1. Introduction

2During the successive governments led by the Workers’ Party (Partido dos Trabalhadores, PT) beginning in 2003, public action changed substantially. New policies were the subject of a controversy that involved two distinct and opposing approaches according to disciplinary approaches. The first is economic; it analyses these changes in terms of the various concepts relating to developmentalist strategies (Amado & Mollo, 2015; Bresser-Pereira, Oreiro & Marconi, 2015; Fritz, Paula & Prates, 2022). The second approach entails a sociological or political analysis of public action. Several economists have focused their study on the deindustrialisation effect of orthodox monetary and exchange rate policy, arguing that it is entirely incompatible with a developmentalist strategy. Conversely, several political scientists and sociologists have emphasised State interventions that break the liberal paradigm. Some consider this break as a result of path dependency: i.e., an institutional continuity whereby the State can reassume its former capacity as a promoter of development by adapting to an open economy (Boschi, 2010; Diniz, 2012; Santana, 2012). Others analyse the competition between social groups in the constitution of a ruling bloc (Boito, 2012; Singer, 2015).

4The regulationist approach to macroeconomics provides an original perspective for investigating something that other heterodox approaches predominantly regard as an error in economic policymaking. In this case, Brazil not only let its trade surplus drive up the value of its currency but specifically accentuated this appreciation by maintaining a very high-interest rate differential that attracted foreign portfolio investments. This policy error is a political issue that we must consider as a strategic government choice, and indicates the importance of concessions to orthodoxy (and thus to interests that would benefit from orthodoxy). The analysis of macroeconomic dynamics in Brazil developed further here shows a de facto fundamental structural compromise: i.e., a coordination of interests obtained through the achievement of a macroeconomic situation in which each group considers that it has reached a satisfactory level in pursuing its interests. This coordination may take place in the absence of any formal negotiation. The distributional compromise is ex-post, resulting from the combination of a restrictive monetary policy and an increase in the wage share of national income.

5Accumulation regime analysis sheds light on the dynamics of income distribution variables and the viability of the social compromise and its characteristics. In the de facto compromise, the conflict is indeed only suspended. The compromise may contain a more explicit or codified form, but we show that this is essentially about economic policy instruments. It is then an instrumental compromise. Moreover, this was the subject of a solemn commitment to the continuity of macroeconomic policy, assumed by the presidential candidate in 2002, Lula da Silva. The former trade union leader thus demonstrated that he could overcome the threat of obstruction from the financial sector. In contrast, when Dilma Rousseff became president in 2011 and challenged financial interests, she lost all support from the ruling classes, and was subsequently forced to change her position and adopt strict austerity as of 2014. This in turn resulted in a loss of legitimacy among the working classes, which, combined with the persistent opposition of the property-owning classes, created the grounds for her removal from office by parliament, on the pretext of hidden public spending.

6The durability of Lula’s distributive compromise depended on Brazil’s macroeconomic dynamics and on its capacity to maintain a unity of the decision-making actors, which turned out to be impossible when the economic situation deteriorated. This observation raises the question of the viability of the distributive trade-off against the central bank’s power. This trade-off aimed at defending the sanctity of Brazil’s “tripod” of “inflation targeting, flexible exchange rate, and a budget surplus”. Michel Foucault’s concept of governmentality seems relevant to studying the way these economic policy instruments fit into an institutional whole which, from an economic point of view, makes it possible to govern populations.

8The concept of governmentality highlights how government choices become technical, implying political decisions based on knowledge regarded as scientific. Such an approach appears to be particularly useful because it investigates the potential irreversibility of these choices. This line of questioning seems relevant given the substantial body of research showing that liberalism has challenged long-standing institutional continuities in advanced capitalist economies (Streeck & Thelen, 2009). It is worth investigating whether this might also be true for Brazil in light of an essential characteristic of the advance of liberalism, namely that a governing technique persists even when political power changes hands, thanks to what can be called “instrumental coalitions” (Baudot, 2014). From this standpoint, the capacity that orthodox theory has demonstrated in Brazil to impose a cognitive framework that prevented a shift from the so-called “macroeconomic tripod” – a primary budget surplus, inflation targeting, and a floating exchange rate – is a liberal governmentality device. If we follow Foucault, the autonomy of this device as a system of action is guaranteed by market forces. Following the regulation theory, it depends on the coherence of an institutional setting that provides a horizon of stability for the accumulation regime, but which ceased to function after 2011.

