Brazilian think tanks and the rise of austerity discourse

Abstract This article discusses the role of think tanks in the production of ideas guiding recent change in Brazil’s economic policy. It claims that think tanks are significant policy-making agents preparing the society for change – via their communicative discourse – but also attempting to influence the interaction between political elites – via their coordinative discourse. The polarization of think tanks’ communicative discourse in regard to austerity during two critical junctures for Brazil is analysed. Discursive institutionalism is applied in order to interpret data from four Brazilian think tanks: the Institute of Applied Economic Research, the Inter-Union Department of Statistics and Socioeconomic Studies, the Fernand Braudel Institute and the Brazilian Institute of Economy. These think tanks have very different organizational and ideological characteristics but a polarization of the discussion around austerity can be observed in the discourse of all four of them. The scale ranges from an active defence of the Brazilian development model to a full-scale endorsement of austerity.


Introduction
Over the past 20 years, Brazil has made important advances in the political, economic and social realms. It has a consolidated political democracy and until recently its economy was growing, inflation was low and poverty was gradually being reduced (Weisbrot, Johnston, & Lefebvre, 2014). The global financial crisis initially served as an opportunity for the Brazilian economy to project stability and maturity, but as of 2014 the economy started contracting and a discussion of a need to change its development model and to introduce austerity measures has been enhanced (Barbosa Filho, 2017). Nevertheless, the country has increased its presence on the international stage, and is widely considered an important emerging power (Hurrel, 2013). Brazil's so-called 'rise' has been observed closely by analysts, media and policy-makers since 2001, when Goldman Sachs included the country in the BRIC (Brazil, Russia, India and China) group of rising economies which would dominate the world in the future. Initially treated with scepticism, the emerging power label endured for over a decade, becoming a determining factor in the identity of twenty-first century Brazil (Stuenkel, 2015).
The starting point of our argument is that two critical junctures for Brazil exist: first, being labelled as a BRIC and a rising global power (2008)(2009), and second, the economic recession eruption (2014)(2015). Critical junctures refer to particular historical moments that have lasting consequences and can be either 'big' events or less significant incidents that happen at the right time and have an enduring impact (Pierson, 2000). Consequently, critical junctures are likely to mark significant transformations in the modes and content of policy-making in the regions where they take place: they are the moments when public policy shifts can more easily occur. Historical institutionalists (HI) and -to a lesser extentdiscursive institutionalists (DI) have thus engaged with the analysis of critical junctures. In this article, it is shown that during critical junctures think tanks can increase their visibility as carriers of new coordinative and communicative discourse (Ladi, 2011). Additionally, a polarization of think tanks' communicative discourse in favour or against austerity can be observed during the second critical juncture.
This article aims to explore the role of think tanks in the strengthening of austerity discourse and policies. Four think tanks with very different organizational and ideological characteristics are discussed: the Institute of Applied Economic Research (Ipea), the Inter-Union Department of Statistics and Socioeconomic Studies (Dieese), the Fernand Braudel Institute (IFB) and the Brazilian Institute of Economy (IBRE). More specifically, following Schmidt's (2008) propositions on discursive institutionalism, these think tanks' participation in the generation of coordinative and communicative discourse in support of or against austerity is analysed. The first part of the paper introduces the theoretical argument and justification for the research design, while the second focuses on the coordinative and communicative discourses of the four think tanks and contextualizes it within the current global and Brazilian environment.

Think tanks and critical junctures: a discursive institutionalist approach
The analysis of critical junctures has been developed in the historical institutionalism strand (e.g. Pierson, 2000). This article claims this concept can also be useful for discursive institutionalism, because if combined with the latter's focus on agency, critical junctures become much more than unexplainable moments when change can be triggered (see Schmidt, 2008). Critical junctures are important because they are the moments in time when changes such as economic policy shifts can more easily occur. Focusing on the way discourse changes during these moments in time can shed light on parameters of the critical juncture (e.g. persistence or not of discourse on austerity) which could have been otherwise omitted. At the same time it is methodologically significant for discursive institutionalism to analyse discourse during critical junctures since these are the signposts where the most interesting discursive shifts can be observed. It has been shown that it is during these critical junctures that think tanks increase their visibility as carriers of new discourse (Ladi, 2011).
Following Schmidt (2008, p. 305), discourse has ideas about policies, programmes and philosophies as its substance but is not confined by them. Discourse as developed in discursive institutionalism describes the substantive content of ideas but also the interactive process by which ideas are spread. Discourse is not just about ideas or 'text' but also about the context in which ideas are developed and promoted. Schmidt (2008) distinguishes between coordinative and communicative discourse. Coordinative discourse refers to the 'creation, elaboration, and justification of policy and programmatic ideas' by policy actors and the way they exchange views and persuade each other (Schmidt, 2002, p. 171). It is concerned with the formation of a common language by policy groups in their attempt to construct a coherent policy programme. Communicative discourse, on the other hand, is concerned with the relationship between policy-makers and the public. It refers to 'the presentation, deliberation, and legitimation of political ideas to the general public' (Schmidt, 2008, p. 310).
