Democracy Lab

How Tradition Remade Rwanda

The secret ingredient in Rwanda's efforts to rebuild its nation after the violence of genocide.

Note: This article is an abridged version of a longer historical case study produced by Innovations for Successful Societies, a research program at Princeton University.

In the wake of the 1994 genocide, the new post-conflict government of the Rwandan Patriotic Front (RPF) inherited the tattered remnants of a highly centralized state. For years, the government engaged in crisis management, attempting to meet the basic needs of a traumatized population with little time to consider long-term matters of governance. It was only in 2000 that the RPF-led government opted for a process of decentralization in an effort to improve local service delivery, alleviate poverty, and create sustainable development. In the new system, local mayors would be responsible for implementing development programs, but there was no built-in mechanism to keep these local leaders accountable. To solve this problem, the government looked to the pre-colonial Rwandan practice of imihigo -- in which leaders or warriors would publicly vow to achieve certain goals, and face public humiliation if they failed -- and adapted it into a tool for modern public service.

The RPF government adopted a national policy of decentralization as a way to rebuild trust, give local people greater voice in governance processes, increase transparency, and maintain political stability -- all of which were vital in post-genocide Rwanda. Protais Musoni, the respected minister of local administration, noted that those high up in the administration were concerned that they were out of touch with the needs of ordinary citizens. Firstly, many of the new government officials were returning exiles who had not lived in Rwanda for years, if not decades. Secondly, many in the RPF felt that people had been manipulated and misled by local leaders into participating in the genocide, and that ordinary citizens had no effective way to influence local leaders or check their behavior.  

The first phase of decentralization began by consolidating administrative boundaries. Previously, the state was comprised of 154 districts. By 2004, the government reduced the number of districts to 30, grouped into six provinces. These new, larger districts had more staff, and the resized territories were easier to administer. Mayors were appointed by the RPF-dominated central government, with plans to shift to a system of direct election of mayors. Until then, the government's initiative was to staff the new posts with competent people and attract university graduates to technocratic positions in the districts.

Decentralization was a way of improving government responsiveness to citizens' needs, while also making local leaders more accountable by giving ordinary citizens a greater role in evaluation. At the same time, government leaders still faced a significant dilemma: How could they ensure that the districts actually delivered on the central government's development program?

Charles Munyaneza, the former director general of Territorial Administration in the Ministry of Local Administration, recalled that while officials were not specifically thinking about imihigo, "what people had in mind was [creating] institutions and to have leaders who would be accountable, who would deliver, and who would be responsible.... The issue of imihigo came as a way of implementing that or achieving that philosophical understanding."

In March 2006, newly-elected district mayors gathered to learn about how their work in the districts fitted into the government's national development plans. During an official induction ceremony with President Paul Kagame present, one mayor, buoyant with his success in the district's first elections, announced that his district would be the best performing district in the country. Another mayor quickly countered, saying that his was going to outperform the rest. The president intervened and pointed out that in Rwandan culture, making bold claims about accomplishments was no small matter. Kagame challenged the mayors to come up with imihigo for their districts that he could then follow up on and evaluate whether the claimed targets were achieved.

President Kagame assigned a small team to adapt the imihigo concept to modern governance. The term quickly came to be known around the country as a combined performance-management system, planning tool, oversight mechanism, and way of using social and traditional pressure to push mayors to greater levels of achievement.

Definitions varied on what constituted a traditional imihigo. Some suggested that it was simply a public vow. Others recalled it as having a basis in war, where warriors would throw a spear into the ground and publicly proclaim the feats they would accomplish in battle. All agreed that the consequences were serious for those who failed to fulfill an imihigo. Alexis Dukundane, of the Ministry of Local Government, put it simply when he said such a person would be considered "incapable" and would be  humiliated within Rwandan society. Government reports suggested that because the practice of imihigo was tradition-based and familiar, the concept generated enthusiasm and a sense of ownership among the population. Dukundane recalled: "People wouldn't understand if you talk about performance contracts, but if you say imihigo, they understand."

Protais Musoni, of the Ministry of Local Administration, said that imihigo targets were intended to be aspirational. "You are not going to be acclaimed for doing what anyone else has done or could do," he said. "An imihigo is supposed to be a stretching of the abilities of both the leader and the people." 

The modern imihigo system incorporated specific measures for planning, monitoring, evaluation, and accountability. The planning process included meetings at every level of government, to give central agencies and district governments a chance to negotiate targets. The procedure aimed to make targets measurable and achievable, and the planning process allowed the central government to ensure projects were consistent with the national government's larger development goals.

