Introduction
Over the last decades, the field of behavioural public policy (BPP) has spread all over the globe. In the literature, BPP is commonly defined as ‘all means and modes of public policy aiming at influencing individual behaviour by using insights from behavioural economics, behavioural sciences, psychology, or neurosciences’ (Straßheim, Reference Straßheim2021:71). BPP is hereby associated with states increasingly turning to scientific expertise to shape and legitimise their policies. One phenomenon that is commonly associated with BPP is that of nudging. Nudges seek to influence the decision-making of individuals through positive reinforcement and indirect suggestions in a ‘choice architecture’, so that ‘desired’ choice-making becomes more likely (Thaler & Sunstein, Reference Thaler and Sunstein2008; Teitelbaum et al., Reference Teitelbaum and Zeiler2018; Whitehead et al., Reference Whitehead, Jones, Lilley, Howell and Pykett2019). Promotors of BPPs hereby often present nudging as an at once cheap and efficient alternative to conventional modes of state governance (e.g. strict regulation), since even slight adjustments in choice architectures can already have a large impact on policy outcomes (Sunstein, Reference Sunstein2014; Jones et al., Reference Jones, Pykett and Whitehead2014). Indeed, as Thaler and Sunstein (Reference Thaler and Sunstein2008) state in their famous book Nudge, nudge thinking emphasises, and draws legitimacy from, the fact that neither coercion nor a reduction of choices should be enforced, and that individuals must be respected in their autonomy – a principle that has been designated as ‘libertarian paternalism’ (ibid.; de Ridder et al., Reference de Ridder, Feitsma, Van Den Hoven, Kroese, Schillemans, Verweij and de Vet2020). At the same time, this linkage to paternalism (in steering individuals towards making decisions that are deemed ‘better’ for themselves and/or society – Banerjee, Reference Banerjee and Mihaila2021) as a complex and manifold paradigm has equally triggered intensive debates regarding the social and ethical implications of behavioural policies (e.g., Curchin, Reference Curchin2017; Watts-Cobbe & Fitzpatrick, Reference Watts-Cobbe and Fitzpatrick2023; Molander & Tosvik, Reference Molander and Torsvik2022).
Despite these ongoing (controversial) debates, behavioural insights have become increasingly adopted by governments around the world and implemented in various social policy fields as a governance model (Baggio, et al., Reference Baggio, Ciriolo, Marandola and van Bavel2021; John, Reference John2018; Curchin, Reference Curchin2017). At the same time, and as this paper seeks to show, BPP should not be regarded as a homogeneous concept. Rather, it always manifests depending on the specificity of (a) the geographical and socio-political context (e.g. Global North vs. Global South) and (b) the specific sector it aims to intervene in (e.g. education). One visible manifestation hereof, is the widespread emergence of behavioural insights units – often colloquially referred to as Nudge units (e.g. the UK’s Behavioural Insights Team (BIT), or the INudgeYou group in Denmark). Such units can be either directly or indirectly connected with governments to foster the implementation of behavioural insights (Afif, Reference Afif2017; Baggio et al., Reference Baggio, Ciriolo, Marandola and van Bavel2021). Even though Nudge units, thus, play a key role in the actual contextual formations of BPPs, they have rarely been investigated from a nuanced, empirical point of view (but see Ball, Reference Ball2022; Feitsma, Reference Feitsma2018, and Mukherjee & Giest, Reference Mukherjee and Giest2020, for some notable exceptions). Such a nuanced view deliberately questions general theoretical – positive or negative – claims regarding the features and characteristics of behavioural policies as outlined above, and instead carefully investigates how such policies are being enacted through particular forms of political, financial, or epistemological contextualisation (Baggio et al., Reference Baggio, Ciriolo, Marandola and van Bavel2021; Banerjee & Mitra, Reference Banerjee and Mitra2023; Hallsworth, Reference Hallsworth2023). Part of that context is, then, equally the specific policy field in which behavioural insights are applied. For instance, making use of nudges in work policies may crucially differ from nudge application in other public policy fields such as health or – as in the case of this paper – education.
In addition, most in-depth studies on behavioural policies and Nudge units that are presently available, primarily focus on Western, Anglo-American, contexts (John, Reference John2019; Hallsworth, Reference Hallsworth2023 – for large-scale, cross-country overviews on acceptance and implementation of nudging that exceed this focus, see Sunstein et al., Reference Sunstein, Reisch and Rauber2018, or Whitehead et al., Reference Whitehead, Jones, Howell, Lilley and Pykett2014). In doing so, research has thus far largely overlooked other regional or cultural contexts (Ball, Reference Ball2022; Decuypere & Hartong, Reference Decuypere and Hartong2023). Consequently, when researching Nudge units, this regional bias ‘limits our ability to account for contextual variation and poses significant challenges to building reliable theories or scaling up behavioural insights’ (Ghai, Reference Ghai2021:971). Moreover, it enforces the emergence of rather generalist claims regarding the effectiveness and relevance of behavioural insights for public policy (ibid.; see also Curchin, Reference Curchin2017), thus calling for research that shifts the focus deliberately towards people, regions, or contexts that are not (predominantly) Western, educated, industrialised, rich, and democratic (‘WEIRD’; see Ghai, Reference Ghai2021; Feitsma, Reference Feitsma2018).
