Social Impact Bonds in São Paulo’s State Public School System: New Modality of Public-Private Partnership in Brazil

This article analyzes the development of a new modality of public-private partnership in Brazilian education, the so-called Social Impact Bond (SIB), known in Brazil as “Contrato de Impacto Social”, whose pilot project, sponsored by the Inter-American Development Bank (IDB), is in its initial stage of implementation in the State of São Paulo. Unlike traditional philanthropy, which legitimizes itself in the alleged non-profit Education Policy Analysis Archives Vol. 26 No. 130 SPECIAL ISSUE 2 nature of its actions, SIB assumes the possibility of obtaining return rates with “social investments,” leading to the creation of a new ecosystem of social finances, which, into this context of privatization of education, incorporates agents interested in capitaliz ing on the provision of public services to vulnerable populations, in this case, to secondary students from public schools in the State of São Paulo. This kind of contract, or bond— paid by means of “delivery” of results—raises a series of ethical, legal, and politicalpedagogical issues related to the accomplishment of “social experiments” with students in a vulnerable condition. From the analysis of internal documents of the Department of Education in the State of São Paulo, in comparison with documentary sources from public schools, we conclude that, from the very beginning of its implementation, the SIB of São Paulo education not only breaks with standards of scientific ethics, but also violates the principle of democratic school management included in the Brazilian Constitution.


Introduction
In this article 1 we analyze the development and the initial process of implementation of a new modality of public-private partnership in Brazilian education, the Social Impact Bond (SIB), translated by its promoters as "Contrato de Impacto Social (CIS)", which pilot project, sponsored by the Inter-American Development Bank (IDB), is in its initial stage of implementation in the State of São Paulo (Brazil).
In November 2017, the Department of Education in the State of São Paulo (SEE-SP) announced the adhesion to SIB and made available a Draft of Public Competition Notice for public consultation (SEE-SP, 2017a). According to the administration's announcement, the purpose is to implement the first experience of SIB in Brazil and the first one on secondary teaching in the world (SEE-SP, 2017b;SEE-SP, 2017c, p. 230), for what the "project bestows to the State of São Paulo the role of national and global leadership in this context. In addition to this, (…) it should contribute to broadening the understanding of the use of SIB in the Brazilian legal context" (SEE-SP, 2017c, p. 230).
The goal is the "hiring of services of execution and management of pedagogical actions aimed at increasing the approval rate and decreasing school dropout, without reducing learning, of public secondary teaching students" (SEE-SP, 2017a, p. 1). As a major differential, it is pointed out that the "hired organization will only receive remuneration if it achieves the impact goals which have been established" (SEE-SP, 2017b), and it is up to it-and not to the State-to provide the initial financial support.
To begin with, for what is considered the pilot project, 122 schools of the Metropolitan Region of São Paulo, located in regions considered vulnerable, will be selected. They will be divided into two groups: 61 schools will receive resources and interventions from the contractor, the socalled "treatment group"; and the other 61 will be the "control group", whose existence, according to the technical guidelines of the project, "is an essential condition to evaluate the effect of the interventions, since it enables to know what would have happened without their implantation in the schools of the treatment group" (SEE-SP, 2017c, p. 107).
A cohort of 12,000 students will be defined, from the schools receiving private intervention: students entering the first year of secondary teaching at the beginning of the project, who will have their school progress indicators monitored and compared to their peers in the "control group". The SIB's goal is to increase secondary teaching completion rate of this cohort by 7%, with no decrease in school performances in the standardized external evaluation applied by the state. In case of failure to achieve the goal, and therefore, the maximum payment, payments proportional to the improvement of completion indicator (in the proportion of 2%-minimum payment-to 7%maximum payment) will be allowed (SEE-SP, 2017a).
In order to increase the attractiveness of the proposal, intermediary payments have been inserted throughout the term of the contract, and in these cases an increase of 5% or more is expected in the progression rates of the cohort, between first and second grades in secondary teaching, and between second and third grades, respectively, during the first and the second year of implementation. Also, in these cases, proportional payments above 2% success are expected, in all the cases, as long as there is no decrease in school performance indicators (SEE-SP, 2017a).
If the goals are fully met, the hired organization or company, called the "master provider", will be remunerated up to R$ 17,774,400.00-up to half of this amount at the end of the three-year period of execution, and up to 25% in the two intermediary payments. Although this is the estimated value of execution of the actions, a calculation based on a simulation of administrative and service costs, the payments are not tied to the proving of the expenses by the provider, but to the achievement of the agreed goals (Pay by Results, denomination of SIBs in USA). In order to measure the results, an external evaluator will be hired (SEE-SP, 2017a).
The SIB model, however, does not require the master provider to have the full capital needed to implement the interventions, for the company is expected to raise it in the financial market, "since [SIB] allows interventions to be financed by private investors, who can be rewarded if the expected social impact is achieved" (SEE-SP, 2017a, p. 14).
