Explosion in model with two deterministic shocks

Hi,

I am running a basic NK model with Government Spending and ZLB. Preferences follow an AR(1) with persistence 0.85. I add an AR(1) process in log government spending with same persistence. In order to replicate the Great Depression output and inflation, I add a large negative shock in preferences and also a small positive shock on Government Spending. However, the system explodes. However, if I give a negative shock to government spending, things work fine.

Any insight would be appreciated.

Thanks,

Please post the mod-file.