Subak farm in Bali, Indonesia

A subak in Bali, Indonesia. (Photo via YouTube)

In the last decade, luminaries in global development like Dani Rodrik, Lant Pritchett, Michael Woolcock, Francis FukuyamaHa-Joon Chang, Peter Evans, and Merillee Grindle have rejected one-size-fits-all policy prescriptions. Instead, in response to the persistence of failed interventions in places ranging from Bali to Afghanistan, they stress the merits of tailoring solutions to each local community's situation.

“Going local” means to account for the specific needs and particularities of a community, whether national or subnational. It also entails using local resources and knowledge in a manner that goes beyond just setting up local offices and hiring local staff.

Although going local is widely embraced in principle, it may not be carried out in practice. For global development agencies to move forward, they must first reform themselves to overcome three major obstacles: lack of knowledge about non-best-practices that work, lack of diversity, and lack of incentives.

If we fail to address these organizational constraints, calls for change risk turning into nothing more than platitudes and buzzwords that exhaust practitioners and drive away true solutions.

Reforming global development agencies is difficult, but not impossible. I propose a number of ideas to help professionals who best know their organizational constraints change the way their teams work.

But first, it's helpful to look at examples of good intentions gone wrong.

Failure to Listen

In the early 1980s, working together with the Indonesian government, the Asian Development Bank (ADB) launched the Bali Irrigation Project. Stoked by the grand vision of modernizing Bali's irrigation infrastructure, the project aimed to replace ancient subaks—a system of water temples and terraced rice fields that had been autonomously managed by local farmers for centuries—with a high-tech, centralized network of canals. It intended to “provide employment,” “reduce rural poverty,” and “strengthen institutional capability.” Increased rice production and exports, planners expected, should adequately offset the projected cost of US$108.9 million.

The reality did not pan out as expected. Stephen Lansing, an anthropologist, found that the Balinese farmers removed newly installed metal gates in the canals as soon as they could.

The farmers weren't resisting modern technology; the new, expensive devices made it impossible for them to schedule water distribution among themselves, which they had done for centuries through temples and religious rituals. From there, the situation worsened. Encouraged by policymakers and experts, the Balinese farmers bought “technology packets”—pesticides and fertilizers—on credit. Whereas the traditional subaks provided natural fertilization and pest control, the technology packets increased the resistance of rice crops to pesticides. The destructive insects' populations exploded. Excess fertilizers flowed from the paddies to the river, clogging the coastline with damaging levels of nitrogen and algae growth.



In his book Perfect Order, Lansing recollects his failed attempts at persuading foreign consultants to recognize the cooperative and ecological functions of subaks:

“Whenever possible I have seized the opportunity to invite them [consultants] to visit a water temple and talk with the farmers directly. This never worked out quite as I hoped. … The views of the farmers, and indeed all the particularities of the Balinese case, are largely irrelevant to this task. When I returned the consultants to their hotels, the image that often came to mind was that of a team of specialists vigorously treating a patient for what might prove to be the wrong disease. Why, I wondered, do the consultants believe that the details don't matter?”

This tragedy is a sobering reminder of a long-standing problem in global development assistance: Interventions that try to help local communities often end up hurting them. Failures similar to those in Bali have occurred all over the world, from ineffective humanitarian efforts in Haiti to damaging resettlement projects in Afghanistan. Why?

We must confront a practical issue: how to reform development agencies internally so that practitioners want to and can effectively localize.

One answer stands out, as Lansing's account underscores: The details that matter to local communities don't matter to the foreign technocrats who are tasked to help them.

Beyond Promises

Fortunately, policymakers and global development practitioners today increasingly stress the value of going local. The “Doing Development Differently” Manifesto, spearheaded by the Harvard Kennedy School, collected signatures from more than 400 professionals who pledged to uphold principles of “solving local problems” and “working through local conveners.” Under the Obama administration, the United States Agency for International Development (USAID) prioritized local ownership and set the goal of distributing more funds through local sources. The United Kingdom's Department of International Development (DFID) now evaluates the performance of aid projects by their relevance to individual country contexts. “Going Local” has become nothing short of a movement in global development and even has its own hashtag: #GoingLocal.

