The posterior distributions are just like spikes

I am using Dynare 4.3.2 to apply Bayesian Estimation to CTW (2010) model. There are puzzles in my results.
The posterior distributions are very narrow when I use estimation with mode_compute = 6, compared to priors. They are just like spikes.
Since I don’t really know how can I solve?I look forward to any hints for this puzzle I encounter. Thank you!

Usually that’s a bad sign. Have you checked the jscale coming out from mode_compute=6? Often it is adjusted to provide a decent acceptance rate and only achieves this by making mh_jscale so small that the MH chain does not wander around in the posterior. If my conjecture is correct, try mode_compute=9. Moreover, try two chains (mh_nblocks=2) and look at the mode check plots and the (mode_check) convergence diagnostics.