According to Schmidt and Radaelli (2004, p. 193), the study of discourse should coexist with the awareness that interests also matter, as well as the material conditions and hard economic variables. Discourse in this sense is the glue between structure and agency. This means that structural conditions (e.g. institutional framework, economic crises) influence the actions of agents such as think tanks. These actions are expressed and can be pinpointed by discourse. Empirically, what is interesting is to outline cases where discourse proves to be central in the development of events and to pick the agents that are responsible for its coordination and communication. Brazil is such a case. The study of the four selected Brazilian think tanks and their changing discourse can be seen as a barometer of government policy as it predates and then accompanies the eventual transition of the Brazilian Government to austerity measures in 2015. In this article it is shown that a polarization of think tanks' communicative discourse in favour or against austerity can be observed during the second critical juncture. Polarization is a characteristic of intense historical moments when capturing public opinion by dismantling the opposition's arguments is often perceived as the most successful communicative strategy. This is why polarization is more evident in the communicative rather than the coordinative discourse of think tanks. Wallace (1994) argues that policy-makers need advisers when they need to rethink the dominant assumptions of policy. Think tanks are there to provide this rethink when necessary. Their aim is to make academic findings more palatable for busy politicians and policy-makers (Stone, 2007). This gives think tanks the opportunity to set the policy agenda and to prioritize some topics over others and thus play an important role in public policy design. The idea that think tanks offer an independent bridge between academia and policy-makers is problematic because it assumes that there are two parts (academia and politics) that need to be linked. In many countries, including Brazil, directors and experts of think tanks are closely related to politicians and bureaucrats (Stone, 2007;Stone & Ladi, 2017). Academics themselves are not distant from think tanks or from politics as they move in and out of official posts. It can be argued that they all belong to the same elite, they have similar word views and their affiliation often changes (Newsom, 1995(Newsom, -1996. In the case of Brazil, the relationship between academics, bureaucrats and policy-makers has at times fluctuated alongside political and economic development, but has become closer as policy decisions become increasingly technical (Kluger, 2015). Think tanks in this article are thus not considered to act as neutral bridges between academia and politics nor to function with public good as their compass (for a detailed discussion see Stone & Ladi, 2017). Think tanks are seen here as part of the elite structure (see also Parmar, 2004). They act as carriers of coordinative and communicative discourse, a role at the heart of political debate. What is illustrated in the next section as well as in Zimmerman and Stone (this issue) is that think tanks increase their opportunities to participate in public policy at critical junctures when both coordinative and communicative discourses are urgently in need. Additionally, in times of crisis the polarization of communicative discourse is evident.

Methodological note
To assess performance of think tanks during critical junctures in Brazil we use Abelson's categories as proxies of coordinative and communicative discourses. Abelson (2009) points out that, although some strategies of think tanks to exert influence are hidden from the public, others can be easily identified. Thus, he differentiates public strategies from private marketing ideas. Private strategies relate to direct, privately held approaches to policy-makers and can be related to coordinative discourse. The public strategies, on the other hand, refer to the broader approaches addressed to the general public through the media, or to focused, but not individual, audiences through events, publications and other activities and can be related to communicative discourse. Indicators of public strategies include: public forums and conferences; public lectures and consultancies; testimonies before congressional committees and subcommittees; books, opinion magazines, newsletters, policy briefs in widely distributed newspapers; mailings and posting of think tanks' studies on internet sites; exposure in the media. Indicators of the private strategy consist of: integration into government positions (in offices, sub-offices and bureaucratic positions in administrations); serving on policy task forces and transition teams during presidential elections and on presidential advisory committees; maintaining liaison offices with policy-making scopes; invitations to selected policy-makers to attend conferences, seminars and workshops; permission for bureaucrats to work in think tanks in a limited period of time; policy-makers working in think tanks; preparation of policy briefs and briefs for policy-makers (Abelson, 2009, pp. 78-82) Accordingly, the public and private strategies of think tanks are assessed to identify the production of communicative and coordinative discourses, respectively. Different kinds of primary data are used in order to understand the identity of the selected think tanks, their approach to austerity and their production of communicative and coordinative discourse. All data collected are internet based (official websites, blogs, social media, online publications, online mass media publications).The collection of budget, staff specialization and size, affiliations and managerial structure information aims to map their profile and institutional capacity. Publications on macroeconomic issues (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) constitute the main source for exploring discursive shifts. Think tanks' communication strategies, public visibility strategies and strategies of interaction with policy-makers have been used as proxies of communicative and coordinative discourses, respectively. More specifically, the previous careers of board members of think tanks are explored and closed meeting and seminars with policy-makers are identified in order to reveal think tanks' private strategies. The content of think tanks' web pages and documents, their media appearances and the organization of public events provide evidence of their public strategies. Combined, these different sets of data lead to conclusions on whether and how their discourses changed or not during the examined critical junctures and which audiences were prioritized.
