I’m using Marco Ratto’s identification toolbox and it used to work fine, but now it crashes when it tries to plot the measures of identification strength… I get the following error:
Error using bar (line 55)
Only complex vectors are supported.
Error in plot_identification (line 74)
Error in dynare_identification (line 341)
If I understand this correctly, the problem is that Matlab tries to plot complex matrices (which gives an error). I understand that the measures of identification strength are defined as the square root of the squared change in the likelihood times 1 - the squared correlation term. Shouldn’t this term be positive by definition, such that the square root is real? How is it possible that I get complex numbers? I’m worried that there is something strange happening in my model…
Any hints are highly appreciated. Thanks a lot!