From this pilot project, we expect the development of a social finance ecosystem in Brazil, consisting of the hiring State, master provider (which assumes legal responsibility before the State), private executing institutions, target public, evaluators, and think tanks. In the case studied, the model proposed by Insper Metricis is described as follows: the Government of the State of São Paulo hires (through SIB) a master provider (hired), responsible for the general goals of the project and for sub-hiring a group of specialized service providers (sub-hired), whose teams must carry out the activities set forth in SIB (…) and directed to the target population of the project (…). In order to finance these providers' activities, investors finance the master provider with funds to sub-hire and supervise specialized providers. These investors are considered "at risk", since they will only get a positive return on their investment if SIB goals are achieved. Throughout the intervention cycle, an independent evaluator should collect data provided by the SEE-SP and determine whether the goals set by the project have been achieved. (SEE-SP, 2017c, p. 235, bolds -SP, 2017c, pp. 224-225) Although the SEE-SP document mentions a new hiring model in abstract, it has been clear from the very beginning that the purpose would be to join SIB with a pilot project for the implementation of the model in Brazil. As soon as IDB gave the "kick-off", the players diversified. Efforts to set up this ecosystem were materialized in 2015, with the creation of the Social Finance Taskforce (FTFS), led by the Institute of Corporate Citizenship (ICE) and by SITAWI Finanças do Bem [SITAWI Good Finances] (an organization for positive financial solutions responsible for SIB in the State of Ceará, the second in implantation in Brazil, in healthcare field), and financed by IDB, FOMIN (Multilateral Investment Fund of the IDB Group), Banco Itaú, ICE, Fundação Telefonica, and Lew'Lara. Following the example of its British peer, the Brazilian Social Finance Taskforce has produced several reports that provide a kind of guideline for the implementation of SIBs in Brazil: survey of possible intermediaries and funding sources, compilation of international measurement experiences, and adaptations to hiring models provided by Brazilian administrative law. 2 The work of Insper Metricis, in this regard, was also aimed at mobilizing potential providers and building a support network. With this purpose, three strategies were adopted. In November 2016, the Seminar Innovation in Social Impact Bonds (SIB): World experience and agenda for Brazil was held (SEE-SP, 2016c). In addition, the work plan agreed by Insper, SG, and SEE-SP suggests the creation of two external support structures-an Advisory Board, consisting of specialists divided into three areas: education, finance/investment, and impact assessment; and a Group of Experts of more sporadic action, also subdivided by specialties: educational, legal, impact assessment, finance/investment, public management, and communications (SEE-SP, 2017c, pp. 30, 32).
In February 2017, two workshops were held for entities potentially interested in acting as master providers or service providers. These events were attended by foundations and business institutes that are active in Brazilian public education. 3 After the event, the organizations Fundação Educar DPaschoal, Fundação ltaú Social, Instituto Aliança, and MGov also participated in "individual meetings," when details were requested on their functioning structures and capacity to support the interventions provided by SIB (SEE-SP, 2017c). Ximenes et al. (2017) point out that there is a great coincidence between the period of divulgation of SIB (Nov. 2017) and the dissemination, in the section Market of the newspaper Folha de S. Paulo, of suggestions for educational technologies applicable to SIB, being tested or deployed in other states and sponsored by the same group of foundations and business institutes mobilized by Insper Metricis. Patton and De Witt (2015) point out some problems in developing countries that could explain the difficulties of implementing SIBs in the Global South, including: political instability, preference for short-term policies, lack of reliable data, scarce public funding, fragmentation of services, difficulties in prioritizing problems (which are often multiple and serious). Interestingly, these are the justifications for the choice of the State of São Paulo as the site of the first Brazilian experience of SIB: the state is rich and financially solvent, produces reliable data (it has 25 years of consistent data in education, for example), has been ruled by the same political party for 24 years, works with relatively centralized policies and is open to "innovation" in public management and partnerships with private agents.
The consolidation of this social finance ecosystem in Brazil has been recently investigated by Gonçalves (2017), who documented a series of perceptions of its players on the expectations regarding the return rates of social investments. Some respondents claim that, at the beginning of the ecosystem development, below-market return rates should be allowed as a kind of "affirmative action" to boost SIBs (2017, pp. 85-86). For others, at lower rates, investments would also be smaller, which is diagnosis of Frederic de Mariz (UBS) at the Insper Metricis Seminar on the usual "appetite" of Brazilian investors for return rates above inflation and basic interest rates.
In this sense, the context of political, economic, and social crisis that has portrayed Brazilian reality for the latest years has been fertile to the experimentation of renewed privatization modalities, especially in basic education. Nationally, Provisional Measure n o 746/2016, edited three weeks after President Dilma Rousseff's impeachment, and later transformed into a Bill (Brazil, 2016), radically reformed the curricular structure of secondary teaching, breaking up this stage into compulsory training itineraries and authorizing, in the case of the "technical and professional training" itinerary, the curricular completion through the establishment of partnerships for the hiring private courses, including distance education, by the State (Ferreira & Silva, 2017).
In the State of Goiás, a process of transferring the management of hundreds of schools to private Social Organizations (SOs) 4 , has been underway since the end of 2015. These SOs directly hire all the administrative staff and almost all the teachers of the schools involved, without a public tendering process (Goiás, 2016).
These reforms are intended to accelerate, within national and subnational education policies, State reform programs based on the concept of what is called New Public Management (NPM). This is understood both in its more general guideline of application of knowledge and instruments of business management in public administration, with a view to maximizing the economic efficiency of services and improving their effectiveness relative to measurable results, as in their common propositions: professional management, standardization of performance goals, control of results, decentralization of services, and competition in the public sector (Verger & Normand, 2015). Verger and Normand (2015), however, point out that, when the central characteristic features of NPM are identified, it presents itself in very different policy configurations; sometimes ambiguous. According to the authors, depending on the national context, on the organization and resistance capacity of workers, and on the hegemonic principles of justice in a given society, "when adopting NPM, some countries put more emphasis on the market and privatization, while others are more concerned with decentralization and accountability" (Verger & Normand, 2015, p. 615).
For this reason, the application of NPM guidelines does not escape ambiguity, as typically occurs in the discussion about democratic management and school autonomy, addressed in this article for the case of adhesion of state schools of São Paulo to SIB. Another aspect is that, under the umbrella of NPM, it is possible to observe different strands and reformist emphases in a same national context. SIB represents a step forward in the known context of privatization of education, by incorporating into this dynamic financial agents interested in capitalizing on "social impact" interventions in public education systems, and which, in a new modality of financial roulette, bet against the State, since the return rates of the project directly depend on the difference in outcomes between schools of the "treatment group" and those of the "control group." Unlike traditional philanthropy, which legitimizes itself in the alleged non-profit character of its actions, SIB opens up a market opportunity, a new "edubusiness" (Robertson & Verger, 2012).
