Abstract
Since post-World War II and especially throughout the 1990s, the globalization of a liberal international order propelled a wave of education reforms around the world. However, recent challenges to the legitimacy of the liberal order may undercut the prevalence of education reform across countries. To reveal how global changes are influencing education, we draw on a newly constructed data set of 6,696 education reforms in 147 countries from 1960 to 2017. Using dynamic negative binomial panel regression models, we find declining levels of reform in recent decades. We also find evidence of changing dynamics of influence among prominent organizational actors: World Bank lending is less associated with education reform over time, whereas the influence of international nongovernmental organizations has grown. This suggests a shifting system of governance, where formal coercive pressures become less palatable and the normative influences of civil society grow stronger. Overall, our findings indicate that education reform arises as a macro-global process as much as a response to local needs and conditions.
Keywords
In recent decades, levels of democracy have been falling in many countries, and populism, authoritarianism, and nationalism have been on the rise (Bonikowski 2017; Moffitt 2016). Civil society organizations are also increasingly restricted and regulated, especially on international dimensions such as foreign funding (Bromley, Schofer, and Longhofer 2020; Dupuy, Ron, and Prakash 2016). Additionally, around the world, there is growing recognition of polarization and inequality on multiple fronts, often attributed to the failures of neoliberal economic policy to live up to its lofty ideals (Ostry, Loungani, and Furceri 2016). Together, these changes indicate a weakening of the globalized liberal institutions and ideologies that characterized the post-World War and, especially, the post-Cold War world order (Mearsheimer 2019). The worldwide expansion of education and the global diffusion of education policies have been core features of the liberal world order in the post-World War II era (Meyer, Ramirez, and Soysal 1992). But as challenges to the liberal world order grow, the status of education as a central vehicle for individual and national development may also weaken, and the roles of key carriers of educational ideologies, such as international organizations, are likely to change.
To shed light on how world changes are influencing education, we constructed a novel database containing the most comprehensive cross-national, longitudinal information about policy changes to date. Specifically, we consider variation in the levels of education reform in 147 countries over the period 1960 to 2017. In total, 6,696 reforms are distributed across 6,939 country-years in the analyses. We use dynamic negative binomial panel regression models to examine the changing determinants of national education reform. We focus on historical trends and the role of two distinct kinds of international organizations (the World Bank and international nongovernmental organizations [INGOs]). Together, our findings indicate that education reform largely emerges out of global processes beyond the specifics of national conditions.
Education reform, in our conceptualization, is a discursive act, separate from implementation or effects (Meyer and Rowan 1977). 1 Reform discourse reveals the educational problems thought to be most salient and the solutions that are perceived as legitimate, shedding light on beliefs about the role of education in society. “Reform” as a solution to educational problems connotes a planned and intentional process of systems change that brings an abstract model or set of blueprints to local settings. Thus, reform is different from other sorts of change, such as crisis response, demographic trends, or local responses to immediate issues.
At the global and national levels, efforts to change policies or laws around schooling partly indicate the intensity of belief in education as a core institution in society and in rationalized approaches to progress. Reform discourse is thus a leading indicator of beliefs about how to pursue progress. Education is central to liberal doctrines, and the globalization of the liberal world order spurred social initiatives like the Education for All movement as well as a worldwide wave of education reforms rooted in neoliberal economic principles through the 1990s and 2000s. As the liberal international order weakens, global efforts to restructure education may falter.
Our study makes three main contributions. First, neo-institutional scholars in transnational sociology and international relations have long documented the influence of Western, liberal cultural norms in shaping the world order and influencing nation-states (Finnemore 1996). We posit that as the liberal world order weakens, education reforms will become less common. A great deal of research has focused on the political consequences of attacks on the liberal world order, but few have examined the social dimensions of these changes. In line with our expectations, we find declining levels of education reform in recent years. Second, our results provide evidence of changing power dynamics among prominent organizational actors in the global system. World Bank loans are now less linked to education reform, whereas INGO influence has grown. This provides evidence of a changing system of governance, where formal coercive pressures, such as the loan conditionalities used by the World Bank, become less palatable and the normative influences of civil society grow stronger. Third, we introduce a new cross-national, longitudinal database of education reforms, and we make it accessible for general use (available at http://www.werd.world). The database is the most comprehensive effort of its kind and can provide a foundation for new research on the causes and consequences of education reform. Together, our findings suggest that education reform only partly arises as a technical response to local conditions; it arises as a macro-global process as much as a response to local needs and conditions.
