An interesting interview of Paul Krugman by Edward Hugh is on A Fistful of Euros:
E.H. : The late Sir Karl Popper used to contrast what he regarded as science with ideologies like Marxism and Psychoanalysis, because there seemed to be no way whatever of consenually agreeing with their practitioners a series of simple tests which would enable their theories to be falsified. Some critics of neoclassical economics – including Popper’s heir Imre Lakatos – have expressed similar frustrations. Do you think we economists are, as a profession, up to the challenge of formulating testable hypotheses in such a way that the public at large might come to have more confidence in what we are up to, or are we a lost cause?
P.K.: I really don’t think that’s a helpful way to pose this question. Economics is about modeling complex systems, and as such the models are always less than fully accurate. What economists do need, however, is some demonstrated ability to get big things right. They had that after the Great Depression, when Keynesian economics clearly made sense of both the depression and the wartime recovery. But now the profession needs to get back on track.