9In the following section of this paper, we analyse how it was temporally possible to combine a regime of accumulation dominated by finance, set up in the 1990s, while implementing a distributive compromise favouring all types of income: i.e. financial revenues, profits, and wages. Section 3 shows how a doctrinal compromise between developmentalist ideas and economic orthodoxy made this distributive growth strategy possible. It explains how a slight change in the regulatory and institutional framework triggered a virtuous macroeconomic dynamic that established the new distributive compromise. The doctrinal compromise also reveals how the virtuous circle was reversed to become a vicious one. Section 4 will decipher the interaction between ideas and economic interests at work in liberal governmentality. It focuses on how the strategic role of the central bank is specified here in locking-in a possible developmentalist coalition and, therefore, in steering the accumulation regime. To conclude, we return to the theoretical lessons of this analysis in explaining the failure of this developmentalist experiment in Brazil.

2. The capital accumulation regime and income distribution

10This section will show that the time of the distributive growth strategy, adopted by the PT government, was running out due to the preservation of financial interests in what we call a de facto social compromise. Therefore, the latter has allowed a finance-dominated accumulation regime to become what we term a finance-led accumulation regime, but this performance could only be provisional. Growth was not sustainable, and the accumulation regime was, in the end, blocked by finance. Stockhammer (2007, p. 4) put forward the concept of a finance-dominated accumulation regime to stress “that financialisation is shaping the pattern of accumulation (or put in another way: the composition of the components of aggregate demand and their volatility)”. The author emphasises the difference with the concept of the finance-led accumulation regime developed by Boyer (2000). We should consider the latter as a particular case whereby financialisation leads to an increase in the growth rate. Regarding the relationship between financialisation and deindustrialisation in Brazil, we will characterise the finance-dominated accumulation regime as a case of an accumulation regime blocked by finance.

11Miguel Bruno (2008) showed that the government of Fernando Henrique Cardoso (1995-2002) institutionalised the increase of financial income, contingent upon a growing external public debt. The outcomes of monetary stabilisation based on a pegged exchange rate and economic liberalisation led to a stagnation of productive accumulation and a reduction of the wage share. Accumulation and wages then progressed under the PT government (2003-2015). This progress raised the question of whether a finance-led accumulation regime had emerged, unlike the regimes in the economies of “the centre”. Beyond the specificities which we attempt to bring to light here, this analysis will show that in the case of Brazil, the viability of such an accumulation regime depends on the capacity of monetary policy to avoid finance driving the economy into an instability zone. This was a significant result of the formalisation of a finance-led accumulation regime that Boyer (2000), derived from the stylised facts of the US economy. The main difference with Brazil is that the expansion of demand could depend only very partially on the wealth effect of financial assets. Increased employment, salaries, and social expenditures were primarily its base and had only a temporary wealth effect on economic growth. We explain this result because the monetary authority was responsible for driving the economy into an instability zone, insofar as its policy only allowed for an accumulation regime blocked by finance in the long run. The monetary authority keeps interest rates high to maintain financial capital inflows consistent with the balance of payments equilibrium. It results in an appreciation of the real exchange rate, and a decline in the share of industry in GDP. As a result, external vulnerability increases with commodity specialisation and an excessive reliance on short-term speculative capital.

12Three new main stylised facts characterise a distributive shift in the finance-dominated growth regime that resulted from the trade deregulation implemented during the 1990s. These facts sum up the main features of the macroeconomic dynamic surrounding the new distributive compromise that restructured the former mediation of interests. These three facts are:

1. the increase in the average real wage surpassed the productivity gains and the inflation rate, while the lower allocation of profits to financial assets enabled a slight acceleration in the productive capital accumulation rate;
2. a substantial expansion of the export sector of agricultural, metal, and energy commodities due to Chinese demand and financial speculation in these commodities;
3. spectacular growth in interest income, paid by the State and the middle classes, the working class and the productive sector, in general, to the banking and financial sector. This expressed a singular form of the financialisation of the economy.

13Nevertheless, financialisation also increased credit supply to households in this context. Although the cost of credit was very high, given the high real interest rates on credit operations in the Brazilian economy, most people repaid loans over time in several instalments. As a result, interest charged per month, for example, may have represented a bearable proportion of household income. People generally do not calculate interest accrued and paid out by the end of the whole financing period.