Education Reform and the Liberal World Order
In the decades following World War II, a world order built on Western, liberal principles of progress and justice became institutionalized and expanded worldwide (Meyer et al. 1997). At the heart of these post-War political, economic, and social institutions rests a belief in individual rights and capabilities: Individuals should elect governments, create markets, and freely associate. Formal schooling emerged as a key means for creating these rights-bearing and responsibly empowered individuals (Meyer et al. 1992). As a result, education systems grew massively around the world at all levels and over the whole life course, especially in countries most integrated into the world society and economy (Jakobi 2012; Schofer and Meyer 2005). At the same time, liberal philosophy “emphasizes Weberian rationality as the means to both justice, defined as equality, and progress, defined as wealth accumulation” (Finnemore 1996:325–26). The bureaucratic nation-state and its core components also expanded dramatically: For instance, the mean number of national ministries in a state nearly doubled between 1945 and 1990 (Kim, Jang, and Hwang 2002). Bringing these trends together, expanding national education systems largely take the form of massive formal organizational structures subject to rationalized planning and reform rather than a looser vision of education occurring naturally in nonformal or informal settings.
Following the collapse of the Soviet Union, the post-War liberal world order intensified. Unchecked by the Cold War, U.S. influence as a hegemonic power increased, and a more aggressive version of liberalism emerged. 2 The earlier liberal focus on education as the key institution intended to help children grow into informed and rational citizens and consumers intensified through the 1990s. As one indicator, education became an increasingly prominent part of political platforms around the world (Jakobi 2011). Access to schooling, already important in earlier decades as part of nation-building, took on newfound force as an individual matter; the 1990s witnessed the rise of a global focus on “Education for All” and the creation of a vast global movement to get all children into school (Chabbott 1998; Mundy 2007). New fields of effort emerged, such as issues of school quality and expanding curricular emphases. Schools worldwide now teach not only numeracy and literacy but also human rights and “twenty-first-century skills” of critical thinking and communication (Kay and Greenhill 2011; Suárez and Ramirez 2007). Guided by the economic side of neoliberal ideologies, school systems were thought to best achieve their educational goals through structures, funding models, curricular programs, and management strategies that reflect the choices of parents and students (Buckner 2017; Renzulli and Roscigno 2005) as well as rationalized management strategies such as ongoing evaluation and assessment (Kamens and McNeely 2010; Ramirez, Schofer, and Meyer 2018). As the liberal world order became dominant following World War II, and especially following the collapse of the Soviet Union, it created a context ripe for a global explosion of education reform.
Education Reform in a Weakening Liberal World Order
There are many reasons to doubt that the expanding visions of progress and justice built into the social, political, and economic institutions of the liberal world order could ever become reality. Serious flaws were evident decades ago. For instance, global financial markets collapsed during the Asian currency crisis in 1997, exposing the volatility of global capitalism and the fact that millions of vulnerable people could suffer from the self-interested decisions of distant and elite investors (Wade 2000). In 1999, protests in Seattle against the World Trade Organization demonstrated both discontent with the instability of global capitalism and the newfound power of transnational activism (Evans 2005; Tarrow 2005). The same year, Boris Yeltsin’s controversy-ridden efforts to bring democracy and capitalism to Russia stopped short with his unexpected resignation on December 31, 1999, which ushered in the increasingly illiberal Putin era (McFaul 2018). Then, on September 11, 2001, the World Trade Center was attacked, and the United States went to war in Iraq and then Afghanistan in what looked like a return to national self-interest to many observers (Dasgupta and Kiely 2006; Desch 2008). The widely criticized wars and the 2008 financial crisis further undermined trust in the global system and, especially, undercut the legitimacy of U.S. hegemony on cultural, political, and economic grounds (Kirshner 2014).
The liberal and neoliberal world order has long had critics, but opposition has increased since the late 2000s (Guillén 2018; Kotz 2015; Norris and Inglehart 2019). The problems of capitalism represent one front of critique. Within-country income inequality has increased on a population-weighted basis due to increases of economic inequality in large countries like China, India, the United States, and Indonesia (Piketty 2013; Shi 2015). Challenges have arisen on multiple other fronts (for a discussion of nationalism, populism, and authoritarianism reactions, see, e.g., Bonikowski 2017). A growing wave of empirical studies document falling levels of democracy worldwide (Fukuyama 2012; Kurlantzick 2013), increased restrictions on civil society around the world (Bromley et al. 2020; Dupuy et al. 2016), growing attacks on higher education (Schofer, Lerch, and Meyer 2019), and the rise of repressive laws against sexual minorities (Hadler and Symons 2018).
Overall, the period following the fall of the USSR is characterized by the globalization of broad cultural principles that put education front and center, asserting that individuals’ choices can provide the key to successful economies, polities, and societies anywhere (Jepperson and Meyer 2021; Meyer 2010). In recent decades, however, the liberal world order has been less dominant, and education may become less central. Based on this discussion, we examine the changes over three time periods: 1960–1991, 1992–2008, and 2009–2017. We expect the following:
Hypothesis 1a: The average number of education reforms in a country will be highest during the heyday of the liberal world order (1992–2008).
