I have tried searching for papers that answer the following types of questions, “which wedges (labor, investment, efficiency, government consumption, etc) matter for business cycle moderations (for a given economy)”. I haven’t found one yet, but applying BCA to identify which wedges matter for moderations is not inappropriate, right? Typically, I see BCA applied to recessions, so that means expansions too. But why not moderations as I have not found one yet?
Or maybe if a particular wedge drives say the 1982 recession of the US economy, then it is likely that it also drives the period of moderation that followed?
One of the problems is the structural interpretation. Wedges will be correlated and are hard to interpret. That’s less of an issue for explaining recessions at the BCA will provide valuable information on what an economic model needs to accomplish. It will typically be less informative for longer periods of time. But a priori, there is no reason to apply the methodology to moderation periods.
Thanks very much!! May I ask this though? So for example, I have a DSGE model that uses data covering the great moderation period, say from 1985-2007. How do I know which class of frictions is appropriate for that part of the business cycle? In Chari, Kehoe, & McGrattan’s BCA model, the argument is that wedges can help us identify a class of frictions appropriate for a given period of the business cycle, right? Since we can map a class of frictions to wedges. But here, the period of interest is longer, not shorter, so BCA may not help.
Are there guides that can give a clue on which frictions to use in your model when your data sample is longer? I heard some researchers do it by trial and error, you know, like you compare the theoretical responses with the empirical ones and if they do not match well, you go back and introduce say habit persistence or some other friction. But sometimes, you do not always have empirical responses, right? And choosing frictions then becomes what you want vs what you don’t want, as against what is likely to be true.
I am not aware of any rule or good guidance. For some exogenous shocks it is obvious (like TFP). But for more complicated models, almost anything goes.