In the next section we will fit this new modality of remuneration in the current context of managerial reforms in Brazilian education, marked by a renewed impetus to the dynamics and experiences of privatization in its different aspects. SIB also raises new questions for the design of educational policies, including ethical, legal and political-pedagogical issues related to the manipulation of vulnerable students, considered "school guinea pigs" ; to the commitment of non-intervention of the State in the schools of the control group, although the latter, by the very criteria of the project, are in equal condition of vulnerability and deficiency; to the potentially deleterious effects of incentives and productivism in ensuring the right to education; and to the obscure form of adhesion of the participating schools. These topics, treated as mere procedural contingencies by the SIB proponents of education in São Paulo, are carefully approached and analyzed in later sections of the text.

Managerial Reform, Renewed Impulse to Privatization, and "Hetero-ruled" School Autonomy in the State of São Paulo
In Brazil, the guidelines of management reform were officially adopted when President Fernando Henrique Cardoso created, in 1995, the Ministry of Administration and State Reform (Mare), which presented the Master Plan for the Reform of State Apparatus (Brazil, 1995). The Plan assumes that the State born in the New Republic produced serious distortions and inefficiencies, and that the Brazilian Constitution of 1998 led to an unprecedented bureaucratic throwback (…) entailed a surprising breach of the state apparatus by extending to the state services and to the state enterprises themselves practically the same rigid bureaucratic rules adopted in the strategic nucleus of the State. (Brazil, 1995, p. 21) It therefore makes a distinction between the strategic core of the State (Legislative, Judiciary, Presidency and Public Ministry) and three other sectors: exclusive activities (regulation, supervision, promotion, public safety and basic social security), non-exclusive services (education, health, research, culture, etc.) and production for the market (public companies). In these three sectors, it would be imperative to migrate to models of managerial administration, nevertheless diversifying the instruments adopted. In the case of non-exclusive services and production, reforming the State means transferring them to the private law regime, either the market, via privatizations and outsourcing, or the "non-state public sector," a legal figure introduced by the Plan. In the latter case, the so-called "publicization" would occur (Brazil, 1995). The recovery of this agenda in Goiás, in addition to the direct privatization and commodification measures promoted by the curricular reform of secondary teaching and SIB, opens a new stage of managerial reforms in basic education, strongly characterized by the dynamics of privatization in its different aspects: privatization stricto sensu, outsourcing, and "publicization," but with emphasis on the strengthening of profitable dynamics and of financialization (Adrião et al., 2016).
Therefore The purpose here is to overcome the ambiguity of social policies under the PT governments, which, while endorsed managerial reforms, promoted different modalities of "hidden privatization"-as suggested by Ball and Youdell (2007)-and encouraged financialization in the offer of higher education, also expanded investments and State participation in education (Marques, Ximenes & Ugino, 2016).
However, even in basic education, where public sector participation is the broad majority, recent expansion policies have opened up new business opportunities, either in direct incentives to partnerships to expand childcare (Adrião & Borghi, 2008), or in the dissemination of private education systems, which consist of low-quality textbooks and administrative and pedagogical management tools sold to municipalities by corporate groups with shares on the stock market (Adrião et al., 2016).
Under the argument of State failure in providing public services, significant portions of the pedagogical tasks originally assigned to schools are explicitly transferred to the private sector and commodified. In the reform of secondary teaching, for example, all the "technical and professional training" itinerary is potentially transferred to private initiative, through partnerships that include distance learning. In the case of SOs, in Goiás, the full administrative management of schools, and the task of hiring, managing and dismissing most teachers is transferred. In the SIB of São Paulo public education, the pedagogical relationship with families and students, as well as control of attendance and academic achievement, are partially transferred to the hired private institution. The explanation here is eminently economic: "actions which reduce dropout and failure allow the government to obtain greater impact with the resources employed in secondary teaching-in other words, it increases the value obtained from the money spent to finance the system (value for money)" (SEE-SP, 2017c, p. 6, bolds in the original).
At this point of managerial inflection, the NPM views on decentralization and school autonomy come into question, since, in the cases mentioned, privatization explicitly occurs on administrative and pedagogical management, tasks traditionally assigned to public schools.
In the propositions of managerial reform in Brazilian education, the offer of a low-quality service would result from the gigantism and the structural inefficiency of the State, wont to horizontalizing policies, which encourage an amplified notion of democracy that is detrimental to the education systems, by creating the illusion that everyone can decide on everything. The decentralization of education systems, in this sense, is disseminated as a mechanism for transferring responsibilities from the higher spheres of the State to those closer to the population, generating greater and better follow-up and participation of the population in the different localities (Gentilli, 1996).
However, this decentralization with an air of democratization displaces the decision-making centralization of the education systems to the extreme ends of the process, which are curriculum and evaluation. The language of large-scale evaluations resignifies the school management: improvements in indicators would result from successful initiatives in pedagogical management (management for results). School management focusing on "innovation". 5 The "innovative" and presumably self-managed school actually sees the breadth and depth of its actions limited by the curriculum and evaluative policies that would give to it the "freedom to create". The insistence on fragmented and measurable goals, which are the measure of monetization of "social impacts", leads school to a kind of management whose rationality is based on the attainment of quantitative goals, reinforcing technocratic instruments and the notion of educational quality as an achievement of numerical targets, presumably consensual.
We are faced with what Lima (2014) calls "hetero-ruled autonomy", which decision-making process can be considered a simulacrum, given the degree of centralization of power, which "is continually stifled and suffocated by intrusive micromanagement and remote control processes, namely electronic, or by market-based regulatory modalities, in the competition between suppliers and in the theories of public choice" (Lima, 2014(Lima, , p. 1074). This, however, does not mean completely disregarding the notions of democracy or participation, but rather adapting them to the necessary mechanisms in order to ensure efficacy and competitiveness in education systems.
This has been the focus in the State of São Paulo since 1995, when, concurrently with the Master Plan for the Reform of State Apparatus, the managerialist reform in public education system begins (Goulart, Rezende Pinto & Camargo, 2017). Back then, it was believed that the quality of education could only be achieved by reorganizing the administrative structure of SEE-SP and seeking new partnerships for the provision of educational services with "entrepreneurs, teachers, parents, associations, universities, etc.-among which the municipalitites will become privileged partners" (SEE-SP, 1995).