One of the main obstacles of assessing Brazilian think tanks, specially searching for data before 2015, is that these organizations have only recently became organized to provide systematized accountability about their performance. Therefore, not only the case selection, but also the research design possibilities get compromised, particularly since the approach relies on internet-based data collection.

Brazil's economic policy and the global financial crisis: resistance and austerity
The analysis of the role of think tanks in the transformation of economic and fiscal policy in Brazil during the global financial crisis within a HI and DI framework necessitates an understanding of the trajectory of Brazilian economy and politics in the past decades. Unlike Western powers, Brazil experienced the global financial crisis of 2007-2009 relatively unscathed. Brazil entered a temporary recession in 2009, but recovered strongly by 2010 with GDP increasing by 7.5%; it was, at the time, referred to as one of the less affected and most resilient economies (Ferrari Filho, 2011). According to the World Bank, between 2006 and 2010 Brazil demonstrated an average annual GDP growth of 4.5%. This period, coincided with the high point of the country's growing relevance as an 'emerging power' and marked a critical juncture in the transformation of its economic policy towards the stimulation of investment, trade and social growth. The effects of the global crisis were felt more strongly six years later. Following the rapid macroeconomic deterioration which began in 2014, with GDP falling by 3.8% in 2015 (World Bank, 2017), the government introduced a shift towards austerity and spending cuts in a second critical juncture.
The global financial crisis overlaps with the first critical juncture for Brazil, which at the outset served as an opportunity for the economy to project stability and maturity: initially, the recession only hit the country for five months in 2008-2009, largely due to the panic spurred in financial markets. The commodities' boom and the growing affluence of Brazil's domestic consumer market at the time, contributed to the fiscal authorities' ability to delay the immediate effects of the external shock and maintain their spending priorities. The speedy recovery was highlighted in the global financial and political circles and in the media as an indication of Brazil's consolidation as an economic power (Economist, 12/11/2009). In 2012, Brazil surpassed the UK and became the world's sixth largest economy (Telegraph, 12/05/2012).
The contrast was striking in comparison to the 1980s and 1990s when inflation, fiscal instability and growing public expenditures and public debt had ravaged the Brazilian economy (Giambagi, 2011). This experience generated the political capital needed to push through market-orientated economic reforms initiated under President Fernando Collor de Mello (1990Mello ( -1992 and strengthened under President Fernando Henrique Cardoso (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002). Under Cardoso, privatization helped decrease public debt (Giambagi, 2011) but fiscal decentralization continued to affect Brazilian primary deficits negatively (Abrúcio, 1998). Nevertheless, the debt continued causing structural problems making the Brazilian economy susceptible to external liquidity crises (Samuels, 2003).
The Cardoso government (Brazilian Social Democratic Party, PSDB) implemented strong fiscal adjustments in 1999 setting rigorous targets that represented effective restrictions to public spending and focused on the fight against corruption (Samuels, 2003). In 2000, the Fiscal Responsibility Law was approved, establishing fiscal discipline and ceilings for expenditures for central government, states and municipalities, and a number of controls on public finances. In the same year, the government adopted an inflation-targeting regime and a flexible exchange rate. These two measures coupled promoted the necessary monetary austerity to keep inflation under control without unwanted consequences on prices. These reforms were combined with the reduction of public spending through the privatization of inefficient public services. Debt was gradually reduced and Brazil became more attractive to foreign investors.
Dissatisfaction with fiscal austerity, as well as growing income and social inequality, led to increasing support for the Workers' Party (PT), and eventually to the victory of its presidential candidate Luiz Inacio Lula da Silva, in 2002 and 2006, and his successor Dilma Vana Rousseff in 2010 and 2014. The first PT government under Lula marked a departure from Cardoso's plan, in that it focused strongly on the expansion of social policies (Pero, 2012), but maintained a number of his macroeconomic policies. The main focus of policy-making then was to improve the national institutional environment to create a fertile ground for sustainable growth (Ministerio da Fazenda, 2004). Inaugurating what would come to be known as a 'responsible left' policy-making was characterized by a more centrist and pragmatic view than the radical left of Hugo Chavez in Venezuela. Under Lula, the PT government pursued domestic reforms and economic growth through a trade-oriented foreign policy. It also combined social policy with free market approaches, leading some to retrospectively refer to its economic policy as 'liberal neo-developmentalism' (Ban, 2013).
