Tuesday, May 18, 2010

Why recent tax rebates did not work

There used to be a time, not long ago, where tax rebates were popular. The idea was that putting money in people's pockets would drive them to spend it all and by multiplier magic the economy would grow sufficiently to recoup the lost taxes. And it did not work, as the Ricardian Equivalence would state, people mostly saved it all up for the looming tax increases. The only real impact came from those households where the Ricardian Equivalence would not hold, the cash constrained ones.

Thomas Bishop and Cheolbeom Park argue that the latter are a dying breed, and this is why the 2001 tax rebates in the US failed even more than the previous ones. The reason is that the increased availability of credit card lines reduced the number of households that were borrowing constraint, and those holding credit card debt simply applied their tax rebate against the current debt. The only ones who would spend the tax rebate are those who maxed out their credit card and still want to spend more.

All of this is rather obvious, and we all know these theoretical results since the works of Mark Huggett, Rao Aiyagari and Christopher Carroll. But it bears repeating, because people, and especially politicians, keep ignoring this.


VIlfredo said...

EL, your post contributes more to understanding the question than the paper you quote, why did you not label it "bad research"? All simulation results are well known, and they offer no empirical analysis...

Agent Continuum said...

Of course, Ricardian Equivalence might also not hold if households can avoid some of the future tax liability, by switching at the margin to household durables (untaxed capital, as it were) or if there's imperfect intergenerational altruism.

Also, under a progressive tax system, a household's future tax liability can be rather different than the rebate received now.

In any case, I agree, "fiscal stimulus" is not a particularly well thought-out idea.

rosserjb@jmu.edu said...


Gosh, you must be right. Declining effectiveness of tax rebates clearly demonstrates the general ineffectiveness of all possible fiscal stimulus policies. This is just so obvious, I do not know why you even bothered to post your comment making the point.

Agent Continuum said...

The first 2 sentences in my comment gave theoretical reasons why the zero-result under Ricardian Equivalence need not hold exactly, since R.E. came up.

My last sentence expresses my view that, even if R.E. does not hold, "fiscal stimulus" is probably still a bad idea. Maybe I should have abstained from just stating that.

Thomas said...

i am very flattered that you cite this old paper, but it models atomistic people with identical preferences and with rational expectations, a model which i no longer believe in.

the relationship between debt, borrowing capacity, the many unobservable preferences that we have, interest rates, our expectations about income growth and other things about the future on the one hand and the willingness to spend from additional income is an interesting one, but more complex than currently available simple models can address.

this paper and a related paper of mine also try to identify and represent mathematically unobservalble preferences and expectations, which might be interesting for theorists, even if it sounds boring for politicians and other policy makers.

in any case, thanks for the invitation to the blog. it looks like it could invite some interesting discussion.

VIlfredo said...

Thomas, I am puzzled why you disavow already a model you published as a working paper a month ago...turshe

Yannick Bourquin said...

As pointed out, these are only simulations and I think we should rely more on real empirical work to assess the effectiveness of fiscal stimuli. See for example : http://www.nber.org/papers/w13694

thomas said...

"Thomas, I am puzzled why you disavow already a model you published as a working paper a month ago..."

well, the paper has been work in progress for several years now, and during that time my excitement for the topic has waned.

the reason for my disavowal of the model is for reasons i stated in the post above: reliance on rational expectations, identical preferences and atomistic consumers where people make consumption decisions independent of others. this is just my current worldview, but others have made similar, but more rigorous, arguments. see for example, Manski, Charles, (2004), "Measuring Expectations," Econometrica, 72, 5, 1329-1376. See also Daniel Bernoulli's original paper (translated in 1954) on how to model individual preferences from consuming goods and services.

but at least the model does throw in an additional representation of the sensitivity of saving and borrowing to changes in interest rates. the simulations predict that this additional measure is not very important, given the construct of the model (under our specification of exogenous parameters). sometimes it's good to do an exercise to find out that that the exercise is not very useful, so that we can move on to more important things.