This is hugely encouraging, but we must confront a practical issue: how to reform development agencies internally so that practitioners want to and can effectively localize. The reality is that many organizations' evaluation criteria, incentives, training, knowledge infrastructure, and more are not set up for the mandate. Facing this situation, practitioners may pledge to do development differently in principle but do nothing differently in practice.

To shift from “Going Local 1.0”—just agreeing that we should tailor development assistance to local conditions—to “Going Local 2.0”—actually practicing it routinely—we need to examine not how international development organizations (IDO) differ from one another but how they are similar. So what do they share? They are nonprofit, hierarchically managed bureaucracies with mandates to foster development in poor countries, and their staffs are largely made up of educated professionals from wealthy democracies.

These commonalities shape the challenge of going local and “making details matter,” so exploring them is a critical step.

Why Don't Details Matter?

As the story from Bali illustrates, donors and technocrats often do not understand or do not care to understand what Thomas Carothers and Diane de Gramon call the “fine grain of actors and interests.”

Why might that be? Or as Lansing puts it, “Why, I wondered, do the consultants believe that the details don't matter?”

For one, the details can't and won't matter to donors and aid professionals who already believe they know the solutions to problems in developing countries. The ideological tyranny of “best practices”—essentially practices found in wealthy democracies— blocks the search for local solutions. Translated into policy, the instinct is to transplant best practices from the wealthy North into the poor South, regardless of whether they fit.

Dani Rodrik sees this as a problem of hubris. As he observes, aid professionals typically enter developing contexts to prescribe, rather than to learn. Urging humility, Rodrik writes, “We can be far more useful when we display greater self-awareness of our shortcomings.”

West Is Not Always Best

On top of hubris, I highlight a different problem: By insisting on one universal standard of good governance, practitioners become blind to numerous possible solutions in developing communities.

For decades, influential development organizations have benchmarked the quality of governance worldwide by a single standard — that of wealthy Western democracies. The World Bank's Worldwide Governance Indicators (WGI) is a prime example. It assigns a single governance score to each country, with countries like Denmark and the United Kingdom predictably scoring the highest, while poor countries always rank bottom. This measurement reinforces the assumption that Western forms of governance are the universal best, and all other aberrations are deficiencies, rather than qualitative differences.

As Victoria Hui aptly expresses, citing historian Bin Wong 's research, “When we take the European experience as the norm and non-Western experiences as abnormal, we are led to search for what went wrong in other parts of the world.” This is exactly the logic seen in the Balinese case. As subaks did not fit Western-modern models of irrigation, they were perceived as a backward anomaly to be removed, rather than a local resource to be used.

The effect of normative blinders is illustrated by another example: the Integrated Diamond Management Program, spearheaded by USAID in Sierra Leone in 2005. The program aimed to liberate diamond miners from exploitation by mine owners — referred to as “supporters” by local communities — by helping the miners form cooperatives, a measure that USAID believed would help peace. In the end, the initiative barely produced any revenue and shut down after only one season. A cause of failure, according to Estelle Levin and Ansumana Turay, was that “the project assumed that supporters are the problem and should be eradicated. ... It did not understand diggers' dependence on their patrons or the level of trust between them.”

When we view governance in binary terms—either you're like a wealthy democracy or something's wrong with you—then existing traditions and practices in non-wealthy-democracies are seen as obstacles to development, be it Balinese subaks, patronage ties in Sierra Leone, or nomadic practices in Afghanistan. Yet, as I argued elsewhere, normatively weak institutions in poor countries may provide the best available resources for kick-starting development, as compellingly illustrated by China, which revolutionized its economy without first establishing conventional best practices.

Too Much of the Same Expertise

The failure to make details matter also occurs when technocrats share similar training and backgrounds—and hence similar assumptions and blind spots.

Although international development organizations (IDOs) come in many varieties, leading organizations like the World Bank are traditionally dominated by economists. As Catherine Weaver describes the bank's Young Professionals Program: “These prestigious slots have until recently been reserved for those with advanced degrees in economics and finance.” Citing another study, she estimated the ratio of economists to non-economists on the bank's staff at 10 to one. “Moreover, these economists tend to share the same theoretical and methodological training,” she adds.