The "separation between the administrative, financial, and pedagogical bodies, in order to empty the deliberative and advisory character of the School Council" (Arelaro, Jacomini & Carneiro, 2016, p. 1151 deepened. On one hand, because the model for strengthening management centered in the figure of the school principal and its management team, with the permanent detaching of students, teachers, and the community, from the conception, planning, follow-up and evaluation of the school political-pedagogical project was replaced by the pre-established assessments at the highest levels of the educational bureaucracy in conjunction with business foundations (Freitas, 2014), producing centralization of pedagogical goals and annulment of school subjects' participation in local decision-making bodies.
On the other hand, because, from a legal point of view, there was no impetus to participatory decision-making processes and decision-making bodies that could actually modify the dynamics of public education. Complementary Bill n o 444/1985, which regulates School Councils in the State of São Paulo, includes, among the attributions of the School Council, deliberating on "(b) alternatives for solving problems of administrative and pedagogical nature; (c) projects of psychopedagogical and material assistance to the student". As long as these Councils are exclusively called to operate standardized curricula and methodologies aimed at meeting the goals of external evaluations, their existence turns out to be merely formal.
Still, even if managerial discourse subordinates democracy to a sparsely controlled participation, legal instruments act to open the decision-making process in schools. "The [constitutional] principle of democratic school management qualifies this question by imposing a procedural delimitation, which aims at allowing the participation of teachers, parents, and students in defining the political-pedagogical project of schools and the participation in the management of their implementation" (Ximenes, 2014, p. 197). In other words, even under a controlled dynamic, the democratic legitimization of certain managerial policies is required. As discussed in section below, we see that the SIB's "hyperbureaucratic" (Lima, 2009) and verticalized dynamics opens a gap between democracy and efficacy, with severe damage for the former.
This, however, does not occur without questioning and some resistance, in view of the contradictions of the educational system itself, which are aggravated in times of fiscal crisis in which the "need for economic and ideological efficiency and stable production tends to be in conflict with the other political needs" (Apple, 1989, p. 72, author's italics). This is what we saw in 2015 and 2016 during the "secondary spring", when hundreds of schools were occupied by their students in São Paulo, Goiás, and other states, with a common agenda of denouncing the aggravation of the precariousness of public schools due to the austerity policies initiated under Dilma Rousseff's second term in government (2015)(2016), radicalized under Michel Temer's government and disseminated in practically all Brazilian states (Campos, Medeiros & Ribeiro, 2016).
Although at that moment the students succeeded in having their demands met, this did not mean an effective restraint of the managerial purpose of downsizing the school system. The closedown of classes did not cease in the following years (Cássio et al., 2016) 6 and new reforms were initiated. Also in 2016, the SIB of São Paulo education (SEE-SP, 2017c, pp. 224-225) started to be developed, although it only reached public awareness at the end of 2017, when its design was already almost complete and operational.

SIB in the Schools of São Paulo: Social Experiments and Scientific Ethics
The "risk transfer" engendered by SIBs is based on the fact that the State's payouts are conditioned to the achievement of goals, which may be of partial results, achieved throughout the term of the contract (outputs), and long-term (outcomes), the social impacts themselves. Investors need to make sure that the impacts can be adequately measured, generating payments. The State, on the other hand, needs to make sure that the desired results will be effectively achieved. Therefore, it is necessary to develop methodologies for the measurement of these results, with a view to their escalating monetization (Pay by Results).
The metrics of SIBs are ruled by an eminently "scientific" logic. This text is not meant to plunge into the epistemological debate of "evidence-based" public policies (Stoesz, 2014), but it is easy to find out, from the many white papers about SIBs in the world, the use of experimental (randomized controlled trials, matching techniques) and quasi-experimental (observational studies) methodologies for the production of "evidences" of causality between the carried out interventions and the social impacts generated by SIBs. 7 In face of the complexity of the social phenomena with which SIBs deal, the "observed" causality is obviously inferred, and the quality of this inference depends on a satisfactory isolation of the variables studied.
The Draft Notice on SIB points out "that the division of schools into 'treatment' and 'control' groups will enable the measurement of the real impact of the interventions in the schools that received it, making possible to condition payment to their results" (SEE-SP, 2017a, pp. 16-17, our italics). The methodology presented includes sample matching (using pre-treatment variables provided by SEE-SP) and random choice of treatment and control schools, as well as balancing variables which consider the effects of other SEE-SP programs carried out in schools, and that could mask the "real" impacts of interventions. The methodology of selection and measurement aims, at one time, to control potential differences between the participating schools and to ensure statistically robust results, enabling to monetize the social impacts achieved and to calculate what would be the payments for partial fulfillment of goals.
The debates on experimental methods in social sciences, in the field of public policies, and also in education, are vast. From a methodological point of view, the idea of "variable control" in social experiments is very problematic (complex, context-dependent phenomena). 8 From the ethical point of view, the very idea of "social experiment" is problematic (Goldstein, 2013), giving way, in practice, to quasi-experimental approaches based on the gathering and interpretation of public data. Matching techniques can be used to construct comparable samples, mimicking, as far as possible, some experimental control conditions (Travitzki & Cássio, 2017, p. 162).
In the SIB in São Paulo, pairing will be used to build sample groups prior to the gathering of data, which qualifies the project as an effective "social experiment." Isabel Opice, an economist who worked in SG until May 2017, confirms the project's purposes by stating that the "main innovation of SIB is to create space for experimentation within the public sector, running on the opposite way of the model of implementing large scale standardized public policies," and that "Alligned with the experimental character of SIB, this first experience in Brazil tries a new contracting mechanism by results" (Opice, 2017, our italics).
