My First Megaproject: 20,000 Schools in Nepal

How to get girls to school, birth rates down, and the economy up, at massive scale. Used by the Bill and Melinda Gates Foundation as an exemplar to follow worldwide.

Bent Flyvbjerg
10 min readMay 13, 2021


Because the Himalayas cover most of Nepal, what geographers call “friction of distance” is extreme here

My first megaproject was Nepal’s Basic and Primary Education Program (BPEP). This was a billion-dollar venture to improve schooling across Nepal, including the construction of 20,000 new schools and classrooms, during the 1990s and 2000s. I had been hired by Danida — the Danish International Development Agency — to work as program planner on the project and now found myself installed at the Hotel Yak and Yeti in Kathmandu, with responsibility for designing and programming the BPEP.

Because the Himalayas cover most of Nepal, what geographers call “friction of distance” is extreme here. In most places it is difficult to get from point A to point B — including for children to get to school. Roads are absent and many children had daylong walks along difficult mountain trails, often over one or two mountain passes, to get to the nearest school. This meant that many children did not attend school or did so only sporadically.

Nothing is more important, however, for developing a nation, or a family, than getting young children to school, and especially girls. This is because primary education is an effective way to start a virtuous circle of better health, lower birth rates, economic growth, and better living standards. Nepalese villag­ers were acutely aware of this, as was their government. They wanted schools (Wal 2006: 62).* It was therefore a main objective of the BPEP to build the 20,000 schools and classrooms as close as possible to children without access to schools.

Nothing is more important for developing a nation, or a family, than getting young children to school, and especially girls

On my first night in Kathmandu, the Danish ambassador to Nepal invited me to a private dinner at his home together with Nepal’s permanent secretary of education. The permanent secretary and I hit it off, sharing good laughs and our joy of fathering daughters. I now had a direct line to the top of the Ministry of Education, which owned the BPEP and was ultimately responsible for its delivery. This greatly facilitated developing the program and having it approved at all levels of government.

Working with my close collaborator, fellow Dane Hans Lauritz Jørgensen, we set out to detail the program. As an architect, Hans Lauritz did the design of individual school buildings and classrooms. As a planner, I did the programming of what had to be built where and when, to deliver the program according to its schedule and objectives. The Nepalese economy is small and fragile. If construction of all 20,000 schools started at the same time, the economy would overheat, a classic error in big construction programs in developing nations. This had to be avoided, pushing the program ahead as swiftly as possible, but not faster than the national economy would allow without being impaired.

After considering different designs for individual school buildings, Hans Lauritz and I settled on three basic prototypes — each type dependent on the gradient of the mountain on which it would be built, with the gradient being a main constraint on design. Turnkey projects would have been an obvious choice, but we decided against them, because studies and site visits showed poor maintenance and lack of local ownership if villagers had not been involved in the construction of their schools. Without proper maintenance, in Nepal’s harsh climate schools would deteriorate fast, which would be counter­productive. It was therefore decided that each school would be built with local labor and would involve the local community in decision making, construction, and maintenance.

We made sure the designs were earthquake proofed. This proved a wise decision when, in April 2015, the biggest earth­quake since 1934 hit Nepal and killed nearly 9,000 people

Finally, we made sure the designs were earthquake proofed, which traditional schools in Nepal were not. If taken into account from the outset, earthquake proofing is inexpensive for this kind of building, so we saw no reason not to do it. This proved a wise decision when, in April 2015, the biggest earth­quake since 1934 hit Nepal and killed nearly 9,000 people and injured close to 22,000, with its epicen­ter in areas with BPEP schools. The news was devastating, but we took solace in the knowledge that children in schools that had been built to our designs would have been well protected, because the buildings would not have collapsed easily.

On behalf of Danida, we negotiated a first draft of the BPEP construction program with JICA, the Japan International Cooperation Agency, which coordinates Japanese governmental development assistance, and which would donate building materials for the schools. After adjustments, the proposal was further negotiated with the World Bank and sufficient funds were secured to begin program deliv­ery. The first 19 districts were covered from 1992 to 1993, adding six more from 1993 to 1994, with a further 15 from 1994 to 1995, and the rest from 1999 to 2004, covering all 75 districts in the country (Chandra 2003: 177).