When causal inference is the holy grail of econometric rigor, we can expect economists to focus on statistical techniques for isolating the causal effects of an intervention on rice production. But how a traditional subak system functions? Or how Balinese farmers cooperate through religious rituals? These details generally don't matter to this particular group of professionals. Yes, economists are essential to development work, but their value decreases as more and more individuals with similar expertise join the same team — consistent with the law of diminishing returns.

Straitjacket of Metrics

Thirdly, “details don't matter” if organizational deliverables constrain aid professionals, even those sympathetic to voices on the ground.

As agents in top-down bureaucracies, aid professionals are evaluated, rewarded, or penalized according to their annual key performance indicators (KPI). Unlike the private sector, where profit easily measures performance, the impact of development work is hard to quantify. Hence, development agencies favor large projects with easily marketable results or interventions whose impact can be precisely measured and proven through randomized experiments. But they have little motivation to pursue local solutions that may fit only particular communities and pose uncertain outcomes.

Practitioners also face strong pressure, from governmental bodies with oversight on their activities to philanthropists eager for results, to design projects aligned with best practices. For aid professionals, the safest defense against criticism is following best practices like everyone else. In an organizational environment where the penalties for deviating from conventions are high and the rewards of conformity are certain, practitioners need protection from the risks of failure and censure before they can seriously explore best-fit solutions.

3 Ways to Make Details Matter

Targeting the problems outlined above, I highlight three ideas that outline broad directions for reform rather than prescribe specific policies. After all, the individuals most qualified to recommend specific organizational solutions are the professionals who work in IDOs.

1. Build a Bank of Knowledge About Non-Best Practices That Work

Although there are already several depositories of case studies, such as the World Bank's Global Delivery Library and the Regional Knowledge Sharing Initiative, I suggest a different initiative: research on markets that have successfully emerged in the absence of good governance. For example, how did some low-income communities harness non-best practices to build markets? Instead of repeating narratives of things going wrong, as described in books like Why Nations Fail, it's time to tackle the different and under-studied question of why some poor communities don't fail. This will expand the scope of development solutions.

The selection of cases for this knowledge bank should be theoretically motivated, rather than randomly listed. We particularly need to draw a sharp distinction between what I term “market-building” and “market-preserving” institutions. Institutions that are conventionally deemed universally best—such as formal property rights, technocratic state agencies, and modern courts and regulations—are necessary for preserving markets that already exist. But building markets from the ground up may require qualitatively different strategies and state capacities. We need cases that illuminate both categories.

Such a knowledge bank must go beyond providing mere “description,” a term implying that answers are readily available and need only to be written down. Practical puzzles with no obvious answers need on-the-ground investigation. For example, how did China's local governments attract massive investments despite the absence of technocratic bureaucracies and rule of law? How did Nigeria's film industry take off without intellectual property rights protection? As Michael Woolcock of the World Bank states, case studies are “not just uplifting success stories,” but “unique data collection tools that can guide policy and practice.” Yet to implement high-quality investigation and data collection, leading organizations must support and recognize the efforts through research grants, awards, publications, high-profile events, and other incentives.

Acknowledging that solutions can come in multiple forms, even in ways that contradict Western best practices, implies that we need to change global measurements of good governance. This is not easy, but it is possible. One encouraging indication is a 2018 conference convened by the World Bank, “Measuring Governance in a Changing World,” which acknowledged that “the way 'governance' is measured is still largely based on how it was understood twenty-five years ago.” Qualitative investigation of non-best practices is an essential complement to recent efforts at revising quantitative measures.

2. Leverage Different Competencies at Different Stages

Much to their credit, many IDOs have taken steps to promote personnel diversity, not only in gender and ethnicity, but also in disciplines and methodological training. This is paired with efforts to expand local offices and increase local hires, as well as to form interdisciplinary teams in the research and program implementation process.

These are encouraging steps, but diversifying expertise should go beyond mixing staff members of different backgrounds on an ad hoc basis. Instead, leaders should think systematically about leveraging different competencies to tackle different stages of their organization's problem-solving process, which I break into four steps:

  • Identify problems that matter to local stakeholders;
  • propose solutions to the problems;
  • test and refine solutions on a small scale;
  • and implement solutions on a large scale.