Hypothesis 1b: The average number of education reforms in a country will decline in the third period (2009–2017) as a reflection of the weakening liberal order.
International Organizations and Education Reform
Macro-historical trends reveal complex, large-scale shifts that unfold over time and cannot be reduced to a single indicator, but we can also look to more concrete mechanisms of change for further insight. International organizations are powerful actors in the global diffusion of policies in general (Barnett and Finnemore 1999; Finnemore 1993) and in education specifically (Benavot and Resnik 2007). Two kinds of international organizations are particularly well-known conduits of global norms around education: the World Bank and INGOs.
The World Bank is a famous (or infamous) catalyst of education reforms (Mundy and Menashy 2014). It has a long history of requiring governments to make dramatic and often contested changes to education systems as a condition for receiving loans (Heyneman 2003; Jones 2007; Klees 2012). Resource dependencies can “generate or alleviate pressure to reform” in ways that align with donor priorities (Steiner-Khamsi 2010:331). Unsurprisingly, the World Bank has pressed countries toward decentralization and privatization, leveraging its financial and technical resources for the promotion of neoliberal education reforms (Mundy and Verger 2015). For example, a study of Nepal from the mid-1980s to 2010 describes how the neoliberal world order was thrust into the small Himalayan nation through the Bank’s educational policy recommendations of marketization, privatization, and decentralization (Regmi 2017).
In part, the Bank’s power to shape national policy directions is straightforwardly coercive, although it has less control over implementation. However, the cultural and ideological elements of the Bank’s work in education are increasingly apparent (Kim and Boyle 2012; Klees 2012). For example, a study of higher-education policy reform in Ethiopia describes how World Bank influence is not only financial but also symbolic (Molla 2013). Following Bourdieu (e.g., Bourdieu and Passeron 1990), the study outlines how the Bank’s great cultural legitimacy during the neoliberal era renders its prescriptions acceptable to (or even welcomed by) local policy actors. In line with these cultural depictions of the World Bank, we focus on its role as a carrier of norms of liberal world society, where a prominent mechanism of influence is financial coercion.
INGOs are also carriers of world society norms, perhaps the most notable enactors and transmitters of global social and cultural principles (Boli and Thomas 1997, 1999). Among other activities, INGOs provide education, develop curricula, build schools, conduct research and monitoring, and propagate educational standards and goals through policy advocacy and campaigns to raise awareness (Bromley 2010; Schafer 1999). INGO memberships are associated with outcomes such as expanded enrollments, persistence in schooling, and education spending, net of countries’ features, such as aid dependence or state strength (Kim and Boyle 2012; Schafer 1999; Schofer and Meyer 2005). INGO memberships are also associated with features of education systems, such as increased accessibility due to reduced school tracking and reduced use of high-stakes exams (Furuta 2020, 2021). Relative to the World Bank, INGOs have reduced capability to use financial coercion to press governments to change education policies. Instead, they rely on normative claims and voluntary action. Although lacking the formally coercive bite of loan conditionalities, social and normative pressures are powerful influences in the global system (Hafner-Burton 2008; Hafner-Burton and Tsutsui 2005).
Overall, both the World Bank and INGOs carry principles of the liberal world order in education. Thus, we posit the following:
Hypothesis 2a: Countries that receive more World Bank loans in education will conduct more education reform.
Hypothesis 2b: Countries that have more memberships in INGOs will conduct more education reform.
Despite their shared linkage to some common global principles, the World Bank and INGOs also embody dimensions of the liberal order that can be at odds with one another (Okun 1975; Rodrik 1997). As outlined in the canonical statement, world society is rife with “internal contradictions and inconsistencies” that “make certain forms of struggle inevitable” (Meyer et al. 1997:168–69; see also Weaver 2008). Whereas the Bank represents a vision of progress with a focus on efficiency and effectiveness rooted in free market ideology, INGOs often represent a rights-based frame emphasizing access and inclusion (Boyle and Kim 2009). Even more sharply, the World Bank has relied on coercion via resource dependency to press countries to change their education policies, whereas INGOs rely on social and cultural controls, such as transnational activism in education (Mundy and Murphy 2001).
The growing challenges to the liberal world order lead us to suspect that the differences between INGOs and the World Bank may create divergent trends in their influence on education reform. Over time, the principles of liberal world society create an expanding number of empowered actors that the Bank increasingly needs to account for and compete with while also undercutting the legitimacy of outright coercion as a control mechanism (Bromley and Meyer 2015; Meyer 2010). In line with our view, Mundy and Verger (2015) outline how the Bank’s role in education is changing due to the increasing power of emerging economies, growing diversity of preferences among Bank borrowers (especially a preference for OECD assistance), and borrower resistance to the historical rigidity of Bank policy prescriptions. Overall, a more complex and contested ideological environment contributes to reduced levels of education reform both because borrower countries resist some Bank reforms and the Bank has fewer clear prescriptions to impose (or offer; Mundy and Verger 2015).
Like the Bank, INGOs face increasing criticism; they are often depicted as interfering foreign agents of hegemonic world society (Bromley et al. 2020), and many countries now seek to restrict INGO activity (Dupuy et al. 2016). However, for the most part, these restrictions aim to prevent political interference and activism around democracy and human rights, not stop all foreign development aid (Carothers and Brechenmacher 2014). Many states try to walk a line that allows them to dampen civil society activism while still enabling development aid (Bush 2015; Van der Borgh and Terwindt 2014). For example, the 2009 Ethiopian Charities and Societies Proclamation Act stipulates that NGOs working on any rights issues must acquire 90 percent of their funding from domestic sources—but it explicitly exempts development organizations from these restrictions (Brechenmacher 2017). In fact, many states encourage INGOs to work on issues that are not considered contentious, as Teets (2014) describes in the case of China on issues of education reform, public health, and disaster relief. In fact, the growing pressure for INGOs to avoid contentious causes around rights and democracy combined with the declining legitimacy of coercive strategies in education reform used by the Bank may be driving additional support for INGO work in education reform in recent decades. Stated formally,
Hypothesis 3a: Over time, the association between World Bank loans and education reform will decline.
Hypothesis 3b: Over time, the association between INGO memberships and education reform will increase.
Additional Country Factors
In addition to the historical and organizational influences that are of primary interest here, country-level conditions are likely to influence levels of reform. To start, prior administrative actions and internal repertoires can predict future organizing efforts (Rao and Greve 2018). Administrative legacies create and maintain “a self-reinforcing cycle of organization building” that influences future organizational capacity (Greve and Rao 2012:638). To control for the dynamics of organizing, we examine whether enacting reform in a prior year shapes the likelihood of undertaking more reform in subsequent years.
Furthermore, a country’s GDP per capita is associated with central features of education systems, such as achievement levels, and may shape the pace of education reform (Hanushek and Woessmann 2014). A country’s wealth could shape reform levels in two ways. Higher-income countries have more resources available to conduct reforms, and we could therefore see more reform in wealthier countries. Alternatively, lower-income countries could perceive the need to do more education reform; improved and expanded education may generate human capital that aggregates to economic growth (Barro 2001; Ramirez et al. 2006).
Similarly, the size of the education system itself may shape levels of reform: Systems with a lower proportion of children enrolled may need more reform, or systems with more students enrolled might require more complex management and therefore higher levels of reform. Relatedly, the size of the population could drive reform, with more populous countries needing larger and more complex education systems that require higher levels of reform to build or maintain.
Finally, features of the political system likely play a role in shaping levels of education reform. Within a country, political interest groups can shape the nature of reform, co-opting education in service of a broader political agenda (Finger 2018). For example, Nitta’s (2007:8) detailed comparative study of structural education reform in Japan and the United States from the 1980s through the 2000s finds repeated evidence of “(a) politicians searching for symbolically significant victories and (b) elite bureaucrats fighting turf battles.” Similarly, a study of school reform in the United Kingdom argues that elite political contests shaped reform agendas in the 2000s more than commitment to shared ideals of high-quality education for all (Gunter 2018). Generally, more competitive political systems may be likely to pursue more education reform—and this relationship is especially likely to exist in a global context where education is valued. 3 In addition, the division of power in a country may matter. Decentralized or federal systems may have more reform because there is greater formal complexity. Our data collection, however, relied on national-level reports, which are more likely to capture country-wide reforms rather than policy changes in all subunits. We would expect countries with greater decentralization or a higher division of power between national and subnational units to report less national reform.
Data and Methods
Database Creation
To test our arguments, we created a database of education reforms in countries around the world over time. A more detailed discussion of the methodology is available with the database at http://www.werd.world. In brief, following the strategy used in Braga, Checchi, and Meschi (2013), we assembled an extensive list of reforms from reports submitted to the major international organizations involved in education. We extend their approach by drawing not only on the OECD’s Education Policy Outlook (EPO; 36 reports) but also on the OECD’s Review of National Policies in Education (RNPE; 47 reports), the World Bank’s Systems Approach for Better Education Results (SABER; 121 reports), World Data on Education (WDE) reports produced by UNESCO and the International Bureau of Education (IBE; 189 reports), and the International Encyclopedia of Education (IEE) compiled by country experts (109 reports). Our database is, to our knowledge, the most comprehensive cross-national, longitudinal compilation of education reforms to date. The reports list domestically initiated reforms and those conducted in partnership with international organizations, often in a section describing key historical changes to education policy. These reports tend to focus on national changes rather than subnational reforms, and they are intended for an international audience, so they may be more likely to reflect policy changes that are valued in the international community. Plausibly, a set of reforms constructed from solely intranational sources could include a greater number of subnational reforms or more reforms that break from global norms.
