Dear Johannes,
My work is based on the non-stationarity of hours worked. I compare the IRs from the actual and the simulated data.
The actual data tells me that hours worked are non-stationary. What I did in my work is following:
1. I estimate hours and GDP using Structural VAR with long run restriction. In VAR model: first variable is hours and second variable is GDP. Also, first shock is labor supply shock and second shock is technology shock. So, the interpretation will be : only a labor supply shock has a permanent effect on hours and both shocks will have a permanent effect on GDP.
2. Then, I build a DSGE model to obtain the non-stationary hours (simulated data) and apply the SVAR procedure that I explain in above. I adopt the DSGE model proposed by Chang et.al (2006) including permanent labor supply shock which yields non-stationary hours. I also have a technology shock. Both shocks have a unit root process. So, this model implies that both technology shocks and labor supply shocks have a permanent effect on GDP; that only labor supply shocks have a permanent effect on hours worked.
I got some feedback in my presentation saying 1. I cannot use the preferences that I have (please see https://www.dropbox.com/s/xn4g6ywicjv1i ... 8.png?dl=0 ).
2. The long run restriction is not valid unless I convince people that the reason why hours are non-stationary in the data is due to labor supply shock.
It is still not clear to me which types of preferences I should you? Where should I look? i would greatly appreciate if you could give me some suggestions about it.
Best