VAR with assymetries

[alvarezcc]

Dear Tom,

I'm trying to measure the degree of passthrough from commodity prices to local inflation in a manner similar as Ferruci, Jimenez-Rodríguez and Onorante (2010) "Food price pass-through in the Euro Area: The role of assymetries and non linearities". My intention is to distinguish between the effects of positive and negative commodity price shocks in order to determine if there are assymetries asymmetries in the response of local inflation to those shocks. However, I'm not sure about how to estimate this effects in a VAR setting. In the aforementioned paper I think the authors separate the commodity price series into positive and negative changes by using dummy variables. Following this path would mean having a VAR with at least three variables (positive and negative changes in commodity prices and local inflation). The problem with this setting is that it is not clear to me whether positive or negative commodity price changes should enter first or second in the VAR ordering. In this case reversing the ordering gives quite different results regarding the sizes of the impulse-response functions.

Another possibility is to estimate two separate VARs, one including positive commodity price changes and local inflation and the other with negative price changes and local inflation. This path has the appealing of getting rid of the problem of deciding which commodity price changes series goes first or second, however my intuition suggests that both positive and negative price changes depend form each other lags, implying a misspecification problem.

Do you have any views about what path is the best to follow in this case. I would really appreciate your advice.

Kind regards

[TomDoan]

I think you're reading that wrong. The endogenous variables don't change. It's that the variables are split into positive and negative changes in the explanatory variables. See Kilian and Vigfusson 2011 for an example.

[alvarezcc]

Thank you very much Tom for your kind reply, reading Kilian and Vigfusson's paper cleared all to me.

Regards