My econometric half

In italiano

I am Full Professor of econometrics at the Department of Economics - Ancona University (officially known as Università Politecnica delle Marche -- yuck). For e-mail communications, please use this address.

Research

My research activity is somewhat "scattered". I've been doing many different things because, for a long time, I was the only econometrician in my department (not anymore, luckily) and, as such, I couldn't be picky. Most of the time, like a country doctor, you work on whatever circumstances dictate. I'm not complaining; quite the contrary, I actually like it. The list of my main works can be seen here. The one thing I would say I'm quite good at is writing code. For my involvement in the gretl project, see below.

SVARs

Perhaps people who are into structural VARs will find my article on identification in structural VARs interesting; the working paper version contains the same idea, but it's decidedly outdated and abounds in typos and assorted mistakes. Here you'll find the ox code to check identification conditions. I also wrote (together with Sven Screiber) a gretl package for estimating SVARs. If you try it out and spot any mistakes, please tell me.

IV Interval Regression

Another article that a few people found useful is the one I wrote together with Giulia Bettin called Interval regression models with endogenous explanatory variables; the title says it all. You will find gretl code for doing IV interval regression in this zip file. It contains the same data that W. Greene uses in hist textbook to illustrate ordinary IV estimation. The file "greene-ex.inp" contains the example code. Unfortunately, we can't provide the data used in the Empirical Economics article, as we have signed a strict non-disclosure agreement, sorry. To run the example, you should not need to modify any of the other files, but of course you're welcome to study the code and any comment is most welcome.

Remittances, IV double hurdle and all that

Giulia Bettin and I also wrote, together with Alberto Zazzaro, a paper on remittances by immigrants to Australia which attracted some interest and was published on the Journal of Development Economics. Apart from its content proper, a few people were interested in the econometric techniques we employed. If you download this zip file, you will find the relevant gretl code on how to implement IV versions of the Heckit and the double-hurdle estimators via LIML (no, we don't do that funky nonparametric stuff here, thank you very much). I'd love to give you the original Australian data to replicate the results in the paper, but the NDA we signed forbids me to. Therefore, in the zip file you will find a file, called "fakedata.gdt", containing artificial data, which are simulated in such a way that the results you get are roughly similar to those you find in the JDE paper.

Dynamic Panel Probit and Logit models

Claudia Pigini and myself wrote a nice little piece of code to estimate quickly and efficiently a variety of dynamic (that is, with the lagged dependent variable among the regressors) binary response models with panel datasets. It's called DPB, comes as a gretl package and the documentation is available here, although a proper journal article is coming soon in the JSS. Among the available estimators you have a nice assortment of RE probit models, plus the QE model by Bartolucci and Nigro (2010), which is more or less a FE estimator for dynamic logit models (not exactly, but to give you the idea). It handles unbalanced and gappy datasets, and it's relatively fast and robust.

Gretl

It is my pride and privilege to contribute to the development of the gretl econometric package. Gretl is different from almost any other econometrics package, as it's released under the GPL; this means that copying and freely distributing the program is not only perfectly legal, it's in fact encouraged. Clearly, this makes gretl ideal for teaching (availability in several languages being an added bonus point), but the most recent versions feature enough stuff to make gretl not utterly useless for research work.

Gretl is available for Windows, Mac OS X, Linux and BSD (click here to download). The software review on gretl that you can find on the Journal of Applied Econometrics is now very outdated; if you want to know what gretl does and how, the easiest thing is to download it and to try it out for yourselves: the user interface is so simple that everything should be more or less obvious. Before downloading the program, you may want to have a look at the manuals; however, the manuals are included in the installer, so you might as well grab the whole lot. Feel free to drop me an email for suggestions and requests.

If you know a bit of C, you can even have a go at contributing yourself (click here to have a look at the source code); not exactly easy-as-pie, though. Moreover, compiling the source requires a reasonably up-to-date Linux distibution (I suggest Debian or Mint).

Teaching

Lecture notes on time series analysis (in Italian -- December 2008)

Here are my lecture notes on time series analysis (in Italian). There will never be a "final version". This one is substantially revised and enlarged and there's even a preliminary attempt to write a chapter on GARCH models. Thanks in advance to anyone who spots and reports any error, omission or general blunder. If you like them, please drop me an email (shameless ego-feeding). Note:I've been working on those lecture notes too little recently. I must find some time to update them. Damn.

Appunti di analisi delle serie storiche (pdf format, about 1.8Mb)

Data in gretl format Here you'll find a zip-compressed file holding the data (in gretl format) used in the examples.

Old tests

"Econometrics"
"Elementi di Econometria" (in Italian)