Er, it's not a real technology, so it's a bit hard to characterize what the mismatch between devices would be!
The whole point of gpdk180 is that it is a sample, vaguely representative, 0.18um technology. You can pick whatever number you like! (I'm not sure what you mean by "mismatch coefficient" anyway, but it's not something that is included in the PDK). There are some basic monte carlo models which should include some mismatch variation - but how realistic they are is debatable (it's a moot point anyway, since the whole process is fictional).