Dynare 4.2.3 and means in second order approximations

I'm obtaining theoretical first moments with Dynare 4.2.3 that are different from those under 4.2.1 and 4.2.2. when solving models with second order approximations. Is this a bug of Dynare 4.2.3? The second moments are the same under the three versions of Dynare.
Thanks.
Thanks.