I’ve spent my last few flights reading The Limits of Institutional Reform in Development, by Matt Andrews. It makes a significant contribution to our understanding of why so many institutional reform efforts fail to accomplish much, and how those that succeed are different.

The book is distinguished by the depth of its analytical backing and the thoroughness of its framework. Andrews marshals a broad body of research to carefully build the case for an emerging approach to institutional reforms. If that sounds intriguing – congratulations, you’re a development nerd! For everyone else, proceed with caution: this book is not for the lay reader with a casual interest in government effectiveness. Andrews is an expert on this topic and he’s writing for those who do this work for a living. The result is not light beach reading.

Current reform limits

Andrews documents the struggles of donor-supported institutional reforms in developing countries. Such reforms seek to improve governance across a variety of sectors: security, public finances, health, education, etc. However, they rarely have the intended effect of strengthening the target institutions. Instead, the efforts incentivize mimicry of international “best practices” but fail to account for context. In short: the institutions look better without actually being better.

Dissecting the many failures and inferring lessons from a few successes, Andrews outlines a better approach to institutional reform.

Better way through PDIA

PDIA stands for “problem-driven iterative adaptation” – though I prefer Andrews’ alternate formulation of “purposive muddling” for its simplicity. Still, it’s worth breaking down his acronym.

The idea of being problem-driven means that, rather than focusing on the solution being implemented, reformers should focus on the problem being addressed and then seek small incremental steps that move in the right direction. It’s worth highlighting one critical question that some other reviews of this book have glossed over: whose problems do the driving? The book brings some nuance to the development shibboleth of “ownership” with the concept of multi-agent leadership. Clearly, reforms will struggle if they target problems that are only seen as such by outside actors; yet having an internal champion may also be insufficient. Andrews makes the case (in the book and elsewhere) that a range of lower-profile actors throughout the reforming institution must be involved in the problem definition and search for solutions.

The iterative nature of PDIA emphasizes the need to tinker, try new things, and see how they work. Those who think in terms of design research principles should find this intuitive: good design isn’t about a perfect definition of the solution in advance, but rather about a process of iterating toward the optimal outcome. We humbly acknowledge that we’ll fail along the way.

Finally, adaptation ties in the longer-term need to promote adaptive capacities within institutions. Reform processes should not simply lead to an improved institution, but to an institution which has its own mechanisms for continual improvement and response to a changing environment. (It’s a critical component, though the book fails to give it quite the depth of support that the rest of the framework receives.)

Beyond “context matters”

The importance of context is a common observation in development. It’s almost become a cliché – which is a good sign, if you ask me, because it means that we’re ready to go a level deeper. After all: if context matters, how do we actually take appropriate account of it?

Andrews underlines the importance of context in explaining past reform failures. Then he takes us to that deeper level by teasing out how successful consideration of context differs from failures. Prior analytical work (needs assessment, political economy analysis, and so on) doesn’t always improve contextual understanding, as it often proves (in retrospect only) to be the wrong kind of analytical work. Some projects even succeed without much in the way of prior research.

The key lies in flexibility. Projects that successfully account for context do so throughout the course of the project, rather than just in advance. These projects are driven by the problems being addressed, which remain rigid even as milestones, goals, and technical content are adjusted. This allows reformers to reveal and muddle through significant elements of context as they proceed.

Even more importantly, context is not simply a constraint on change. It’s also a target of change. Politics, institutional norms, and other elements of context are precisely what reforms aim to change. A problem-driven learning approach will lead reformers to address these elements as their role in obstructing reforms becomes apparent.

Here’s how Andrews describes it:

Reforms should focus on the disruptive problems that emerge in particular countries, and designs should allow contextual complexities to reveal themselves while addressing these problems. This requires an approach to institutional reform that acknowledges its endogenous nature and seeks to develop and implement designs actively through a problem-driven incremental process in the context, with local agents occupying the driving seat. This is proposed as the only way in which contextual factors can be effectively accounted for in institutional reforms.

PDIAing the development sector

The final chapter turns PDIA on the development sector itself. After all, bilateral and multilateral donors provide significant inputs to both successful and failed reform efforts. These outside actors structure many of the incentives that encourage poor practice: the “projectization” of aid, the international acclaim that encourages isomorphic mimicry, etc.

Andrews sounds a hopeful note that a 2008 report from the World Bank Independent Evaluation Group could serve to disrupt the development sector’s approach to institutional reform. He’s also cautious given the strength of the dominant logic. More work is needed to force reflection on the current approach. This book contributes to that. The entrance of new players (an anthropologist running the World Bank?) could also shake things up.

A story without characters

Throughout the book, I noticed a conspicuous absence of characters – that is, descriptions of the individual people involved in the reform processes studied. The absence has both stylistic and methodological ramifications, but that’s more than I’ll get into here. Head over to Reboot’s blog for the full story on that.

Summing up

PDIA isn’t an entirely new approach. Andrews admits as much. Elements of it exist throughout current thinking and practice in the development sector. You also see overlapping principles in other fields: designers value iteration, as mentioned; community organizers know problems must be locally defined; and corporate managers know that any change process requires buy-in throughout the organization.

Where this book succeeds is in pulling the pieces together into an analytical framework, and supporting it with an evidence base strong enough to ensure that its implications for development practice are taken seriously. It represents a big step toward better practice.

Related posts:

Full disclaimer: Many thanks to Matt Andrews for graciously providing me a review copy of the book, and apologies to him for the six months it took me to post this review! I ended up buying a Kindle version so I could read while traveling. I’m sure my views would not have been tainted in any case!

  1. […] you’re interested in digging further into PDIA, start with this post and consider reading the book. Fair warning though: this is not a pop development book for the […]


  2. […] at the Kennedy School. The focus of the event was recent work on governance reforms, including Matt Andrews’ book. The various talks and panels sparked several draft blog posts, which I’ll hopefully get out […]


  3. […] is just one manifestation of the reform problems that Matt Andrews has detailed. I think what I took away from the discussion at the Kennedy School that day is the need to be […]


  4. […] of iteration and adaptive management. Here the principles of “fail fast” and “purposive muddling” seek to embrace the lessons being learned. A recent movement to “admit failure” […]


  5. […] The need for both kinds of indicators also points to the need for humility about what indicators can accomplish. Indicators are great tools for managing linear, cause-and-effect relationships, but governance and system reforms are far from linear. Relationships between system elements are often complex, dynamic, and dependent on feedback loops that can be easily misunderstood. In such a context, indicators merely point the way to problems. Designing solutions requires deeper engagement, including a certain amount of iteration and muddling through. […]


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: