I have considerable belief in microefficiency of liquid organized markets. I am doubtful about any great macroefficiency. General Motors' stock gets priced before and after dividend payments to preclude easy pickings. GM convertibles sell about right in ratio to GM common and GM preferred. (For a small firm, followed by few investors, I'd have to modify this plug for microefficiency.)
Why the shouldn't I believe that the level of the whole equity index gets priced right in good times and bad? The difference is that when Franco Modigliani sees a mispricing of GM common and preferred, he and others can make profits doing what corrects that discrepancy. And that's why the microefficiency gets wiped out as soon as it becomes significant enough to become recognizable.
How does the story run when, in the late 1970's Professor Modigliani opined that the Dow was below 1,000 "irrationally," in the sense that be believed it would be at least 1,400 if investors and speculator understood how not to double count the bad effects on corporations of then-current inflation? All he could do was write about it. Arguing with the tape by selling the general index short could be costly, and in any case ineffective, while animal spirits were what they were and analysts' shortcomings were what they were.
That is why I stand with Shiller (1986) as a doubter of market macroefficiency.
Caveat: I suspect that macro-mispricings will, as Japan during the period 1987-1992, create some tension toward correction. Therefore, I would expect a kind of bounded degree of macro-efficiency to prevail most of the time in the long run. Although I personally am cautious in trying to "time," when Tobin's Q wanders too far from its historic haunts, I find myself making modest changes in degree of equity exposure. Who's being human now?