15In this equation, r^e is the profit rate of non-financial corporations; ρ_k is capital productivity; w is the average real wage; ρ_L is labour productivity; and φ is the rentier share (or finance income share). We calculate the finance share as total financial income (interest, dividends, and other income distributed by corporations) divided by GDP. The deduction of this general formula of the corporate profit rate is in the appendix of this article.

16Based on the national accounts data for Brazil, Figure 1 shows the determinants of the profit rate, in an index of values, according to this equation. We can draw the following conclusions.

17The profit rate of non-financial corporations peaked in 2008Q3 (the index was 81.5) and then entered a rapid decline until 2016Q1. We can also observe that, despite the initial drop caused by the impact of the American-led global financial crisis of 2008/2009, the profit rate of non-financial corporations continued to oscillate around a relatively stable level between 2008Q1 and 2012Q4. We can explain this stylised fact by the robust external demand for Brazilian commodities. However, as soon as the boom in international commodity prices ended, the rate of profit began a rapid fall that would intensify the distributive conflict between capital and labour, creating the political and institutional conditions for the 2016 impeachment and the removal of the Workers’ Party from the Brazilian government.

18Despite growing, labour productivity remained below the strong growth in real wages, which explains the expansion of the wage share in this period.

19This trend in the profit rate of non-financial corporations was strengthened by the decline of the finance income share of GDP until 2012, expressed by the variable.

20From 2014 on, φ began to grow as an expression of the increased share of national income taken by the financial sector and its rentiers in interest income. In this macroeconomic environment, with the profit rate declining and increased interest rates, the government’s counter-cyclical policies were ineffective, as we observed in 2014. In 2015, Rousseff’s second government adopted a strongly deflationary fiscal adjustment with cuts in public spending. The Central Bank of Brazil nevertheless raised the policy interest rate, the so-called SELIC (Sistema Especial de Liquidação e Custodia – Special Clearance and Escrow System).

21From August 2016, the Central Bank of Brazil began to reduce the SELIC interest rate until February 2018, when it stabilised at an average level of 0.5% per month. This explains the drop in the finance share of national income, observed from 2016 onwards, as financial income is predominantly interest income. This is a specific feature of the financialisation of the Brazilian economy. The drastic increase in unemployment due to the fiscal adjustment and the reduction in finance income share led to a recovery in the profit rate. We can see that since 2015, the wage share has remained stagnant.

22This new arbitrage in the reallocation of profits between real and financial assets may have laid the groundwork for a finance-led growth regime. However, the confirmation of this new regime would require an econometric analysis that we have deliberately left for future research. In this context, we can define two variables that will help us understand the structural factors that determine Brazil’s capital accumulation dynamics.

23The first definition is the ratio between financial income flows and gross fixed capital formation. We can call it the financial-domination rate – a ratio of two flow variables. Financial income (interest and dividends) is a deduction from the profit of non-financial corporations, especially those companies in the productive sector. So, the greater the weight of financial income as a proportion of gross fixed capital formation, the lower the rate of fixed capital accumulation and, therefore, the lower the economic growth rates tend to be.

24The second definition corresponds to the ratio between two stock variables: the stock of non-monetary financial assets (the difference between the monetary aggregates M4 and M1) divided by the stock of fixed capital. We can call it the financialisation rate and its growth expresses the speed at which productive fixed assets are replaced by financial assets.

25The trend growth of one of these two ratios or both expresses a financialisation process of the economy. We may then consider that the first ratio – the rate of financial income flows to the stock of productive capital – reveals a decrease in the domination of finance over the accumulation regime (from now on referred to as “financial domination rate”), despite a still growing “financialisation rate”.

26The collapse in the first ratio indicates a lower effect of financialisation on saving and investment decisions, as highlighted by Aglietta (2001). The constraint on growth originated with the unproductive allocation of savings motivated by financial income, related to the transfer of property rights (interest and dividends), in contrast to their productive assignment to generate gross fixed capital stock. Moreover, the strong correlation between the rise in the accumulation rate and the increase in the wage share during the period must be considered (see Figure 2). The additional demand caused by wage increases, credit expansion, and commodity export growth pulled growth in general.

