Irf in second order approximation with pruning and without pruning

I find a question as i plot irf in second order approximation. There is a productiviy shock in the model. When i do not use pruning algorithm, the irf graphs of postive shock and negative shock are symmetric as expected. However, when i use pruning algorithm, the graphs are asymmetric. Could anyone explain it?

I have to use a algorithm like pruning or nlme in second order approximaiton, because the shock size is so small if they are not applied. what should i do to get a symmetric irf graphs under the same shock with different directions.

The code is attached as follows. You can run Main.M file to get the graph
Renjun.rar (3.5 KB)

Many Thanks

Why should you get symmetric IRFs? The model is not symmetric and you are solving it nonlinearly.

Thanks for your kind reply.

I think it should be symmetric as it is exactly symmetric in second order approximation without pruning. So asymmetry comes from pruning algorithm, not the model itself i guess. Why you think the model is not symmetric?

The IRFs are not perfectly symmetric. The differ, if only a bit for the small shock size.

ok! Thanks very much!