Responding to both mentioned research gaps – the general call for more context-sensitive empirical research on the enactment of behavioural policies, and the specific call for research that includes greater sample diversity and (equally) focuses on contexts of such policies that equally incorporate the Global South – this article provides an in-depth qualitative study of MineduLAB, a Nudge unit active within the Peruvian Ministry of Education that was founded in 2013, and the (self-proclaimed) first public institution in Latin America with the purpose to systematically employ behavioural insights in the specific sector of education (MineduLAB, n.d.). MineduLAB’s specific contextualities hereby partly reflect images commonly associated with behavioural policies in the literature (e.g. through a shared narrative of behavioural insights as a coherent concept and a reliance on nudging as the primary behavioural insights technique used – Ball & Feitsma, Reference Ball and Feitsma2020), but – as we will show – equally manifest in a multiplicity of paradoxes.
Our study builds on documents and reports from the official website of MineduLAB, complemented by semi-structured interviews with five current and former members of the unit (conducted in the first half of 2022). The interviews covered questions on MineduLAB´s creation, operationality, reflections on the specific context of Peru, as well as the specific policy sector the unit is targeting (education). All interviews were transcribed and thematically analysed following a mix of inductive-recursive and deductive-theoretical coding (Ravitch & Carl, Reference Ravitch and Carl2021). The analysis revealed a number of important contextualities of MineduLAB. First, while the unit is centrally situated within the government, it is hardly financed by governmental budget. Rather, it is ‘enacted’ by unpaid contributions of a wide network of public and private partners outside that government. Second, while policy knowledge is heavily imported from global actors (including the BIT in UK), this knowledge is simultaneously adapted and shifted in a highly localised manner. Third, the staff of MineduLAB expresses a strong orientation towards educational aims and values, which actually question the suitability of experimental design to properly address ‘true’ educational problems.
Taken together, this study showcases the crucial importance of paying close attention to empirical manifoldness, ambiguities, and contextuality regarding behavioural policies and the role of Nudge units herein. Through a detailed empirical analysis of such manifoldness, ambiguities, and contextualities, the study aims to provide a baseline for future research devoted to such a context-sensitive approach.
Nudge units and the critical exploration of behavioural public policies
Triggered by the political legitimacy crises of the ‘1960s’/‘1970s’, as well as by the growing complexity of policymaking, states around the world have increasingly adopted governance models that move away from hierarchical control and intervention, and towards more collaborative and indirect forms of regulation (Mayntz, Reference Mayntz2017). Behavioural policies can be regarded as a particular instance of this ‘soft’ mode of governance (Jones & Whitehead, Reference Jones and Whitehead2018; Doyuran, Reference Doyuran2024). Some of many areas where such behavioural policies have been applied over the past decades are tax declaration, energy consumption, nutrition, or vaccination (European Commission, n.d.; Thaler & Sunstein, Reference Thaler and Sunstein2008).
Behavioural policies have not only been promoted as an effective means to ‘steer’ people towards desired behaviour; they have equally been promoted to be relatively low in costs and comparatively high in acceptance by citizens (Baggio et al., Reference Baggio, Ciriolo, Marandola and van Bavel2021; Tor, Reference Tor2022; Szaszi et al., Reference Szaszi, Higney, Charlton, Gelman, Ziano, Aczel and Tipton2022). However, behavioural insights have equally been subject of substantial critique (see Hallsworth, Reference Hallsworth2023, for a recent overview). For instance, nudging has been criticised for ‘[…] neglecting the ability of humans to actually reflect on their actions or to learn from failure (which nudges actively prevent) or for implementing soft, unconscious forms of coercion and behaviourist manipulation (which neither include democratic participation nor a structural addressing of social problems)’ (Decuypere & Hartong, Reference Decuypere and Hartong2023: 139; see also Mols et al., Reference Mols, Haslam, Jetten and Steffens2015; Schmidt & Engelen, Reference Schmidt and Engelen2020). From that standpoint of critique, the ‘behavioural state’ is often said not to aim at changing fundamental socio-economic, political, and institutional structures through profound reforms at the macro level, but rather addresses the micro level of the individual to modify (problematic) behaviours, whilst at the same time guiding them in making better choices (Curchin, Reference Curchin2017). Similarly, critiques of neoliberalism have argued that, whereas behavioural economics originally aimed at establishing a more correct and realistic view on humans than the liberal image of the homo economicus (Thaler & Sunstein, Reference Thaler and Sunstein2008), the political economy of behavioural policies found in many national contexts has actually furthered neoliberal agendas (‘neuroliberalism’; Whitehead et al., Reference Whitehead, Jones, Lilley, Howell and Pykett2019; see also Gandy & Nemorin, Reference Gandy and Nemorin2019). Moreover, from the quantitative literature on nudging, we can learn that nudge interventions overall show mixed results, and point to some serious issues such as ineffective interventions and publication bias (e.g. Damgaard & Nielsen, Reference Damgaard and Nielsen2018; Maier, et al., Reference Maier, Bartoš, Stanley, Shanks, Harris and Wagenmakers2022; Mertens et al, Reference Mertens, Herberz, Hahnel and Brosch2022; Hummel & Maedche, Reference Hummel and Maedche2019). In that respect, there has been a rising call for more nuanced approaches towards behavioural policies, triggered by both conceptual work (e.g. Banerjee, Reference Banerjee and Mihaila2021; Schmidt & Engelen, Reference Schmidt and Engelen2020; Watts-Cobbe & Fitzpatrick, Reference Watts-Cobbe and Fitzpatrick2023; Molander & Torsvik, Reference Molander and Torsvik2022; Curchin, Reference Curchin2017) and empirical investigations (e.g. Ball, Reference Ball2022; Feitsma, Reference Feitsma2018; Mukherjee & Giest, Reference Mukherjee and Giest2020).