27As the real average wage and consumer credit grew together, an econometric analysis would be necessary to determine which elasticities boosted demand the most. As the nature and logic of financial accumulation dominated the Brazilian growth regime, we could consider it a growth regime driven by the State and the primary export sector. The State stimulates aggregate demand growth, even at low rates, despite the financialisation of the economy. If the State leaves aside fiscal and monetary stimuli or if external demand weakens, then the regime assumes a finance-blocked character, because the income and the stock of wealth are highly concentrated. The taxonomy of a blocked-growth regime is unusual in the economic literature on accumulation and economic growth patterns. However, we can find it in Bruno (2008 and 2007).

28Figure 3 shows the significant and almost continuous upward trend in the financialisation rate, as defined above. We observe that this trend, beginning in 1990 with liberalisation, contrasts with the minimal upward trend in the accumulation rate during the 2000s. The deceleration of the latter after 2010 suggests that its long-term downward trend since the mid-1970s is not over, and that the Brazilian economy has fallen into the trap of a finance blocking accumulation regime.

29In Figure 3, the financialisation rate is growing, while in Figure 2, the financial domination rate is declining. Why does this occur? Because Figure 3 shows the ratio of two stock variables: the total stock of financial assets as a percentage of the total stock of productive fixed assets. The numerator of this ratio, like any stock variable, can continue to grow even if interest flows have declined, as shown in the previous Figure 2. Therefore, there is no contradiction between growth in the financialisation rate in Figure 3, and the financial domination rate in Figure 2, because the latter is a ratio between flow variables.

30These facts constitute a significant macroeconomic change at the heart of the new distributive compromise, namely, the proportion of gross macroeconomic profit flows (i.e. its overall volume including financial returns) reallocated to productive investments (i.e. gross fixed capital formation as a percent of profit) increased from 2003Q3 to 2013Q3 (see Figure 4).

31However, the increases in investment occurred more in the industrial sectors linked to the exports of commodities, agribusiness, and natural resource sectors (see Figure 5).

32From Figure 5, we can see the positive and strong correlation of investment with the prices of commodities exported by the Brazilian economy. It suggests that the sectors linked to commodity exports invested the most under PT governments.

33Investment in the manufacturing industry remained below the levels required to raise external competitiveness, as the priority was exporting metal, energy, and agricultural commodities.

34However, this new trade-off only lasted until mid-2011. After 2008, counter-cyclical measures temporarily stopped the decrease in profit and accumulation rates that followed the global financial crisis. Yet, the worsening of external and fiscal balances made these policies erratic after 2011. The distributive growth strategy reconciling a rising wage share with a productive reallocation of profits was over (see Figure 2 above).

35These stylised facts structured the macroeconomic dynamics that we discuss in the next section, which explains that the main demand-side contribution to growth, in terms of national accounting, has been final consumption. Furthermore, it is worth noting that the contribution of exports to growth declined after 2005 (see Figure 6), while it increased by 2.5 percentage points of GDP from 1995 to 2004 (IPEADATA).

36In sum, the increase in the wage share, contributing to the rise in private consumption and gross capital formation, was reconciled with the still-increasing financialisation of capital, thanks to a lower finance-domination of the accumulation regime. However, the persistence of financialisation marks the fragility of the distributive orientation of the growth regime. The domination by finance was only regulated, but not at all challenged fundamentally.

37From 1995 to 1999, there was an exchange rate band regime, in which the Central Bank established a range in which the rate could vary. This was a crawling peg regime. Then in January 1999, Brazil adopted a floating exchange rate system.

3. The macroeconomic dynamic and its reversal

38From 2003 to 2011, monetary and exchange rate policies drove a positive macroeconomic dynamic, attesting to a virtuous circle between the trends mentioned above, as shown in Figure 7 below. These policies were rooted in a doctrinal compromise between monetary orthodoxy and growing consumption that fitted the PT’s political agenda. This compromise was made public by the academic controversy between Finance Minister Guido Mantega (2006-2014) and the Central Bank of Brazil. Mantega adopted the new-developmental criticism of currency appreciation to counter the central bank’s neoclassical criticism of lax fiscal policy. What the central bank denounced as a non-cooperative game, forcing it to maintain high-interest rates that fuelled currency appreciation, was in fact the rhetoric of a doctrinal compromise that merely staged the appearance of non-cooperation. However, the doctrinal compromise was unsustainable and could not be projected into the long-term. Its expansionary impact remained dependent on a positive current account balance.