Among SIBs enthusiasts, the debate about the ethical implications of adopting control groups is typically underestimated. 9 A white paper published by the Organisation for Economic Cooperation and Development (OECD), for example, states that the control group methodology is a "more challenging" measurement process (Galitopoulou & Noya, 2016, p. 17). Along the same line, a UK investment fund pontificates that randomized controlled trials (RCTs) are generally considered as the "gold standard," although not always feasible, for reasons of cost and "practicality" (Bridges Ventures LLP & Bank of America, 2014, p. 17).
OECD also welcomes the use of control groups, but points out their disadvantages: (1) technical complexity; (2) identification of an appropriate control group; and (3) difficulty in knowing whether the control group is receiving other services (OECD, 2015, p. 11). The Draft Notice on SIB suggests that interventions from other SEE-SP programs already underway (such as MMR) will be taken into account to "isolate" the causal effect of SIB interventions in treatment schools (SEE-SP, 2017a, p. 59). There is, however, no guarantee that the proposed balancing is capable of solving the problem of the "statistical strength" of the method. 10  This excerpt is not part of the publicly disclosed Draft Notice on SIB (SEE-SP, 2017a), but it is interesting to note that the private agent is not embarrassed to suggest that the State signs a contract and commits to not implement any relevant programs in the schools managed by it, something that runs against its constitutional obligation to ensure the social right to education with a quality standard. 12 In fact, the strictest way of isolating the effects of interventions in the treatment group would be the cancellation of any other State program in control schools. Nonetheless, this would probably be the least ethical way of conceiving the design of public education policies. Although they advocate the SIBs' model, Gustafsson-Wright et al. recognize that, ultimately, "the extent to which evidence must be rigorous is very dependent on the risk appetite of the investors and the requirements of outcome funders" (2015, p. 5). In other words, the more conservative investors are about financial risk, the more guarantees of accuracy in measurement (more scientificity) will be required. Ron Davies also defends SIBs, but acknowledges that the methodology of control groups may entail accusations of discrimination from those who belong to these groups, and who, of course, will not benefit from the interventions (Davies, 2014, p. 6).
Indeed, recent surveys on SIBs around the world have shown that social impact assessments using purely experimental, even quasi-experimental, methodologies have become less significant than the analyzes of validated administrative data, the use of historical series, interviews, and questionnaires (Gustafsson-Wright et al., 2015;Mudaliar et al., 2017). Although it can be said euphemistically that this is for "practical" reasons, the use of control groups in SIBs implies important ethical issues.
In this sense, the case of SIB which started in 2012 in Rikers Island prison complex (New York City) is quite illustrative. It was a program focused on inmate teenagers, which aimed at reducing their recidivism in criminal activities. The wardens were not able to separate the teenagers assigned to participate in the cognitive-behavioral therapy program that was part of the intervention (treatment group) from those who should not participate in the program (control group). Sampaio and Munday consider that SIB in Rikers Island was not a complete disaster as New York City "tried a new scale intervention with the benefit of a rigorous assessment, without spending any taxpayer money" (2016, p. 34). Like those authors, in the white papers of social organizations and think tanks which have specialized in SIBs around the world, the success or failure of contracts refer solely to the saving of public resources and the return to investors. There is no moral dilemma in the act of, in name of the statistical rigor of the metric, depriving imprisoned youngsters from having psychological treatment or participating in educational programs. Julia Morley reports that an article published in The Economist in 2011 about the inaugural SIB, implemented in Peterborough Prison (UK), was illustrated with a photograph of an inmate in a cell with the following subtitle, "Fancy a flutter?", in reference to the ethical problems associated with gambling in the results of experiments with people. She points out that this kind of financial roulette disrespects the individual, which would be considered unethical by many who take a deontological approach (Morley, 2017, p. 18). She also states that "The existence of incentives for gaming activities renders the SIB ethically questionable, particularly because the subjects are not in a position to opt out of the intervention and select an alternative that is not profit-based" (Morley, 2017, p. 20). James Ronicle underlines Morley's arguments that SIBs result in the view of people as commodities, questioning the concept of electing people to invest in and people to be left behind (Ronicle, 2016).
Michael J. Roy et al. argue that the proliferation of SIBs has occurred without sufficiently considering the implications of these instruments or the potential ethical problems emerging from them. They raise a more general ethical problem caused by SIBs, stating that they imply that the rehabilitation of offenders is no longer prompted by moral motivations but incentivised by the prospect of profit, a profound shift which changes its normative nature. One implication of such a way of thinking is that there would be no principled objection from a SIB perspective to meet a target to reduce re-offending by simply paying people not to commit offences. (Roy, McHugh & Sinclair, 2017, p. 10, authors' italics) The metrics of results by frequency correspond fairly well to the professional ethos of social service and medical areas. Dealing with the most difficult patients (or students) is considered a sign of ability in many professions (Maier & Meyer, 2017). Binary and goal-oriented metrics, on the other hand, may encourage the deliberate choice of simpler or cheaper cases to solve. 13 Morley (2017) points out that the provision of services associated with SIB, an activity that generates profit, can be considered inconsistent with the professional identity of social service. We could go further: with the professional identity of teachers, school managers, health workers, doctors, etc.
If the implantation of SIBs in Brazil has repercussions on the vacuities of Brazilian administrative law, it also impacts the highly regulated field of scientific ethics. In view of its methods of sampling, measurement, evidence production, scalability, and reproducibility, this kind of intervention should be subjected to the same scrutiny of scientific research involving human beings. The fact that SIBs in São Paulo and Ceará are the first in Brazil reinforces its experimental character.
Marta Garcia, who leads Social Finance UK's work in Latin America, supports the need for extreme flexibility in SIBs so that the services offered by the provider can be modified until the desired result is reached (in other words, a trial-and-error process; Innovation in Social Impact Bonds Seminar, 2016). The State, in Garcia's view, should give investors (and their service providers) maximum autonomy to decide which interventions should be implemented, and which, if necessary, should be aborted. If the contracted sampling and measurement methods and practices by themselves give the SIBs a character of social experimentation, the abdication of State responsibilities for overseeing service provision-though in the specific field covered by the contract-gives investors the power to "change experimental conditions" at any time. Thinking of public schools in the State of São Paulo, this means implementing and discontinuing actions aimed solely at achieving the desired results, without considering the expectations of students, families, and education professionals involved.
