In recent collaboration together with Johannes Pfeifer, we replicate the study done by Gali (2010) : Monetary policy and unemployment, which can be found here.
In his paper, Gali builds on the New Keynesian model and relies on the search and literature to integrate the labor market frictions in such a model. The main finding is that labor market frictions matter, both for the shape of the economy and for the design of monetary policy.
But I'm not writing this blog about the paper and the topic of monetary policy and unemployment. This post is more on the experience of replication, what it teaches and what it reveals.
This experience -of replicating other's work- was interesting in more than aspect. First, the deep understanding of a paper cannot be achieved only by reading. More you get involved, more you enhance your understanding. Second, we learn important things: how to build a model with all the difficulties that go with, how to ask a questions about the plausibility of approach used, how to go further and extend the model, etc. Always, when reading an article we tend to have the impression of that we can do the same easily. But when engaged really in the work lot of difficulties arises, which are challenging and of course learning-enhancing.
More interestingly, we found that we cannot replicate exactly the model of Gali (by using his calibration values) due to some inconsistencies in the calibration and in the definition of some variables. Lesson to learn is not to always take the results of paper as granted!
Gali (2010), is it a case of the 'crisis of replication' ?
Enregistrer un commentaire