.Seasonal wrote: ↑Tue May 18, 2021 10:07 amIf not just Monte Carlo simulations that are flawed if we don't know the distribution of 10-year future equity returns.Northern Flicker wrote: ↑Tue May 18, 2021 1:57 am Monte carlo simulations to forecast 10-yr equity returns are quite flawed. The similulation must use some probability distribution for 10-year returns, but we don't know the distribution of 10-yr equity returns. This makes it a bit like trying to forecast the weather in Chicago using historical climate data for Los Angeles.
Instead, the model performs the simulations using a dynamic model incorporating macroeconomic and financial risk factors for each region like yield curves, inflations expectations, various other leading economic indicators, and other historic risk factors by asset type. The dynamic statistical relationship between these risk factors and asset returns is based on historic data.
It then runs the Monte Carlo simulation using current local market conditions, and then also applies a currency overlay. In doing so, the model assumes a long-term convergence on what they call their global capital markets equilibrium, which represents an economic theory of how global capital markets should operate over the very long term, under normal conditions. But the simulations are designed to allow for many different paths to long-term convergence.
If that all sounds pretty dicey, it is! But they are quite right that the idea of having an economic theory of what markets should do over the long term given normal conditions, and applying that to a historically-derived model of risk factors but with a generous allowance for different paths, is better than simply trying to form expectations based on averages derived from some limited set of historical data.
Better in the sense of less bad, but by no means guaranteed to be right!
Tactical asset allocation models that were all the rage in the 1980's also used linear factor models with macroeconomic variables to forecast future return and volatility. They also failed.
Re: Are Vanguard's 10-year forecasts flawed in any way?
if it was right i would but 100% intl and value, why hold bonds ?
Thanks!
Re: Are Vanguard's 10-year forecasts flawed in any way?
The forecast paper models 100% stock portfolios, too.
Going 'no bonds' is an option.
Global stocks, IG/HY bonds, gold & digital assets at market weights 75% / 19% / 6% || LMP: TIPS ladder
-
- Posts: 9244
- Joined: Sun Dec 26, 2010 11:47 am
Re: Are Vanguard's 10-year forecasts flawed in any way?
I would like to see a 10 year forecast where they also publish an accepted margin of error, so at least they are putting something on the line. Without that such a forecast is meaningless and cannot be looked at retrospectively to see how good they have been at guessing.
4% isn't a horrible guess for somebody who knows nothing and is basing it on nothing. If they are off by 50% and the return is 6%, will you call that a "good forecast"?
And I am curious....what did they say in their prior 10 year forecasts (going back as far as possible)? How have those predictions worked out?
And yes, they are deeply flawed. On what are they basing their predictions? What is the data they are using to make them? (Because such a prediction would be exquisitely sensitive to initial conditions). 7.1% is pretty specific. Not "0-10%" or even "7-8%", but they have it down to tenths of a percentage point. Wow.
I did find this...Vanguard's outlook published in 2014 :https://www.institutionalinvestor.com/i ... _FINAL.pdf
I opened randomly to a page on predicted stock market returns (not cherry-picked) where they stated: "we estimate a 50% likelihood that a global
equity portfolio will fail to produce a 5% average real return over the decade 2014–2024."
Well, first, a "50% likelihood"? That's a lot of wiggle room. I guess they didn't have the technical expertise way back in 2014 to nail it to a tenth of a percentage point like they do now. I would estimate there is a 50% likelihood that when I flip this coin it will land on heads.
But anyway, I checked the 7 year performance of the FTSE All-World Index (5/19/14 to today) and the average annual return has been 9.79%. Only almost double their 2014 prediction. Well, yeah, I guess we are in for a crash in the next 3 years....or wait....maybe not....they did say there was a 50% chance the return would be greater than 5%/yr so wow! They nailed it! There is a 50% chance it will be less than 5%/year, and thus a 50% chance it will be equal to or greater than 5%/year. I concur that is 100% accurate.
OP, did I answer your question?
4% isn't a horrible guess for somebody who knows nothing and is basing it on nothing. If they are off by 50% and the return is 6%, will you call that a "good forecast"?
And I am curious....what did they say in their prior 10 year forecasts (going back as far as possible)? How have those predictions worked out?
And yes, they are deeply flawed. On what are they basing their predictions? What is the data they are using to make them? (Because such a prediction would be exquisitely sensitive to initial conditions). 7.1% is pretty specific. Not "0-10%" or even "7-8%", but they have it down to tenths of a percentage point. Wow.
I did find this...Vanguard's outlook published in 2014 :https://www.institutionalinvestor.com/i ... _FINAL.pdf
I opened randomly to a page on predicted stock market returns (not cherry-picked) where they stated: "we estimate a 50% likelihood that a global
equity portfolio will fail to produce a 5% average real return over the decade 2014–2024."
Well, first, a "50% likelihood"? That's a lot of wiggle room. I guess they didn't have the technical expertise way back in 2014 to nail it to a tenth of a percentage point like they do now. I would estimate there is a 50% likelihood that when I flip this coin it will land on heads.
But anyway, I checked the 7 year performance of the FTSE All-World Index (5/19/14 to today) and the average annual return has been 9.79%. Only almost double their 2014 prediction. Well, yeah, I guess we are in for a crash in the next 3 years....or wait....maybe not....they did say there was a 50% chance the return would be greater than 5%/yr so wow! They nailed it! There is a 50% chance it will be less than 5%/year, and thus a 50% chance it will be equal to or greater than 5%/year. I concur that is 100% accurate.
OP, did I answer your question?
Last edited by protagonist on Tue May 18, 2021 7:40 pm, edited 5 times in total.