A link from Marginal Revolution took me to a paper called "An Empirical Guide to Hiring Assistant Professors in Economics." It is as interesting for what it doesn't say as for what it does. It concludes that "top 30" Ph.D. programs in economics, which accept a bunch of quite bright college graduates every year, do a terrible job in making those who graduate capable of publishing work that academic economists find sufficiently worthwhile to accept for publication in academic journals. From all the top 30 programs combined, the average number of economics Ph.D.s in the period the authors studied was 460. Only 143 (31 percent) had at least one publication in any academic journal ranked by the authors six years after graduating. Even at Harvard, Chicago, or Berkeley, the bottom half of the class essentially published nothing. The paper is talking about graduates, and is excluding students who didn't complete their degrees. Another finding: for graduates who were not at the top of their Ph.D. cohort, Princeton, Rochester, and the University of California-San Diego seem to have provided the best preparation for writing publishable academic papers.
An interesting thing that the paper leaves unsaid is that books count for little or nothing. Journal articles are all that count. I found that curious because if you want to leave an impact in the sphere of ideas that is lasting, which let us arbitrarily define as 50 years in the future, a book is better than a journal article for trying to do it. Larry White and George Selgin's books on free banking from the 1980s will, I am convinced, still be read in the 2030s, by economists and by some interested noneconomists. Indeed, it is a strength of the idea of free banking that its audience is not purely academic. I cannot think of a single academic article from the 1980s that will be read by noneconomists (though I welcome your proposed counterexamples), and not very many that will be read even by economists.
Another interesting thing the paper leaves unsaid is what Ph.D.s who don't write academic journal articles do. The Dutch economist Arjo Klamer once wrote an essay called "Academic Dogs." (It is in David Colander and Reuven Brenner, editors, Educating Economists, Ann Arbor: University of Michigan Press, 1992; I haven't found it online to give a link.) He compared academic publishing to a dog show. Owners spend a lot of time training their dogs to run through some paces to impress the judges at the dog show, but how useful is that outside the arena? Klamer decided that both he and his dog would be happier by staying out of dog shows. It is implausible to me that more than 300 Ph.D.s a year from the top 30 programs have nothing to say. Rather, I strongly suspect many of them have decided that the dog show of academic publishing does not interest them. Some prefer teaching students to publishing. Others go work in nonacademic settings–consulting firms, central banks, government agencies, banks, think tanks, international organizations, etc.–and write for audiences other than the dozen people in the world who care about some narrow academic topic and are unable to make anything come of it in the world.
For recent or future Ph.D.'s who like the academic dog show, fine: you will probably do well at it. A few of you will even do work that is truly important, that changes the world a little bit or at least changes what teachers of economics teach their students about the world. For those who don't like the dog show, take heart: you were wise to have studied for your Ph.D. in economics, not English, and you have many ways of using your knowledge in a worthwhile way outside of academia, at a good salary.