The monetary and fiscal history of Brazil, 1960-2016

Ayres, Joao; Garcia, Márcio Gomes Pinto; Guillen, Diogo; Kehoe, Patrick J. Working Paper The monetary and fiscal history of Brazil, 1960-2016 IDB Working Paper Series, No. IDB-WP-990 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Ayres, Joao; Garcia, Márcio Gomes Pinto; Guillen, Diogo; Kehoe, Patrick J. (2019) : The monetary and fiscal history of Brazil, 1960-2016, IDB Working Paper Series, No. IDB-WP-990, Inter-American Development Bank (IDB), Washington, DC, https://doi.org/10.18235/0001672 This Version is available at: https://hdl.handle.net/10419/208175 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/igo/legalcode The Monetary and Fiscal History of Brazil, 1960-2016 João Ayres Marcio Garcia Diogo Guillen Patrick Kehoe IDB WORKING PAPER SERIES Nº IDB-WP-990 A pril 2019 Department of Research and Chief Economist Inter-American Development Bank A pril 2019 The Monetary and Fiscal History of Brazil, 1960-2016 João Ayres* Marcio Garcia** Diogo Guillen*** Patrick Kehoe**** * Inter-American Development Bank ** Pontifical Catholic University of Rio de Janeiro (PUC-Rio), CNPq, and FAPERJ *** Itau-Unibanco Asset Management **** Stanford University, Federal Reserve Bank of Minneapolis, and University College London Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library The monetary and fiscal history of Brazil, 1960-2016 / João Ayres, Marcio Garcia, Diogo Guillen, Patrick Kehoe. p. cm. — (IDB Working Paper Series ; 990) Includes bibliographic references. 1. Inflation (Finance)-Brazil. 2. Monetary policy-Brazil. 3. Fiscal policy-Brazil. 4. Debts, Public-Brazil. I. Ayres, João. II. Garcia, Marcio. III. Guillen, Diogo. IV. Kehoe, Patrick J. V. Inter-American Development Bank. Department of Research and Chief Economist. VI. Series. IDB-WP-990 Copyright © Inter-American Development Bank. This work is licensed under a Creative Commons IGO 3.0 AttributionNonCommercial-NoDerivatives (CC-IGO BY-NC-ND 3.0 IGO) license (http://creativecommons.org/licenses/by-nc-nd/3.0/igo/ legalcode) and may be reproduced with attribution to the IDB and for any non-commercial purpose, as provided below. No derivative work is allowed. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the UNCITRAL rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-IGO license. Following a peer review process, and with previous written consent by the Inter-American Development Bank (IDB), a revised version of this work may also be reproduced in any academic journal, including those indexed by the American Economic Association's EconLit, provided that the IDB is credited and that the author(s) receive no income from the publication. Therefore, the restriction to receive income from such publication shall only extend to the publication's author(s). With regard to such restriction, in case of any inconsistency between the Creative Commons IGO 3.0 Attribution-NonCommercial-NoDerivatives license and these statements, the latter shall prevail. Note that link provided above includes additional terms and conditions of the license. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. http://www.iadb.org 2019 Abstract* Brazil has had a long period of high inflation. It peaked at around 100 percent per year in 1964, decreased until the first oil shock (1973), but accelerated again after ward, reaching levels above 100 percent on average between 1980 and 1994. This last period coincided with severe balance of payments problems and economic stagnation that followed the external debt crisis in the early 1980s. This paper shows that the high-inflation period (1960-1994) was characterized by a combination of fiscal deficits, passive monetary policy, and constraints on debt financing. The transition to the low-inflation period (1995-2016) was characterized by improvements in all of these features, but it did not lead to significant improvements in economic growth. In addition, this paper documents a strong positive correlation between inflation rates and seigniorage revenues, although inflation rates are relatively high for modest levels of seigniorage revenues. Finally, this paper discusses the role of the weak institutional framework surrounding the fiscal and monetary authorities and the role of monetary passiveness and inflation indexation in accounting for the unique features of inflation dynamics in Brazil. JEL classifications: E42, E63, H62, H63 Keywords: Brazil’s hyperinflation, Stabilization plans, Fiscal deficits * This is a chapter in the book The Monetary and Fiscal History of Latin America, forthcoming in 2019 from the University of Chicago Press. We would like to thank Marcelo Abreu, Pérsio Arida, Edmar Bacha, Marco Bassetto, Tiago Berriel, Afonso Bevilaqua, Amaury Bier, Claudio Considera, Gustavo Franco, Fabio Giambiagi, Claudio Jaloretto, Joaquim Levy, Eduardo Loyo, Timothy Kehoe, Ana Maria Jul, Randy Kroszner, Pedro Malan, Rodolfo Manuelli, Andy Neumeyer, Juan Pablo Nicolini, Affonso Pastore, Murilo Portugal, Thomas Sargent, Teresa TerMinassian, José Scheinkman, Rogério Werneck, and participants at the workshops held in the University of Chicago, LACEA-LAMES 2017, PUC-Rio, Central Bank of Chile, and Inter-American Development Bank. Marcio Garcia coordinated this project. 