Though it's largely a reprinting of a group of previously published essays that fit well together, One Economics, Many Recipes: Globalization, Institutions, and Economic Growth by Dani Rodrik helped me understand how he would apply his principles of "diagnostic development." The first section (3 chapters) lays out the case that successful countries develop using different strategies that may or may not follow standard prescriptions, with the emphasis on NOT. This sets up the second section (2 lengthy chapters plus a very short summary) on what industrial policy and governance institutions look like in the context of diagnostic development. The final section - a little less-well integrated with the other two - expounds on the lessons of different policies for the case of globalization.
I find myself in large agreement with the overall diagnostic principle, which says that we as economists should identify the most binding constraints on an economy's growth and develop context-appropriate policies to deal with that particular constraint. He argues that one of three basic problems likely to lead to low investment: low social returns, low private appropriability of those returns, or high costs of finance. Each of these might have several causes. You go through the country's (or sector's) statistics to determine where the constraint is likely to be. If interest rates, capital account deficits, and the returns to investment are very high, like Brazil, it's probably the high cost of finance. He narrows it down further to the low availability of domestic savings. In El Salvador, on the other hand, interest rates and returns to human capital are low, taxes and corruption are low, there's macro stability and good property right protection ... leaving by a process of elimination that there is a lack of entrepreneurship and new ideas that is inhibiting growth there.
I'm more reticent to accept his push for industrial policy, however. While he can cite a number of successful cases where direct government support of a sector paid large developmental dividends, it seems as a practice to have as many failures as the Washington Consensus. In part he calls this a virtue: a government that has only successes probably hasn't been doing enough, he claims. I'm less convinced.
Among the things I liked from his discussion were the institutions that need to be in place in order for good outcomes to be more likely: "Effective industrial policy is predicated less on the ability to pick winners than on the ability to cut losses short once mistakes have been made," addressing bureaucratic capacity as a scarce resource (which is just as scarce a resource for employing Wash. Cons. reforms too), and an acknowledgement that this is not a story of "omniscient planners ..., but of an interactive process of strategic cooperation between the private and public sectors that ... elicits information on business opportunities and constraints and ... generates policy initiatives in response." He puts a large premium on transparent, accountable, participatory governance, another point in his favor. Since he argues throughout that higher-level economic principles do not translate obviously one-to-one into particular institutional frameworks -- multiple institutional arrangements can generate substantively similar results -- he focuses on those higher level principles ... and then oddly enough spends a large chunk of the chapter on cross-country large-n regressions, which are nice and supportive but perilous, particularly when he criticizes similar work by other high-caliber economists (Jeff Sachs and Daron Acemoglu and their co-authors, for instance).
His discussion of globalization largely focuses on preserving policy space for countries that want to use industrial policy or other heterodox methods, and the difficulties (he says impossibilities) of simultaneously preserving national sovereignty, mass politics, and global economic integration. There's a brief plug for expanded migration as part of a multilateral framework.
The globalization section shines in discussing the international governance architecture. 1) Instead of trying to maximize trade, put the emphasis back on maximizing economic growth and poverty reduction; 2) Instead of trying to harmonize (i.e. make uniform) trade, finance, etc. policies, the global institutions should preserve nations' policy space while reducing the transaction costs between them.
Things I thought were missing: aid institutions, more discussion of when Washington Consensus reforms are likely to be the binding constraints, doing more to answer why import substitution industrialization largely failed (he just asserts it worked and is done). He lists 83 episodes of sustained growth spurts, and I would have liked more information about more of them, particularly in Africa.