In a 1996 article, William Milberg surveyed all international trade articles appearing in the AER, JPE, RES, and JIE from 1988 to 1992. He described two notable shortcomings:

Of articles containing no policy conclusions, 55.9 per cent included empirical studies. Of articles with policy relevance, only 16.1 per cent had empirical content. This is precisely the opposite of the expectation that policy-relevant research will tend to be grounded on empirical support, and that when such relevance is not at stake, empirical support will be less important. According to the survey, empirical analysis is often used to explore positive issues, whereas policy-related issues are most often analyzed using purely theoretical arguments…

[I]n a field where writing structure and even methodology are otherwise extremely uniform, the rhetoric of policy relevance is diverse, flexible and unrigorous. Paradigmatic convention seems not to dictate this aspect of the discourse. Norms of systematic and logical analysis break down precisely where the most is at stake in economic analysis – the legitimacy of its policy conclusions.

It’d be fascinating to see such an analysis of the recent literature.

Hat tip: Matthew Eagleton-Pierce.