Does a trivial econometric error explain why Andersson and Jonung (2014) get different estimates of a Swedish Phillips curve than the very robust estimates that I get in Svensson (2015)? Yes, their trivial error is not to have done the standard test for weak instruments when using regressions with instrumental variables. Their instruments soundly fail the standard Cragg-Donald F-test. Their instruments are weak and as a consequence it is their estimates of the Phillips curve that are unreliable, not mine. See this note.
(A previous response in English to their criticism is available here.)