1 Introduction This paper presents the monetary and fiscal history of Brazil between 1960 and 2016, with emphasis on episodes of hyperinflation. It describes the evolution of the Brazilian monetary and fiscal policy institutions and how they relate to episodes of macroeconomic instability and growth, focusing on the high-inflation period (pre-1994) and two stabilization plans: the Government Economic Action Plan (PAEG, an abbreviation for Plano de A¸c˜ao Econˆomica do Governo) and the Real Plan. The PAEG, in 1964, stabilized inflation around 100 percent per year, whereas the Real Plan, in 1994, stabilized inflation around 90 percent per month after six failed attempts in over a decade. The analysis follows the conceptual framework in Chapter 2 of The Monetary and Fiscal History of Latin America (henceforth MFHLA) by focusing on the government budget constraint. A summary of the period is illustrated in Figure 1, which shows the evolution of real GDP per capita, inflation, and government deficits for the 1960–2016 period.1Three subperiods are identified: (i) 1960–1980: fast economic growth with high inflation and moderate deficits; (ii) 1981–1994: slow growth with hyperinflation and high deficits; and (iii) 1995– 2016: moderate growth with low inflation and low deficits.2The 1981–1994 subperiod stands out not only for its poor growth performance and hyperinflation but also for severe balance of payments problems, a common feature among highly indebted Latin American countries affected by the increase in international interest rates and the slowdown in international economic growth. When relating the episodes of macroeconomic instability to the government fiscal and monetary policies, we observe the following: (i) both stabilization plans, PAEG in 1964 and the Real Plan in 1994, included measures to improve fiscal balances and were followed by increased access to debt financing; (ii) the government policy to increase public investment in the wake of the first oil shock in 1973 explains the rapid increase in external debt that preceded the external debt crisis of 1983 seen in Figure 2; and (iii) the high-inflation periods (pre-1994) were characterized by the combination of fiscal deficits, passive monetary policy, and constraints on debt financing, while the transition to the low-inflation period (1995– 2016) was associated with improvements in government fiscal balances and higher de facto independence of the monetary authority (as of this writing, Brazil still lacks a formally independent central bank), as well as much greater access to debt financing. In comparison to other Latin American countries, the following two characteristics make 1Appendix Cdiscusses the data and methodology. Our definition of “deficit” is the primary deficit plus real interest payments on debt discounting for real GDP growth (see Chapter 2 of MFHLA), and throughout the chapter, we use the General Price Index from the Getulio Vargas Foundation, IGP-DI, as our benchmark. 2One must bear in mind that the quality of fiscal statistics decreases as we move back in time. 2 the Brazilian experience unique: (i) a long period of high inflation, with annual inflation rates above 100 percent between 1980 and 1994; and (ii) modest levels of deficits for very high underlying inflation rates. We discuss two features that may explain these unique characteristics of Brazilian hyperinflation. The first is a poor institutional framework in which other public entities besides the monetary authority had indirect control over money issuance. We discuss that framework in Section 4.1. The second is the combination of a high degree of indexation in the economy to past inflation with a passive monetary policy.3 Together, these features created what was called at the time inflation inertia, which could explain why Brazilian hyperinflation was a much more protracted process than elsewhere and gave many the illusion that it could be cured without major improvements in the fiscal stance. We discuss that factor in Section 4.2. This paper is organized as follows. In Section 2, we present a summary of the government budget constraint, and in Section 3, we provide a historical description of each of the subperiods 1960–1980, 1981–1994, and 1995–2016. In Section 4we discuss the evolution of the institutional framework involving both fiscal and monetary authorities and the genesis of inflation inertia, and in Section 5, we present our final remarks and conclusion. 