Mayors talked to members of community councils to learn local demands, and then created five-year, three-year, and annual action plans. Annual action plans contained all the activities that an individual mayor and his staff pledged to complete during the year. These included rote administrative tasks such as preparing budgets, hiring staff, and completing paperwork and reports; the real priorities, however, were economic development, social welfare, and governance. Activities included building schools, equipping health centers, building marketplaces, paving roads, building bridges, and electrifying villages. Larger projects included resettling whole communities that lived in regions prone to landslides or flashfloods.

Before finalizing annual action plans, district mayors engaged in an extensive process of consultation with  a vice mayor for social welfare, a vice mayor for economic development, and an executive secretary. Together, they constituted the district executive committee. Once a year, all district executive committees met with representatives from the central government to establish a clear idea of their goals and limitations. They would also address budgetary constraints that might affect their planning; for example, Ministry of Education representatives would inform the districts of the money and resources available to build schools or hire teachers. While districts did have the ability to collect certain taxes, tax collection on a local basis was a struggle. Districts were therefore often financially dependent on allocation from the central budget, so the annual meetings were critical in the planning process and in the production of realistic imihigo.

District executive committees also held extensive consultations with various local-level councils, which wielded significant authority in their local areas. Each district was divided into sectors, each sector into cells, and each cell into a cluster of villages. Committees were elected at every level, and representatives from each sub-level were sent to the level above. (Each village committee, made up of the heads of the 10 households in the village, put forward a representative to sit on the cell committee. Each cell committee in turn put forward a representative to sit on the sector-level committee, and so on.) Individuals who sat on these councils were not full-time administrators, but respected members of the community. The district executive committees sought input from all the levels. The draft imihigo had to be approved by the district councils.

Imihigo targets were linked to performance metrics and given a score. If, for example, a district might commit to plant a million trees, the number of trees actually planted by the end of the year would determine the percentage of the activity completed. At the end of the consultation process each year, having settled upon an imihigo and measurement metrics, district mayors publicly signed their imihigo with President Kagame. The ceremonies were aired on TV and radio, and the imihigo documents made public online and in the local press, so that citizens in knew exactly what their leaders had committed to.

While imihigos were built in close consultation with local figures, the central government still played an assertive role by monitoring implementation. District executive committees were required to keep accurate records of activities and expenditures, andprepared a detailed progress report for the central government halfway through the fiscal year. At the end of the fiscal year, two teams were assembled, each with a representative from the Ministry of Local Administration, the prime minister's office, the Ministry of Finance and Economic Planning, the Rwandan Association of Local Government Authorities, the implementation secretariat, and an umbrella civil-society group. Each team was responsible for 15 districts, visiting each district for two days. Leonard Rugwabiza of the economic planning and finance ministry remarked that the presence of high-level government officials ensured that mayors treated the teams' visits with the appropriate degree of respect and preparation.

A typical two-day evaluation would begin with auditing the district's paperwork at the head office, often going late into the night. A vice mayor explained: "If we report that we have held a meeting [with NGOs working in the area], they will want to see minutes from the meeting, a list of who attended, a list of what decisions were taken, and a report on the outcomes of the meeting." On the second day, the evaluation team would meet with the executive committee, giving them a chance to explain any problems they may have found in the paperwork, and to consider any difficulties the mayor may have encountered while attempting to implement programs. The team then would present the mayor with a list outlining the particular imihigo targets that the evaluation team would spot-check, such as the building of a health center or electrification projects.

Evaluation scores were based on the completeness of imihigo activities. An activity that was 90-100 percent complete would be scored a 10; 80-89 percent would be scored a 9; 70-79 percent, an 8; and so on. These scores were then part of a weighted final score that assigned 60 percent to economic activities, 30 percent to social-welfare activities, and 10 percent to governance activities. (The relatively small percentage allotted to governance reflected the belief that effective governance was inherent to successfully complete a project that contributed to economic development.)

The evaluation teams provided recommendations on how any failures could be addressed in the future, with particular emphasis on how they should be carried forward into the next year's imihigo. Final reports were officially present to the mayors and also publicly disseminated through the districts and on various government websites and forums.

Districts were then ranked based on their imihigo performance, both within their respective provinces and within Rwanda as a whole. A high ranking translated into bragging rights for district mayors and governors. Fred Mufulukye, the director general of the Territorial Administration, said that one of the significant achievements of the system was fostering competition between districts. "Now districts -- both citizens and mayors -- want to compete and want to be the best," he said. "This competition drives development." According to Leonard Rugwabiza at the economic planning and finance ministry, high-performing districts could receive special funding for projects.

