Subject: Re: OT: S&P versus T-Bills?
Everything in this long thread makes a lot of sense and without disagreeing with any of it I think the piece missing is return on equity of the underlying businesses. Over a long enough period that is the key determinator (is that a word?) of your return.
12% over say 60 years is 1000x your money. If valuations go down 75% over that same period you end up with 250x your money...

You're right that the return on the equity is a key determinant, though that isn't quite the right math.
That would assume that the 12% rate is sustained in real terms, and that all owner earnings are successfully reinvested at that rate. Lots of money is paid out as dividends, and lots of money is invested badly, and the 12% number is a little dubious.

A simpler view: I would take it as a reasonable axiom that the value of a collection of equities is some direct function of the (cyclically adjusted) earning power. Relatively few operating firms are appropriately valued based on net assets, so for a broad portfolio they could be ignored. There are complications in that taxes and interest rates and labour relations push the margins up and down over time, but in the end it's really the earnings that matter. Smoothed real earnings are up inflation + 3.83%/year in the last 30 years, which is a pretty good proxy for how much teh value of the index itself has risen. Add the average dividend yield of 1.88% to get a real total return of 5.71%/year, and that's a pretty good idea of how much value was created in that period. It's a fine number, but not close to 12%, nominal or after inflation.

I am a big fan of ROE, though, even if you look at nothing else.
e.g., in the same last 30 years, an equally weighted portfolio of the 50 US-listed firms with the highest ROE beat the S&P by 4.8%/year.
(after trading costs, taking them from the Value Line database, reconstructed and rebalanced quarterly)
I run some money in a quant portfolio that is, more than anything else, basically this strategy.

Jim