Munger took some time before getting his eyes lasered today to say this:
To have an effect on economic activity, tax cuts have to be credibly permanent.
The assumption here is that people will recognize that present tax cuts will be offset by future tax increases, so unfunded tax cuts will have no effect on behavior: people will just sit on the money until they have to pay it back. And that's true if people are atomistic Friedmanite permanent-income-smoothing fully-forward-looking individuals with complete and perfect information about their future incomes. We know this from theory that is deduced from those assumptions, and if those assumptions hold, then the conclusions also hold.
But the assumptions of theory do not necessarily hold. All of those descriptors above are wrong or at least (I'm being very generous) probabilistic rather than certain, so the question becomes empirical, not theoretical. And the empirical record, as far as I can tell, is completely bollocks-upped.
I can easily imagine a situation in which someone offers to give me money today, on the condition that I have to pay it back tomorrow, and I nevertheless alter my behavior and spend it rather than save it. That's how credit -- the foundation of the capitalist economy -- works. Mastercard does not have to credibly promise me a free $500 that I will never have to pay back in order for me to spend. They only have to promise me $500 today. If I think spending $500 today is worth paying back $525 tomorrow -- say, because I'm unemployed, or my house is at risk of foreclosure, or I just got furloughed at work, or... -- then I just might do it.
Example: the cash-for-clunkers program is routinely criticized for doing nothing more than pulling future spending into the present. But that was the whole point. And it worked! There were a lot of car purchases during that window that would have been spread out otherwise. (IIRC, Munger took advantage of this himself, but I can't find the post now so it could have been someone else in my RSS.) Of course people with clunkers would eventually buy a new car, but the government decided that they wanted them to buy a car today, not tomorrow. As it happens I think that was a poor choice of priorities, as I think most (not all) of this grand Obama/GOP bargain to be a poor choice of priorities, but the problem wasn't that cash for clunkers didn't work; the problem was that it was a silly and expensive tradeoff.
This is something that has always confused me about rational expectations econ, but then my econ education ended in undergrad. I can't even pretend to have a great grasp on the advanced literature, so possibly all these objections are easily overcome. If so, I'm sure someone will set me straight.
UPDATE: I've been grading finals all day (All Done!) so I didn't see Angus' relevant reply until just now. Angus believes in rational expectations, PLUS seems to have a zero percent discount rate, and still finds (to his apparent surprise) that he himself doesn't act as if these were his true beliefs. People, doesn't this tell you all you need to know? If a true believer rat expectations doesn't act like he should, who will? Yikes!
UPDATE 2: I'm still catching up on my RSS, and now I see Munger wrote a follow-up post in which he describes a scenario in which a car dealer offers a customer a loan, to be repaid with interest, if the customer is willing to purchase a car today rather than after he has saved up enough to pay the full sticker price in cash. Munger views this scenario as absurd. I view this scenario as basically the single most typical durable goods transaction in the world. One of us has to be wrong.