12 thoughts on “A DSGE primer”

  1. Thanks. I had needed just such an article but was too lazy and ill-endowed to go find one myself.

    Question. Do they have more advanced models that separate out sectors of the economy? In particular, it would be nice if the product, service, and technology sectors were modeled explicitly with effects between them. The reason I ask is that I’m wondering if they can separate out the effects of Kling’s “recalculation” from Sumner’s tight money.

  2. Just look at what the paper considers to be the issues facing the economy, and look around now. The major puzzle is an increase in inflation from 1.5 to 2 (we’d call them the good old days now), and the cause? Management of inflation expectations by the Fed. I recall that oil hit 140 around that time, I doubt that had anything to do with the Fed “communicating” or “non-communicating”. A representative household, no financial anything, not even a hint that there are other things going on in the economy. I like to call it fin-de-siecle macroeconomics.

  3. Yes, I’m not sure what to say about this…

    As a basic building block, it’s OK, I guess, for some particular questions (ones which I don’t find all that interesting, but whatever).

    I’m not exactly sure who would benefit from this paper.

    Just to be on the record, my best bet is that the future will look like Bewley + Mortensen-Pissarides with a small menu of private and public assets.

  4. Kevin, the shock to productivity in this model can be more explicitly modeled as a sector specific shock to a multi-sector economy (I’m thinking of a presentation Kehoe gave at the 2009 annual meetings). Namely, you can do all the work of modeling extra sectors, but you get the same results.

    ssendam, no where in that paper do they say that was the most important macroeconomic event. They use that particular event for expository reasons; a simple case where the model can be used for policy analysis.

    AC, it depends on the question that the researcher is trying to ask. If you’re looking at the risk sharing or liquidity constraints then Bewley and if you’re looking at unemployment then M-P. Both of these sets of questions have been explored in the DSGE framework (check google scholar).

  5. This is what I don’t like. You build a model for this, you build a model for that… But what if I happen to suspect that both self-insurance and labor market frictions and a bunch of other stuff are all essential for the cyclical behavior of the economy?

    Everyone’s warning you against building “one big model” of the economy, but if you heed the Lucas critique and if you take structural, quantitative modeling seriously, that’s exactly what’s called for. Doubly so if you want to estimate it.

  6. Nothing personal, but I’m very suspicious when someone says that the behavior of a more complex dynamic model will be replicated by a less complex dynamic model. I’ve done enough modeling to see this assertion disproven rather often and am familiar enough with the complexity literature to know there are fundamental reasons for believing this not to be true.

    So are there multi-sector DSGE models? I’d love to take a look at what is considered to be a good one.

  7. AC, if you’re hypothesis is that both of those things interact in interesting ways, then yeah you need that bigger model. If not, then you don’t need “one big model” to answer all questions.

    Kevin, all DSGE are multi-sector… there’s a household sector, a monetary authority, etc, etc. Those sectors are different in ways that make modelling them explicitly an important thing to do. In your previous comment you asked about different sectors in the sense of industries. Unless you think there is some interesting asymmetry (in which case model it!), its a fact that those sectors don’t need to be modeled separately. An interesting asymmetry may be that policy interacts with the sectors different, for example. So in general its hard to answer the question, “are there good dsge’s with multi-sectors” because I need to know what sort of problem you have in mind.

    Is it possible to have more than one sector in this framework? Yes.

  8. Ahh, got it. Thanks. Interesting overloading of the term “sector”.

    I’m precisely interested in industries where there is an asymmetry that needs to be modeled. For example, declining vs static vs emerging industries; service based versus product based industries; and industries whose output increases the productivity of other industries (e.g., “Technology”).

    Another way to do it might be lifecycle stage. Back to the graph of job creation you pointed to. New, young, medium, and old companies. I could imagine incorporating something like a Markov survival model.

    I’m just trying to get a feel for how far the DSGE framework has pushed out.

  9. Hmm… I dunno. My friend was working on something like this. I think he found that some big names where basically doing the same thing so he shelved that paper. I’ll ask him about it.

Comments are closed.