2 The Government Budget Constraint We are interested in analyzing the evolution of the government budget constraint for Brazil in the period 1960–2016. We attempt to match the variation in stocks (debt figures) with flows (fiscal deficits), duly accounting for valuation effects.4Table 1presents a summary of the results. In order to finance interest payments and primary deficits, the government can either issue domestic and external debt or issue money and receive seigniorage revenues. Transfers account for the residual.5 We divide the 1960–1980 subperiod into three parts: 1960–1964, 1965–1972, and 1973– 1980. In 1960–1964, markets for government debt securities were still underdeveloped, and the government faced restrictions on both domestic and external debt financing. Interest payments were low but primary deficits were on the rise, and had to be financed with seigniorage revenues. In 1964–1967, the PAEG stabilization plan implemented both fiscal and financial reforms, which reduced primary deficits and allowed the government to issue domestic debt securities. Those reforms account for the increase in domestic debt financing 3Most prices, wages, taxes, and the exchange rate were indexed to past inflation. 4Mainly the effect of devaluations on foreign-currency-denominated or indexed debt. 5The sums of primary deficits and transfers is close to the measure of the primary deficit reported by the Central Bank of Brazil starting in 1985, which is based on public-sector borrowing requirements and is usually referred to as the primary deficit below the line. See Appendix C. 3 and the reduction in seigniorage revenues that we observe in the 1965–1972 period. In 1973– 1980, on the other hand, we observe a rise in both debt financing in external markets and seigniorage revenues, which were associated with higher interest payments on external debt and a significant rise in transfers, the residual.6Fortunately, in this case, we can explain what most of these transfers are. In the wake of the first oil crisis of 1973, the government implemented policies that aimed at boosting investment through external borrowing, and that was done mainly through state-owned enterprises (SOEs). The debt series that was used to compute the government budget constraint includes SOEs, but the primary deficit series does not. The increase in investment by SOEs, documented in Table 2, accounts for a large share of the increase in transfers.7Therefore, we argue that deficits at the time are better represented by adding the transfers to the reported primary deficits, which is reported in row 9 of Table 1, labeled “primary deficits + transfers.”8 In 1981–1994, debt financing in external markets decreased sharply, and interest payments on external debt increased as a reflection of the debt crisis that followed the increase in international interest rates. In that period, seigniorage revenues were used to finance the payments of both principal and interest of the external debt as well as the primary deficits. The 1995–2002 period followed the agreement on the external debt renegotiations and the end of the hyperinflation in 1994. It showed a significant reduction in seigniorage revenues and a large improvement in primary and fiscal balances. Interest payments on external debt decreased, whereas interest payments on domestic debt increased. The 2003–2011 period continued to show primary and fiscal surpluses and low seigniorage revenues, and the external debt was replaced by domestic debt. As we will discuss, the pattern that we observe in 1994–2011 reflects changes in both monetary and fiscal policy institutions, with higher de facto independence of the central bank and greater control over the government budget. However, in the most recent period, 2012–2016, we observe a deterioration in fiscal balances that have been financed by a rapid increase in domestic debt. In the sections that follow, we provide a detailed historical background that describes the fiscal and monetary policies adopted in the 1960–2016 period and that account for the evolution of the government budget constraint. 6Interest payments might be negative because we are discounting for growth rates in real GDP and for the monetary correction of the debt. 7According to Werneck (1991), the average capital expenditures of SOEs for the 1973–1980 period made up 7.4 percent of GDP, so our figures might be underestimating their importance. Nevertheless, both sources of data indicate a rapid increase in investments by SOEs in that period. 