It seems to me that one should be an economist (or sociologist, or whatever) who evaluates the impact, rather than an impact evaluator. Chris Blattman worries about overspecialization:
Chris Blattman's Blog: So you want to be an impact evaluator? A cautionary tale: Several aspiring graduate students have written me about becoming an impact evaluator. 'What degree to get and where?' is the most common question. I think the best advice is: don't get a PhD to do evaluations. The randomized evaluation is just one tool in the knowledge toolbox. It's currently the rage, but that means it will probably be old news by the time you finish your PhD.
Yes, the randomized evaluation remains the "gold standard" for important (albeit narrow) questions. Social science, however, has a much bigger toolbox for a much broader (and often more interesting) realm of inquiry. If you want to know the effects of small binary treatments, you are in business. If you find any other question in the world interesting, you have some more work to do.... The best advice I can give to aspiring researchers: apply to PhD programs that will give you the best all-round training in as many different tools and pools of knowledge as possible.
Also: use your schooling time to tech up in formal theory and statistics (plus qualitative and comparative methods if you are a political scientist). Once you are finished, you won't have time to acquire these skills. From the day you finish your PhD, it is a slow but steady descent into technical obsolescence.
If you're interested in becoming a professional evaluator, rather than an academic, my advice changes little. It is now possible to be an evaluation consultant in much the same way that you can be an accountant or a lawyer--a highly specialized professional, with interesting and rewarding work. A PhD helps, but I don't think it's requisite. Yet I would still make the same case: you will be a better consultant, manager, and professional if you have a broad range of skills and knowledge...
If your goal is to improve the delivery of aid, and truly advance development, many more skills and knowledge are involved than the randomized evaluation. See here for more. But in short: a well-identified causal impact that arrives two years after the program does not performance management make.
For aspiring professionals, a masters in statistics with an MBA or MPA may be preferable, along with plenty of experience on the ground--preferably working inside a developing country government, not an aid agency. If nothing else, you may need a different set of skills if the fad ever fades.