The SIB contractual structure, which does not regulate the relationship between providers and investors 14 , allows investors to demand changes in the strategies of the providers working in 13 This is part of a warning list from several white papers about SIBs, which describe the mechanisms for selecting the simplest, and probably the cheapest cases (creaming, cream-skimming, cherry-picking), or deliberate exclusion of the most difficult cases (parking). See, for instance: Davies (2014, p. 6); Galitopoulou and Noya (2016, pp. 16-17); Gustafsson-Wright et al. (2015, p. 40);Liebman and Sellman (2013, p. 19); OECD (2015, pp. 10; 13); Sampaio and Munday (2016, p. 35); So and Jagelewski (2013, p. 9);Social Finance UK (2013, p. 13;2016, pp. 59-60). From a critical perspective on SIBs, see: Maier and Meyer (2017, p. 5); Morley (2017, p. 19); Roy et al. (2017, p. 9);Warner (2013). 14 This is a particularly problematic issue in SIBs. In the documents of Insper Metricis attached to Administrative Process SE n o 1967/2017, for example, there are several mentions to "bonus" mechanisms to providers which have been deleted from the Draft Notice (which regulates solely the public-private schools, given the possibility of financial losses. In the United Kingdom, where there is a consolidated social finance ecosystem, government recognizes the need for investors to control the delivery of SIB because of the risk they take, and suggests that this can be done in different ways, such as taking a seat on the service provider's board or replacing the providers that do not achieve the goals. (Morley, 2017, p. 23) The unending asymmetry of powers in SIBs is also linked to the vulnerability of subjects who undergo the interventions, in a condition of subordination and weakened in their capacity to question the agents of recognized "competence" (Chauí, 2014), authorized by a supposed technical knowledge. Contracted and contractor do not acknowledge the agency of these subjects, considering them vulnerable because of their assumed inability to handle their own difficulties 15 , people who should be protected by groups which have greater moral and technical legitimacy (Chauí, 2014;Morley, 2017, p. 18). It is not by chance that the first SIBs were carried out in one of the most extreme contexts of human vulnerability, one in which bodies are under the custody of the State: prisons.
All the problematic points raised so far for SIB in São Paulo-the right to choose to participate in the project (or to quit at any moment), the vulnerability of the subjects of the interventions 16 , the lack of prior information on the existence of treatment and control groups, the extreme flexibility of interventions (which frustrates expectations and places school subjects at the service of trial-and-error)-should be observed in protocols of research involving human beings in Brazil, regardless if in the areas of health or of human and social sciences.
Research projects involving human subjects must be approved by institutional Research Ethics Committees (Comitês de Ética em Pesquisa, CEP) accredited by the National Commission for Research Ethics (CONEP), linked to the National Health Council (CNS). Resolution CNS/CONEP n o 466/2012 determines that the ethicity in research involving human beings implies: a) respect to the research participants in their dignity and autonomy, recognizing their vulnerability, reassuring their willingness to contribute and to remain or not in the research, through purported, free, and informed manifestation; b) balance between risks and benefits, both known and potential, individual or collective, committing to maximum benefits and minimum harms and risks; The Resolution also states that research involving human beings, in any field of knowledge, must meet certain requirements, among which: relationship). Therefore, during the term of a future bond, there is no guarantee that, within the scope of the private relationship between investors and providers, such reward mechanisms will not be implemented without any possibility of control by the State, much less by schools. 15 In the first work plan submitted by Insper Metricis to SEE-SP, it is stated that the target population of the study must meet three main criteria: (a) It should have concrete shortcomings leading to poor performance in the long term due to individual circumstances and the possibility of improving the quality of services. Usually, but not always, this situation correlates with the long-term costs this low performance imposes on society; (b) To have needs that can be addressed through existing interventions; and (c) To be clearly identifiable and accessible at a low cost (SEE-SP, 2017c, p. 25). As we will see in the next section, the principals' arguments to persuade School Councils to join SIB are based on a presumable inability of schools to reverse the misfortunes pointed out by indicators of school flow and learning. 16 Add to this the fact that the SIB in São Paulo focuses on secondary school students, who, typically, are not yet legally responsible for their consent to participate in a social experiment. f) if there is a need for random distribution of research participants in experimental and control groups, ensure that, a priori, it is not possible to establish the advantages of one procedure over another, through literature, observational methods or methods that do not involve human beings; g) obtain free and informed consent from the research participants and/or their legal representatives, including cases in which the research, by its nature, justifiably entails a posteriori consent; j) be developed preferably in individuals with full autonomy. Vulnerable individuals or groups should not participate in the research when the desired information can be obtained through participants with full autonomy, unless this investigation can bring benefits to vulnerable individuals or groups; Resolution CNS/CONEP n o 510/2016, which is specific to research in the areas of human and social sciences involving human beings, states that: Art. 12. There should be a justification for the choice of children, teenagers, and people with diminished decision-making capacity in the protocol to be approved by the CEP/CONEP system. Single paragraph. In the cases provided for in the caput, the participant's free and informed consent should be obtained through his/her legal representatives, preserving the participant's right to information and autonomy, according to his/her capacity.
None of the SEE-SP documents shows any concern that interventions in schools, despite their undeniable character of social experimentation, are actually circumscribed within the framework of ethics in research with human subjects. With no legal framework regarding hiring models, the secondary education SIB in São Paulo will have even fewer normative contours in the field of scientific ethics. Let's see next how the selection of school units participating in the project, in addition to bearing serious ethical problems, also violates the constitutional principle of democratic school management.