As noted in the introduction, over the past decades, many governments around the world have created ‘Nudge units’ (or similar institutional settings) responsible for the design, implementation, and evaluation of behavioural policies, either based directly in the central government (as the UK’s Behavioural Insight Team, BIT, was for a long time), or operating in a more autonomous, project-based manner (such as the iNudgeYou consultancy agency in Denmark) (Afif, Reference Afif2017). Some international institutions have equally created behavioural insights units to support their policies at supranational level, such as the World Bank´s eMBeD (The Mind, Behaviour, and Development Unit). Yet, while their quick global diffusion may evoke assumptions of Nudge units as all working in similar manners, research has started to point to important differences and a high influence of local contextuality, for instance regarding the actual and dynamic positionality of these units within specific political economies. One example is Neuhaus & Curley’s (Reference Neuhaus and Curley2022) recent study on the UK’s Behavioural Insights Team (BIT) – which is the Nudge unit most commonly discussed in the literature (see also Delaney, Reference Delaney2018). Neuhaus and Curley trace the BIT’s ongoing transformation from a department in the British Prime Minister’s Cabinet Office, over an autonomous agency, into a global, privatised service provider that is now owned by the self-proclaimed ‘innovation agency’ Nesta (Nesta, 2021). Each period was substantially triggered by particular actors (either from within the government, or from within the BIT), and strongly affected not only the structure, but also the overall positionality of the BIT and, hence, the contextualised manifestations of behavioural policies in the UK (ibid.).
In one of the few existent qualitative empirical studies on the actual operationalities of Nudge units (in this case, in the Netherlands), Feitsma (Reference Feitsma2018), in turn, shows that claims regarding nudging as emblematic of ‘technocracy’ (meaning that the state is ‘engineering’ citizens from above through expert knowledge) or ‘psychocracy’ (meaning that expertise is drawn primarily from psychological knowledge and methods) turn out to be very generalist and theoretical when put to the empirical test. More specifically, Feitsma shows, contra simple technocratic and/or psychocratic visions, how the Dutch behavioural insights team frequently established collaborative democratic work with citizens in discussing and developing nudge designs, hereby building on heterogeneous forms of knowledge consolidation (that greatly extended psychological insights alone). Put differently, while behavioural policies are oftentimes associated with experimental Randomized Control Trials (RCT’s)Footnote 1 as gold standard of knowledge production (Ariel et al. Reference Ariel, Bland and Sutherland2022) – also because the BIT heavily promoted this method in their (global marketing) work (Delaney, Reference Delaney2018; Monaghan & Ingold, Reference Monaghan and Ingold2019) – knowledge building and filtering within Nudge units may also go well beyond that method.
Similarly, in the case of an Australian behavioural insights unit, Ball (Reference Ball2022) has argued that behavioural insights are not uniformly understood by policymakers. Indeed, different traditions interpret behavioural insights differently, ranging from an instrumental focus on ‘what works’ towards equating behavioural insights with nudging. Like Feitsma, Ball (Reference Ball2022:326) equally shows that people inside Nudge units are not necessarily only targeting the level of the individual and often actively aim to minimise concerns that are associated with adopting behavioural insights (such as the often untransparent deploying of nudges), as well as to address critiques that are often given to Nudge units (e.g. technocratic and top-down decision making). Such arguments closely align with recommendations made by scholars that seek to push the field of BPP towards more transparency and public deliberation, such as nudge plus (Stoker & John, Reference John2019; Banerjee, Reference Banerjee and Mihaila2021), ‘advanced’ BPP (Ewert, Reference Ewert2020; Ewert et al., Reference Ewert, Loer and Thomann2021), and a greater focus on the collective societal constellations in which individual behaviour is always necessarily embedded (Banerjee & Mitra, Reference Banerjee and Mitra2023).