Because the rankings were presented at public forums, citizens had the information to judge the performance of their elected mayors. If a mayor was consistently underperforming, he could be removed by the local District Council (whose members were indirectly elected). According to one observer, 75 percent of mayors were removed from their positions between 2007 and 2009, though  it was unclear how many were removed for imihigo-performance related reasons. Enige Rugamba, the director general of planning at the Ministry of Local Administration, explained that during the same period, there was an overall reform in governance procedures, such as strengthening the auditor-general's office, the office that investigates cases of maladministration, and the Public Procurement Agency. Many mayors were ousted, and in some cases jailed, due to corruption and irregularities in procedures as part of the new standards of accountability and increased scrutiny.

While no comprehensive data is available to determine whether the imihigo process resulted in better service delivery in the districts, anecdotal accounts suggested there was improvement. For example, a 2010 U.S. Agency for International Development report on family planning in Rwanda asserted that the inclusion of family-planning issues in the imihigo for the district of Nyamagabe more than doubled the adoption rate of modern contraception. Protais Musoni, in his paper on decentralized service delivery, noted that the introduction of imihigo resulted in fewer administrative conflicts in the local government from 2006 to 2007 and in improvements in local development projects, such as malaria eradication and tree planting. Fred Mufulukye, director general of Territorial Administration, observed that preliminary results showed great general performance, especially in education (such as a new system for nine-year basic education), health (health insurance coverage above 90 percent nationwide), road construction, high agriculture production through land consolidation, environmental protection, and governance.

The system did have critics. Apollinaire Mushinzimana, coordinator of the decentralization implementation secretariat, noted that the system had created "a strong perception from local government that the central government is 'looking at us.'" Leonard Rugwabiza recounted a story he heard about the pressures of scrutiny as a mayor: "People were saying, 'You know, when you're the mayor, you've got a contract of one-day renewable. Every 24 hours you can be removed.'" Frank Habineza, chairman of the opposition Democratic Green Party of Rwanda, drew a sharp distinction between decentralization and deconcentration. He argued that institutions may have been decentralized, but that decision-making and planning were still very much a central-government activity, and not as localized as the process might appear.

In 2010, the success of district councils also was unclear. The idea behind vesting so much authority in an indirectly elected local committee was that such a group would be in touch with local demands and wield significant oversight powers. However, these committees were voluntary, and some members failed to attend the council's quarterly meetings; others actually worked in Kigali and were out of touch with local demands; and others still were employed by local governments, for example, as teachers, and because schools were dependent on allocations from mayors, providing real oversight presented potential conflict.

Other critics argue that because the success of imihigo was tied to the RPF party's popularity at local levels, the system did not guarantee stability. But despite these criticisms, the ambitious imihigo process appeared to be delivering on much-needed development targets in Rwanda's districts.


Democracy Lab

The Changemakers

How reformers made the Bangladeshi civil service more responsive to people's needs.

On an August night in 2007, hundreds of people gathered outside the regional passport office in Comilla, Bangladesh, to secure a spot in what they knew would become a long queue by daybreak. A fight soon erupted between people trying to maintain their places in line. When the police intervened, citizens pelted them with stones and then turned their anger on nearby vehicles. Police then fired rubber bullets to disperse the crowd, and 25 people were injured before the army could move in to restore order.

This incident illustrated the kinds of frequent problems and tensions caused by ineffective and strained public services in Bangladesh, the world's seventh most-populous country. Earlier, in 2006, the Bangladeshi government, in collaboration with Britain's international aid agency, the Department for International Development (DFID), launched an innovative program called Managing At The Top 2, or MATT 2 for short. The program targeted problems in the public service that were making everyday life more difficult for Bangladeshi residents, including the nearly 50 million people living in deprivation or extreme poverty. Conceived by a small group of insiders and advisers, the program aimed to change the bureaucratic behavior of top-level civil servants and thereby enable them to demand long-lasting change in civil service operations.

In 2000, Bangladesh had roughly one million civil servants, or about six for every 1,000 citizens. The ratio was far lower than that of many other countries. For instance, in the 1990s, the average for Asia was 26 civil servants per 1,000 citizens and the average for OECD countries was 77 per 1,000.

Inefficiency had long been a hallmark of Bangladesh's public services. A 2001 World Bank study rated the country's bureaucratic efficiency as 4.7 on a ten-point scale, with 10 being peak efficiency. Other South Asian countries fared better: Sri Lanka notched 6.7 while India came in at 5.5.

