Thursday, February 25, 2010

Avoid Bad Science!


In particular, bad science by extremes of the political spectrum - those are the ones that generally tend to produce it to try and win you over to their rather untenable beliefs.

An example is given today via the above plot, produced in American by a lady called Veronique de Rugy, who is somewhat anti-government, and hence would fit in nicely with the 20 economists, and George Osborne's Tories.

The plot shows that private sector unemployment fell in the US since the stimulus package (that injection of government cash into the circular flow of income), and even public sector (government) employment fell too.

This, it is claimed, is clear evidence that the stimulus failed. Please, as a result of your university education, if you take one thing out of your degree take out the ability to analyse bad science like this!

To show properly whether or not the stimulus worked we would need to analyse a parallel version of the US (or UK) economy, in which the stimulus package was not launched. This is how in an experiment you would analyse the effects of some treatment (say a new cure for cancer): You give some subjects the treatment, and restrict the treatment from another group, the control group. Then you compare the outcomes between these two groups.

Of course, the problem is the control group for the US or UK economy, the version of it that didn't get the treatment. It doesn't exist.

So we actually have no basis on which to analyse the stimulus package, because we have no idea how many jobs would have been lost had the package not been announced and implemented. Of course that kind of caution doesn't tend to hold back economic fundamentalists like de Rugy.

It may be that the stimulus failed, it may be job losses were higher under it than they might have been without it - but the important thing is that we'll never know this. This is somewhat unsettling, but we're not 100% without hope.

Economic theory, the kind of stuff you're learning this term, allows us to start trying to analyse things like the effect of policy. Because in economic theory we build up a simplified theoretical model of the economy. We then can, in that model economy, look at what happens if we increase government spending, and what happens if we don't increase government spending. It's not perfect, since our model are simplified. But it's a much better attempt at the problem than the bad science example above from de Rugy.

No comments:

Post a Comment