Analytical framework
As the last section showed, there is a growing, yet still limited, body of research that has developed a more context-sensitive empirical understanding of Nudge units as parts (and emblems) of BPPs. It is that kind of research that the present article seeks to further contribute to, by investigating the self-proclaimed ‘Nudge unit’ and ‘innovation lab’ MineduLAB (Afif, Reference Afif2017).
Conceptually, the article is situated within an interpretive approach towards BPP (Rhodes, Reference Rhodes and Massey2019). Rather than focusing on a priori existing structures or functions in an organisation (i.c. MineduLAB) and aiming to provide a unified understanding of ‘the’ institution, an interpretive approach is interested in actions and practices of people involved in making the organisation whilst navigating heterogeneous contextual influences. An interpretive approach thus focuses on how people make, and act on, meaning, and the analytical narrative that is the result of such an undertaking ‘provides an account of complex specificity in context’ (Rhodes, Reference Rhodes and Massey2019).
It is such an interpretative approach in which our analytical framework is situated. On the one hand, we developed a preliminary theoretical understanding of BBPs nature and contextuality through a thorough engagement with the literature (and with particular emphasis on deploying BI in education), which informed the structuring of the data collection (e.g. interview questions) and gave structure and meaning to the data analysis. On the other hand, we sharpened this theoretical understanding in the course of the actual data analysis in a more inductive manner. As Ravitch and Carl (Reference Ravitch and Carl2021) argue, the goal of such an approach is to bridge and overcome rigid borders between the conceptual, the theoretical, and the methodological dimensions of qualitative research, while at the same time ensuring validity and theoretical robustness.
In our case, the result of this combined approach is a framework that distinguishes three dimensions in order to capture the differential enactment of MineduLAB: (1) the political-financial context, (2) the epistemic community, and (3) the sector specificity (here: education). The first dimension – the political-financial context – particularly addresses the regional situatedness of MineduLAB in the Global South, specifically in the politically unstable as well as financially challenging environment of the Peruvian system. At the same time, countries such as Peru oftentimes serve as a ‘target’ for international actors such as the World Bank to promote Western public policy models in the Latin American context (which also showed as being highly relevant for the diffusion of BPP). Furthermore, MineduLAB’s specific focus on the education sector relates to the substantial political-financial challenges the system is ongoingly confronted with, including high rates of poverty-related school dropout, distrust in the public education system, and poor quality of teaching and infrastructure (Saavedra & Gutierrez, Reference Saavedra and Gutierrez2020).
The second dimension refers to the epistemic community around the enactment of MineduLAB. Epistemic communities are characterised by their members being ‘glued together by internal beliefs that make them politically autonomous and credible’, as such sharing ideas and conceptions about what valuable and legitimate knowledge is (Straßheim, Reference Straßheim2021:70). In our case, this pertains to how MineduLAB adopts, designs, develops, and distributes, specific ideas and (behavioural) insights about what constitutes, and should be consolidated as, valid knowledge about learning, development, and the overall Peruvian educational system.
Lastly, the educational dimension refers to the specific policy sector MineduLAB is addressing. More specifically, the dimension helps bring attention to the particular connections between education and nudge thinking, which are at least twofold. On the one hand, it can be argued that there is a traditional closeness of the field of education to the field of behavioural insights (e.g. with regards to matters of disciplining students – Gigerenzer, Reference Gigerenzer2015) and BPP more generally. At the same time, it can equally be noted that there is often a tension between how learning is conceptualised in both fields – the field of behavioural insights traditionally building on more behaviourist learning theories; the field of education employing more constructivist ones (Decuypere & Hartong, Reference Decuypere and Hartong2023; Damgaard & Nielsen, Reference Damgaard and Nielsen2018). On the other hand, however, the field of behavioural insights has increasingly started to design interventions (including ‘boosts’ and ‘nudge pluses’) that are reflective in nature, that foster ‘perspective transformation’, and that are for that reason designated as educational strategies to school individuals beyond a behaviourist mindset (Banerjee, Reference Banerjee and Mihaila2021; Banerjee & John, Reference Banerjee and John2021). It is, hence, to be expected that this focus on education (and its complex relationship towards nudging) equally affects the operationality of MineduLAB in specific ways.
In the next sections, we provide more specific insights into our findings regarding each of the three dimensions.