SIB in the Schools of São Paulo: Violations to the Principle of Democratic Management
In an official letter dated October 26, 2017, the CGEB (Coordination Office of Basic Education Management-an internal body of the SEE-SP), defined the target regions of SIB, which include 11 region Schools' Administration Departments (SAD) in the Metropolitan Region of São Paulo: Carapicuíba, Itapevi, Taboão da Serra, Itaquaquecetuba, Suzano, Osasco, Santo André, Norte 1, Centro-Sul, Sul 2 and Sul 3 (the latest four located in the city of São Paulo). After videoconferences with teaching supervisors and school principals from these regions (October 4 and 16, 2017), a total of 180 school units expressed interest in participating in the project, 122 of which will be in the treatment and control groups (SEE-SP, 2017c, pp. 354-355).
Until then, however, no public document about the SIB had been made available. Thus, the "schools" which were willing to participate did so in the strict figure of their principal. The broad dissemination of SIB was made only on November 23, 2017, through a note on the SEE-SP website (2017b) 17 , which published the Draft of Public Competition Notice on SIB (SEE-SP, 2017a) and opened online public consultation, wherewith the interested parties could e-mail "requesting elucidation, as well as submitting criticisms, suggestions, opinions or ideas to change the text." It is not possible to know how (and if) SEE-SP provided feedback to the consultation, which neither included meetings nor public hearings with other sectors of society interested in knowing and discussing the proposition.
In face of the late disclosure of information-because the process had already been in progress for two years-and the fact that the Administrative Process SE n o 1967/2017 presented terms of acceptance signed, most of them before November 20, 2017, it is convenient to investigate how the decision-making process to join SIB was made in schools, and what kind of explanation about the project was actually offered to the communities. And this for the simple reason that this new modality of hiring-as the extension of the present text implies up to this point-is not the kind of subject that can be elucidated in a brief discussion.
During field research, we had access to the minutes of six School Councils in SADs of Santo André and Sul 2. 18 In all the schools investigated, the Councils' extraordinary meetings 19 were held before the Draft Notice by SEE-SP (Nov. 23, 2017).
The analysis of the minutes indicates that the schools' procedure to join SIB was purely formal, without providing a set of qualified and timely information for a satisfactory discussion of the project. The Administrative Process SE n o 1967/2017 provides clear evidence in this sense: the dates of signature of the schools' terms of acceptance within the same SAD are always the same (SEE-SP, 2017c, pp. 417-599). For example, all the terms of acceptance of SAD of Itaquaquecetuba (20 adhesions) were signed on November 17, 2017, while those of SAD of Santo André (13 adhesions) were signed on the 14th, and so on. The internal discussion processes about SIB in schools-already hampered by the lack of information about the project-were still limited by the deadlines determined by the SADs, in turn, driven by the SEE-SP deadlines.
The minutes analyzed suggest other distortions in the decision-making process, such as the absence of information on which school segments were present at the Council' meetings, the scanty presence of counselors, and the absence of large discussion among those present (at least in what is reported in most minutes). The emptying of the sessions, in contrast to the sensitivity of the agenda, indicates the absence of an adequate call for School Councils by the principals.
In the case of SAD Sul 2, the delivery of SIB's term of acceptance was requested to schools by e-mail. In a message sent to the principal of José Porphyrio da Paz State School (Oct. 16,2017), the school was informed that it had been selected for its high rates of students' non-attendance in the last five years, thanking for the school's interest in participating in the project, and reinforcing that it is "essential" to sign the term of adhesion, as well as holding a extraordinary meeting of the School Council within two days (Oct 18, 2017), being up to the management staff to "clarify" the reasons for which the school was selected to participate in the project.
The same can be observed in an e-mail dated October 14, 2017, sent by SAD Sul 2 to Amelia Kerr State School, forwarding the term of acceptance, "which must be completed and signed by the School Principal, scanned and forwarded until (…) 11/16/2017", and emphasizing the need to meet the deadline, due to the divulgation by SEE-SP, on November 17, 2017, of public consultation with "the purpose of ensuring transparency in all stages of SIB".
Several minutes of School Councils that decided to join SIB emphasize the insistence of principals on arguments based on large-scale assessments and school flow indicators: "our school presents flow issues" (José Porphyrio da Paz State School) "the school has high rates of nonattendance in Primary and secondary teaching (Prof. Herculano de Freitas State School), "our dropout rate in 2016 was very high-9.49%" (Edevaldo Perassi State School).
In Edevaldo Perassi State School, where the School Council approved the adhesion to SIB with only one contrary vote, it is clear that the decision was made without the minimum piece of information about the project, and the councilors were persuaded with arguments of authority. The minute shows how the dissent was gradually persuaded by the school's principal throughout the meeting: Only a part of the development of this project was made clear to the [Council] members, as we did not have more concise information that could further clarify such action. (…) In the discussion about the adhesion offer, there was a manifestation of refusal (…), since we did not have much information about this project. Even the intentions of the State Government about this implantation were questioned (…). Faced with this controversy, it was stated that performance in our secondary teaching is unsatisfactory. (…) In spite of that, we believe that, by joining the project, we will be able to conclude secondary teaching activities with better quality in the future. (Edevaldo Perassi State School) None of the minutes mentions that adhesion to SIB could represent the entry of the school into the control group, since the principals' defenses of the project were directed to the beneficial actions of SIB, mitigating flow and learning problems (always in the logic of large-scale evaluations). It should be noted, however, that all the attributions of SEE-SP regarding schools during the term of the contract relate to the schools in the treatment group (SEE-SP, 2017a, pp. 83-84).
It is understandable that the dissemination of the information that, due to sample procedures, 61 schools would not suffer any intervention in the "SIB Project", could jeopardize the success of the project due to a potential breach of expectations at the time of adhesion. It is at this point that the violations of the principle of democratic school management 20 -information misuse, manipulation of school decision-making bodies, and tightness of deadlines-converge with the violations of the ethicity in "social experiments" discussed in the previous section.
