How the world’s most improved school systems keep getting better

By Michael Barber, Chinezi Chijioke, and Mona Mourshed

How does a school system with poor performance become good? And how does one with good performance become excellent? Our education report explains.

How does a school system with poor performance become good? And how does one with good performance become excellent? These were the questions policymakers and education leaders asked us in the wake of our 2007 report How the world’s best performing school systems come out on top, in which we examined the common attributes of high-performing school systems.

In this new report, How the world’s most improved school systems keep getting better (PDF–4.5MB), we attempt to answer these questions. We analyzed 20 systems from around the world, all with improving but differing levels of performance, examining how each has achieved significant, sustained, and widespread gains in student outcomes, as measured by international and national assessments. Based on over 200 interviews with system stakeholders and analysis of some 600 interventions carried out by these systems—together comprising what we believe is the most comprehensive database of global school system reform ever assembled—this report identifies the reform elements that are replicable for school systems elsewhere as they move from poor to fair to good to great to excellent performance.

The systems we studied were: Armenia, Aspire (a US charter school system), Boston (Massachusetts), Chile, England, Ghana, Hong Kong, Jordan, Latvia, Lithuania, Long Beach (California), Madhya Pradesh (India), Minas Gerais (Brazil), Ontario (Canada), Poland, Saxony (Germany), Singapore, Slovenia, South Korea, and Western Cape (South Africa).

The report’s findings include the following eight highlights:

  1. A system can make significant gains from wherever it starts—and these gains can be achieved in six years or less. Student outcomes in a large number of systems have either stagnated or regressed over the last ten years. However, our sample shows that substantial improvement can be achieved relatively quickly. For instance, Latvian students in 2006 demonstrated performance that was half a school-year advanced to that of students in 2000. In Long Beach, six years of interventions increased student performance in grade four and five math by 50 percent and 75 percent respectively. Even systems starting from low levels of performance, such as Madhya Pradesh in India, Minas Gerais in Brazil, and Western Cape in South Africa, have significantly improved their literacy and numeracy levels within just two to four years, while making strides in narrowing the achievement gap between students from different socioeconomic backgrounds. Improvement can start from any student outcome level, whatever the geography, culture, or income.

  2. There is too little focus on “process” in the debate today. Improving system performance ultimately comes down to improving the learning experience of students in their classrooms. School systems do three types of things to achieve this goal—they change their structure by establishing new institutions or school types, altering school years and levels, or decentralizing system responsibilities; they change their resources by adding more education staff to schools or by increasing system funding; and, they change their processes by modifying curriculum and improving the way that teachers instruct and principals lead. All three of these intervention types—structure, resources, and process—are important along the improvement journey. The public debate, however, often centers on structure and resource due to their stakeholder implications. However, we find that the vast majority of interventions made by the improving systems in our sample are “process” in nature; and, within this area, improving systems generally spend more of their activity on improving how instruction is delivered than on changing the content of what is delivered.

  3. Each particular stage of the school system improvement journey is associated with a unique set of interventions. Our research suggests all improving systems implement similar sets of interventions to move from one particular performance level to the next, irrespective of culture, geography, politics, or history. For example, the interventions undertaken by Madhya Pradesh (India), Minas Gerais (Brazil), and Western Cape (South Africa) on the path from poor to fair performance have striking similarities. There is a consistent cluster of interventions that moves systems from poor performance to fair, a second cluster of interventions does the same from fair performance to good, a third cluster from good performance to great, and yet another from great performance to excellent. For example, systems moving from fair performance to good focused on establishing the foundations of data gathering, organization, finances, pedagogy, while systems on the path from good performance to great focused on shaping the teaching profession such that its requirements, practices, and career paths are as clearly defined as those in medicine and law. This suggests that systems would do well to learn from those at a similar stage of the journey, rather than from those that are at significantly different levels of performance. It also shows that systems cannot continue to improve by simply doing more of what brought them past success.

  4. A system’s context might not determine what needs to be done, but it does determine how it is done. Though each performance stage is associated with a common set of interventions, there is substantial variation in how a system implements these interventions with regard to their sequence, timing, and rollout—there is little or no evidence of a “one-size-fits-all” approach to reform implementation. Our interviews with system leaders suggests that one of the most important implementation decisions is the emphasis a system places on mandating versus persuading stakeholders to comply with reforms. For example, while all improving systems make substantial use of data to inform their reform programs, only a subset of our sample systems translate this into quantitative targets at both school and classroom level, and then share this information publicly (US, England, Canada, Madhya Pradesh, and Minas Gerais). In contrast, Asian and Eastern European systems refrain from target-setting and only make system-level data available publicly. Instead, they prefer to share performance data with individual schools, engaging them in a private dialogue about how they can improve. The systems we studied have adopted different combinations of mandating and persuading to implement the same set of interventions. For example, a system will tend towards persuasion when there are stark winners and losers as a result of the change, it can afford a longer implementation timeline, the desired change is not a precursor for other changes, the system and national leadership is at a tenuous moment of credibility and stability, and/or the historical legacy of the nation makes enforcement of top-down decisions difficult.

