The Community-Driven Development (CDD) Global Solutions Group (GSG) series CDD Voiceshighlights how CDD approaches are used throughout the world to benefit poor communities.

Rob.jpgThis month, we’re speaking with Robert Wrobel, Senior Social Development Specialist and TTL of Indonesia’s PNPM Generasi. Launched in 2007 by the Government of Indonesia, PNPM Generasi aims to empower local communities in poor, rural sub-districts in project provinces to increase utilization of health and education services. An impact evaluation completed in 2011 showed that the project had improved immunization coverage, prenatal care visits and assisted deliveries by 10 to 11 percent from 2007 to 2010. Since 2014, the program has targeted maternal and child nutrition services and behaviors, and supports a Ministry of Education-led effort to increase participation in early childhood education and development (ECED) services.

You have worked on CDD programs in Afghanistan, Mindanao, and Indonesia.  Are there any commonalities in terms of successfully operating CDD in these diverse contexts?

Three commonalities come to mind:

Most communities where I’ve worked were concerned with ensuring that everyone can benefit from the project. In parts of Afghanistan, villages are made up of 10 to 12 hamlets spread across mountains. In the first couple of years of the National Solidarity Project (NSP), communities would generally propose smaller, individual sub-projects—instead of pooling resources together for larger infrastructure such as a village access road—so that each hamlet could benefit. I’ve seen this happen in southern Philippines and Indonesia as well. Over time, however, as communities gain experience with the project and begin to trust one another, large cross-hamlet and even cross-village sub-projects start to be developed. I’ve seen sub-projects in Afghanistan where 12 villages pool resources together to build a tertiary road spanning an entire valley.

This can create challenges when implementing a project that targets specific groups or relates to private goods, such as providing startup funds or loans to village entrepreneurs. In Indonesia, work related to revolving loans has not been as successful because communities struggle with the idea of providing resources to a specific group in the context of a program focused on delivering public goods. Teams working on CDD projects must consider what the project is trying to achieve and how these goals interact with a community’s idea of fairness and justice.

Another commonality is that the administrative systems underpinning CDD projects, once they’re up and running, have strong momentum. These programs are very resilient and can withstand shocks such as disasters and conflict, which is why CDD is often used in these contexts.

However, this momentum can make it difficult to change core design features over time.  Needs in communities change, as do priorities in national and regional capitals, and project managers quickly gain experience of what works and what doesn’t.  Task teams must take particular care in planning systems that are resilient yet not impossible to evolve and adapt. PNPM Generasi is an example on how a core national CDD platform managed to evolve, and continues to evolve, to target needs that were not envisaged at the time of its design.

Third, simplicity of a system is key to its success. There’s a tendency to overregulate once a system has been set up and has been operating for a while because of the World Bank’s fiduciary, social, and environmental safeguards.

However, CDD projects work best in low capacity, low literacy communities who find it hard to manage additional, more complex requirements. The more that operations manuals and forms “balloon” out, the less effective a CDD project becomes, especially when it comes to community empowerment objectives. Task teams should keep this in mind when considering adding requirements to meet internal reporting needs.

You are currently TTL for PNPM Generasi, which is an innovative CDD program designed to improve health and education outcomes. What do you see as the main advantages and disadvantages of this approach?

PNPM Generasi works by combining elements of household-based conditional cash transfer programs to incentivize communities to improve the uptake of health and education services; most traditional CDD programs work on an “open menu” of services where communities are generally free to invest in what they want, with some exceptions. In Generasi, communities monitor and measure performance targets on health and education-related indicators, and communities are awarded more resources if targets are met or surpassed.

This approach has several advantages. Governments—national, local, and village-level—like the clarity that PNPM Generasi provides on what communities are trying to achieve. Accounting for results also becomes easier, and implementing agencies can then take these results to other stakeholders responsible for local service delivery. Agencies are often hesitant to commit to the program and be accountable for results without knowing what the community will select under an “open menu.”

Second, PNPM Generasi’s performance-based design provides signals to people to focus on specific challenges. Community facilitators can develop expertise on selected services such as early childhood education. They can get in depth with community members when diagnosing problems and identifying specific constraints and solutions in a way that might not be available for general “open menu”-type programs.

Third, and my favorite part, is that PNPM Generasi encourages communities to work with their local health and education departments and vice versa. Communities, once empowered with project resources, have a strong interest in understanding and coordinating with service providers to complement the provider’s plans with community investments. This also gives communities more leverage when working with service providers: they initiate and build relationships with these providers and bring their own resources to the table instead of passively waiting for providers to act or provide information.

There are also disadvantages. Indonesia is a massive country and the type and quality of services varies by region. Communities on Java are more concerned with early childhood education while eastern Indonesian provinces face gaps in primary school education. Coming up with a simple, standard delivery model aligned with varied local needs is a real challenge. To respond to this, PNPM Generasi expanded its education focus from just primary school education to include early childhood education.

Another disadvantage is that PNPM Generasi’s performance bonus system requires large amounts of data, which is collected primarily by community members with some knowledge of health and education. This becomes increasingly difficult to operate the lower the capacity of a community is. To help with this, the program allows communities to invest more project resources in operational support if needed.

And does this approach work well only in Indonesia?

PNPM Generasi’s approach can work in other countries—Morocco’s National Initiative for Human Development, the Bank’s first Program-for-Results operation, adopts a similar approach, where investments are linked to particular service delivery outcomes.

A few considerations for task teams looking to adopt PNPM Generasi’s model:

The model requires a functioning “supply side”—in this case, a health and education sector—to support the demand built through the program and for the community to coordinate with. You cannot incentivize people to use services or coordinate with providers that don’t exist or are of poor quality. To implement PNPM Generasi’s model where there is a limited “supply side,” I’d advise teams to focus on a limited set of services that the community would operate independently, although these would require strong inputs on technical content to ensure quality.

Task teams should ideally keep the set of services as focused and simple as possible because of high data requirements. For PNPM Generasi, we have 10 maternal and child health and two education indicators; communities lack resources to invest in all of these services or gather data for every indicators. I’d recommend around 5 to 6 indicators/services instead, linking actual performance bonuses to a maximum of 4 to 5.

Indonesia could pivot to PNPM Generasi’s new model because the government and facilitators were already familiar with CDD from the previous program, which had been operating for about a decade. Countries going into a second- or third-generation CDD program would absolutely be able to do this; some countries innovating in CDD are Laos, the Philippines, and Yemen. Starting without such a set of experiences, on the other hand, would be much more challenging for a first-generation program.


Tell us your favorite development story from one of the countries you’ve worked.

The story that comes to mind is from Afghanistan, around 2003 when the first round of NSP grants were going out.

We were visiting a village in rural Parwan province, about six hours outside of Kabul, where the community had built a cement foot-bridge to replace a rickety wooden structure that would often become damaged or wash away during the winter. I remember clearly the Afghan villagers told us that this was the first time that anyone—be it the government, NGOs, or other development actors—had entrusted them with the resources and responsibility to use as they saw fit.

It sounds cliché, but I found this to be sincere and powerful expression of frustration with standard ways of doing business. Rural villagers throughout Afghanistan thought that they would benefit greatly from the program and were willing to go the extra mile to participate, often at great personal risk.

By 2004, the program was disbursing $20 million in block grants per month. I saw that NSP was having a huge impact throughout the country, and that this change was happening quickly. From that point on, I fell in love with the CDD model and wanted to work on it in one way or another for the rest of my career.