Unpacking the local enactment of behavioural public policies: the case of MineduLAB
The political-financial dimension of MineduLAB
MineduLAB is the first out of three so-called innovations labs ‘created by the Peruvian government aimed at making positive changes by tweaking policy with insights gleaned from the study of behavioural economics’ (Bernstein, Reference Bernstein2017). A combination of two incidents can be regarded as crucial for the foundation of MineduLAB: the assignment of economist Jaime Saavedra as Minister of Education in 2013, and the publication of the PISA (Programme for International Student Assessment) results in the same year that ranked Peru last regarding its students´ performance (OECD, 2013). Both incidents together evoked a major education reform, which was at the same time strongly impacted by Saavedra:
“Jaime Saavedra had a strong technical view on policy decision making. He created the Strategic Monitoring and Evaluation Office, where MineduLAB is installed.” (Nancy, current member of MineduLAB) Footnote 2
As this quote shows, MineduLAB should be regarded as one crucial piece within an education reform that targeted performance increase through installing a stronger ‘technical’ social policy approach, characterised by an increasing relying on behavioural insights, of which nudging formed one component. Hence, MineduLAB became operationally directly attached to the government’s Office of Strategic Planning (OSP) (MineduLAB n.d.); that is to say, it became part of Peru’s wider monitoring and evaluation policy package. The position of MineduLAB within this office enabled convenient access to administrative data, which was regarded as supporting knowledge consolidation and evaluation significantly (cf. infra). As different interviewees explained, MineduLAB’s central positioning within the government was also crucial because of credibility reasons:
“(…) it is important to be in the government because it gives you the trust of the public sector. They usually have more clarity on the real impact or potential of the innovation. If they were implemented in a private organization, it would be faster, because there are no hindrances from the state. If they happen within the state, it takes a bit longer, but it is clearer. Also, the escalation or scope that the state can reach is way broader than what the private sector could cover.” (Amanda, former member of MineduLAB).
Given the specific economic context of Peru, however, the implementation of MineduLAB also became immediately affected by a second requirement: cost efficiency. As shown in a World Bank (2018) overview, Peru was, at that point in time, one of the only emerging economies that installed a Nudge unit, resulting in substantially limited resources as compared to Nudge units in the Global North (see also Afif et al., Reference Afif, Islan, Calvo-Gonzalez and Dalton2019). For instance, whereas the BIT’s annual budget was in 2015–2016 already estimated to amount to half a million pounds (Ball & Head, Reference Ball and Head2021), MineduLAB, conversely, could only afford a team of four members, with minimal budget for the design, implementation, evaluation, and eventual adaptation, of actual nudge instruments. As a result, on the one hand, MineduLAB largely built on student data that were generated within the governmental data systems anyway, to inform its operations and to form the backbone of the behavioural insights adopted. On the other hand – and this will be further explored in the next section – both the creation and the daily operations of MineduLAB have only been rendered possible through the technical (and, thus, indirect financial) support of a wide range of external organisations. Such organisations particularly include international policy actors such as IPA (Innovations for Poverty Action), J-PAL (Abdul Latif Jameel Poverty Action Lab), FORGE (a Canadian fund to strengthen initiatives of evidence generation in education), and GRADE (Group for the Analysis of Development; a Peruvian non-profit organisation that aims to conduct applied research to inform policy decisions), but equally researchers from diverse universities that became in charge of designing, piloting, evaluating, and implementing the instruments subsequently rolled out by MineduLAB (often designated as ‘innovations’) free of charge, yet in exchange for ‘reputation’ and/or ‘information’. Put differently – and here we see an interesting first paradox in MineduLAB’s enactment of behavioural policies – while the Nudge unit has been deliberately situated in the government, it has always worked through a wide actor-network that stretched from local to global, and across various political agendas. Likewise, a large set of non-monetary resources (e.g. reputation) has always been traded between MineduLAB, international and local academic partners (including scholars associated with Vanderbilt, Princeton, Oxford, the MIT, the World Bank and the BIT), local public servants, and the Peruvian Ministry of Education, in order to enact MineduLAB’s work (MineduLAB, n.d.). While MineduLAB has hereby been receiving technical expertise, access to data, collaborators to design and implement nudges, as well as credibility, our interviews made clear that academic partners received opportunities to conduct and publish cutting edge research, and that MineduLAB’s public servants actually welcomed scientific ideas that could potentially facilitate their work. These mutual benefits indeed represent a unique characteristic of the operationalisation of MineduLAB, and contribute to further understanding the dynamics of the diversified political economy on which the lab thrives. That is, while we may talk of ‘the lab’ as a single governmental entity, in doing so, we actually disregard the massive decentralisation of tasks behind it.
In addition, there is another interesting dimension about the low-budget setup of MineduLAB which became clear in the interviews, namely an intention to safeguard the unit from corruption and, in doing so, to retain its evidence-based orientation. Given that corruption forms a significant threat to the Peruvian political context (e.g. Beesley & Hawkins, Reference Beesley and Hawkins2022), our interviews indicate that the low budget could, consequently, be understood as setting the unit at least partly ‘free’ from this context. This makes it easier to understand that, even though MineduLAB has undoubtedly been affected by Peru’s tumultuous political environment, it has, until today, survived six presidents and twelve ministers of education (Acuña, Reference Acuña2023). Or, as a former member of MineduLAB put it:
“It keeps running despite the government, not thanks to it.” (Amanda, former member of MineduLAB)
This does not mean, however, that the lab did not need to restart activities from scratch several times, and to work alongside thematic priorities of the different governments in order to hold its position:
“They were developing an innovation to address gender gap reduction in accessibility of women in STEM careers, but changes in management led to a reducing of this priority.” (Camila, former member of MineduLAB)
Given this still substantial governmental dependence on determining which priorities to set – in addition to the crucial importance of external network support – transferring behavioural insights to local contexts, and thus extending MineduLAB’s epistemic community, has become a high priority of MineduLAB. That is to say, while MineduLAB initially designs and tests nudges, the goal is also to empower different regional leaderships to autonomously implement and evaluate these ‘interventions’ as part of their journey towards evidence-based policy, as the next section equally shows.