The private sector, particularly leading technological companies, provide useful lessons. Firms like IBM, Xerox, Intel, Google, and Netflix routinely house a design ethnography team, alongside engineers and data scientists. For example, PARC, a Xerox subsidiary, features 15 core competencies. They include not just ubiquitous computing, socio-psychological modeling, and big data, but also ethnography, a branch of immersive qualitative research derived from anthropology.

Why do high-tech companies need ethnography? As I discuss in my forthcoming IBM report, ethnography uncovers “hidden unmet needs and problems” and “the mechanisms behind observed patterns.” Ethnographers do not assume that problems are already known and that hypothesized solutions need only to be tested and implemented. They instead ask: What's the problem? What do people in this particular context really need? What should we measure? Why do we observe these patterns? Their inductive inquiry is an essential complement to deductive reasoning.

Experts such as these, in conjunction with local partners, are best positioned to identify the concerns of stakeholders in aid-receiving contexts, a perspective that foreign observers often miss. Their definition of local problems and gathering of local solutions should occur prior to designing, testing, and implementing specific interventions on the ground. With their input, solutions can go beyond simplistic assumptions, such as “Bali's irrigation system is backward, let's modernize it,” or “patronage ties in Sierra Leone are problematic, let's get rid of them.”

3. Experiment and Redefine Success

To make details matter, empower special teams within IDOs to try out localized or unorthodox solutions without fear of reprisal for failure. With this protected pocket of flexibility, these teams can unearth valuable information on an initiative's scalability that might otherwise remain buried in more traditional operations and protocols. Surprisingly, China widely practices this organizational strategy, which is a key source of its adaptive governance.

The idea of experimental pockets is not new. Many IDOs have created innovation groups, labs, and hubs. Yet organizations often fail to change the evaluation criteria for these teams. Without tailoring the metrics of success to the locally variant, less predictable, and less quantifiable nature of the special teams' mandates, they are unlikely to have impact. At one IDO, I encountered an innovation group bound to the exact same metrics and evaluation process as every other department. Despite the team's mission of sparking innovative thinking and experimenting with new solutions, it slogged through the same rigid protocol, performance indicators, and constraints as other parts of the organization. Rather than innovating, it felt compelled to reduce its “innovation” to bits of quantifiable output, which defeats the point of promoting experimentation and radical change.

Designing metrics for IDOs is never easy, especially in the uncharted areas of innovation and localized solutions. But it is meaningless to create experimental teams if their evaluation criteria is not tailored for experimentation. Here are a few pointers:

  • Instead of giving special teams detailed guidelines on what they should deliver, leaders should specify a few red lines — policies or priorities that must be enforced — and grant autonomy in the remaining areas. Not to be confused with simply “embracing messiness,” this approach gives a clear sense of the boundaries of experimentation.
  • Methods for evaluating originality and innovation should not be the same as those for evaluating routine, passable work. Success shouldn't be about hitting a quantity of trivial, pre-determined goals. It should instead deliver a few or even just one path-breaking result.
  • Track progress with narratives—from the team and its partners and clients—not just numerical targets.
  • The team's narrative of its deliverables should include both successes and failures. This proposal may seem odd, given that organizations usually prize only success. But one purpose of these experimental teams is to generate knowledge of locally tailored solutions that work and fail, for the benefit of the entire organization. Teams that never fail haven't really experimented or innovated.

Getting to Going Local 2.0

My ideas echo various adaptive movements in global development, most notably Problem Driven Iterative Adaptation (PDIA) from Matt Andrews, Michael Woolcock, and Lant Pritchett at Harvard Kennedy School. PDIA makes a powerful impact by introducing a different set of problem-solving principles to development professionals around the world.

My take on adaptive management and localization complements PDIA by going a step further to underscore the organizational reforms that incentivize and empower practitioners to really go local and do development differently. This is the meta-challenge of adaptive development—designing institutions to alter the process of problem-solving itself, which will in turn affect every aspect of development assistance.

The application of these reforms must vary across organizations. Some may be able to hit every mark of changing measurement standards, creating new depositories of knowledge, diversifying expertise, matching different competences to different problem-solving tasks, forming experimental teams, changing evaluation criteria, permitting failures, and more. Others may not. But organizational reform must occur for global aid agencies to move beyond buzzwords and slogans and truly transition to “Going Local 2.0.”