8By doing so, we approximate what the Central Bank of Brazil has done in its fiscal statistics starting in 1985. See Appendix C. 5 3 Historical Description 3.1 1960–1980: Fast Growth with Macroeconomic Instability Brazil went through important transformations during the first subperiod of our analysis. It moved from being a rural society, in which 55 percent of the population lived in rural areas, to an urban society, with 68 percent of the population living in cities. Its production structure shifted toward the manufacturing sector, which increased its share of GDP from 32 to 41 percent, while the agricultural sector saw its share of GDP decline from 18 to 10 percent.9 It was a period of fast economic growth, with real GDP per capita increasing 4.6 percent per year on average (Figure 1a). However, it was also a period of macroeconomic instability, with a deep recession in the early 1960s, increasing external indebtedness following the first oil crisis in 1973, and nominal instability. Inflation rates rose in the beginning and reached levels around 100 percent in 1964, when the PAEG stabilization plan was implemented after a military coup.10 Inflation rates fell significantly but started to accelerate again around the first oil crisis in 1973, returning to three-digit levels in 1980.11 To understand the fiscal and monetary policy institutions that were in place during these years, one should note that it was a period of heated debate regarding the role of the state in promoting economic development, during which the government undertook major national development plans, such as the Targets Plan in 1956–1961, the National Development Plan I in 1972–1974, and the National Development Plan II in 1975–1979. That process also led to a surge in the number of public banks, with nine out of 23 states creating their own banks between 1960 and 1964 to finance their fiscal deficits, and to the creation of some of the largest Brazilian SOEs, such as Eletrobras in 1962 and Telebras in 1972.12 As we discuss below, they would all play an important role in explaining the dynamics of the government budget constraint. 3.1.1 1960–1964 Before 1964, the government Treasury had direct control over money issuance through the Bank of Brazil (BB), which was both the bank of the government and a commercial bank.13 The main monetary policy instruments in use were the control over the monetary base, 9Data from the Brazilian Institute of Geography and Statistics (IBGE). 10The military dictatorship would last until 1985. 11For thorough analyses of that period, we refer to Orenstein and Sochaczewski (2014), Mesquita (2014), Lago (2014), and Carneiro (2014). 12The other well-known Brazilian SOEs, Companhia Sider´urgica Nacional (CSN), Companhia Vale do Rio Doce, and Petrobras, had been created in 1941, 1942, and 1953, respectively. 13Section 4.1 discusses at length the institutional framework in which the central bank operated. 6 readjustment cycle. Moreover, unemployment benefits were introduced, and the minimum wage was raised by 8 percent in real terms. The exchange rate regime also changed, with the domestic currency now pegged to the US dollar. Fiscal and monetary policies were put under the discretion of the policymakers, but there was an important change: the end of the Conta de Movimento between the central bank and the Bank of Brazil. In practice, however, that only took place after 1988 because another account between the central bank and Bank of Brazil, Conta de Suprimentos Especiais, replaced Conta de Movimento until its elimination in 1988 (see Section 4). Another important measure was the creation of the National Treasury Secretariat, which would take control over both the administration of the domestic public debt and the government budget.27 At first, the Cruzado Plan was very successful in reducing inflation. The average monthly inflation from March to July of 1986 was 0.9 percent (11 percent per year). Moreover, the claim to freeze prices had a civic impact since the population was encouraged to “audit” prices; but that led to overheating. Sales increased 23 percent in the first six months of 1986 compared to the first six months of 1985, and real wages increased 14 percent from March to September of 1986 (Figure 18). One story that is consistent with such evidence is that even though prices were not allowed to change, equilibrium prices were increasing, which produced overheating since posted prices were too low. Therefore, production increased to meet the higher