We relied on a team of undergraduate research assistants to conduct the coding, following a coding definition that described a reform as “planned, systemic, and nonroutine” (a detailed set of coder instructions is available upon request). We conceptualize education reform as intentional (“planned”) change (“nonroutine”) occurring at an administrative level above an individual school (part of a “system”), distinguishing it from unplanned, nonsystemic, or nonformal education fads or small-scale initiatives. The intraclass correlation for coding reforms using these criteria is .85, which is considered “excellent” by common standards of interrater reliability (Cicchetti 1994).
As an example, the 2017 SABER 30-page report from Liberia contains five reforms: 1989 Act Establishing a National Commission of Higher Education, 2011 Education Reform Act, 2014 National Policy on Higher Education in Liberia, 2016 university admissions reform, and the 2017–2021 Sector Plan (excerpt shown in Figure 1). In our analyses, the dependent variable includes reforms for Liberia in 1989, 2011, 2014, 2016, and 2017. The year and type of report are included as control variables.
Example of education reform from the World Bank’s Systems Approach for Better Education Results (SABER) 2017 Liberia report: “Getting to Best” Education Sector Plan 2017–2021.
Analytic Strategy
Our dependent variable is a count of the number of reforms reported in a country in a year. The modal response is 0, with a mean of .95 and a maximum of 27. Given that the count data are overdispersed, we use a negative binomial count model for panel data to consider factors associated with the outcome and include country fixed-effects (Long 1997). As our argument describes, we also incorporate a lagged dependent variable to account for how prior levels of reform may shape current levels in any given year.
Independent and Control Variables
To identify the historical periods relevant to our argument, we include a series of dichotomous variables that identify the time periods from 1960–1991, 1992–2008, and 2009–2017. We use 1992 to mark the post-Cold War era and 2009 to mark the post-global financial crisis because these two events likely intensified longer-term shifts in the expansion and weakening liberal world order. In our analyses, the period from 1960 to 1991 is the reference category.
A country’s World Bank loans for education projects identifies the cumulative number of loans a country has received from the World Bank for an education-related project up to a given year (Furuta 2020). To code this variable, we drew on publicly available records of all World Bank projects funded from 1947 to 2017 (World Bank 2017), and we identified whether a project was education-related by the project title. A country like Romania, for example, received funding for several projects over this time period, including projects titled “Education Reform Project” in 1994, “Reform of Higher Education and Research Project” in 1996, “Rural Education Project” in 2003, and “Romania Secondary Education Project” in 2015. This measure is logged to address the positive skewness of the distribution and mean-centered in all of our analyses.
A country’s memberships in INGOs is a widely used measure that captures a country’s diffuse linkages to global institutional models (e.g., Boli and Thomas 1999). This variable is coded using the Union of International Associations (1960–2018) Yearbook of Organizations, using the natural log to reduce the positive skewness of the distribution. The variable is also mean-centered in our analyses. We also include several control variables that reflect more nation-specific and functional processes that shape the pace of education reform in a given country. As a measure of administrative legacies, we include a variable capturing the number of education reforms adopted globally in the prior year, and we logged this to account for skewness of the distribution.
We include a country’s GDP per 10,000 capita to identify a country’s level of economic development using data from the Penn World Tables (Feenstra, Inklaar, and Timmer 2015); this measure is logged to reduce the skewness of the distribution. To examine arguments that higher- and lower-income countries are both more likely to engage in education reform, we include a variable that identifies a country’s GDP per 10,000 capita squared (logged).
The size of a country’s education system is coded as its gross secondary enrollment ratio using data from the World Development Indicators (World Bank 2020). A country’s population is also measured using data from the World Development Indicators (World Bank 2020) and logged to reduce the skewness of the distribution.
The degree of political competition in a country is measured using a variable from the Varieties of Democracy data set that jointly captures the institutionalization of political competition and government restrictions on political competition (Coppedge et al. 2020). This combined variable is measured on a scale from 1 to 10, where 1 indicates “repressed competition” (i.e., almost all political activity conducted outside of a country’s hegemonic regime is not permitted) and 10 indicates “institutionalized open electoral participation” (i.e., political groups compete for influence without the use of coercion, and no groups are excluded from politics).
