Ten things on development that conflicts with conventional wisdom

For an insider looking at the development discourse, there are several misconceptions about the theory of change in development, about how ideas get adopted and implemented. 

This post will highlight ten important aspects of the development discourse where reality runs contrary to conventional wisdom. Some of them have been very clearly articulated by Lant Pritchett in his works.

1. Evidence-based policymaking may be the most misleading phrase in today's development discourse. In reality, evidence has only a very marginal role in policymaking. 

Political economy and other factors prevent the termination or significant modification of any flagship program even if there is clear evidence that it's not generating impact. And new policies and programs in any major area, as we will see next, are very few.

Further, evidence of headline efficacy of an intervention is almost completely irrelevant, given the countless confounders associated with any implementation. Instead, the importance of evidence lies in triggering and contributing to debates that gradually change opinions. In other words, evidence contributes to the gradual creation of a body of knowledge on the issue. 

The most effective evidence is that which is of proximate relevance as decision-support in policy making and, especially, in implementation. This comes from the analysis of administrative data from implementation, comparative data, benchmarks and norms, historical records, etc. Unfortunately, these are also not considered rigorous or serious enough to be accepted as 'evidence' in the rarefied environs of development research. I don't recollect a single peer-reviewed paper in a reputed journal in recent years that has its findings based on administrative data analysis. 

2. Effective development is mostly about doing things better and only marginally about doing new things. 

Even with all the technological advances and other changes over time, the basic processes of most development interventions - learning and skilling, maintenance of cleanliness and hygiene, maternal and child health care, children's nutrition, agriculture productivity enhancement, women's empowerment, maintenance of law and order, targeting and selection of welfare beneficiaries, etc - remain the same and cannot also be done any differently. For sure, the contextual changes can marginally impact the core activities. However, given the low baseline of implementation quality and outputs in all these areas, instead of struggling to figure out the implementation of a new approach (which also requires strong state capabilities), it's better to improve capabilities to execute the familiar processes better. This is likely to generate the greatest marginal impact. In general, to move a system from poor or average to good, doing things better may be the most appropriate strategy. 

Further, in reality, those ideas presented as new interventions are either tinkering with an ongoing intervention or marginal addition to the main intervention. So, for example, new interventions like nudges to change behaviours and increase uptake are almost always marginal complements and cannot be a substitute for the core tasks. 

Wanting to change and do new things is a common human reflex when things are not going well. This urge is reinforced by modern management theories and the practices followed in the private sector. However, like with the other New Public Management ideas, this too has its limitations in the context of development.

Therefore, on average, new is required only as a distant secondary addition to primarily doing more of the same better. 

3. In the aggregate, it's more useful to be banal than be innovative in development implementation. 

Innovations are mostly of relevance in the field of product development, whereas in the case of economic development improvisations may have greater relevance. While development also involves products (IT systems, medical drugs, etc), it's mostly concerned with processes and their implementation. Unlike innovations which generally tend to emerge suddenly as mutations, improvisations tend to happen iteratively. 

Further, most often, innovations are a distraction, which displaces effort and resources from real tasks. The design and implementation of innovations require considerable systemic bandwidth. 

4. Policy implementation is much much more important than policy formulation, except in a handful of areas. And the main problem in development is pervasive implementation deficits. Most policies and programs fail due to bad implementation (and not bad design). 

The design space for policy formulation is much smaller than the possible degrees of freedom in its implementation. The scope for getting things badly wrong in implementation is far greater than getting the design completely wrong. In fact, it's rarely the case that the policy design is very bad unless it's a conscious choice. But implementation strategy can be flawed or weak in several ways, and its execution can go wrong in even more ways. 

5. The adoption and implementation quality of an intervention/program differ widely, translating into variable outcomes. This conflicts with the assumption and requirement of uniform outcome expectations. 

This expectation of uniform adoption, implementation, and outcomes shrinks the space available for experimentation and failures. It encourages officials to game the processes and manipulate data to mislead monitoring. 

Instead, we should be comfortable with variable implementation and outcomes. The strategy should be to engage actively with the implementation, collect feedback on what's going right and wrong, rectify the mistakes and adapt implementation processes, and iterate continuously. Successful implementation is almost completely dependent on the quality of this iterative adaptation.

Such iterative adaptation is also one reason to discount the obsession with getting the best policy design since the design features can be refined during implementation. 

6. A very important requirement for successful implementation is not the accounting of the accountability, but the account within the organisation of the accountability. 

Any implementation is monitored using a logical framework of accounting that's used to fix accountability. It's believed that such tight top-down monitoring can ensure effective implementation. But this belief is flawed on at least two counts. One, this log frame cannot capture the important small details of implementation, especially involving the aspects of quality, which are critical to the success of the intervention. Two, even if they can be captured for monitoring, in interventions where quality is of the essence, these aspects are not amenable to top-down direction. 

Instead, successful implementation requires organisational commitment to the cause or the account of accountability. This depends on the ownership, motivation, and intent of the individual officials and the collective organisational purpose. Both are endogenous and feed on each other. 

In fact, the inherent top-down control preference of a bureaucracy ends up conflicting with the bottom-up account cultivation requirement for successful implementation.

7. Effective implementation of new ideas, technologies, and interventions requires high state capability. State capability is the binding constraint to the adoption of reforms and execution of new ideas. 

Except in a few cases like certain legislative enactments, deregulation, and new drugs, where implementation is like turning on a tap, the vast majority of new development interventions require continuous and high-quality engagement by the system. This is particularly so with IT interventions where the initially installed design version will need to be debugged and iteratively improved for some time before we can get it right. This iteration will also help overcome the inevitable gaming by the entrenched vested interests who stand to lose from the new intervention. 

8. Most policy design and implementation decisions are exercises of judgment and not about making technical choices. Technocracy is unlikely to substantively improve development outcomes. 

There are very few cut-and-dry issues and contexts in development policy design and implementation. Decisions will have to be made under conditions of considerable uncertainty. These decisions are then about weighing the different technical and other factors and exercising good judgment. Good judgment is also about the experience. It requires the application of both expertise and experience and not merely expertise (technocratic). 

9. The most important factor in the process of getting policies approved or pushing reforms is not its technical merits but the management of the policy-making process and environment, and the policymakers. 

In democratic polities, even initiatives with broad political support must go through the elaborate bureaucratic process of consultations with different departments. There's a process of negotiations and bargaining, involving mainly the Finance, Law, and Personnel Departments. This tends to constrain the program's flexibility, local discretion, resource availability, etc in the guise of retaining bureaucratic control and limiting expenditure. Navigating this system and getting approval without seriously compromising the design elements require considerable skill and effort.

Apart from this, there's the need to articulate and present the ideas or policies, or reforms in a manner that is able to convince important decision-makers. Most often it's only one or two individuals who matter in the entire decision process. And narrative, framing, articulation, timing, and trust matters more in these decisions than evidence and logic. This is the much under-appreciated specialised expertise of the generalist.

10. In any successful implementation of a large quality-dictated intervention, the greatest contributor to success is invariably plain good luck! This good luck comes by way of the fortuitous confluence of several requisite factors - political resolve, bureaucratic leadership and commitment, ownership and active engagement among the implementing officials, community and societal support, stakeholder/user receptivity and interest, etc. 

As can be seen, these are all non-technical factors. None of them can be consciously planned and managed into being. Therefore the need for good luck.

Interestingly, the most important determinant of all these aforesaid factors is economic growth and the stage of national development. One more reason why getting economic growth right is the most effective development strategy.