Message Font: Serif | Sans-Serif

No. of Recommendations: 4
"Pick a stable metric, estimate where it will be, and estimate what multiple it's most likely to get."

Trying to forecast return this way can be humbling, but in general I think it works reasonably well. It's not theory; it's just algebra. The more accurately we can predict growth and final price multiple, the more accurate the forecast.

One problem we run into is that the stable metric is not always stable, or even smoothly changing. In the case of Berkshire Hathaway , if you plot BV versus year over the period 1965 to present, it's not a smooth curve, but rather a curve with a couple of knees. BV growth was more or less constant from 1965 to 1980, then from 1980 to 1999, and then from 2000 to present, but with markedly different growth rates during each period. From Oct 1965 to Dec 1980 the BV growth rate (least squares) was 18.6%/yr; from Dec 1980 to Dec 1999 it was 28.8%, and from Dec 1999 to Sept 2016 it was 10.5%/yr. (Updating the latest period, from Dec 1999 to Dec 2017 it was 10.6%). The S&P 500 index exhibited similar knees at roughly the same times. From Oct 1965 to Dec 1980 the growth rate of the S&P 500 index (least squares) was 1.4%; from Dec 1980 to Dec 1999 it was 13.0%, and from Dec 1999-Sept 2016 it was 3.1%. Predicting the growth rate, like predicting the final price multiple, requires some caution.
No. of Recommendations: 1
My thinking for total return over the next 10 years is something like this:

security: growth, dividend, change in price multiple, total return (all annualized)

S&P 500: 5-6% EPS growth, 2% dividend yield, (15/24)^(1/10)= -4%/yr multiple change, 3-4% total return
BRK: 8-9% BV/share growth, 0% dividend, (1.2/1.36)^(1/10) = -1%/yr multiple change, 7-8% total return

There's that dangerous prediction :)

Yogi Berra
No. of Recommendations: 12
What do you think the P/B or P/IV will be for BRK stock in 5-10 years, or whatever your expected holding time is?
I ask because increasingly I am thinking that the P/B will be lower, not higher, than it is today, even though BRK stock is currently undervalued.

I'll predict a bit higher.

Not because the stock is hugely undervalued now, but more because the fair P/B will continue to rise
ever so slowly over time as it gradually becomes an ordinary company with operating business and profits.

Think of it this way:
The typical really large US firm is accorded about \$4.50 in market value for each \$1 in book.
How far does Berkshire have to go towards being a "normal" company (as opposed to an investment fund) before a slight move in that direction is warranted?

Jim
No. of Recommendations: 6
S&P 500: 5-6% EPS growth, 2% dividend yield, (15/24)^(1/10)= -4%/yr multiple change, 3-4% total return
BRK: 8-9% BV/share growth, 0% dividend, (1.2/1.36)^(1/10) = -1%/yr multiple change, 7-8% total return

But why a forecast of 5-6% EPS growth for the S&P 500?
Seems awfully optimistic. Who's ever seen that, except in a stretch during the up part of a cycle?
Even before inflation adjustment, rolling EPS for the S&P 500 is up only 2.5%/year in the last 11 years.
(a pre-crunch start date near that cyclical top, but still...)

The long run earnings growth rate has been roughly inflation+2%, so 5-6% would seem to anticipate either 3-4% inflation or unusually fast earnings growth.
Not to mention allowing for the possibility that current earnings are perhaps ahead of trend these days.

Jim
No. of Recommendations: 0
"I'll predict a bit higher. Not because the stock is hugely undervalued now, but more because the fair P/B will continue to rise ever so slowly over time as it gradually becomes an ordinary company with operating business and profits."

I sure hope you're right. I've normally predicted a reversion to fair value, but that's what I'm having second thoughts about. One concern is the eventual loss of Warren Buffett, whose skill at capital allocation is unmatched. A friend of mine voices another concern, and that is that large conglomerates tend to sell at discounts to net asset value. He cites the example of Jardine Strategic Holdings, which perpetually sells at a large discount to NAV even though a large fraction of its NAV is in publicly traded companies. (He concedes that Jardine Strategic Holdings may not be the best example. Their large cross-holdings with Jardine Matheson makes takeover by anyone other than Jardine Matheson impossible.)
No. of Recommendations: 1
"But why a forecast of 5-6% EPS growth for the S&P 500?"

You're right. 6% EPS growth is the average since World war II, but since 2000 it's been considerably lower. Real GDP growth, which real EPS tracks fairly closely, is down by a percentage point or so versus the post WW II average, and down by a couple of percentage points from the 1950s. On top of that inflation is down by a percentage point or so. You may have seen Jeremy Grantham and company's latest 7-year forecast for US large cap stocks. Their forecast is down another 0.2 percentage points from their last forecast, and now stands at negative 4.4%. They assume reversion to the mean valuation multiple over seven years, which they say is typical. Scary stuff.
No. of Recommendations: 8
"But why a forecast of 5-6% EPS growth for the S&P 500?"

You're right. 6% EPS growth is the average since World war II....

Without an inflation adjustment, that's probably not a great number to use.

Here are the rolling 30 and 40 year rates of change of smoothed annual EPS per S&P 500 point, adjusted for inflation, using the S&UP 500 and cap weight predecessors.

`Ending   30yr   40yr2018     2.6%   1.9%2008     1.7%   1.6%1998     1.1%   1.5%1988     1.3%   2.1%1978     2.9%   2.5%1968     3.0%   2.1%1958     1.9%   0.9%`

Basically, something in the vicinity of inflation+2%/year has been a not bad best guess.
My latest figure for smoothed real earnings is \$92-93, so that would put them around \$112-113ish in ten years, in today's money.
That's what history suggests.
If the recent tax changes stay in effect and offer a permanent one-time boost to net earnings of (say) 10%,
that would put the ten year expectation somewhere around \$124-125 in today's money.
The tech bubble, the first of the recent three big valuation excursions, started to take off around 1995.
The average earnings yield calculated the same way since then has been about 4.0%.
So, if that were the trend earnings yield ten years from now, terminal trend earnings around \$124-125
suggests one might expect the S&P index to rise to about 3100 and change in ten years.
The return from now till then would therefore be inflation + 1.2%/year + the dividend yield, currently around 1.8%, for a real total return of about 3.0%.

That's just one set of assumptions.
It's easy enough to plug in different ones, to match one's expectations.
But (like your approach) it's a very sensible one.
Pick a stable metric, estimate where it will be, and estimate what multiple it's most likely to get.

Jim
No. of Recommendations: 1
"Pick a stable metric, estimate where it will be, and estimate what multiple it's most likely to get."

For most of the companies that I've looked at the smoothest metric is sales. it's also the one that market cap tracks most closely. Operating earnings generally work quite well, too. Net earnings work OK, although less so that do sales or op earnings. I suspect that for a company like Exxon Mobil the metric that market cap might track most closely might be proven reserves because that's where most of Exxon's value lies, but I have not tried it. Of course one is not restricted to one metric for any given company. For example, for Berkshire one could use investments and operating earnings, but market cap tracks that less well than it does simply BV. The r^2 for market cap versus BV is 0.99. The r^2 for John Kish's IV estimates versus BV over the 33-year period 1981-2014 is 0.997. For Berkshire BV works quite well as a metric for forecasting market cap, as long as we can predict BV and P/BV :)
No. of Recommendations: 1
"For Berkshire BV works quite well as a metric for forecasting market cap"

At least it's worked extremely well for explaining market cap over the past 52 years. How well it will continue to work could be affected by things like buybacks, the accounting method used for mergers, etc.
No. of Recommendations: 4
"Pick a stable metric, estimate where it will be, and estimate what multiple it's most likely to get."

Trying to forecast return this way can be humbling, but in general I think it works reasonably well. It's not theory; it's just algebra. The more accurately we can predict growth and final price multiple, the more accurate the forecast.

One problem we run into is that the stable metric is not always stable, or even smoothly changing. In the case of Berkshire Hathaway , if you plot BV versus year over the period 1965 to present, it's not a smooth curve, but rather a curve with a couple of knees. BV growth was more or less constant from 1965 to 1980, then from 1980 to 1999, and then from 2000 to present, but with markedly different growth rates during each period. From Oct 1965 to Dec 1980 the BV growth rate (least squares) was 18.6%/yr; from Dec 1980 to Dec 1999 it was 28.8%, and from Dec 1999 to Sept 2016 it was 10.5%/yr. (Updating the latest period, from Dec 1999 to Dec 2017 it was 10.6%). The S&P 500 index exhibited similar knees at roughly the same times. From Oct 1965 to Dec 1980 the growth rate of the S&P 500 index (least squares) was 1.4%; from Dec 1980 to Dec 1999 it was 13.0%, and from Dec 1999-Sept 2016 it was 3.1%. Predicting the growth rate, like predicting the final price multiple, requires some caution.