In an article where the term ‘dominant discourse’ is used almost 30 times, a reviewer finds it hard to avoid situating himself inside or outside the discourse. Ha-Joon Chang's (2011) take-no-prisoners approach to the application of institutional economics to economic development leaves little ground for impartial observers. While I agree with all three of Chang's main difficulties linking institutional economics with development economics, my hopes of remaining neutral have been dashed by my concerns about Chang's interpretation of the dominant discourse.

Here are the agreements. Chang makes three basic points about the intersection of institutional economics and development economics. First, institutions are endogenous: endogenous in the narrow sense that economic growth is both a result of institutional change and a cause of institutional change, and in the broader sense that economic development encompasses simultaneous changes in political and economic institutions. Second, the problematic use of expert opinions about the quality of institutions as proxies for institutions in empirical work. Third, the difficulty in moving from ideas about what institutions are and how they work, to an understanding of how and why institutions change over time. I agree that these are fundamental difficulties in understanding what institutions are, how they change, how they affect economies and societies, and how economies and societies affect institutions. I do not, however, think that any of these three problems are fundamental criticisms of the current conversation about institutions and development unless we take a cramped, narrow and ungracious reading of the ‘dominant discourse’. I address each area in turn and then close with some larger thoughts.

‘The currently dominant view is that institutions are the ultimate determinants of economic performance (e.g., for the latest statements along this line, see Acemoglu et al., Reference Acemoglu, Johnson, Robinson, Aghion and Durlauf2005; North, Reference North2005). However, the causality in the other direction – that is, from economic development to institutions – is usually neglected.’ (Chang, Reference Chang2011: 4) What do Douglass North (Reference North2005) and Daron Acemoglu et al. (Reference Acemoglu, Johnson, Robinson, Aghion and Durlauf2005) actually say about the endogenous relationships between growth and institutions? North (Reference North2005), in his book Understanding the Process of Economic Change, tries to integrate what we know about human cognition into a theory of institutional change. Perhaps North's Reference North1981 or Reference North1990 books (North, Reference North1981, Reference North1990) would have been better choices to characterize his thinking about institutions. Regardless of the choice, however, I cannot see how any plausible reading of the three books supports Chang's assertion that North treats institutions as exogenous to the growth process: one of North's central points is that institutions are endogenous.

Similarly, the Acemoglu et al. (Reference Acemoglu, Johnson, Robinson, Aghion and Durlauf2005) citation is to their essay in the handbook of economic growth. Again, it is difficult to read their essay as arguing that institutions are fundamentally exogenous to the growth process:

Acemoglu et al. (Reference Acemoglu, Johnson, Robinson, Aghion and Durlauf2005) are not just hand waving; the following section of their paper is devoted to laying out a conceptual framework in which institutions are endogenous outcomes. It is difficult to find the dominant discourse, as defined by Chang in this manner, as guilty of ignoring the endogeneity of institutions and development.

Chang's second point, his criticisms of cross-country growth regressions, are obviously correct although he is hardly the first person to make the points. The cross-country income data enabled a kind of empirical answer to the question ‘are institutions important’ for the first time in the 1990s, when a much wider cross-section of national income and product data appeared for societies around the world. Combining measures of economic performance with survey data on the quality of institutions produced clear evidence that institutions had a significant effect on income (Knack and Keefer, Reference Knack and Keefer1997). The question about whether high-income countries had better institutions because they had higher incomes, and not the reverse, remains controversial. But it is not a question that has been ignored. Acemoglu et al. (Reference Acemoglu, Johnson, Robinson, Aghion and Durlauf2005), Dani Rodrik et al. (Reference Rodrik, Subramanian and Trebbi2004), Rafael LaPorta et al. (Reference La Porta, Lopez-de-Salinas and Schleifer2008), Stanley Engerman and Kenneth Sokoloff (Reference Engerman, Sokoloff and Haber1997), and many others have addressed the endogeneity question and resolved the basic empirical question of whether institutions matter: they do. That is a contribution of the new ‘institutional econometrics’ and it is a very important contribution indeed.

That is about all the cross-country growth regressions do; however, they tell us nothing about how or why institutions change. Chang's third point concerns the process of institutional change, how easy or difficult it is to change institutions, and how the dominant discourse ignores the costs of constructing and maintaining institutions. The new institutional economics is built upon Ronald Coase, Oliver Williamson and North's insistence on the importance of transaction costs. Indeed, Coase's fundamental contribution is to show that institutions matter whenever transactions costs are positive (the inverse of the way the Coase theorem is often stated). For Chang to suggest that this work assumes away the cost of running institutions rings false.