Rousseff took over under a rapidly deteriorating global economic environment. Her government attempted to stimulate the economy twice in response to the 2009 crisis, by introducing two investment boosting plans: the second Growth Acceleration Plan (PAC) in 2010, and the Bigger Brazil (Brasil Maior) plan in 2011. These sought to encourage domestic investment in key industries such as petroleum and gas exploration, electric power generation, agribusiness, nano and biotechnology, sanitation and housing, through soft loans, electricity rates subsidies, payroll reductions, and other tax incentives. Other initiatives aimed at stimulating investment included a simplification of the tax system, the introduction of tax-advantaged bonds for infrastructure and research and development investment, as well as labour skilling/upgrading.
In spite of economic stimulus measures, the toll of the protracted recovery of global markets and of falling global commodity prices, was felt increasingly after 2012. The economy entered into recession in 2014 (ECB, 2016). This, we argue here, was the beginning of the second critical juncture for Brazil. In the second quarter of 2015, the country fell into a technical recession, with Standard & Poor's downgrading it to BB plus 'junk' status. Public debt and current account deficit grew rapidly, while the trade balance declined. Inflation rose steadily and Brazil recorded an annual primary deficit for the first time since 1997. The Brazilian currency (Real) hit its weakest point since its introduction in 1994, losing over 35% of its value against the dollar.
The rapid increase of the real effects of the financial crisis in 2014-2015 triggered a distinct economic policy shift. In response, the Brazilian Government and the Central Bank introduced a tighter monetary stance and a fiscal adjustment to contain inflation and stabilize public debt (Banco Central do Brasil, 2015). Three constitutional reforms aiming to address the crisis were debated in Congress: the reform of Social Security in order to reduce costs; the re-imposition of a financial transactions tax (CPMF); and the need for more flexibility in budgetary management. In September 2015, Finance Minister, Joaquim Levy announced a new fiscal austerity package in order to reduce expenditure by $USD 6.8 billion. The package included measures such as freezing public sector salaries and curtailing social programmes, cutting back access to subsidized loans, scaling back tax breaks and subsidies, increasing social security and other taxes.

Brazilian think tanks during the two critical junctures
Brazilian think tanks have been progressively playing a significant role in national politics, but is still an arising phenomenon in the country. A historically highly insulated governing model during the twentieth century, combined with a disorganized civil society, resulted in the dominance of governmental or government-affiliated policy consultancy organizations. By the end of 1980s, the re-democratization led to the emergence of a larger number of think tanks, however, still relying considerably on governmental financial resources due to a weak philanthropy culture (Hauck, 2015).
Historically, we can observe episodes of think tank-like organizations standing out and considerably intervening in political processes during the twentieth century. The most prominent examples are think tanks organized to oppose government in the 50s and 60s democratic interregnums as well as in the neoliberal diffusion wave of the 80s and 90s (Birkner, 2002;Correa, 2005;Durand, 1997;Gros, 2004;Lopes, 1991;Ramírez, 2005). In the past 15 years, similarly to those moments, funding from enterprises mainly affiliated to conservative forces, has nurtured Brazilian think tank's proliferation as an alternative path to the political scene that has been extremely government centred. While 'progressive' think tanks have been strengthened by successive presidential terms led by the PT, a network of conservative think tanks has been evolving sharing funders' values and a growing conservative public opinion.
Four think tanks are selected in this article in order to illustrate think tanks' capacity of carrying communicative and coordinative discourse. The main concern in the case selection, thus, is their role as providers of political economy knowledge through a distinctive set of strategies (Hauck, 2017). Therefore, all four are mainly dedicated to economic issues; they are representative of diverse perspectives, and they have a systematic online publication production as well as an academic profile.
The Ipea is a governmental think tank. 1 Despite the stability and supposed independence provided to its research staff as civil servants, its survival and decision-making depends on its relationship with political agents (Rigolin & Hayashi, 2013, p. 414). The Inter-union Department of Statistics and Socio-economic Studies (Dieese) is representative of the progressive think tanks. It was created during the democratic period (1955) and has been 1 the case selection adopts a think tank definition approach that includes governmental public policy research institutes because of the centralized nature of the Brazilian political system (Fieschi & Gaffney, 2004;Freres, Seabra, & Moraes, 2002;Garcé & Uña, 2010;McGann & Johnson, 2005;thunert, 2011). significantly nourished under the recent PT administrations. The Institute of Brazilian Economy (IBRE) is a consolidated conservative think tank, affiliated to a private independent university, but with a long history of receiving government funding, especially grants destined to the production of economic information of public interest. The IFB is mainly financed by private companies and represents the conservative think tanks' reaction to successive left-wing presidential governments, aiming to push for a shift in political economic thinking. Although it is relatively small institute, its choice serves the purposes of this article since it has a significant production of academic papers and it does not produce only policy briefs. Additionally, it has an excellent internet presence which facilitates the collection of comparable data.
