Note that firms are also often liquidity constrained (and often behave as if they were even when they aren’t; pecking order theories of capital structure and similar models under which propensity to invest out of current profits is greater than out of external finance are quite mainstream). A tax on profitable firms redistributed to loss-making companies would also probably have a multiplier.

Remember that the whole PIH came out of an effort to reconcile contradictory time series and cross section evidence on consumption and income. And it was roughly successful at this. Since then the analysis got more refined and currently as far as I know a “PIH with liquidity constraints” does very well at explaining empirical data (for example as in Muellbauer’s work). You control for some demographic effects and you’re golden. So you got a consistent theory, which doesn’t rely on ad hoc assumptions, which takes into account the fact that the future matters, which takes into account ‘real world’ institutional constraints and market imperfections and which does very well when tested against actual data. Why throw that out and rely on a hit or miss unrealistic heuristic (I mean seriously, do you really think that wage earners never save? And that investors never spend their dividends?)?

Just because it comes from a school of thought which has a certain ideological pedigree?

Here, we simply disagree. I think the assumption that profits are saved and wages are consumed is much more realistic—and more importantly, more useful as a heuristic—than the assumption that households are maximizers of income, consumption or “utility”.

Have you read much Lance Taylor? Check out Chapter 1 of Reconstructing Macroeconomics—it makes the case much better than I can.

And come on, arguing that saving behavior follows from some kind of hard wired class consciousness or something is wayyyyyy more unrealistic.

No. Because credit markets are far from perfect, because the future is fundamentally uncertain, and because lenders are concerned about liquidity. Your implicit model, where all future income flows have a known probability distribution, is not relevant to the real world.

Sure, that’s what I said.

“You can only argue that varying propensities to consume reduce to the permanent income hypotheisis if you think liquidity constraints are relatively unimportant.

Not quite. But I do think that the PIH is the natural way to talk about different MPCs. Why are some people credit constrained? Because their future (permanent) income is low. What are some other reasons for different MPCs? Because people are at different points along their lifetime income path.

“”the source of one’s income determines what one does with that income. Which seems silly.”

It only seems silly in a model—as in mainstream macro, where I guess you’re coming from—that doesn’t include firms, only households. ”

No, no, we got firms and households. But by their nature firms don’t consume and save, they invest, pay out dividends, and pay out income to factors. Again, what should matter is how much income a person has (both temp and perm) not what the source of that income is. Whether or not there are firms or just households’ is sort of irrelevant here.

“In the real world, the division of income between wages and profits is even more important in terms of the MPC than that between households of different income levels.”

It’s really not, and to the extent it is, it is only because the division of income between wages and profits correlates with the differences between household’s incomes. Hence, the assumption that workers have high mpc and capitalists don’t can sometimes be a useful simplifying assumption. But that’s all it is.

Ok, that’s what I was trying to clarify. Things got confusing because there were like two or more arguments criss crossing each other.
So, let me clarify again. If you take ‘basic Keynesian’ framework (like c=a+b(y-t)) and you compare it to a PIH framework you get a smaller multiplier. If you take a ‘basic Keynesian’ framework with different MPCs (reflecting liquidity constraints) you still get a smaller multiplier than from a PIH framework with roughly similar liquidity constraints. But yes, you can still get a multiplier out of PIH with liquidity constraints.
To put it another way, in ‘basic Keynesian’ framework (yes, the bastard, mainstream version) all that matters for consumption is temporary, present, income. In pure PIH with no liquidity constraints all that matters is permanent income. PIH with liquidity constraints is in between the two, where both matter (the fact that consumption depends at all on present income is sometimes even called the “excess sensitivity”).

Of course with a quadratic utility function, it’s all just a random walk, per Hall’s famous “He Must’ve Been on Drugs When He Wrote It” paper.

The confusion arose because Alex said that with different MPCs it’s no longer true that tax financed spending multiplier is 1 and smaller than a deficit spending multiplier. This is false. I didn’t have a piece of paper handy then but a quick run through the algebra confirms that even with different MPCs it’s still the case that if dG=dT then dY=dG, unlike the deficit spending in which case dY>dG. Then I made the off hand comment that if you’re gonna talk about different MPCs then you pretty much got to go with PIH since that’s the framework in which that makes the most sense. (Liquidity constraints in ‘basic Keynesian’ framework being an ad-hoc quick and dirty version of PI effects here).

So I don’t think we disagree here except for some minor quibbles.

“Just simplistically, a straightforward transfer tax from top to bottom deciles would increase the average marginal propensity to consume out of an increase in aggregate income.”

Not necessarily under PIH with no liquidity constraints.

I don’t see how you get from here to the no-multiplier conclusion. Surely progressively-financed government spending would increase consumption in such an economy? Just simplistically, a straightforward transfer tax from top to bottom deciles would increase the average marginal propensity to consume out of an increase in aggregate income.

Also, the real heroic and ubiquitous assumption here is that of generic “tax”. Pay for your infrastructure spending with a windfall tax on the banking system (that good old Thatcherite policy) and you can surely have a fiscally neutral spending with a multiplier.

In other words, liquidity cnstraints lead to a departure from permanent income. You can only argue that varying propensities to consume reduce to the permanent income hypotheisis if you think liquidity constraints are relatively unimportant.

the source of one’s income determines what one does with that income. Which seems silly.

It only seems silly in a model—as in mainstream macro, where I guess you’re coming from—that doesn’t include firms, only households. In the real world, the division of income between wages and profits is even more important in terms of the MPC than that between households of different income levels.

The classical assumption of “workers eat their wages, capitalists save their profits” if taken at face value would imply that somehow the source of one’s income determines what one does with that income. Which seems silly. Rather it was probably intended as a very simple basic model of the permanent income effect.

Why do you say that?