The epistemic community dimension of MineduLAB
As shown in the last section, MineduLAB’s work has always been embedded in a wide network of actors that not only provides resources for the design and implementation of its instruments (i.e. nudges), but also and in particular shapes which kind of knowledge the unit has been drawing on – and, thus, created what Monaghan and Ingold (2018) describe as ‘evidence hierarchies’ – and which kind of knowledge has, in turn, been produced. In general, despite the turbulent political and financial environment, most interviewees emphasised how MineduLAB has successfully contributed to shifting the epistemic ‘mindset’ towards evidence-based policies in the field of Peruvian education. Not only has MineduLAB developed a set of instruments that can, in their view, help improving education through ‘innovations’; equally, it has inspired other public institutions to develop and apply similar evidence-oriented methods:
“There have been changes in the way people [policy makers in public institutions] manage and use evidence; not only in education, but equally in other areas” (Daniela, educational policy evaluator).
As shown above, Saavedra, in his role as education minister when MineduLAB was founded, radically opted for this evidence-based approach in order to overcome the ‘mediocrity’ (Saavedra & Gutierrez, Reference Saavedra and Gutierrez2020) of the Peruvian system. At the same time, with a background in behavioural economics, Saavedra equally strongly drew on international ‘role models’ of countries that had successfully implemented behavioural insights and evidence-based approaches in their BPPs. More specifically, our interviewees argued that his idea was to design a Peruvian Nudge unit based on the UK’s BIT model (Behavioural Insights Team, see also below), and more particularly on their understanding of what proper behavioural knowledge and insights are precisely (see Nesta, 2021).
Regarding knowledge consolidation, many international and local collaborators did not only support MineduLAB, but the nudges themselves were largely generated outside of the unit. While the Ministry of Education itself hereby developed some proposals for policy ideas (MineduLAB, n.d.), other proposals directly came from researchers, social entrepreneurs, and the private sector. In any case, large parts of the design development, evaluation, and technical implementation support have been conducted by external partners, who, consequently, brought in their specific knowledge and ideas on different levels, and as such contributed to the establishment of a heterogeneous epistemic community. While in the first few years, researchers supporting the lab could directly invite colleagues to identify areas of relevant nudge research, later onwards MineduLAB published broader public calls for proposals. As noted in the previous section, researchers commonly do not receive any financial reward. Instead, their incentive is to gain experience within a governmental institution to develop a project, and to collect data that can subsequently be used for academic research and publishing.
Interestingly – and here a second paradox can be identified – despite this strong reliance on the BIT model, both our interviewees as well as the documents we analysed do not use the word ‘nudge’ at all, and instead refer to ‘innovations’ and ‘interventions’ to describe what they do. A reason for this might lie in MineduLAB’s central role within a broader educational reform movement directed at building an innovative future (as such extending behavioural rationales); another might be a larger attractiveness for external actors who more likely seek to contribute to (larger) innovative solutions than only to (rather narrow) nudge instruments; and a last one might simply lie in lacking options of Spanish translation for ‘nudge terminology’ (which would have implied to only work with the English term, and hence a break with the conventions of the Spanish epistemic community). Whatever the reason, this ‘break’ with one of the core components on which the lab has based itself (i.e. the BIT model) is at least remarkable.
In addition, the key role of international and local partners to contribute to the nudge design, implementation, and evaluation process does not mean that knowledge consolidation within MineduLAB’s epistemic community is not equally affected by local educational actors. In this regard, interviewees for instance identified a high relevance of local schools/school coordinators, principals, or teachers who are responsible for formally organising the implementation of MineduLAB’s designed nudges. More specifically, once a (nudge) design is ready to undergo a pilot study within selected schools, MineduLAB first has to present the proposal to the Ministry of Education, which then decides positively or negatively on the rollout of the ‘innovation’. Even though this partly constitutes centralised decision-making, negotiations between the Ministry and the affected schools hereby typically take place via decentralised school governance units, including the UGEL (Local Educational Management Units), and the DRE (Regional Directorates of Education). Regarding these negotiations, we clearly see that there is far from only ‘top-down’ regulation taking place within these Nudge units, but rather a strong influence of open discussion and collaboration with the implementing institutions (Feitsma, Reference Feitsma2018). At the same time, however, within that collaboration, MineduLAB strictly limits its own position to that of an ‘experimental evaluator’ of the nudge output, while schools or local actors are deliberately not assisted in the actual implementation process (in practice). In other words, we can note a clear split here between actors providing experimental/evaluative knowledge (in hands of MineduLAB) on the one hand, and other kinds of knowledge needed to implement nudges (in hands of local actors) on the other hand. In turn, if the pilot study has yielded significant results (in whichever form), they are typically extensively discussed by MineduLAB and/or by its academic partners in various forms of publications; that is, ‘reclaimed’ by these actors, for purposes of knowledge sharing.