The organizational characteristics of the think tanks analysed differ significantly, which implies that, a priori, they have different capacities to carry communicative and coordinative discourses and to affect political debate. Notably, Ipea has the most robust organizational structure financed by an USD 106 million budget in 2015 and managed by more than 1000 staff of which around half is directly dedicated to research. As a governmental agency, the organization holds close ties to policy-makers and, as its administration changes with presidential terms, that can have an impact on their discourses. Their studies are widely disseminated by several public opinion strategies, through press conferences, internet channels (social media, blogs), a non-academic journal and a TV programme (Panorama Ipea). Ipea, therefore, focuses on the production of both communicative and coordinative discourses.
Dieese is an initiative of the trade union movement tasked to conduct research in support of workers claims. Accordingly, the organization survives mainly because of union contributions, accounting for an USD 12.2 million budget in 2015 and around 270 employees. They have close relations and share ideological alignment with PT governments. They also have a specialized communications team to organize press conferences (1200 interviews in 2015). Daily presence in a radio station and in specialized labour union communication channels are additional strategies. Like Ipea, Dieese is organized to disseminate ideas both to the public and to policy-makers and specialists. IBRE, on the other hand, is part of Brazil's most prominent think tank cluster (Truitt, 2002) -Fundação Getúlio Vargas (McGann, 2016. IBRE produces a highly referenced temporal series of economic indicators to supply both the government and the private sector. Charging for part of the data and information produced and for consultancy contracts -600 in 2015 (IBRE, 2014) -they are a significantly autonomous think tank. Around 40 researchers work permanently and exclusively for IBRE. Its main communication channels are newsletters and press conferences that take place on a daily basis. Nevertheless, IBRE does not seem to have a deliberate and strategically organized front for delivering communicative discourses even though its website has the most intensive traffic because of FGV's prominence -it is between the 350 most accessed websites in Brazil and has more than 5500 external linkages to it. 2 IFB has financial support from foundations, private companies and international agencies. The permanent staff only accounts for 7 researchers; however, 97 external members collaborate in articles and papers. The organization's most fruitful channel to spread their ideas is its 'Brazil Economy and Government' (BEG) blog with 60 articles per year, mostly 2 these websites assessments were provided by http://www.alexa.com/. all think tanks' media data were collected on 31 august 2016. written by economists. BEG is the IFB's main instrument for communicative discourse by providing 'explanations on economic issues that would help voters understand the choices made by the public sector in the management of the economy' (BEG, 2016). Newsfeed mechanisms and social media fan pages are used as well as mass media publication, gaining massive idea spread results. IFB seems to be primarily dedicated to communicate its discourse to the general public, hoping to influence public opinion.
Three of the four organizations were founded in the 1950s/1960s -IBRE (1951), Dieese (1955) and Ipea (1964). This longevity has transformed them into reference organizations in macroeconomic matters. Dieese is considered the main reliable source for economic indexes and rates, which are used for macroeconomic analyses, including those produced by the government (Portal Brasil, 29/01/2014). Similarly, IBRE has become a reference in indicators for market prices and several economic indexes that are widely used by the private and public sectors. As a governmental agency, Ipea's studies have also been extensively used as a basis for economic planning. Cross-referencing shows that Dieese's and IBRE's indicators are profusely used by all four think tanks, attesting to their mutual recognition of credibility. Ipea's studies also appear as a recurrent reference in all think tanks publications, including IFB's (Conceição, 2014;Dieese, 2014b;Nery, 2015;Rocha & Lopes, 2015). Their interaction is also shaped by ideological factors. Even though IBRE frequently uses Dieese's products and some of Ipea's research, it leans towards a more orthodox macroeconomic position. The same holds for IFB. These ideological alignments are clearer in the first critical juncture. In the second, however, subtle shifts take place in their discourse strategies and consequently in their interactions.
In the following sections, the analysis of the think tanks' main ideas and strategies regarding macroeconomic policies is structured around the communicative and coordinative discourse produced during the two aforementioned critical junctures.