41Figure 7 sums up the causal relationships whereby choosing currency appreciation enabled a reduction in the policy interest rate (the SELIC). It helped lower the public debt to GDP ratio (despite growth in the stock of internal public debt), thus providing leeway for higher public expenditure. This process, combined with the new wage regime and upswings in consumer credit and residential mortgage lending, resulted in faster growth, raising tax returns.

42In January 2016, the proportion of the total stock of the Brazilian government’s public debt indexed to the SELIC interest rate corresponded to 86.4%. In May of 2017, it stood at 75%. With such high levels, it is understandable that public debt should grow in pace with the variation in the policy interest rate. What is worth noting is that the growing stock of public debt (Figure 8) played an ambivalent role in this virtuous circle. On the one hand, it helped refinance servicing of previously-issued debt. It thus lowered the level of the primary fiscal surplus, maximizing the capacity for public expenditure. The reduced strain of growing public indebtedness thanks to the expected appreciation of the real was an indirect lever of growth. This was the main feature of Brazil’s finance-led accumulation regime.

43Moreover, as the debt to GDP ratio was declining, growing public expenditure did not dent the credibility needed to attract foreign portfolio investments to offset the rising current account deficit from 2008. Yet there was a flipside: the increase in new debt issuance at interest rates far higher than prevailing international ones drove the increased financialisation rate. The result of this ambivalent situation was that the demand impulse allowed the productive accumulation rate to increase as it reduced idle capacities, but without triggering a sufficient surge in investment to enable a sustainable industrial surplus to emerge, and thus allow Brazil to escape from the reversal of this virtuous macroeconomic dynamic. We can argue that high returns on Treasury bonds, as they are quasi-liquid financial assets, strengthen the preference for liquidity to the detriment of productive investments. Probably, only those on a short-term horizon of growing anticipated demand occurred.

45The above analysis underscores the threat that the high servicing costs of public debt will grow again not only in absolute terms but as a ratio of GDP, in the future. Moreover, when US long-term interest rates begin to rise once more, this will force Brazil’s central bank to raise its policy interest rate even higher. Under the PT government, high international currency reserves allowed financing the growing public debt. But higher debt requires a higher fiscal surplus once tax revenues decline, as occurred in 2014. The issue of high reserves is crucial to the debate, as nobody knows the degree of dependence on foreign exchange reserves due to the deterioration in the current account. This worsening trend on reserves was reversed in 2015, but only thanks to the recession at the time. Still, in Figure 8, we may note that despite the downward trajectory of the policy interest rate from 2016 onwards (under the Temer and Bolsonaro governments), the accumulated impact of this real interest rate continued to grow, because of the compound capitalisation of this interest rate. This expresses the rentier and financial accumulation that has characterised the financialisation of the Brazilian economy.

4. The construction of liberal governmentality makes the central bank the pilot of the accumulation regime

46The above analysis of macroeconomic management accelerating growth reveals a fundamental mechanism of Brazil’s political economy which supported the PT government. However, the temporary nature of its positive growth impact also casts light on how “steering” the accumulation regime was the ultimate cause of the cycle’s reversal. The deterioration in the external and fiscal constraints was a direct by-product of the political decision not to cross the red line whereby the market sets exchange rates: a de facto independent central bank fixes the policy interest rate.

47The final macroeconomic outcome highlights the ambivalence of economic policy as a practical experiment in political economy, revealing the extent to which financial power has constrained the institutional changes implemented by the PT, and thus made the steering of the accumulation regime contradictory. This issue raises a crucial debate about the nature and mechanisms of institutional change, including the importance of knowledge regimes. On this precise matter, countering the interests of the financial rentier class was not even conceivable, if we regard the predominant macroeconomic knowledge regime as a cognitive framework for public action governing economic beliefs and behaviours. There could be no government coalition to oppose financial interests in concrete terms. As a presidential candidate, Lula therefore had to promise solemnly the continuity of orthodox economic policy.

48We share here the definition of Campbell and Pedersen (2014, p. 3): “Knowledge regimes are the organisational and institutional machinery that generates data, research, policy recommendations, and other ideas that influence public debate and policymaking”. Drawing on Boyer, these authors also hypothesise that the perception of economic performance linked to the knowledge regime is a fundamental factor of institutional change. The theory of government by instruments enables an analysis of public action in macroeconomics, leading to a hypothesis that the incremental nature of institutional change reveals transformative results with a more liberal than a developmental focus. This was the case because the central bank’s lengthy efforts to assemble an instrumental coalition prevented any heterodox theory from becoming a predominant knowledge regime. Furthermore, this analysis characterises a lock-in of institutional change in its political and cognitive dimension, eliminating the current government’s leeway for resuming efforts to transform public action in favour of a developmental State.

