Does not compute

Suppose A \Rightarrow B. Steve Waldman says recent events have shown !A:

Economists talk about consumption smoothing, how it may be optimal for a consumer whose income is volatile to borrow during periods of low income and repay (or save) during periods of high income in order to maintain a constant standard of living. That’s very well in models where consumers know the true distribution of their future income, where the spread between borrowing and lending interest rates is not very large, and where consumer preferences are time-consistent. In practice, none of these conditions hold even approximately.

He concludes !B:

consumers ought to borrow only to counter severe downward shocks to income, pay off borrowings quickly, and build buffers of precautionary savings, since the cost of dissaving is much less then the cost of borrowing.

Dude, (A \Rightarrow B) \not \Rightarrow (!A \Rightarrow !B)!

7 thoughts on “Does not compute”

  1. Yeah. Logic is hard. Do you know anyone who is doing empirical work on how people estimate and update estimates of permanent income? Because I feel like if anybody could come up with some useful empirical generalizations about this (1) they could get really, really rich and (2) it might be pretty useful for economic policy.

  2. If by “people” you mean econometricians (as in “how do econometricians estimate lifetime income?”): I’m not “in” that literature but I know of this paper that tries to correct for the biases introduced by using current income as a proxy for lifetime income.

    I wrote about a reproduction of that paper using Swedish data here.

    Re making money: I suspect the errors from these regressions are highly correlated with locations of payday loan outfits.

  3. Yeah. By “people” I emphatically did not meant “econometricians,” though some people are econometricians. I meant, assuming that savings, consumption, debt choices are in fact in part determined by some kind of internal estimate of lifetime income as economists say (which is not obviously true), how is it that folks make this estimate in the first place, and how do events like recessions, bubbles, etc. cause them to update it? For example, if there is a “wealth effect” from rising home prices, how exactly does that work psychologicially? SOMEBODY ought to be working in that, right? Or do I need to start a start-up to do psychological research that we can then sell to hedge funds?

  4. You mean people don’t act like econometricians when they update their expectations? Weird.

    The running assumption in economics is that any error between what the econometrician estimates and how agents actually behave is random and uncorrelated with the things econometricians observe. This is assumed true even though econometricians observe lifetime incomes after the fact and agents don’t when they have to form expectations of their lifetime income. Rational expectations baby.

    Even if this assumption is true, there’s still tons of variation unexplained. We’re happy to get R-squares of 0.3 (i.e. 30% of variation is explained by what we observe). Agents know a lot more about themselves (e.g. their own productivity or motivation) so they may be better at predicting their lifetime incomes. But they may be subject to all sorts of biases too.

    The test of whether or not those biases wash each other out on average is just a test of the lifetime income hypothesis (i.e. do people smooth consumption). My opinion is that they do, usually. Just look at the consumption graphs I posted last week. Also, most of the exceptions to the lifetime income hypothesis that have been found (e.g. people’s consumption drops a lot more than it should at retirement, people consumptions doesn’t rise as much as it should with anticipated wage increases, etc) have been rationalized in one way or another. Still, I suspect individuals are subject to biases even if they do average each other out when aggregated.

    You’ll have to talk to the psychologists if you want to exploit individuals’ quirky behavior. I doubt they’ll have ways to systematically exploit this quirkiness. More likely, they’ll have elaborate ways to say its quirky.

  5. I wouldn’t call behavioral deviations from those predicted by economic models “quirky.” I would call them “the way people behave.” I strongly suspect there are regularities in the way people make sharp downward or upward updates in estimates of lifetime income, and that understanding this is pretty important. I mean, in the current debate over the stimulus, you hear many otherwise brilliant people making a lot of baseless conjectures about mass psychology — about consumer and creditor “fear” and “uncertainty” and what to do about it. But as far as I can tell, none of them have even a rudimentary theory about the causes of micro-fear or how it scales up to aggregate consumer demand/credit supply, etc. What I hear are people talking out of their asses about a subject they’ve never studied and pretending it is economics. Maybe I need to write a blog post about this…

  6. Your post on this rocks…

    The key question, I think, is are these biases systematically important? Economists don’t assume away the biases; we assume the biases are random.

Comments are closed.