Welcome to European Tribune. It's gone a bit quiet around here these days, but it's still going.
Quite apart from the strict time demands (I was sitting an exam on Friday and had to look after my kids alone for the rest of the day. Also, I need to find some work assignments. Those will tend to take precedence):

A correction, retraction or apology? What for?

My graph is a plot of table 3.2 of your paper. So that seems to fit the description that I'm plotting your data. In fact, the most recent data points in your sample -those that could, by the way, be seen as the most relevant to a paper pretending to give some insight into the future.
That you then decide not to use them in a regression that should not have been produced at all is neither here nor there. It would raise strong suspicions as to the motives for leaving those data points out, except that the conclusion is not supported by the 2008-11 sample either.

We're not at analytical stage yet with that. Practical assessment suggested that having your own currency was very important. You suggested as much but still included euro countries despite claiming that you had restricted the data to countries that had their own currency.
Then, moving on to the graphical assessment, it is good practice not to confuse stage effects from stratification ones. So, a time series that shows that things change after 2008 (and that's not passing a debt threshold since it does for countries with a rather low debt too), and a scatterplot with a single point per country, which will always be, unless there is a pressing reason to do otherwise, the most recent, with the obvious previously (at the practical stage) identified stratification is in order.

Yet, despite having highlighted it when wrongly claiming that you had only taken data from countries with their own currency, and while having the data right there, you don't do this simple stratification. In a way nevermind: any experienced data analyst will guess that there are two populations from just a passing glance at such a graph. But that's the thing: no regression should be attempted in that situation. Not on the whole data sample anyway.

Then, there is the little rule of not using a model outside of its scope, so using a correlation (probably not causation, the cause is far more probably to be found in current account imbalances) derived purely from euro countries to make policies recommendations in the USA is, as Eurogreen noted, to be avoided.

As for your implied assertion that my post does not follow the scientific process, well, I don't feel I'd need to defend that at all. It was posted on my blog, and cross-posted here. Neither is a noted scientific publication. I have never, in fact, published in a scientific journal. So, no, this did not have to follow this process, although posting anything where people like JakeS or Migeru are lurking involves some of the strictest kind of peer review -and it happens in public, as you can see.

Earth provides enough to satisfy every man's need, but not every man's greed. Gandhi

by Cyrille (cyrillev domain yahoo.fr) on Sat Mar 16th, 2013 at 04:01:04 AM EST
[ Parent ]

Others have rated this comment as follows:


Top Diaries

Occasional Series