From 2008-2009 to 2014-2015: a march towards polarization
The 2008-2009 critical juncture, which combined the outbreak of a world economic crisis with a growing confidence in Brazil's economy, had diverse effects in think tanks discourse. On the one hand, Ipea and Dieese, strongly affiliated to the government, progressively intensified their support of counter-cyclical measures, advising for an increase in governmental expenditures. Ipea focused on income transfer policies and credit expansion through public banks. However, some of their studies combined Keynesian assumptions with ones that remain under the aegis of the neoliberal tripod regime of inflation targeting, floating exchange rate and fiscal surpluses (Sicsú, 2008). Dieese's initial main concern was to maintain employment levels without diminishing workers' rights. That protection would come through increased financial market regulation and government incentives (Dieese, 2008(Dieese, , 2009).
On the other hand, IBRE argued that austerity measures should be progressively applied, focusing on specific public investments in infrastructure and productivity. IBRE advised heavily on the need for fiscal reform combined with a reduction in public consumption, especially through pension and social security reform. At that time, IFB evaluated Brazil's position in the crisis in a very punctual way, yet, aligned with IBRE's critique about government's strong fiscal and credit stimuli, seizing Brazil's emergence as an economic power (Afonso, 2011;Afonso & Mussi, 2009;Bonelli, 2010). Until 2011, IFB had weak channels of discourse delivery. In that year, the creation of the BEG blog aimed to enhance its dissemination capacity. Therefore, the focus on communicative discourses became pronounced. Even though BEG's editors maintained connections from their past and present careers in governmental institutions (Federal Revenue, Central Bank, Ipea), IFB's board of directors was and remains, mainly formed by business executives. The linkages to the political realm are restricted to its president, a former politician, and a few members of the Board. Their right-wing affiliations provided them with less political access to the Executive Branch in the period. In addition, IFB had almost no traceable interaction with the other analysed think tanks.
IBRE's strong trajectory as a national supplier of economic indexes provides for considerably more opportunities to carry coordinative discourses, regardless of which government is in power. Like IFB, IBRE also includes former politicians in its Board. However, their actual capacity to articulate coordinative discourse is enhanced by several contracts with the government through which they can deliver their main ideas. They do not seem to have the general public as a target audience. This is evidenced by their scarce use of channels that provide communicative discourse. For example, IBRE has considerably less followers than any of the four think tanks on Facebook and a low average of shares to its posts -Ipea, Dieese and IFB had in average 17, 26 and 04 shares per post in 2015, respectively, while IBRE had a .5 average of posts shares.
As a governmental institution, Ipea is naturally a carrier of coordinative discourse. The various technical cooperation agreements for joint research between Ipea and ministries, Senate and other governmental institutions during this period, attest to that. Moreover, the frequency of seminars and meetings on fiscal issues with government members is significant. Ipea has also continuously used numerous channels for the dissemination of its publications to the general public. Its official Youtube Channel has 88 uploads on average per year, 510 only in 2015 (Agência Ipea).
Due to its role in the conduct of labour union movement negotiations with the government, Dieese has gained considerable political access in the PT, enabling its role in coordinative discourse production. DIESE holds meetings with governmental institutions on a regular basis, on labour union negotiations -on average 850 meetings a year during the analysed period -but also technical cooperation meetings -on average more than 70 per year. In the 2008-2009 period, Dieese did not have a social media strategy, but used the website and press conferences as a channel for communicative discourses -680 thousand website accesses and 1322 press conferences per year, according to the organizations' Annual Reports .
Between the 2008-2009 critical juncture and 2013, Ipea and Dieese progressively enhanced their support for counter-cyclical measures, which translated into increased governmental expenditures. Even before the effects of the crisis became noticeable in Brazil's economy, Ipea adopted an even more active position, arguing in favour of public spending expansion through an income transfer programme. For the Institute, the fiscal and monetary counter-cyclical actions taken in 2008-2009 had been too frugal and the low levels of public spending were responsible for low economic growth (Ipea, 2010a(Ipea, , 2010b(Ipea, , 2010cAraujo & Gentil, 2011). Panorama Ipea broadcasted several publications defending the institution's policy proposals and disseminated them in its social media channels. Ipea also organized numerous seminars for governmental and international organization members on this topic (Ipea Agenda, 2016). In that same path, Dieese's advice on fiscal adjustment inflection, resumption of credit extension policies and public spending, and an explicit neoliberal critique started appearing in 2011 (Dieese, , 2011b(Dieese, , 2012). An example of Dieese's coordinative discourse in the matter is its participation in the Fiscal Policy Sustainability Work Group (led by Economic Issues Commission in Senate).