The public had little confidence in the integrity of civil servants who ran the country's nearly 60 ministries, responsible for health, education, agriculture, and transport services, among many others. A 2002 survey by Transparency International found that 33 percent of households had paid bribes for an electricity connection, 65 percent said it was almost impossible to get a trade license without money or influence, and 70 percent reported unethical practices in getting patients admitted to public hospitals.

Government ineffectiveness had several causes. In the 1990s, sharp political divisions led the two main competing parties (the Bangladesh National Party and its rival, the Awami League) to make wholesale staffing changes every time government control changed hands. The frequent shuffling crippled the civil service sector, as the government in power would promote party followers to the top ranks of the bureaucracy, and transfer non-supporters to lower positions. Political machinations dampened creativity and initiative across the civil service. Civil servants with no political allegiance learned to keep a low profile, as any effective government program might be seen as a success story for the party in power, and could become a target for retribution when the opposition party took over. Moreover, civil servants who improved ways of doing things often found it difficult to sustain their efforts. In many cases, even modest reforms ended when the reformer left the department or ministry, as each new government would scrap programs and suspend projects instituted by its predecessor.

A rigid hierarchy weighed on attitudes and behavior in the civil service. Rizwan Khair, a Bangladeshi civil servant who served as academic coordinator at BRAC University, a research and training institute in Bangladesh, explained: "Once you enter that structure, the so-called ‘iron framework,' your whole mentality changes. You go by processes, rules, and regulations. You become unresponsive to the needs of the citizens." The hierarchical nature of the civil service also created strict boundaries between senior and junior civil servants that impeded the exchange of ideas and the ability to make decisions.

Another challenge was Bangladesh's market for public office, in which influential officials acting as middlemen secured high-level posts for those willing to pay. Positions that provided greater opportunities for graft, such as infrastructure, commanded premium prices. The market for public offices made performance unimportant for promotion and accelerated the shuffling of high-level positions, as brokers' earnings depended largely on how many appointments and transfers they handled.

Numerous efforts to overhaul the civil service collapsed. Between 1982 and 2002, 17 different commissions issued reports that urged reform, but none produced significant changes. In 1999, the government tried an incremental approach with Managing At The Top 1 (MATT 1), a program conceived by officials at the Ministry of Public Administration, with funding and technical support from DFID, to provide project management training to some of the country's most senior civil servants. MATT 1 ran for three years and engaged 100 senior civil servants in a series of six-week training sessions held at universities in Britain. During the sessions, participants prepared project proposals to introduce reforms within a department or ministry, and at the end of the training, submitted the proposals to the heads of their ministry. Observers deemed MATT 1 a limited success. John Wallace, a former Ireland civil servant who had prior work experience in Bangladesh, explained that the initial program lacked sustainability: "The MATT 1 participants emerged with new ideas, new thoughts, new concepts, but stepped back into work roles and work environments which were totally unchanged."

When MATT 1 ended in 2002, a group of alumni and administrators proposed a follow-up program that would improve on MATT 1's shortcomings. The new organizers included high-placed persons such as Secretary of Public Administration Anwarul Bar Chowdhury and Mir Obaidur Rahman, one of the senior directors at the government's civil service training college and the program manager of MATT 1. They wanted a successor program that would build confidence and expertise by allowing participants to develop and implement their own small-scale projects. Chowdhury lobbied Bangladesh's Prime Minister Khaleda Zia, who gave the program her blessing. DFID approved funding and worked with senior civil servants to design the program. In October 2005, after prolonged discussions, officials at the Ministry of Public Administration and DFID agreed on an ambitious seven-year, $25 million project that would target a significant number of senior-level managers in the civil service.

The framework had two key components. The first was a development program that would allow participants to design and implement small-scale reform projects; the second called for MATT 2 consultants to provide performance-management and strategic-planning advice to the Ministry of Public Administration. As part of the development program, facilitators would train 40 to 45 civil servants to design performance improvement projects (PIPs) targeting problems they faced on a daily basis. Graduates of the course then had to implement seven or eight PIPs within the next four months. One of every four of these first-level participants -- chosen based on their performance -- then participated in an eight-week additional course. These high performers would develop and implement six-month projects known as super performance improvement projects (SPIPs).

