For your calculation on dividends, what was the assumption for change in stock price due to change in demand resulting from the issue of the dividend?

"Demand" doesn't generally change stock prices past the trading noise window. (prices are set purely by supply and demand for capital goods, but not so for purely investment securities, where it's better viewed as changing estimation of available return than supply/demand per se). Weighing machine and all that. Consequently I didn't do anything in terms of estimating a change in demand per se. US dividend payers pretty consistently have slightly lower average long run total return even before tax than non-dividend-payers, but I ignore that and assume it's meaningless and therefore a tie.

I assumed that the smallish reduction in cash did not change the appreciation of the value of the firm's operations and assets--the cash pile would still be huge today--and therefore did not change the valuation multiples the market applied to those assets. It's almost the same firm, so we'd expect about the same multiples of assets, revenues, and earnings. In other words, I calculated how much book per share would have dropped, and applied the same multiple of book observed on each day in the past to the slightly lower adjusted book per share. So a 1% drop in cash per share was assumed to correspond to a 1% drop in the share price.

There's another way you could go:

In theory a rational market should assign a high P/B to the operating assets and a separate P/B of (or near) 1.0 to the cash in excess of operational needs, giving a weighted average of (it seems) around 1.4. (incidentally, this theory does not seem to fit the data very well for Berkshire, but I digress). So a slight reduction in cash should cause a rational market to assign a slightly higher overall P/B: same high multiple on op assets, same 1.0 on a little less cash, slightly higher weighted average multiple, times a lower number. In this case the market price of a share would be lower by only the amount of cash paid out. (this is normal for ex-div day, but we're talking about the long run cumulative effect). On that assumption, the wealth penalty of 20 years of 1% average yield is only -2.6%.

For both methods, the figures I posted assume 25% tax on dividends. I pay 30%, but I'm a little atypical.

Jim