The BPEP has been, and is, a significant success, according to independent evaluations (Chandra 2003; Jerve, Shimomura, and Hansen 2008; Little 2007; Skar and Cederroth 2005; Wal 2006). A review of development aid to Nepal, commissioned by the Norwegian Foreign Ministry and undertak­en by the Nordic Institute of Asian Studies at Copenhagen University, concludes for the first 40 districts of the program:

“The BPEP has had an enormous impact in a short space of time. It has provided addi­tional classroom space for about 300,000 students and non-formal education for more than 170,000 adults and children. With such achievements it is not surprising that the government is eager to finance a second stage of this project” (Skar and Cederroth 2005: no pagination).

Chandra (2003: 177) evaluated the BPEP management structure and found it has “led to the efficient delivery of project inputs.” School enrollment has increased, as intended, and it has increased more for girls than for boys, as also intended, although girls still lag behind in terms of enrollment. Repetition rates (having to repeat a class due to poor performance) have decreased, and again more for girls than for boys (Wal 2006: 63, 77).

Finally, Skar and Cederroth (2005: 49) found that “DANIDA’s involvement in the BPEP is regarded as a model by other bilateral donors” and that “donors have flocked” to the BPEP for this reason, helping to raise the necessary funds to make the program a success, including funding from UNICEF, UNDP, the European Union, Norway, and Finland, in addition to the Nepalese government and the funding already mentioned. In total, the program has impacted education for more than a million children and grownups across Nepal.

To be honest, I did not know much about scale-up when Hans Lauritz Jørgensen and I planned the BPEP construction program for the Nepalese government. Later, when I learned from Danida that the program was showing signs of success, I was delighted but did not think much about it. I thought this was normal, just another day at the office for the megaproject planner. Only when I began to research megaproject performance in my job as a university professor did I realize that success was the excep­tion for such projects and failure the norm. “Over budget, over time, under benefits, over and over again,” as I coined the situation in the Iron Law of Megaproject Management, when I later had the data for a full review (Flyvbjerg 2017a: 12).

I then returned to my experience in Nepal and began to ponder what had happened here. Was there a method to our success with the BPEP? Or had we just been lucky? Was it the plentiful and timely funding from donors that did it? Or the unwavering support from the permanent secretary and his ministry? Or the way Danida had managed things?

It took me a while to figure out the answers to these questions. First, I needed experience and research from more projects. Ultimately, I found that the BPEP success was explained by none of the above factors, although they are all important. Instead, I found that the following two features of the BPEP were the keys to its success, more than anything else:

1. The program was designed and delivered in a modular, replicable fashion.

2. The program was designed and delivered at speed.

First, regarding modularity and replicability, the constraints of the Nepalese geography, the scarcity of labor and materials, and the vast scale of the program pushed us to be frugal, working with just a few standard designs that could be easily built and replicated over and over, as LEGOs in the Himalayas.† In essence, officials and villagers would choose between two or three standard designs of schools — which would be the basic building blocks of the BPEP — literally determined by the slope of the moun­tain on the site where the school would be built.

You would then build the chosen design, following relevant instructions, and populate the building with teachers, teaching materials, and children. You would repeat this until a district was covered. Then you would do the same in the next district, and the next, until all 75 districts in the country had been included. Replicability applied not only at the level of the building modules (the schools and classrooms), but also at the district level, where experience from one district would be replicated in the next, with the district now being the module to repeat over and over.

Research shows that replicability is crucial to effective learning (Baldwin and Clark 2000). Replicability creates a feedback loop where you can use the experience from delivering one module to improve the delivery of the next, repeatedly, ensuring that the quality of delivery constantly improves as you go along, due to learning-by-doing. Replicability is also conducive to experimentation. Instead of going full scale immediately, you experiment with a few modules and use your experience from the experiments to improve the next modules, and you repeat this until you master delivery, which is when you go full scale.