To consider the extent to which political centralization shapes reform, we use the division of power index from the Varieties of Democracy data set (Coppedge et al. 2020). The score ranges from 0 to 1, where the lowest score indicates a country that has no elected local or regional governments. The highest scores are found in countries in which both local and regional governments are elected and able to operate without restrictions from unelected actors at the local or regional level (except for judicial bodies).
Our models also control for the number of each of the five report types listed previously (in a country in a year):EPO, IEE, RNPE, SABER, and WDE. We “lead” this variable by one year to account for the fact that country reports identify more reforms in the year immediately prior to their publication date, and we consider alternative specifications in our sensitivity checks. Table 1 presents the descriptive statistics for each of these variables.
Descriptive Statistics (n = 6,939 country-years).
Note: For logged variables with zero values, a small amount (.001) was added to the raw number before logging. INGO = international nongovernmental organization.
Variable is scaled so the mean equals zero to facilitate meaningful interpretation of main effects in an interaction in the analyses.
To provide a more concrete understanding of the measure of World Bank education loans, note that the 1960–1991 period had an average of 1.3 cumulative loans per country-year (and a range of 0 to 39), the 1992–2008 period had an average of 5.04 cumulative loans per country-year (and a range of 0 to 85), and the 2009–2017 period had an average of 8.5 cumulative loans per country-year (and a range of 0 to 95).
Findings
Figure 2 shows the number of reported reforms over time as a global total and as an average per country in three periods. From 1960 to 1991, countries reported an average of one reform every two years; from 1992 to 2008, the average increased to 1.65 reforms per year; and from 2009 to 2017, the average drops to less than one reform per country per year. These periods capture rough historical eras in the world system: The fall of the Soviet Union and the global financial crisis were shocks to the global system that exacerbated the trends of neoliberalism. We also show the average number of reforms per country per year because the number of countries in our sample each year varies. However, the country average trends closely with the global count; fluctuations are not driven by changes in the number of countries in the sample. 4
Number of reforms over time (n reforms = 6,700; n countries = 147).
These descriptive findings align with our first hypothesis and provide cross-national, large-scale evidence in support of existing case studies and historical descriptions of the growth of education reform activity through the 1990s. For example, Pogrow (1996) describes a period of “hyper-reform” in the U.S. education system in the mid-1990s. Prior work has noted an explosion of education reform activity and the globalization of education policy during the 1990s (Ball 1994; Mundy et al. 2016; Ramirez, Meyer, and Lerch 2016). Going beyond existing studies, our data also reveal a decline in levels of reform in the most recent decades. This decline is in line with arguments that the weakening liberal world order may be resulting in a reduced focus on education reform.
In addition to change over time, there is substantial variation between countries. The count of reforms reported for each country over the entire period ranges from a low of just one reform to a high of 215 reforms. The countries with the highest levels of reform are spread across the world and across several levels of economic development. Our regression analyses help explain the longitudinal and cross-national variation observed. We start with three models showing our control variables in Table 2. Model 1 presents the controls for the numbers of each type of report in a year. Model 2 introduces a lagged dependent variable to control for administrative legacies; as expected, reform levels in a prior year are positively associated with subsequent reform. In Model 3, we include additional country conditions that could shape levels of education reform. GDP per capita has a curvilinear relationship with the level of education reform in a country, with countries doing more reform at higher and lower levels of GDP per capita. Plausibly, perceived need and ability both play a role in supporting education reform. As a country gains resources, it may have greater capability to take on reform, and lower levels of GDP may promote reforms intended to stimulate economic productivity. In addition, the size of the education system matters. As the proportion of students enrolled in secondary school increases, countries articulate more education reform. 5 A larger population is also associated with more reform in these initial models.
Negative Binomial Regression Predicting Number of Education Reforms with Controls, 1960 to 2017 (n = 6,939 country-years; 147 countries).
Note: Standard errors are in parentheses. EPO = Education Policy Outlook; IEE = International Encyclopedia of Education; RNPE = Review of National Policies in Education; SABER = the World Bank’s Systems Approach for Better Education Results; WDE = World Data on Education; BIC = Bayesian information criterion.
p < .1. *p < .05. **p < .01. ***p < .001.
