Is a steep Phillips curve anti-Keynesian?