A final point regarding the knowledge consolidation context of MineduLAB, is the usage of data for nudge evaluation. As noted earlier, MineduLAB’s central location within the government provides the unit with direct access to student data, which was equally seen as a cost-effective way to evaluate the success of nudges through tracking changes in these data. More specifically, MineduLAB has commonly drawn on SIAGIE (Sistema de Información de Apoyo a la Gestión de la Institución Educativa - Information and assistance service for the management of student information), which is a state-administered tool to keep records of students’ educational trajectories in both public and private education. In other words, the SIAGIE allows the Ministry of Education to manage information on the processes of enrollment, attendance, and academic achievement of students. MineduLAB uses these data to inform their nudge designs; that is to say, to diagnose problems, define the characteristics of the experiment, analyse its impact, and evaluate the results after the implementation. However, in practice, different interviewees stated that the data systems are not only oftentimes incomplete, but also that more nuanced data about student learning cannot be evaluated through the system. Equally, the data are usually collected and transmitted to MineduLAB at a rather low frequency.
The educational dimension of MineduLAB
The third dimension refers to the relation between behavioural insights and the specific policy field that is targeted by MineduLAB – the domain of education – for instance, regarding the relevance of particular understandings of education in designing specific instruments (‘innovations’) that are informed by behavioural insights. Equally regarding this last dimension, the case of MineduLABs reveals different paradoxes.
To begin with, while much literature on the adoption of behavioural insights in the education sector has identified a strong focus on (improving) the learning outcomes of students (Decuypere & Hartong, Reference Decuypere and Hartong2023), the staff of MineduLAB reports that their work is oriented at much broader social challenges, including teacher absenteeism and dissatisfaction, student dropout, or problems of school maintenance. In that regard, one interviewee made very clear that MineduLAB, even though it thematically focuses on education, seeks to improve social policies and not the learning outcomes of individual students. Here, one of our interviewees mentioned a different unit – FONDEP (National Fund for the Development of Peruvian Education) – that is responsible for assisting and promoting innovations in specific classrooms. In contrast to FONDEP’s central area of working, MineduLAB’s focus is on designing policies that are amenable to nationwide implementation and, hopefully, nationwide social improvement. This is why the policies of MineduLAB must be easily adaptable to different school contexts. As already noted above, the lab distributes material for building behavioural interventions into schools. This material has already gone through experimentation to prove its effectiveness (in the sense of ‘being able to scale’). However, adapting this material to the concrete contexts and demands of the school, is something schools themselves are expected to do. Interestingly, several interviewees emphasise that this material still needs pedagogical contextualisation (for instance, the innovation ‘grow your mind’ aims at activating the potential of students through psychological resources, but always still requires adaptation by the local school), but argue that MineduLAB (with its focus on large-scale and scalable solutions) cannot do this work for the schools. Moreover, the limited ability of MineduLAB to track the potential success of interventions, particularly over a longer time, also plays a role in this deliberate ‘pedagogical limitation’ of its own work. This occurs because administrative data provided by the OSP are sometimes reported to be incomplete or outdated, at times making it an unreliable tool to evaluate success of the developed ‘innovations’.
However, while these limitations seemingly shift MineduLAB’s focus away from educational matters such as student learning, children’s development and teacher pedagogy, many of our interviewees were deeply concerned with such matters. This frustration equally includes awareness of the actual non-suitability of low-cost experimental designs to fully address complex educational questions like the proper development of Peruvian children:
“Given the low-cost interventions and experimental designs, the range of programs [MineduLAB] could evaluate was limited. It did not involve interventions aimed at improving children’s development in border or rural areas, for example. Experimental designs do not address more complex problems of the Peruvian education […].” (Alvaro, educational researcher).
Moreover, when looking at the actual innovations developed from MineduLAB, many of them are strongly directed at evoking conscious educational decisions from people, for instance through encouraging them to pursue formal education, through reminding them of the values of education, or through teaching them mechanisms of learning. For instance, the nudge ‘Choosing for a better future’ illuminates the social and monetary returns of secondary and higher education to students, making use of videos and infographics. The ‘Grow your mind!’ nudge, in contrast, uses scientific articles about the working mechanism of the brain and discusses them with high school students to change their perception of their own abilities. Both nudges are, thus, clearly ‘educative interventions’, in the sense that they aim to lead students and parents to the path of ‘perspective transformation’ (Banerjee & John, Reference Banerjee and John2021). The attempt to design such ‘educative nudges’ that can ‘school’ people, strongly resonates with other behavioural insights and techniques presently advanced (such as ‘nudge pluses’, ‘boosts’ and ‘thinks’ – Banerjee, Reference Banerjee and Mihaila2021; Oliver, Reference Oliver2018). As we will argue in the conclusion, this clearly showcases the need for further empirical and conceptual work on the close relation between the fields of behavioural insights and education.
