“Economics department abandoned econometrics because it was rejecting its models”!

Paul Krugman writes:

OK, several correspondents have weighed in on the story I’d heard about the economics department that abandoned econometrics because it was rejecting its models. It wasn’t quite as alleged, but close enough.

The department in question was the University of Minnesota. For those readers new to this discussion, “freshwater-saltwater” was a distinction originally due to Bob Hall, who noted that the economics departments that had rejected Keynes and anything reminiscent of Keynes were inland schools like Minnesota, Chicago, and Rochester, whereas the places that retained a belief in the usefulness of monetary and fiscal policy were places like MIT, Princeton, and Berkeley.

So the story as I now have it was that there was harsh conflict between the macroeconomic theorists at UMinn, especially Prescott, and the econometricians who had the nasty habit of showing that those models didn’t work. And for at least some period econometrics was dropped as a required course for the Ph.D. — I don’t know whether it has been restored.

Now that is pure madness…

 

 

update:

There is no specific econometrics requirement for graduation, but most economists confront issues of extracting information from data throughout their careers, and the course sequence in applied econometrics will introduce you to some of the important issues in doing so. Moreover, econometrics faculty are unsympathetic to students who skip the course but then seek help with econometric issues in writing a thesis.

 

Excerpt from UMinn department of economics graduate student handbook

Tagged , ,

6 thoughts on ““Economics department abandoned econometrics because it was rejecting its models”!

  1. Costas says:

    Two interesting facts:
    Prescott started as an econometrician. See for example his paper here on estimation of stochastic parameters http://www.minneapolisfed.org/research/prescott/papers/estimation.pdf
    After he abandoned econometrics (in favor of calibration) he used to say jokingly that “it is better to progress than to regress”.

    • epanechnikov says:

      The only joke here is calibration. At least when used as a substitute of estimation. Calibrations are useful only when used to facilitate thought experiments with a particular model. They can not be used to test our hypotheses or to form any sort of beliefs.

      Hopefully they will soon realize that statistics is indispensable to economics research. That is unless they are happy doing their thought experiments instead of doing science

      • Costas says:

        Indeed, it is a mystery to me how eyeballing real and simulated impulse response functions to judge how close one is to the other passes as a scientific method of evaluating theories. Of course one can argue that the choice of confidence intervals is also arbitrary, but at least it is more informative as it allows you to know the probabilities or Type I and Type II errors.

      • epanechnikov says:

        Calibration implies choosing the underlying parameters in a way so that our model generates a path which looks similar to what we’ve observed in the real world. The value we conveniently impose on the parameters could deviate heavily to any data driven estimates of the same parameters. Hence our model could be completely wrong but bingo… we generated something that looks plausible.

        On the other side statistical techniques allow us to assess the fundamental assumptions of the model as all our inputs are data driven and not conditional on a particular macro model.

  2. Costas says:

    Epanmechnikov we agree, and your description of the two methods is excellent. The point I was trying to make is that there is no hard rule of deciding when a path looks similar enough. It is completely up to subjective interpretation.

    • epanechnikov says:

      Yep you are right Costas. There is no hard rule of deciding how close you are to the real path. No divergence measures, no error evaluation, no loss functions, no likelihood ratios… But still some people think they can do science by avoiding getting their hands dirty with statistics

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s