While I agree that buybacks of very overvalued stocks are not in the shareholder's interests, I also think the danger of this is perhaps overstated.

In a general sense, both Graham and Buffett agree that the market is a weighing machine over time. That it is generally efficient. This doesn't mean that price accurately reflects value in the short term, but that it provides a close approximation over the long term.

Thus companies that have a continuing and relatively consistent, buyback activity over time should be getting approximately the correct value over time. This should particularly hold in the case of large, long established companies that are closely followed by Wall Street analysts.

As one example, Chevron just reported that their buybacks over the past two decades happened in three out of every four years. Total was $65 billion. And that their prices averaged $2 under market price over that period. That suggests that Chevron management did pay some attention to value in their buybacks, but the net effect is probably under 2% over time.

Would AXP, KO, JNJ, etc. be expected to be much different?

Obviously, there are periods to not be buying - BRK in 1998 is one example. And periods to be actively buying - BRK in 2000 another (missed) example. And if we really believe that BRK is consistently undervalued by the marketplace, a consistent buyback program should be in the shareholders interest.

I'm not saying that management shouldn't use some judgment at market extremes, especially for the kind of companies BRK invests in. But I do say that I think the risk to shareholders is overstated over time, especially for blue chips.

Other opinions may differ.