49The evolution in the visible knowledge regime, in what we have called a doctrinal compromise, shows the fully reversible nature of this reconciliation of conflicting interests as soon as recession or poor growth prospects weaken the regime. The question then becomes whether or not this trajectory can cause a change in economic policy acceptable in the eyes of policy-makers, chiefly the central bank. This would mean a move to another growth regime, no longer dominated by finance.

50“The people are not the problem; they are the solution,” was a famous statement by Lula. It is a rhetorical statement illustrating how, throughout the 2000s, there was a prevailing will by government to find a way to make the domestic market a growth driver again as part of an open economy with competitiveness gains. Nevertheless, the design of public activism also partly used several liberal policies from the toolbox recommended by international bodies (Ban, 2013). Recent research on this toolbox has suggested that the so-called neoliberal turn should be regarded more as a new form of economic governance than as a withdrawal of the State per se.

51This hypothesis is now commonly accepted in political economy, following Michel Foucault, who was undoubtedly the author to show first its analytical consequences for understanding power beyond the State. Foucault showed how economic liberalism, far more than just promoting market self-regulation, actually mobilises the knowledge of government that enables changes in mentalities and behaviours. Liberalism thus redefines the exercise of power according to what Foucault called governmentality (1991, 2004). His research is at the origin of understanding neoliberalism as the building of a new power model based on what has been called “government by instruments or government at a distance”. These instruments support socio-technical systems which transform the government’s intentions into the behaviour of the governed. The concepts used in this theory, along with the ideas of the political sociology of institutional change, have prompted us to hypothesise the existence of a twofold political and institutional obstacle (respectively the impossibility of an alternative coalition, and the autonomy of the central bank), blocking the construction of a “new-developmental” coalition.

52We assume that this blockage reflects the devices of governmentality prevailing in monetary and exchange rate policy. The macroeconomic analysis of Section 3 above suggests that the central bank was the organisational cornerstone since the trade-off between the policy interest rate and the exchange rate was the vector of a growth plan aimed at redefining the class structure of Brazilian society. The distributive compromise obtained by macroeconomic management expressed the Lula government’s intentionality. Its form was a doctrinal compromise. The macroeconomic analysis above has shown that it was not sustainable. The central bank’s power jeopardised this intentionality, and so guaranteed the incremental advancement of pure liberal governmentality. The rest of this section looks at the historical construction of the central bank’s power. Our analysis relies on a remarkable study by Carlos Santana (2012) which highlights the interaction between the consolidation of a deflationist coalition and the capacity of the State.

53Santana’s research describes the trajectory of this interaction as a two-stage incremental process. The first stage in forming this interaction was the Real Plan, named after the new currency instituted in 1994 by the Cardoso government (ibid., p. 40-49). This plan launched privatisations and set up a fixed exchange rate regime, ending the hyperinflation that had plagued the country for 15 years. Santana shows that the autonomy acquired by the Central Bank of Brazil came about because the State could act as a political entrepreneur. He succeeded in doing so “by taking advantage of the organisational weakness and lack of resources of a set of strategic actors to impose a solution where they all had to yield to a certain extent to end inflation”.

54This State capacity began institutionalising the insularity of economic policy-makers at the central bank. The second stage of the “disembeddedness” of monetary authorities consolidated this insularity, accentuating “the normalisation and bargaining capacity of the community of orthodox economists” (ibid., p. 57).

55Since the 1990s, this bargaining has mainly involved the remarkable predominance of the mobility circuit of economic decision-makers (holding PhDs in economics from North American universities) among the Brazil’s State organisations (chiefly the central bank) and private-sector banks (national or foreign), as well as the IMF and World Bank. The second channel for economists (possibly even trained by heterodox research centres) to be recruited as high-ranking decision-makers has been to climb the State bureaucracy. Santana’s survey showed that this recruitment channel was marginalised in the 1990s, but then regained substantial influence in the Ministry of Finance and the Brazilian Development Bank (BNDES).