  5. Six interventions occur equally at every performance stage for all systems. Our research suggests that six interventions are common to all performance stages across the entire improvement journey: building the instructional skills of teachers and management skills of principals, assessing students, improving data systems, facilitating improvement through the introduction of policy documents and education laws, revising standards and curriculum, and ensuring an appropriate reward and remuneration structure for teachers and principals. Though these interventions occur at all performance stages, they manifest differently at each stage. Taking the example of teacher training, for instance: while Armenia (on the journey from fair to good) relied on centrally-driven, cascaded teacher training programs, Singapore (on the journey from good to great) allowed teachers flexibility in selecting the topics that were most relevant to their development needs.

  6. Systems further along the journey sustain improvement by balancing school autonomy with consistent teaching practice. While our study shows that systems in poor and fair performance achieve improvement through a center that increases and scripts instructional practice for schools and teachers, such an approach does not work for systems in “good” performance onwards. Rather, these systems achieve improvement by the center increasing the responsibilities and flexibilities of schools and teachers to shape instructional practice—one-third of the systems in the “good to great” journey and just less than two-thirds of the systems in the “great to excellent” journey decentralize pedagogical rights to the middle layer (e.g. districts) or schools. However, in parallel, the center mitigates the risk of these freedoms resulting in wide and uncontrolled performance variations across schools by establishing mechanisms that make teachers responsible to each other as professionals for both their own performance and that of their colleagues. For example, these systems establish teacher career paths whereby higher skill teachers increasingly take on responsibility for supporting their juniors to achieve instructional excellence first within the school, then across the system. These systems also establish collaborative practices between teachers within and across schools that emphasize making practice public—such as weekly lesson-planning for all teachers in the same subject, required lesson observations, and joint-teaching—that serve to perpetuate and further develop the established pedagogy. Although teachers receive 56 percent of all support interventions in our studied systems, they receive only 3 percent of accountability interventions. In other words, collaborative practice becomes the main mechanism both for improving teaching practice and making teachers accountable to each other.

  7. Leaders take advantage of changed circumstances to ignite reforms. Across all the systems we studied, one or more of three circumstances produced the conditions that triggered reform: a socioeconomic crisis; a high profile, critical report of system performance; or a change in leadership. In 15 out of the 20 systems studied, two or more of these “ignition” events were present prior to the launch of the reform efforts. By far, the most common event to spark the drive to reform is a change in leadership: every system we studied relied upon the presence and energy of a new leader, either political or strategic, to jumpstart their reforms. New strategic leaders were present in all of our sample systems, and new political leaders present in half. Critically, being new in and of itself is insufficient for success—these new leaders tend to follow a consistent “playbook” of practices upon entering office to lay the foundations for their improvement journey.

  8. Leadership continuity is essential. Leadership is essential not only in sparking reform but in sustaining it. Two things stand out about the leaders of improving systems. Firstly, their longevity: the median tenure of the new strategic leaders is six years and that of the new political leaders is seven years. This is in stark contrast to a norm: for example, the average tenure for superintendents of urban school districts in the US is just three years; the average tenure of education secretaries in England just two years; similarly, that of education ministers in France is two years. Secondly, improving systems actively cultivate the next generation of system leaders, ensuring a smooth transition of leadership and the longer-term continuity in reform goals. This second observation lies at the heart of how a handful of our studied systems (e.g. Armenia, Western Cape, Lithuania) have managed reform continuity despite regular changes of political leadership. The stability of reform direction is critical to achieving the quick gains in student outcomes outlined above.

The fundamental challenge school system leaders face is how to shepherd their system through a journey to higher student outcomes. This journey is all the more complex because system starting points are different, contextual realities vary, and system leaders face multiple choices and combinations of what to do along the way—a single misplaced step can result in system leaders inadvertently taking a path that cannot get them to their desired destination. While there is no single path to improving school system performance, the experiences of all the 20 improving school systems we studied show that strong commonalities exist in the nature of their journeys. This report outlines the aspects of these journeys that are universal, those that are context-specific, and how the two interact. We hope these experiences benefit school systems around the world in navigating their own path to improvement.

Download the full report on which this article is based, How the world’s most improved school systems keep getting better (PDF–4.5MB).

About the author(s)

Michael Barber is an alumnus of McKinsey’s London office, Chinezi Chijioke is an alumnus of the Johannesburg office, and Mona Mourshed is a director in the Washington, DC, office.
More on Social Sector

The next-generation operating model for the digital world

Article - McKinsey Quarterly

What makes a CEO ‘exceptional’?

Article - McKinsey Quarterly

Three game changers for energy

Article - McKinsey Quarterly

How functional leaders become CEOs