But how institutions change has proved to be a much harder nut to crack. It is here that Chang's criticisms have the most traction, but also generate confusion. The last 30 years have shown that a theory of institutions based solely on transaction costs is inadequate. What goes for a theory of institutions goes double for a theory of institutional change. The use of neoclassical theory to help understand how institutions work and change has been fruitful, but it also has serious limitations. If we think of a narrow neoclassical theory as a simple property rights/transactions cost theory of institutions, then it tells us very little about how institutions change over time or why some institutions persist and others do not.

Again, I agree with Chang's points about the complexity of institutional change. But when Chang concludes his criticism of institutional economics in the plea that it ‘pay more attention to the real world, both of the present and historical’ (Chang, Reference Chang2011: 22), I disagree. I agree with sentiment, I am an economic historian, but Chang's criticisms really concern the lack of a general theory of institutions and institutional change. Chang's language about the dominant discourse gets in the way of seeing this for two reasons. First, Chang does not identify the dominant discourse in a clear and useful way. He starts out by focusing on the development policy and donor communities. The Washington consensus views represent something of a dominant discourse within the development community. Like all political agendas, the development community requires simple messages, not nuanced and subtle ones. The development community listened to the academic community when the academics said that institutions were important. When the policy makers asked the academic community how they should go about changing institutions in a way that could be sustained, the academic community did not have any good answers. In the absence of an academic consensus on what policies would work, in the sense of being feasible to implement, the policy community went about trying to implement a simple and common-sense version of ‘good’ institutions directly. The simplicity of the Washington consensus and good governance agenda draws most of Chang's ire.

Do the policy implications of the Washington consensus or the current good governance agenda really reflect an underlying simplicity in the way institutional economists, such as Douglass North or Daron Acemoglu and James Robinson, think about these problems? I do not think so. Chang tags a simple and reduced set of policy implications as the dominant discourse, lumping together the ideas of people ranging from conceptual thinkers to development practitioners at the World Bank into a common dominant discourse. This is not productive. They do not share the same ideas at the same level of abstraction or concreteness. It is not easy to characterize the academic spectrum, however. At the most policy-oriented end of the academic community are people pushing specific policy agendas, including the good governance agenda. Where should we place the dominant discourse in this spectrum? While there is no easy answer to these questions, Chang owes us a bit more nuance.

This brings us to the second problem with Chang's framing of the dominant discourse. The new institutional economics did not provide a complete theory of institutions because its primary raison d'être was to criticize the neoclassical economics of the 1960s and 1970s. For over a century institutional economics, both old and new, has made its living by criticizing the simplifying assumptions of standard economics (whatever the variant). Chang is firmly in this tradition when he argues that ‘reality is often stranger than fiction and therefore our theories need to be more richly informed by real-world experiences – both history and modern-day events’ (ibid.: 23). While it is always wise to remind ourselves that the world is more complicated than our models, that does not always tell us where we should work to extend our understanding to make better sense of the complexity. It may not be wise, for example, to say that a general theory of how institutions work, which is a necessary element in a general theory of how institutions change, must be really complex and stranger than fiction.

Chang performs a rhetorical slight-of-hand when he conflates the dominant discourse with a general theory of institutions, then criticizes the purported general theory for its weakness. There is no workable general theory of institutions available to us. The attempt to construct a neoclassical theory of institutions has not failed completely, but it has not succeeded either. Both North (Reference North2005) and Acemoglu et al. (Reference Acemoglu, Johnson, Robinson, Aghion and Durlauf2005) and their subsequent work (particularly Acemoglu and Robinson, Reference Acemoglu and Robinson2006, and North et al., Reference North, Wallis and Weingast2009) are attempts to move beyond neoclassical theories of institutions. These are attempts to grapple with endogenous institutions and the process of institutional change. We may be dissatisfied with the results, but it does not help to accuse them of ignoring endogeneity or change. This is part of the dominant discourse that Chang should applaud.

I read most of Chang's complaints about the use of institutional economics in development economics as a charge that there is no general theory of institutions, and I agree with the charge. Ideas such as the centrality of property rights and naive beliefs in their power to solve social problems, the power of democracy to do the same, or the inability of institutional theorists to decide among themselves whether institutions are really hard or really easy to change are simplifying rules-of-thumb that reflect the lack of a coherent general theory. Chang sets up the Global Standard Institutions and other shards of the debate as if the participants in the dominant discourse really claim that this is a general theory of institutions, and I disagree with that set up. Chang takes the simplest ideas about institutions and portrays them as if their proponents think they are complex, then criticizes them for their naiveté.

These are not easy problems to solve. While it is wise to remember that the world is more complex than our models, it is also wise to remember that no one has a workable general theory of institutions, particularly a theory of institutional change. Raising and arguing these questions, as Chang does, continues to be a valuable and necessary part of moving forward.