56As for the normalisation capacity, Santana’s analysis describes setting policy interest rates using the inflation targeting technique. The description of the interaction between market actors and decision-makers characterises the socio-technical system constituted by this instrumentation of economic policy, as designed by mainstream economists. The level of interest rates should maintain price inflation at a level consistent with the NAIRU (non-accelerating inflation rate of unemployment), a concept rooted in monetarist theory. To achieve this objective, the central bank collects information on inflation expectations from the research departments of the leading financial institutions. Moreover, Santana’s press survey reveals that this particular power of the market over monetary policymaking is conveyed by the prominent roles of the chief economists of these institutions, as economic public opinion leaders, insofar as they maintain the “discourse authorised by science”, to borrow an expression from Bourdieu (1975).

57The orthodox turn that Dilma Rousseff implemented in 2014 demonstrated that the influential power of the orthodox epistemic community was still strong. Its status as an advocacy community enabled it to safeguard what might be called an “instrumental coalition”, to emphasise the autonomy that a system of government by instruments can acquire, thus taking on its historicity, as Baudot (2014) highlighted. Indeed, this appears to be the case for Brazil’s so-called economic policy “tripod” (inflation targeting, a primary budget surplus, and a floating exchange rate), which has remained in place since 1999.

58We can trace the institutional change regarding economic policy that constitutes this tripod back to the Real Plan. Inflation targeting was only a marginal adaptation of the instrumentation of economic policy, switching out of a fixed exchange rate regime as an inflation anchor and replacing it with interest rate management, whereby inflation-targeting warrants the exchange rate anchor. Our analysis shows that the government’s strategic instrumentation of economic ideas mainly matters. The institutional macroeconomic research above demonstrates the lack of political will to tackle the power of finance. In hindsight, this political decision appears to result from an incremental consolidation process for the liberal programme. Its continued application means a particular form of the progress of liberalism, causing the "conversion of institutions redirected to new goals, functions or purposes" (Streeck & Thelen, 2009). The autonomy granted to the central bank took the form of new functions attributed to it (apart from guaranteeing price stability), thus making the economic policy tripod irreversible. Hence, we can consider the instrumental coalition gathered by the central bank to be standing at the centre of a set of institutional arrangements, procedures, analyses and calculations, as well as tactics that together constitute governmentality (Foucault, 1991). This implies a political and cognitive lock-in of the developmental State’s constituency.

59Lastly, we must emphasise the unique interest of our analysis in terms of governmentality. In the political economy literature, the importance of a central bank’s de facto autonomy in financialisation seems established well enough as to be common knowledge. The governmentality approach has prompted us to deal with this question in parallel to constructivist institutionalism. The latter concept was set out especially Mark Blyth, one of one of the founders of this theory, and emphasises the causality of ideas giving content to agents’ interests in periods of crisis: during such periods, uncertainty reigns, resembling a knowledge void). Colin Hay compares this approach to historical institutionalism and to rational institutionalism (Hay, 2008, p. 17). He considers it is most appropriate for explaining institutional change “under imbalance conditions”. The governmentality analysis adds a valuable complement by shedding light on the State’s role as a critical site for the institutional integration of power technologies (Jessop, 2012).

60Therefore, the real effect of ideas is not necessarily to impose them on an entire population, thus ending uncertainty. We must consider this effect in comparison to the question of the government of populations associated with a set of instruments. The economic policy tripod was the instrument that was a critical marker defining the profound logic of the PT’s governmental policy, alongside the heterodox aspects of policy. The issue here is not whether the entire population adhered to this interpretive framework for the economy. Instead, the concept of governmentality contains a mix of ideas and governance instruments (Foucault, 2001, p. 299), and in the case of Brazil, this mix generated two potential contradictory types of governmentality: neo-developmentalism and liberal governmentality. This tension resulted in a non-viable macroeconomic trend, as we have seen. The economic downturn merely revealed the limits of the PT’s unique adaptation of liberal governance. It proved a failure, not due to the hybrid nature of the economic policy regime (Ban, 2013) per se, but due to the solidity of the instrumental coalition as the device of central government. The analysis in terms of governmentality establishes the economic conditions for the processes of variation, selection, and retention of techniques of government by the State (Jessop, 2012). The autonomy of the central bank of Brazil fits perfectly into this perspective. Its power was used as a government strategy to break the inflation, imposed by the price-makers. However, this initiative made it possible to introduce the interests of public debt-holders and, more broadly, of creditors into the heart of government action, by instituting the mechanism for setting its interest rate, based on the inflation expectations of the leading banks. It was not just a monetary policy device. It was an instrument of remote government.