Conclusion
The goal of this study was to disentangle and analyse the local contextual enactment of BPPs, as operationalised in Nudge units, through a case study of MineduLAB. We did so to improve our understanding of what such behavioural policies are and how they manifest in specific regional (i.e. Peruvian) and specific policy (i.e. education) contexts (e.g. Feitsma, Reference Feitsma2018; Schubert, Reference Schubert2017; Gandy & Nemorin, Reference Gandy and Nemorin2019; Schnellenbach, Reference Schnellenbach2012). Building on a growing body of literature in this field, we hereby argued that it is important not to follow simplified or generalised assumptions about Nudge units, and behavioural policies in general, and that the way forward is to acknowledge – and analytically address – manifoldness, ambiguities, or paradoxes (Hallsworth, Reference Hallsworth2023; Molander & Torsvik, Reference Molander and Torsvik2022).
In line with that argumentation, in studying a Latin American Nudge unit, we aimed to contribute to countering the still existing bias in sample diversity in research around behavioural policies in general, and Nudge units in particular (Decuypere & Hartong, Reference Decuypere and Hartong2023). At the same time, our aim was to learn more about how the specificities of a particular social sector (education) manifest in the work of Nudge units. Indeed, as our findings illuminate, Peru’s unique political-financial context revealed a high impact on how MineduLAB operates and, consequently, how behavioural policies are concretely given shape in a specific context. Moreover, our analysis has shown that this impact is not to be understood unidirectionally, but rather manifests as a series of paradoxes which can be found in all three dimensions of our framework. It is these paradoxes which, in their ongoing interplay, constitute the specificity of the MineduLAB case.
More specifically, regarding the political-financial dimension, MineduLAB appears as a Nudge unit that is situated in the government and at the same time ‘enacted’ through the collaboration of various external actors from other sectors. Regarding the related dimension of epistemic community, we found that MineduLAB shows a significant orientation towards the ‘BIT model’, at the same time equally strongly shying away from using any ‘Nudge’ terminology at all. This ambivalence is also visible in Minedulab’s dealing with experimental evidence production (i.e. RCTs), which is at once heavily employed and criticised. Particularly the latter equally affects MineduLAB’s engagement with the specificities of education as a policy sector. That is to say, while MineduLAB is considered to be very distinct from (local) educational practices, our participants simultaneously expressed strong educational aims and values, thereby actually questioning the suitability of experimental design to properly address educational problems. Moreover, some of the nudges that Minedulab developed, are explicitly seeking to foster reflection and be of ‘educative’ nature. At the same time, such educative nudges seem to be very much akin to didactic techniques and teacher work that is commonly deployed in the classroom, and we argue that further research should further unpack the (possible) linkages between the fields of behavioural insights and didactics, both conceptually as well as on an empirical level (Decuypere & Hartong, Reference Decuypere and Hartong2023; Gigerenzer, Reference Gigerenzer2015).
In sum, while the three dimensions substantially influence the enactment of behavioural policies in the case of MineduLAB, we do not suggest that other Nudge units reveal the same structural conditions. Rather, our study precisely shows that contextual factors affect the fundamental structural dimensions under which a Nudge unit functions, suggesting there might be other constituent elements that fell out of the scope of this study, and that may become manifest when including other analytical entry points or case studies, for instance other Nudge units in the Global South (Halpern, Reference Halpern2015; Feitsma, Reference Feitsma2018; Decuypere & Hartong, Reference Decuypere and Hartong2023). In that respect, the insights yielded by our study should still be deepened in future work. Equally, some themes that were absent in this study should still be explored in further research, such as the change in mentality that MineduLAB has triggered within the Peruvian education system, and the magnitude of that change, as this potentially could have large implications on various policy issues in the context of an emerging economy in the Global South.
Competing interests
The authors declare none.
Paola Ñañez is a Colombian educationalist living in Belgium. Her international outlook led her to recently finish her MSc in Educational Sciences at KU Leuven. In her Master’s Thesis, she investigated the political economy of MineduLAB and was invited as a guest lecturer at the Helmut Schmidt University in Hamburg to present her research project.
Mathias Decuypere is professor of school development and governance at the Zurich University of Teacher Education. His research adopts an international, macro-level perspective on educational policymaking and governance. It is interested in how distinct global developments in governance (e.g., behavioral governance; platform governance; synthetic governance) affect local school practices.
Sigrid Hartong is professor for sociology at the Helmut-Schmidt-University in Hamburg. Her research focus is on the transformation of governance in education and society, covering topics such as algorithmic, anticipatory, or behavioral governance.
Open access