The importance of the availability of qualified information, the time for maturing ideas, and the possibility of schools' free decision is confirmed by the fact that 25 schools of SAD Sul 2 recalled their School Councils and reversed the initial decision of joining SIB after reading the Draft Notice (SEE-SP, 2017a), and by a period of discussions among teachers representing Sul/Santo Amaro sub-headquarters of Teachers' Union of Official Teaching in the State of São Paulo (Apeoesp). The result of this movement of clarification was a meeting of teachers from the schools, which called off the adhesion, representatives of the union, and the leader of SAD Sul 2 for the delivery of the minutes, in which they demanded from the SAD the respect for the sovereign decision of the School Councils. The terms of acceptance of SAD Sul 2 schools, signed between November 14 and 17, 2017, are still included in the Administrative Process SE n o 1967/2017, but they eventually called off the adhesion to SIB in the following month. The justifications for this calloff include: the short time for discussion with the community, the transfer of public money to private entities, the lack of physical space for the project in the period after classes, the participation of the school in other projects with similar objectives.
Some schools did not join SIB from the beginning, such as 16 de Julho State School (SAD Santo André), whose reasons for refuse coincide with the justifications for reversion we observed in SAD Sul 2: there was no presentation (…) details of the agreement that would be signed between the school and the company that would provide the services; the degree of interference the company would have in the school routine was not clear; besides, the establishment of such a contract would be an intrusion of the private sector; neither were clear the functions that the group of employees of this school should perform during the term of the contract, and, finally, the urgency for definition of adhesion or not to the project prevents a deeper and wider analysis and discussion, more accessible to the whole community. SIB introduces new groups of interest into a previously established relationship between the direct beneficiaries of the public service and the State, which is responsible for providing services (in the case of Brazilian education, by ensuring a social right). The relation of these groups to School Councils-school's decision-making instance-is irrelevant in the documents of SIB in São Paulo (SEE-SP, 2017a;SEE-SP, 2017c). No interference is expected from the School Councils in the actions of the hired, and there is no possibility of suspension or revocation of the contract if the school subjects understand that there is risk or diversion, or if they express disagreement with the implemented actions.
The contracted will report only to the school principal, as indicated in the Draft Notice on SIB PAULO , item 6.3.6-"The actions proposed by SIB must start from a joint dialogue with the management in a complementary way to the actions foreseen in the political-pedagogical project (SEE-SP, 2017a, p. 20), subordinating the school's political-pedagogical project to the actions of SIB. The summary table of SIB's activities (SEE-SP, 2017a, pp. 21-22) include textually, among the actions planned, "To review and continuously improve the pedagogical political project in the light of the activities foreseen in SIB", and "Include the activities (…) in the political-pedagogical project." From what has been exposed so far, it is expected that the participation of School Councils in the decisions on the alterations of the political-pedagogical projects induced by SIB will be as protocolary as they were when joining the project.

Concluding Remarks
The process of centralizing decisions within the educational policies of the State of São Paulo has been presented for at least two decades (Adrião, 2006), balancing the choice of the state as the site of the first Brazilian SIB experience. In the São Paulo government narrative, SIB is the epitome of innovation in public management, which lacks flexibility and adaptability to the "customers" needs, in contrast to an alleged breach of traditional universalist policies.
The implementation of SIB in São Paulo collaborates to inaugurate a new stage of managerial reforms in Brazilian public basic education, in which monetization of educational evaluation results takes place, and foundations and institutions linked to banks can fulfill new roles as service providers in an ecosystem of social finance. The "good finances" of social investments give a new moral stature to profitable activities in public education.
The SIB design in São Paulo education tries to follow, as much as possible, the most rigorous standards of evidence production, using treatment and control groups, sample matching, and balancing the effects of other programs that already exist in the education system. The analysis of the public documents on the project (SEE-SP, 2017a; SEE-SP, 2017c) reveals a great concern of the proponents with the methods' statistical robustness, so that the first experience of SIB in Brazil is not only successful from the financial point of view, but scalable and reproducible in other places and in other contexts.
The payment for results, which characterizes SIB, implies ethical and legal problems. Particularly in education, it also implies political-pedagogical problems related to the schools' decision-making autonomy. The analysis presented shows that the methodological design of SIB in São Paulo ignores the ethical implications of adopting control groups in social experiments, without any information in this regard being known by the schools, which joined the project. The ethical problem of offering differentiated services to equally vulnerable populations is aggravated by the fact that the state contracts these differences, breaking with the universalist standard which establishes education as a social right in the Brazilian Constitution. In addition, the analysis reveals a series of distortions in the decision-making processes of schools' adhesion to SIB, suggesting what will happen when they are effectively implemented (in treatment schools). Not unambiguously, the maintenance of the play of forces within State institutions also ensures that the population that uses public service can, through organization and representation, know, monitor, follow, suggest, and participate in the construction and implementation of policies, even if in a limited way, in terms of an hetero-ruled autonomy (Lima, 2014).
The SIB in São Paulo schools, a social experiment in the broad sense, is based on the canons of the NPM described by Verger and Normand (2015), and is offered by SEE-SP as "a solution to improve educational results of secondary teaching students and innovate in the hiring of services" (SEE-SP, 2017c, p. 9). Here we have looked at a general description of its institutional design and a more careful analysis of its ethical and political-pedagogical implications. In future works on this new form of privatization, an analysis of its legal framework in Brazilian administrative law deserves attention. A more detailed characterization of this new Brazilian social finance ecosystem, especially its evolution after the implementation of the first SIB experiences, as well as investigations into the effects of the "emphasis on results" in daily school life, "impacts" that are difficult to measure by the instruments of SIBs, is needed.

SPECIAL ISSUE Management and Management Policies in Lusophone Countries
education policy analysis archives Volume 26 Number 130 October 15, 2018ISSN 1068-2341 Readers are free to copy, display, and distribute this article, as long as the work is attributed to the author(s) and Education Policy Analysis Archives, it is distributed for noncommercial purposes only, and no alteration or transformation is made in the work. More details of this Creative Commons license are available at http://creativecommons.org/licenses/by-nc-sa/3.0/. All other uses must be approved by the author (s)