61This analysis thus leads us to conclude that we can understand better the shift from success to catastrophe in terms of the relationship between government (associated with a knowledge regime and a set of instruments of power), an instrumental coalition of interests, and economic processes. In this respect, we may consider that the predominant knowledge regime contributed to the correlation of forces favourable to the rentiers. Viewed in this perspective, the weight of liberal doctrine answers the question raised by the analysis of the ruling blocs as to why the industrialists rallied to denounce Dilma’s economic policy, which they had initially endorsed (Singer, 2015).

5. Conclusion: the governmentality of “liberal neo-developmentalism” is unable to overcome the contradictions of a financialised accumulation regime

62A primary outcome of this analysis is that the likely continuity of an instrumental coalition favouring financial rentiers implies financialisation of capital at an accelerating pace, correlated with decreasing productive investment. This enshrines a persistent lack of competitiveness and maintains low growth.

63The central bank played a decisive role in consolidating a pattern of international integration that prioritises financial accumulation to the detriment of industrial accumulation. This political choice jeopardises the implementation of the new developmental proposal. From a Keynesian perspective, the recession during the second mandate of Dilma Rousseff (2014-2016) was the result of a shift in economic policy, curbing the expansion of lending to households and public expenditure, halting the rise of the minimum wage, and thereby stopping domestic demand growth (Serrano & Summa, 2015).

64In the light of our analysis, this new commitment means that the priority became to keep Brazil’s current account deficits and the public debt within manageable limits. Again, we should emphasise that industrial accumulation was insufficient, no matter how much it increased a growing part of the overall profit volume (including financial returns) allocated to productive investment. The government’s decision to break the trend in demand growth responded to the imperative to demonstrate its commitment to reining in public debt and, more precisely, to pursuing this commitment with conventional means. The central bank assembled a virtual coalition that we can describe as instrumental, to the extent that it convinced the government to adopt (authorise) “scientific instruments” that would supposedly guarantee a return to growth, namely a fiscal squeeze and structural reforms. This cognitive dimension of the political economy results from an unshaken institutional commitment inherited from Cardoso’s liberal government.

65This political-institutional legacy confirms the incremental dynamics of the neoliberal institutional changes as a lock-in of any possibility for building a regulatory framework conducive to the new developmental proposal. Demand-led growth is unsustainable in a finance-dominated accumulation regime. Demand-led growth regime arises when the expansion of aggregate demand promotes investment and economic growth. In this context, domestic consumption may grow due to real average wage growth or income distribution measures favouring the salaried population. The de facto autonomy of the central bank ensured the permanence of the financialisation of capital. This particular governmentality device jeopardised the sustainability of the institutional changes by which Lula’s government stimulated demand growth (mainly through growing wages, social benefits, and lending). However, the central bank consolidated the neoliberal formatting of a finance-dominated growth regime, which led to finance blocking the growth regime. Therefore, the stagnation at 0.3% over the 2012-2019 period, with a 4.9% drop in investment, and a 0.9% drop in household consumption (see Figure 5), appeared to be the price of a structural incompatibility between the pure liberal governmentality’s device (that represented by the “tripod” model) and the intentionality of a new government of the population.

66This incompatibility was hidden by what could be called a “liberal neo-developmental governmentality”, which combined a Keynesian and neoclassical corpus. Nevertheless, the result could not be a sustainable finance-led accumulation regime. A growing financialisation rate of capital (due to an overvalued currency) was incompatible with progress in Brazil’s economic specialisation. However, this is not only an issue of pure macroeconomic restriction. The main result of this analysis is to show that macroeconomic dynamics are also a question of power, mediated by governmentality devices. It is a fundamental dimension of the fragility of Brazil’s accumulation regime, which is blocked by finance. The solvency of public debt defines its zone of stability and acts as a stagnation factor correlated with the lack of industrial competitiveness. This analysis has shown that the instrumental coalition built by the central bank had acquired a kind of veto power over the government’s political project. This finding raises a new question about the political impact of dominant economic ideas. More specifically, it may be asked to what extent the dominance of such thinking and of the principal stakeholder – the Central Bank of Brazil – were accountable for the loss of support for the Workers’ Party through the impeachment of President Dilma Rousseff, which she herself denounced as an institutional coup d’état.