This piece is part of Apolitical’s spotlight series on scaling social impact, in partnership with the Bernard Van Leer Foundation. It also appears in our government innovation newsfeed.

The Graduation Approach is the only proven way out of extreme poverty. It started in Bangladesh in 2002, and pilots have now spread to 99 countries. But if governments are to hit the United Nations’ Sustainable Development Goal to eradicate all poverty by 2030, it needs to scale enormously: a billion people live in extreme poverty.

The program starts with consumption support, dealing with survival-level issues such as food security. Participants are then helped to identify a sustainable livelihood. This could be anything from pig farming to tailoring. The program helps them develop that livelihood through financial and technical training, as well as providing seed capital to get going. And the final component is in-person mentoring. The program lasts two to three years.

The idea is that this mix of interventions gives people the first foothold they need to "graduate" out of extreme poverty and keep climbing. The catch is that these interventions are complex, personalised, expensive and long-running.

That means the program runs counter to the accepted principles on how to design interventions for scale. And indeed, BRAC, the NGO that developed it, is vast, employing more than 100,000 people. Is also has huge expertise: after four decades of experience in Bangladesh, it is deeply embedded in the communities it serves.

The Graduation Approach worked in Bangladesh. But could a program like this be made to work elsewhere without BRAC in charge? And what would you need an organisation with millions of staff to provide it?

A universal solution

That’s the question that the Ford Foundation and the Consultative Group to Assist the Poor (CGAP) sought to answer.

They tested the Graduation Approach through pilot projects implemented by NGOs in 10 sites across eight countries in Asia, Africa and Latin America. Aude de Montesquiou from the World Bank helped lead these efforts. “Because we were doing these RCTs to demonstrate that the BRAC model could work elsewhere, we had to stay pretty close to the core components,” said de Montesquiou. Adaptation was limited to small changes, such as matching the food support to a country’s lean season, or suggesting careers that matched local markets.

The trials involved over 10,000 people in Ethiopia, Ghana, Honduras, India, Pakistan and Peru — a deliberately diverse range of countries. And the results, described in a paper in the magazine Science, showed that the approach could, in fact, be replicated in different contexts by different hands.

There was some variation between countries. In Honduras, for example, the impact was insignificant. “That’s where people chose to invest in chickens and the chickens died,” said de Montesquiou. “It was sad, but it’s also a proof of concept: this model works if people make a livelihood out of the opportunity.”

But where there was impact, it continued to grow after the program finished — like in Bangladesh. “The Science article included three-year results; we now have six- and seven-year results,” said de Montesquiou. “Not only are the results sustainable, but the impact actually increases over time compared to the control group.”

The pilots demonstrated that the Graduation Approach could be a universal solution to extreme poverty. But they also left many questions unanswered.

For one, the pilots did not tinker with the formula. They showed that the full package worked, but not what the relative importance of the different components was.

Nor did they look for opportunities to cut costs. Take the life coaching, which is one of the most expensive components. The pilots didn’t test whether it was necessary, or whether it could be reduced or digitised. “Can we do less of that — can we do 20% and get 80% of the results?” asked de Montesquiou.

And how would a scaled up version of the Graduation Approach warp local economies?

“Our pilots were sufficiently small that we didn’t see general equilibrium effects. As they scale up we’re very conscious of that risk,” said de Montesquiou. “After all, how many carpenters can you have?”

Eradicating extreme poverty

BRAC developed a very complete package; CGAP and the Ford Foundation showed that it it could work elsewhere.

But these pilots were small, and there are almost a billion people living in extreme poverty. The question remained: can the Graduation Approach hit that kind of scale?

The first phase of pilots did not provide much evidence in this regard. And that’s partly because the design they tested is in most cases not one that will be scaled.

When BRAC designed the program, they did a better job making it replicable than making it scalable. It’s a holistic program, with wrap-around services delivered with great attention to the individual’s needs. These aspects make it more likely to succeed in small-scale pilots, wherever they take place. But they are also expensive, labour-intensive and time-consuming — and those make it harder to scale up.