Finally, features of the political system shape levels of education reform. An index of the division of power in a country is negative and significant in early models, suggesting more decentralized polities report less reform. This finding likely reflects the fact that our data sources capture national levels of reform more than subnational trends, so levels of reform appear lower in decentralized or federal systems. For example, in the United States, the national reforms triggered by the 1983 A Nation at Risk report and the 2002 No Child Left Behind law appear in our data, but the multitudes of state-level reforms of curricular standards, teacher training, and other matters are not present. Political competition has a positive and significant correlation with levels of reform, consistent with case studies documenting the powerful effect of political forces on education reform (Finger 2018; Gunter 2018; Nitta 2007). For example, reforms focused on expansion at the primary level have been initiated to gain the support of median voters to increase state legitimacy (Paglayan 2021). These studies argue that education reforms and policies often reflect government efforts to satisfy citizens’ preferences (Lü 2014; Stasavage 2005). It follows, then, that political competition will likely matter most in contexts that value education, as in the neoliberal era of our study. Indeed, in a corollary analysis, we find that after including controls for historical eras, political competition is only significant in the peak of the liberal era from 1992 to 2008 (results available upon request). Subsequent models include all the controls. Levels of prior reform, GDP per capita and the squared term, and secondary enrollment remain significant throughout the analyses.
We turn to our core arguments in Table 3. Model 1 supports the descriptive historical trend shown earlier. After controlling for levels of reporting, prior reform, and relevant country conditions, we see an increase in reforms in the neoliberal era relative to the liberal period (Hypothesis 1a), on average, and then a recent decline (Hypothesis 1b). One possible interpretation of the recent slowdown in reform activity is that the world has reached a natural saturation point of education reform. However, the “ceiling effect” argument is hard to sustain in the face of long-term, unsolved, wicked problems and the rise of new issues such as socioemotional learning and increasingly urgent matters of sustainable development and climate change. In addition, the idea of a “natural” level of reform is hard to mesh with the great cross-national variation observed. The average country conducts less than one reform per year, but some countries conduct upward of a dozen reforms in a year, suggesting very high rates are possible.
Negative Binomial Regression Predicting Number of Education Reforms with Controls Plus Core Arguments, 1960 to 2017 (n = 6,939 country-years; 147 countries).
Note: Standard errors are in parentheses. WB = World Bank; BIC = Bayesian information criterion.
p < .05. **p < .01. ***p < .001.
A more likely explanation is that the 1990s, as a period of unprecedented liberal dominance, ushered in principles celebrating individual development and rational bureaucratic management. These principles spurred a wave of education reforms, encompassing expanded content (e.g., including curricula on human rights), expanded access (e.g., the Education for All movement), and structural changes (e.g., decentralization and privatization; for a diffusion of “decentralization” reforms, see Hossain 2022). But today, the legitimacy of the liberal international order is being challenged, which dampens enthusiasm for national education reform. Interrelated mechanisms could be operating simultaneously to slow the pace of reform over time. Nation-states may feel less compulsion to follow recommendations of liberal international institutions and the professionals that carry globalized scripts, or there may be less agreement around the principles of the liberal world order (including the centrality of education). Future qualitative research should examine the possible nuanced forms of weakening of the world order that underpin the statistical associations observed here.
Model 2 examines the influence of international organizations as key carriers of liberal principles. We expected international organizations to be key catalysts of education reform, and we find support for our second set of hypotheses. There are positive associations between World Bank loans and INGO memberships and increased education reform net of relevant country characteristics (Hypotheses 2a and 2b). It is no coincidence that the rise of neoliberal approaches to education policy pushed by the World Bank grew together with stronger INGO networks in the 1990s (Mundy 1998; see also Boyle and Kim 2009). With decentralization, private organizations are encouraged to flourish and provide greater choice and social services in replacement of the state. Together, these organizational actors carry principles of the liberal world order into national settings.
In Models 3 and 4, we consider how the influence of international organizations might change over time using interactions with historical era. For interpretability, Figure 3 depicts average marginal effects of the interactions in these models. As hypothesized, we see a declining role of World Bank loans in driving education reform over time (Hypothesis 3a). 6 At least since the mid-1990s, the World Bank has faced growing criticism of its policies (Ogbu and Gallagher 1991; Sen 1999; Vavrus 2005). As a result, the Bank is increasingly trying to position itself as a “Knowledge Bank” instead of a traditional lender, turning away from coercive lending mechanisms and focusing on providing research or technical advice over policy prescriptions (Zapp 2017). These findings support arguments that as the liberal world order weakens, the Bank’s coercive strategy of linking loans to public-sector reforms becomes less palatable (Bonal 2002).
Illustration of interactions between international organizations and time period in predicting number of reforms.
In contrast to the Bank’s influence through loans, the association between INGO memberships and education reform grows stronger over time, as shown in Model 4 (Hypothesis 3b). Although there is pushback against INGOs as challenges to the liberal world order grow, our empirical findings show that INGOs continue to drive education reforms. In part, these recent attacks have directed INGOs toward the safety of development work, amplifying efforts in arenas like education instead of more controversial matters like democracy or rights. 7 In addition, the declining legitimacy of coercive strategies in education reform, like the loans used by the Bank, may drive additional support for INGO work (for a discussion of how the World Bank increasingly uses civil society groups as “partners” to increase buy-in, see Nelson 2000).