It is worth noticing that these two institutions developed intensive joint actions during that period, geared predominantly towards the production of coordinative discourse pushing for the adoption of their policy advice. They co-hosted events and courses (CUT Notícias, 10/11/2009), produced joint publications (Ipea Notícias, 09/04/2012) demonstrating alignment regarding the economic crisis until 2013. They jointly defended a constitutional amendment to reduce working hours regardless of the fiscal critique (Câmara Notícias, 19/05/2009). No such cooperation beyond cross-referencing was identified between any other combinations of the think tanks studied.
From 2009 onward, IBRE's publications focused on pro-fiscal adjustment recommendations. The core argument was the limit reached by the model that favoured Brazil's growth performance from 2006 to 2011 (Pinheiro & Bonelli, 2012;Barros, 2013). IBRE maintained its usual way of discourse channelling: coordinative through consultancy contracts and communicative in a less engaged fashion, primarily through press conferences.
IFB's strategy, on the other hand, became more focused on communicative discourse with the creation of BEG, in which several publications argued that, despite the BRIC brand, austerity measures should be progressively applied -infrastructure public investment and the neoliberal tripod to correct imbalances caused by the crisis (Mendes, 2011a(Mendes, , 2011bRocha & Αlvares, 2011). The credit for expansion of demand, tariff barriers, special tax exemptions and excessive labour market regulation were the main targets of critics to Keynesian measures (Kohler, 2012). To create more opportunity for coordinative discourse, IFB often held open events for small audiences predominantly formed by business executives, regional politicians and students. Frequently mid-level technical members of government led the debates that became progressively more critical of the heterodox governmental policies (IFB, 2016) ( Table 1).
As the delayed effects of the crisis hit Brazil's economy in 2014-2015 and swept away its emerging economic power promise, an enhancement of strategies and discourse content already in place from 2011 has taken place in all think tanks, except Ipea. Dieese's publications maintained the concern about the recovery of socio-economic development without significantly changing public investment and incentives for private investment and credit supply (Dieese, 2014a(Dieese, , 2014b. Their discourse grew progressively more politicized (Dieese Comunica, 03/12/2015) against austerity by the time the political situation became critical. Dieese in this period intensified its effort in directing its communicative discourse beyond workers' fora and towards the general public. To illustrate, the activity on Dieese's Facebook page in the month of December 2015 increased by 54% in 2015 compared to 2014 which was 600% higher than 2013. In order to preserve its access to policy-makers, Dieese also maintains its relationships with several governmental institutions and ministries (e.g. Ministry of Cities, Ministry of Planning, Ministry of Work, Ministry of Science and Technology). In 2015, Dieese organized a special programme targeting legislative institutions aiming to discuss the public debt and the centrality of workers to economic development .
The necessity of regulating the Fiscal Responsibility Law, amplify austerity and reestablish realism in fiscal goals in order to regain credibility became common ground for IBRE and IFB (Appy, Lisboa, Mendes, & Lazzarini, 2015;Barros & Afonso, 2013). 'Mistakes from the Xs mark the strategies explicitly used by think tanks in that period as a clear and extensive part of their discourse agenda. the absence of marking doesn't mean think tanks have never used that strategy but only that our analysis showed it was only occasional. briefs and media exposure and events take a pro-Keynesian orientation to economic policies. the position is enhanced until 2013, defending an income transfer programme as a counter-cyclical measure no study, brief, media exposure or event defending austerity could be found. the main focus is on the benefits of high employment rates and minimum wage valorization to overcome economic turmoil Some studies, briefs and events focus on fiscal adjustment and defend the adoption of neoliberal tripod recommendations. Seminars discussing those issues with the participation of former politicians Studies, briefs, media exposure and events constantly discuss the necessity of controlling public spending, and fiscal reforms. But the incentives of public spending on infrastructure and productivity is strongly present the past: how to get out of fiscal meltdown' (2015) and a series of symposiums promoted by IFB had a common agenda: the fiscal adjustment, proposing a limit fixation to federal government debt and, therefore, spending. In 2015, IBRE took part for the first time in at least three meetings of the Economic Assessment Committee in Senate promoting the same ideas. One of them was about re-opening a legislative amendment, which would bring changes in fiscal limits including a ceiling to the consolidated gross debt (Senado, 2016). Consequently, Dieese, IFB and IBRE displayed a deeper commitment to spreading their ideas more aggressively and through communicative and coordinative fora. For Ipea, however, 2014 marked the moment when it adopted a more moderate position and the traditional alignment with Dieese diminishes significantly. Their interaction remains, but also begins involving IBRE (Ipea Notícias, 15 September 2015;Dieese Comunica, 26 May 2014). From 2014 on, Ipea's and IBRE's interaction to discuss fiscal balance, fiscal rules, productivity and minimum wage increased considerably (Ipea Notícias, 30 September 2015;Ipea Notícias, 11 December 2015;Ipea Notícias, 17 December 2015;Ipea Notícias, 27August 2015). These discussions were backed up by studies posing the problem of fiscal adjustment in the existence of structural rules of public spending, as compulsory expenditures (Gobetti, 2015;Gobetti & Orair, 2015). Some relevant IFB publications were aligned to that argument therefore advising for reforms in taxation (Gall, 2014). IFB's publications indicated a structural problem in relation to public spending growth, to fiscal sustainability and to stagnant productivity, which went beyond short-term fiscal problems (Barros, 2014;Salto, Afonso, JR Biasoto, & Kohler, 2015).