DFID contracted with Tribal-Helm Corp. Ltd., a Britain-based business and management consultancy, to manage MATT 2. John Wallace, who worked at Tribal-Helm, headed the implementation along with Roger Fernando, a human resources-management specialist. They soon teamed up with Mosharraf Hossain, a former civil servant with experience at the Bangladesh Public Administration Training College, to run the program. Eventually, other facilitators from the training college helped lead the workshops. 

Wallace, Fernando, and Hossain faced a stern task in trying to make MATT 2 work. They had to overcome skepticism and inertia at all levels of the civil service, identify key areas requiring reform, and then implement the ambitious effort across a broad span of Bangladesh's government.

The team members recognized that they had to enlist support and build a mandate for reform. They therefore organized a series of meetings at which over 230 senior civil servants identified policies and processes that were ripe for change. Participants listed impediments to doing non-routine work or showing initiative, identified their own development needs, and provided examples of policies or projects they had tried. Based on these conversations, Fernando and Hossain selected 14 target areas for reform, including strengthening leadership, creating accountability, managing resources, and delivering pro-poor and pro-women services effectively.

To lobby high-level support, Hossain and Fernando then took their case to the government's top decision makers. At a two-day conference, 33 secretaries, along with additional and joint secretaries from the remaining 29 ministries, endorsed MATT 2. "We said: ‘We have done a development needs analysis by asking your colleagues a list of questions and this is what they have told us. Is it your view that they are on the right path, particularly with these priorities?'... Every single ministry was represented, and they unanimously endorsed the 14 items and the priorities," Fernando said.

Between September and November 2006, the MATT 2 team held the program's first training workshop. Over the course of two weeks, facilitators asked five-member teams to discuss the problems they faced in fulfilling their job responsibilities. They then asked each team to focus on one issue and develop a PIP to solve the problem. In most cases, the PIPs focused on issues of service delivery, governance or land and environment. Many PIPs included a gender or poverty component designed to provide better services to women, children, and poor families.

The participants then traveled to Singapore or Thailand, two neighboring countries with records of efficient service delivery, for another two-week session to learn about issues relevant to their own projects. Upon their return to Bangladesh, the project teams spent a week working in low-income areas, exploring grassroots service-delivery issues and citizens' service needs.

For many participants, this week was transformative. A joint secretary in the Prime Minister's Office, Afzal Hossain, and his team visited a slum in Dhaka in 2011. In sessions with residents, he and his team discussed ways to prevent fires that frequently ravaged the city's poor areas. He said: "High-ranking civil servants sit high above and have privileges. But when completing the PIPs, they are compelled to work with common people and compelled to see their problems. ... They go to rural areas to serve the people and learn to become more service oriented. ... Earlier they would say to citizens, ‘Who are you?' and now they might say, ‘How can I help you?'"

Project teams spent the sixth and final week refining their project proposals and identifying the resources needed to implement their PIPs. At the same time, participants also developed individual action plans that would help them improve their own job performance and effectiveness in bringing about changes in their workplaces. The goal was to help participants change their own behavior in ways that were personally fulfilling as well as helpful to their organizations. 

Before the teams began implementation, participants sought approval from the secretary of the Ministry of Public Administration to ensure continued high-level support. Once a team completed a PIP successfully, its members presented the results and their findings to other participants and mentors.

Although not all proposed projects were successful or sustained, during the next several years, MATT 2 participants produced a number of innovative PIPs. The 2007 passport services PIP was an example of a pilot project that was implemented, sustained, and replicated. The team that implemented the project assessed existing passport services and designed a PIP to farm out passport submission and collection to post offices around Dhaka. The PIP reduced the waiting time for passport services from an average of six hours, to under an hour. The department replicated the experiment in five other city post offices and 70 district post offices in 2008.

Media support for the PIPs was instrumental to sustaining successful initiatives because it spurred citizen awareness and demand for better services. Fernando explained, "A vast majority of successful PIPs involved the mobilization of citizen clients. We asked the PIP team to make sure that they got the public and newspapers on their side." For instance, the Ministry of Home Affairs extended and replicated the 2007 passport PIP when the pilot project ended; Fernando attributed the ministry's speedy response to a public clamor for continuing passport services in post offices.

In July 2008, the MATT 2 team and Ministry of Public Administration officials launched the second stage of the program. They selected 10 to 12 participants who had chosen innovative PIPs, implemented projects particularly well, and exhibited good English language skills for additional training in Britain to develop SPIPs. During their time abroad, project teams interned at host organizations. Steve Taplin, who was responsible for placing participants in host institutions in Britain, said: "The value that I saw was that the program extended their thinking, challenged them, helped them think through the project development cycle. Once they inculcated that, they could translate it to a whole range of settings." Upon their return to Bangladesh, the participants worked with mentors to implement their SPIPs. Examples of SPIPs included projects to improve the delivery system of a pension plan and mobilizing local resources for a 100 primary schools.

