Should instrumental variables be used as matching variables?

Z bias