Ipea's discourse during this period focused on the mix of a macroeconomic tripod restoration combined with government spending on infrastructure and productivity, but became progressively moderate and leaned towards an austerity design, resembling the discourse of the first critical juncture (Bahia, 2015;Camarano, 2014;Ipea, 2014). As usual, the publications were launched in open events and broadcasted by Ipea's TV programme, seeking for public repercussion. It is interesting to note a decrease of almost 65% in the average number of events where coordinative and communicative discourse could be promoted in 2014-2015, compared to the 2008-2013 average, such as meetings and audiences with policy-makers and decision-makers and lectures and seminars open to public.
In summary, from the first to the second critical juncture a polarization can be observed in terms of discourse. IBRE and IFB present progressively more austerity-aligned discourses. To the contrary, Dieese strengthens its position against austerity and increases its creation of communicative discourse channels. Interestingly, given its necessity to align with the government's response to the economic and political crisis, Ipea shifts its position towards a more austerity-friendly perspective, even though during all the period analysed the administration of IPEA is subordinated to a left-wing presidential government. A clear indicator of that is its approximation to IBRE from 2014 on. During this period, IBRE, Ipea and Dieese enhance their communicative discourse while continuing their coordinative efforts which also include inter-think tanks cooperation. IFB attempted to reach public opinion by promoting right-wing prescriptions but remained isolated from the rest of the think tank community. It primarily focused on social and mass media (Table 2).

Conclusions
The four think tanks in Brazil changed their discourse in relation to austerity during two critical junctures for Brazil (2008-2009 and 2014-2015). Two organizations are ideologically and institutionally, respectively, linked to government (Ipea) and to the trade unions movement (Dieese) and two institutions are linked to business circles and can be characterized as conservative (IFB and IBRE). All four think tanks have been carriers of coordinative and communicative discourse for or against austerity during the two critical junctures for Brazil. The divisions in the discourse intensified and the debate became more polarized during the second critical juncture which was when the economic situation worsened and austerity measures were actually introduced. No major differences can be found between the substance of the coordinative and communicative discourse of the four think tanks. What changes, is the tone and the style of their discourse. Communicative discourse which has as its main audience the general public is characterized by simpler language and stronger messages and thus is more polarized especially in times of crisis while coordinative discourse which mainly targets the political and business elites is more technical and often evidence-based.
Our main argument is that think tanks neither act as neutral bridges between academia and politics nor always function having public good as a compass. Think tanks are carriers of discourse which reflects their institutional affiliation and their political and ideological preferences. During critical junctures think tanks not only increase their visibility but they also, when the circumstances demand it, produce more polarized discourse in order to more clearly convey their message and differentiate themselves from other think thanks and policy actors. It is during critical junctures that interaction among think tanks and between think tanks and policy-makers increases. This is a significant finding because it proves think tanks can be significant agents of policy change which act as mediators between society and governments -via their communicative discourse -but also between political elites -via their coordinative discourse. This finding confirms that the study of discourse during critical junctures can reveal the background of policy shifts which could have otherwise remained unnoticed.
The 'revolving doors' phenomenon is quite pronounced in Brazil and provides think tanks with access to political and business elites and allows them to act as carriers of coordinative discourse via the creation, elaboration and justification of policy and programmatic ideas. They often provide the framing of policy issues and the arguments for the debate. They also act as carriers of communicative discourse via multiple means especially during critical junctures by presenting and legitimizing political ideas such as austerity to the general public when decisions have already been taken. Investigating whether this is also true in other parts of the world and whether the way austerity discourse is diffused is any different to how other types of discourse are diffused could lead to interesting generalizations.
To conclude, think tanks are agents that should be closely studied especially during critical junctures since their role as carriers of discourse gives them the power to articulate and thus influence the content of ideas and arguments. Studying shifts in the discourse of think tanks cannot answer questions about their overall influence upon policy but it can provide information on how the climate of ideas changes (Stone, 2001). Comparative research of think tanks and their role across their world can shed light on the actual mechanisms used by them but can also give us a better picture of their organizational nature and the interests they represent.