Kevin Hoover quote on microfoundations based on representative agent modeling. Aggregation based on similarities of individuals assume similarity of individuals. In some cases that hold, in many others, not, and where it doesn't hold are the points of interest.
Not only does the representative-agent model fail to provide an analysis of those interactions, but it seems likely that that they will defy an analysis that insists on starting with the individual, and it is certain that no one knows at this point how to begin to provide an empirically relevant analysis on that basis.You can't get there from here.
Lars P. Syll’s Blog
On the poverty of microfoundationalist fantasies
Lars P. Syll | Professor, Malmo University
5 comments:
It's is a fourier transform of the population. The simplistic categories people are simplified into when becoming representative agents removes the interesting part of economics. It's the same operation as compressing a beautiful picture of Pluto into a compact JPEG for the news paper and 90% of the detail is gone and people fail to understand why scientists were so giddy about a few blurry mountainous dots on a picture.
I'm not so sure about the critique.
The problem is unless you understand the aggregation procedure you end up curve fitting at the macro level. So you end up treating Philips curves as sacrosanct when they are merely empirical regularities at a particular point in time.
You *must* have a modelling system that allows the aggregate to emerge from the micro inputs. Half the problems I've seen with the Post Keynesian dynamic modellers is that at best they are using Block Diagram System Dynamics simulation techniques rather than Agent Based modelling.
And that means they struggle to have the aggregate emerge. Instead they have to assume and assume that the aggregate remains constant under change.
System Dynamics suffers from the Lucas Critique.
They choose a block diagram of their favorite explanations and imagined rationales of how an economy 'should work' using their simple limited human imagination, usually with little real experience out side of academics, then attribute observed activities to "agents" in those goofy imaginary categories, what could possibly go wrong? Most of the categories are designed based on information easily had from national accounting identities and other measures that are usually extraordinarily flawed. Even the best statistical agencies in the world, like Stats Canada have awful data. Places like China have imaginary data. Corrupt places like the US and Nigeria have anomalies and model adjustments during election years with results that are so far out normal patterns of deviation but economists pretend that everything is hunky dory rosy peachy. Minor details like compiling aggregates from data reported where companies or banks are trying to avoid taxes and reporting income are out of sight and out of mind. In places like the UK, the entire country's productivity numbers are suspiciously low because the top companies and top 1% claim the earnings and income made in the country, on things as innocuous as a cup of coffee occurred in far away tax havens, much less income from their primary industry of finance where it is much harder to track income. By the time economists try to fit their imaginary models to the data... LOL, a bit of a joke.
Seems to me there is a fundamental difference between constructing an "aggregate" from the individual level, and obtaining an aggregate from data and deconstructing to the degree possible. It won't always be practical to get down to the individual level with great specificity. But the aggregate is supposedly based on data that originated as the individual level.
However, modeling "aggregation" by assuming uniformity of individuals is just clearly wrong when that uniformity doesn't exist.
Combining my comment and Ryan's, macro is at best a stab in the dark.
Post a Comment