Hi Moz,

The measurement issue is so fundamental for economics. There are three issue with the data:

1) You require a theory to understand what the data says in the first place – I would not call social data objective.
2) The data is viciously incomplete.
3) The data is very untimely.

I talk about the issues in more detail here:

http://www.tvhe.co.nz/2009/02/18/the-issue-of-data/

The Lucas critique tells us that we can’t just look at empirical relationships to figure out what is good policy – as policy changes these relationships by influencing the behaviour of individuals. The structure of the game is not policy invariant.

As a result, economists that just grab a whole lot of data and say X is related to Y so we should do Z are ignoring this critique. This implies that just grabbing the data and asking it what we should do is not the way to make good policy.

We need some type of “microfoundations” – we need a theory that tells us how agents behaviour will change when policy changes, rather than just looking at correlations in the data.

Hi Greg,

I agree that representative agents don’t really imply true microfoundations in the sense that we are aiming for here. A good post on the issue was here:

http://yetanothersheep.blogspot.com/2007/01/what-are-microfoundations-good-for.html

And a more recent and interesting critique was here (ht rauparaha):

http://www.voxeu.org/index.php?q=node/3210

However, I am unsure that ANY economic subdiscipline has been able to derive sufficient micro-conditions for the macro economy – as it just doesn’t seem possible. Even if it was – I severely doubt that you would gain a unique result.

Well yes, measurement is critically important to science. That’s why I qualified my comments with the remarks about science as distinct form politics or religion. If you want to talk about economics as politics you can move away from the observable world and talk about immeasurable stuff as much as you like. Or you can talk about economics as religion and (say) the epiphany of St Friedman.

Why do you think engineers believe that only measurable things matter? In my experience most engineers are well aware that critically important things can’t be measured. That’s one reason why so few of them dabble in politics, because they’re aware that their approach tends to play down non-quantifiable values. Even some economists do this – “Counting for Nothing” being one critique I hope you’re familiar with.

The Lucas Critique sounds interesting, albeit I’m not sure to what extent it boils down to “models have to assume that actors know the rules”. I assume I’m wrong to link this back to the whole “rational economic actor with perfect knowledge of the market” idea and ask how anyone could ever imagine that said actor would not react to a change in the market? Or is Lucas just the guy who’s recognised as pointing that out?

Simple examples about in software (my field), because we have to deal with users who are human. One classic is users ignoring dialog boxes. For a while there was a war between ever more irritating dialog boxes and users becoming more intent on “make it go away” and hence less likely to even realise that there was a message. These days we tend to go for automated error reporting and so on, specifically because dialog boxes only work if there are very few of them (a tragedy of the commons, even).

We even have the joy of public pronouncements from technically illiterate people who the public regards as experts for no reason other than that they always have something to say.

You are a really rare bird if you are actually serious about being serious about this.

– “K”

– “representative agent”

– a math function defining risk or uncertainty

– AS / AD

– IS / LM

The logic and understanding provided by “micro” is a joke for all the the main stuff where micro is actually needed in macroeconomics.

And even if they can be measured how to we decide how to interpret them, very true Kimble 🙂

“Where macroeconomics works is where it ignores microeconomics and works directly with statistical measures of real societies”

No, not at all. There is this thing in economics called the Lucas Critique – as soon as we have a measurable relationship between economic variables, and we try to develop policies based on it, we often find that the relationship evaporates.

The usefulness of statistics only comes about when we have a lens to view them through – microeconomics provides that lens.

Once again moz lets his engineering bias show. Wouldnt life be grand if everything was run by engineers? So the only things that matter are those that can be measured. If something cant be measured then it isnt worth considering.

Wait, applying an engineering methodology to economics is what is wrong with modern economics, right? Doesnt the scientism of macro-economics cause people to overestimate its usefullness?