From the archives:
[Krugman] uses theoretical modeling and observation of the real world to discipline and reinforce each other. In the introduction to Currency and Crises  , his second collection of articles, he includes a “personal manifesto” of his approach: “[T]he essential purpose of this kind of economic analysis is to build intuition … (and get] greater insight into an economic problem, an insight whose major content is often that of providing a language to discuss that problem. … This methodology brings with it a natural attitude toward modeling style: a very strong preference for maximum simplicity. … The point … is to wear one’s technique lightly.” But modeling and technique are essential components. When the world presents us with a genuinely new problem, “guesswork is all that we have to go on, and those who discipline their guesses with models are more reliable than those who fly by the seat of their pants, no matter how well tailored
by Dixit. See more here.