The ability to experiment and learn — something humans are inherently good at, probably because it enhances our chances of survival in evolutionary terms — is the most basic explanation why a venture that is based on modular replicability is more likely to succeed than a venture that depends on a one-off, bespoke construct that can only be delivered in one go — something humans are inherently bad at, having difficulty getting things right the first time. Experimentation and learning are main reasons the BPEP succeeded instead of failing like the majority of megaprojects.

Second, regarding speed, it was clear that Nepal immediately needed more schools, which the perma­nent secretary had hammered into me when we first met at the ambassador’s house. Speed was there­fore of the essence. Neither design, negotiations, decision making, nor delivery could be allowed to drag on for years, as is common for large programs. It took us only a few weeks to develop the first draft of the construction program, not including negotiations for funding. The other parts of the BPEP were similarly accelerated and developed in parallel. Raising funds and final decisions took a few months and construction began immediately after this, with 19 districts covered in just a year, which is exceptionally fast for this type of project.

It should be emphasized that despite the accelerated pace, the BPEP was not fast-tracked. With fast-tracking, construction starts before designs and plans are completed, which is common when speedy delivery is a must. However, fast-tracking is notoriously a high-risk strategy, because the chances of making wrong decisions multiply without a firm design (Williams and Samset 2009, Tighe 1991). For the BPEP such risk was not an option, so designs and plans were completed before construction began and proved sufficiently robust to keep the program on track throughout.

Two things, in particular, contributed to the speedy delivery: (a) the simple, modular designs for schools, which were easy and quick to build, making replicability a facilitator of speed, and (b) the fact that a number of districts were completed within just a year, after which finishing the program became a matter of repeating the experience from these districts in other districts, over and over, until the whole country had been covered, again making replicability pave the way for speed. In fact, replicabil­ity at the level of both individual buildings and individual districts worked so well that the whole BPEP was delivered years ahead of schedule, a rare feat for programs of this size (Flyvbjerg 2017a).

Today, the Bill and Melinda Gates Foundation and many others use the BPEP as an exemplar to follow, worldwide. The success was beginner’s luck in the sense that I was fortunate that my first mega­project lent itself to modular replication and accelerated speed. We did not realize at the time just how lucky we were and how rare success is in megaproject management, because nobody knew at the time, for lack of systematic evidence. However, the BPEP success was not beginner’s luck in the sense that we quickly realized that modularity, replicability, and speed would be key to success, and we proactively and very deliberately designed the program around these concepts, to great effect.

Today, I have learned — through my experience from working on dozens more megaprojects and from my research at Aalborg, Delft, and Oxford on the performance of thousands of projects — that modu­larity, replicability, and speed are the key determinants of success, not just in programs like Nepal’s BPEP, but in any type of megaproject or other large-scale venture.

Whether you’re a small startup or Elon Musk trying to grow Tesla and SpaceX, or Jeff Bezos scaling up Amazon, or Larry Page and Sergei Brin trying to cover the planet with Internet and Google servers — or you’re the US, UK, Chinese, or other government trying to increase power production, expand your infrastructure, or make your health, education, and social services work better — modularity, replica­bility, and speed is the answer to effective delivery. How well you deal with these three issues will decide whether your efforts succeed or fail.

Today, the first thing I do when assessing whether a prospective megaproject — or other outsized venture — is likely to succeed, is to assess the degree to which it lends itself to modularization, replica­tion, and speed. If a project can be delivered fast in a replicable, modular manner, it is likely to do well. If it cannot, it is likely to be troubled or fail.


*) See full paper with references, here.

†) Modularity in construction often conjures up images of unattractive, low-quality buildings, like 1960’s Soviet, US, and UK social housing. Here we maintain that modular construction can and should be aesthetically pleasing and of high quality, and that there is no good reason for this to not be the case, but quite the opposite. Aesthetics and quality were not compromised by modularity for the 20,000 schools in Nepal.



Bent Flyvbjerg

Professor at University of Oxford and IT University of Copenhagen. Writes about project management.