And then there’s the question of who would deliver it. The first phase of pilots was implemented by NGOs, but few NGOs could deliver the program at the scale required to hit the SDGs. That job falls to governments, and that means the program needs adaptation to their systems and structures. Nonetheless, almost a hundred countries are moulding it to their needs now, and most of their changes revolve around improving cost-effectiveness.

Fundación Capital, an international development organisation, is working with the governments of Colombia, Paraguay, Mexico, Honduras and Tanzania to make the Graduation Approach a national policy.

They identified three challenges. The first was to see what components of the Graduation Program were already available as government programs, and how they could be integrated. The second was to maintain the quality of implementation while scaling up rapidly. And the third was to reduce the cost.

Their two main innovations have been switching from in-kind asset transfers to cash transfers, and making use of technology to deliver education. These have targeted the two big expenses of the program: the seed capital and the life coaching. Respectively, those make up 50% and 35% of the program costs in Latin America.

Switching to cash transfers is a risk. It clearly involves less handholding, but it is also empowering and cheaper. “Before, it was: do you want chickens, goats, cows — take your pick from this menu of items,” said Ana Pantelic, Chief Strategy Officer at Fundación Capital. “And we said: here’s cash. Make the decision for yourself, whether you want to be a barber or a seamstress, whatever it is.”

Other governments are experimenting with cash transfers too, and there’s no concrete consensus on which works better. The answer will likely lie in greater segmentation: categorising participants according to their needs and treating them differently. “And that’s going to change based on countries, based on people’s profiles: urban or rural, age, gender, location, economic activity,” said Pantelic.

One thing that’s different for every participant is the coaching — and that’s what makes it so costly. Compared to food parcels or cash transfers, coaching is hard to standardise. It also tends to be the part governments are least equipped to provide. In Lesotho, for example, each social worker covers 2,000 households. “The coaching and the life skills component is definitely the most challenging part to scale up,” de Montesquiou admitted.

Fundación Capital see this as an opportunity to bring technology into the process. They are distributing tablets to communities and designing apps that teach participants about, for example, financial literacy and business building. Part of the challenge is designing apps for people who may be illiterate, and may have never used a digital device.

The idea is not to replace the in-person coaching, but to reduce it while maintaining the impact. It could greatly reduce costs and expand the reach of the program, and would also have the effect of standardising a high level of training in the app.

And then there are the benefits of data. GPS data can track the tablet and coach’s movements; how people use the app will help the designers refine it; nudges can be incorporated for those with phones; and integrating, for example, mobile money data with data from the app will paint a picture of how the program is changing local commerce.

The extreme poor are normally off the grid; bringing tech into the Graduation Approach can bring them into focus.

Graduation approaches

The Graduation Approach has had the collective will of some of the world’s biggest NGOs and foundations driving it, but getting it to scale worldwide is a slow, ongoing process. Some lessons have emerged.

One is that it is useful to have a neutral, global platform to share best practices. CGAP provided that service for the first wave of pilots, and now the Partnership for Economic Inclusion at the World Bank has taken the job over. It is helping governments scale their programs, while encouraging and tracking the results of their innovations.

A second point is that the program is scaling almost in spite of itself. As an expensive, bespoke package, it’s perhaps not surprising that the pilots worked so well, and worked almost everywhere they went. But maybe it could have been designed in a more scalable way to start with.

Nonetheless, almost 100 countries are now shaping the program to their needs. Technology will certainly feature. It is already changing the scale-up process at every step, providing opportunities to cut costs while expanding reach and enhancing engagement, as well as to monitor quality.

And finally, the Graduation Program demonstrates how massive scale will be achieved: through partnerships. Government will be the backbone, while specialised non-profit and private sector organisations step in for specific tasks.

This will require accommodation on all sides. Working with governments is a huge opportunity, but it introduces problems with politics, electoral cycles and annual budgets. Take Fundación Capital: in the space of just over half a year from December 2017, their partners in Honduras, Paraguay, Colombia and Mexico will all have had elections. Plans for scale-up are on ice until new governments get settled. “It’s been stressful,” said Pantelic, with some understatement.

(Picture credit: Flickr/Asian Development Bank)

Tom Graham