Shock decomposition: Dynare output graph versus graph with stored results

Dear all,

After estimating my model, I tell Dynare to produce the shock decomposition by calling the argument shock_decomposition inside the command estimation. For completeness, I also plot the shock decomposition in Matlab by using bar and calling the results stored in oo_.shock decomposition. The problem, however, is that the plot produced by Dynare and the one I produce manually seem not to coincide in terms of magnitude. The results are the same for technology shocks, but not for preference and interest rate shocks. Does anyone know why this is happening?

Many thanks in advance for your time and consideration.

Without the files it is impossible to tell.