Finally, we conducted many additional sensitivity and robustness checks, some of which are worth mentioning here. For example, we considered an array of additional country-level factors, such as country spending on education, that might be associated with reform but are excluded from our final analyses. Generally, these were insignificant after including relevant controls, and in some cases, the sample size became quite small due to missing data. In a few cases, the measures were collinear with other variables we included in the model, so we opted to exclude them. For instance, we considered a standard measure of democracy from the Polity V database, but it is highly collinear with political competition, and only one of those measures can be included (Coppedge et al. 2020; Marshall, Gurr, and Jaggers 2014). We include political competition because it aligned more closely with existing arguments in the literature about the political processes that generate reform. We also considered diffusion measures capturing the number of education reforms adopted both globally and regionally in the prior year. 8 These are both positive and statistically significant. However, we opted to focus directly on the mechanisms envisioned to facilitate diffusion (i.e., World Bank loans and INGO memberships) rather than on a general indicator for peer adoption. 9
As with any study, ours has limitations. We focus on the amount of reform and the carriers of reform rather than the content of the policy changes. However, our study and the database we introduce provides a starting point to build future research on the nature of education reform. Future work could analyze the contents of the policy titles and descriptions we compiled by education level, subject, or thematic areas.
Conclusion
The rapid and global diffusion of economic, political, and social policies rooted in the liberal valorization of individuals and choice was “the defining feature of the late twentieth century” (Simmons, Dobbin, and Garrett 2006:781). Reforming education was a central project for nation-states, as schooling provided the key vehicle for creating productive individuals under the liberal paradigm. Since the late 2000s, however, the liberal world order has faced growing criticism. Our findings show that levels of education reform within countries declined in recent decades, on average, which we argue was driven in part by the global cultural backlash against the liberal world order. 10 As the liberal world order and its celebration of the individual becomes less hegemonic, education reform becomes a less central feature of nation-state rhetoric. In addition to the broad historical and cultural influences that shape education reform, specific actors also influence nation-states. World Bank loans and INGO memberships promote education reforms. But we observed a lessened role for World Bank loans in promoting education reforms over time and a growing role of INGO memberships, pointing to a changing system of governance. Together, these findings show that macro, global processes play a key role in the adoption of national education reforms.
Declining levels of education reform portend several possible scenarios. On the one hand, less reform could lead to reduced access and deteriorating school quality. Enrollment growth has already slowed (Furuta, Meyer, and Bromley forthcoming). We may also see budget cutbacks, lower student achievement, and increases in student and teacher absenteeism. On the other hand, the link between education reform and access or quality is unclear. Although many actors advocate for more reform, accusations of “reformitis” and widespread doubts about the efficacy of education reforms are also commonplace (Debeauvais and Livesey 1986; Zaff 2011). Some scholars have observed a “predictable” and “persistent” failure of education reform (Payne 2008; Sarason 1990). To this point, Farrell (2007) argues that gains in education are the result of economic growth and social structural change outside of schools rather than a product of education reform itself. Moreover, some might see a reduction in some kinds of reforms as a positive and welcome shift (e.g., the much-criticized spate of neoliberal reforms around accountability and monitoring). Importantly, our database creates the possibility to systematically explore the association between education reform and measures of education quality and access around the world over time.
Institutions are sticky (Thelen and Mahoney 2010), so these future scenarios would likely take several decades to materialize. But we suspect declining reform efforts are a leading indicator of transformation. Much attention has been directed toward attacks on democracy and capitalism, but the social consequences of challenges to the liberal world order (e.g., changes to national education systems) are similarly immense.
Footnotes
Acknowledgements
We wish to thank John W. Meyer, Francisco O. Ramirez, Evan Schofer, and Chris Thomas for insightful feedback, and we thank undergraduates at Stanford University and the University of Toronto for research assistance. We also benefitted from comments by participants at the 2020 and 2021 meetings of the Comparative and International Education Society, Stanford’s workshops in Comparative Institutional Analysis and Philanthropy & Civil Society, the Scandinavian Consortium for Organizational Research, and participants in the 2021 workshop on Developments and Changes in Education Systems across Global Cultural Spheres at the University of Bremen.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding support came from the Institute for Research in the Social Sciences and the Center on Philanthropy & Civil Society at Stanford University, the Social Sciences and Humanities Research Council of Canada, and Victoria College at the University of Toronto.
Research Ethics
This research used publicly available anonymized survey data and thus does not constitute human subjects research.