As MATT 2 progressed, the team institutionalized the 2006 conference with ministerial secretaries, making it an annual event. Secretaries or their deputies gathered for two days to hear about accomplishments from the past year and discuss next steps. Wallace noted: "It re-empowered the MATT 2 process. So there was never any question of it not happening: The precedent was established, and once a precedent is established in Bangladesh, it is difficult for anybody to step back from it." The annual secretaries' workshops were an important means of bolstering support for MATT 2. Hossain of the MATT 2 team recalled, "Each year we got the secretaries' endorsement, and this was the most powerful thing."

The program's design imposed limitations, however. MATT 2 largely recruited civil servants from the influential administrative cadre, triggering resentment among members of the 26 civil service groups in foreign affairs, police, taxation, trade, and others. Observers such as Mohammad Mohabbat Khan, a professor of public administration at Dhaka University, criticized the foreign tours to Singapore, Thailand, and the United Kingdom as the wrong kind of incentive for participating in the program. Moreover, the duration of MATT 2 strained resources. The government found it difficult to spare a large number of senior civil servants for the training. 

Others pointed out that MATT 2 did not build linkages with similar programs in Bangladesh, nor did it reach out to civil servants who had participated in MATT 1. Khair of BRAC noted, "They did not build on the learning and did not evaluate and learn." He added, "MATT 2 is run in isolation and there are no linkages with programs like ours at BRAC institute. There are no exchanges of the ideas and experiences learned from abroad." Doubts also arose over the sustainability of the PIPs and SPIPs that the teams had already implemented.

The program's people-centered approach, geared towards changing the behavior of civil servants rather than overhauling operations, did not produce deep reforms. MATT 2 might have had a greater impact if civil servants had used the project implementation skills acquired through PIPs and SPIPs to introduce and institutionalize changes within their ministries and to influence the behavior of junior civil servants. Although some graduates did so, the Ministry of Public Administration failed to capitalize on their efforts.

In addition, MATT 2's advisory role was underused. The government did not push structural changes or seek guidance from MATT 2 advisers to tackle the underlying causes of bureaucratic paralysis: the ability of party leaders to promote, demote, or transfer civil servants; the tendency of party divisions to quash successful initiatives or even attempts at reforms; and the existence of a market for public office. These embedded issues made lasting reform unlikely.

Still, MATT 2's achievements were noteworthy. Between 2006 and 2011, the program trained 1,323 senior civil servants, including 195 women. By June 2011, MATT 2 trainees had completed more than 200 PIPs in the areas of health, education, governance, environment, and poverty alleviation, and an additional 32 PIPs were in progress.

Although the program had not solved many deep-rooted problems of Bangladesh's civil service, it had worked to overcome the civil service's historic resistance to change by embedding projects within a training program and creating a network of civil servants capable of implementing a broader agenda of reform. More importantly, it provided a large number of senior civil servants with substantial hands-on experience in understanding problems and developing solutions.

The MATT 2 training groups and project teams, made up of different grades of civil servants, challenged traditional thinking and hierarchical decision-making. Manowar Islam, a MATT 2 alumnus and director-general at the Department of Environment in 2011, praised the innovative approach: "MATT 2 was totally different [from civil service trainings]. It changed mindsets. Training was only a part of it. The major part was that you thought about your work and your department. It changed my attitude. Wherever I went, I tried to think about what to change. That was MATT 2's contribution." Members of the MATT 2 team measured their most important achievements relative to the enhanced abilities of civil servants. The program's 1,323 graduates knew how to design, test, and implement reforms. They shared a common experience and were capable of mobilizing for deeper reforms.

MATT 2 made significant strides in a political context that discouraged strong performance and innovation on the part of civil servants. The program taught a large number of senior-level officials to think about creating and sustaining change, with some alumni risking their careers to institutionalize or replicate projects. As Iqbal Mahmood, secretary of the Ministry of Public Administration, put it, "There is a mindset change in many bureaucrats, and a kind of cultural and behavioral change in the process. ... You know that there is a problem and you feel that you need to address the problem within the shortest possible time. ... There is also the knowledge that you cannot address the problem alone and need to have the cooperation of others. So, it is a kind of blending of coordination, teamwork, and leadership."

Lalage Snow/AFP/Getty Images