demand in the beginning, but then production decreased and stores started to run out of stock. Meanwhile, the Central Bank of Brazil tried to keep interest rates low to induce low expectations. One huge imbalance was the inconsistency of the plan for inflation and the monetary base: the monetary base was increasing much faster than inflation itself. In July 1986, the government implemented a timid fiscal package, Cruzadinho, focusing on increasing government revenues. But in reality, Cruzadinho had the opposite result form what policymakers expected. Expecting prices to be allowed to change again, demand increased and the overheating problem became even more dramatic. Inflation remained low, but it was not truly representative because products were scarce. Because of high demand, imports kept increasing while exports declined (Figure 19), thereby exacerbating the trade deficit. A rumor of a large devaluation in the near future reinforced that pattern. This expectation lead to a postponement of exports and an acceleration of imports, which increased the problems with the balance of payments.28 Facing all these challenges, in November 1986, the government opted for a fiscal plan, Cruzado II, trying to increase revenues through the readjustment of some public prices and some indirect taxes, which led to a high inflationary 27Before that, the Central Bank of Brazil managed both the domestic and external public debt, which included issuances, amortizations, and interest payments. 28The government kept the mini-devaluations based on an indicator of the exchange rate–wage (crawlingpeg) ratio. However, this same indicator was suggesting that the exchange rate was appreciated. 13 shock. Once again, the environment was one of high inflation (17 percent per month in January 1987). Meanwhile, the external crisis was just getting worse. In February 1987, the government suspended interest payments on external debt for an indeterminate amount of time (Figure 16). The idea was to stop the losses of international reserves and to start a new phase of the renegotiation of the debt with the support of the population. Bresser Plan: In July 1987, the government implemented the Bresser Plan, named after Finance Minister Luiz Carlos Bresser-Pereira. It was presented as a hybrid plan, with fiscal and monetary policies as well as features intended to address inflation inertia. As in the Cruzado Plan, prices were frozen. As usual, the moment in which the price freeze took place was important because the relative prices would remain stuck and possibly off-equilibrium. Trying to obtain a better result than the Cruzado Plan in this aspect, after the price freeze the government increased the prices of public services and some administered prices to correct for misalignments in relative prices. The elimination of the automatic trigger in wage resetting if inflation surpassed a 20 percent threshold was perceived as another improvement. But the economic team created another kind of wage indexation, the URP (Price Reference Unit), in which in each quarter the government would specify the readjustment for the next three months based on the average inflation of the period. This would keep a monthly readjustment, but a gap would remain between the readjustment and current inflation. In contrast to the Cruzado Plan, monetary and fiscal policies were active. Real interest rates remained positive in the short term. In the fiscal policy arena, the government aimed to reduce the operational deficit from the expected 6.7 to 3.5 percent of GDP.29 The plan did not address (or seek to address) the consequences of the previous default with creditors. Another interesting aspect of this plan is that it did not target zero inflation, but was instead meant to be only a deflationary shock. Bresser-Pereira’s main purpose was to introduce a fiscal reform to reduce inflation. However, the reform was not successful. In 1987, the deficit was much higher than promised. Unlike the Cruzado Plan, which had popular support, the Bresser Plan was not popular, and in February 1988, some liberalization of prices took place, reducing the effectiveness of the price freeze. A third problem of the plan was that it led to a fall in gross fixed capital formation. After Minister Bresser-Pereira left, Ma´ılson da N´obrega, the second in command, took his position. In January 1988, the government adopted an economic policy referred to as the Feij˜ao-com-Arroz policy, which can be translated to English as the “Black-Beans-and-Rice” 29At the time, the government used the public-sector borrowing requirement as a measure of the nominal deficit. However, nominal deficits were very high because of the monetary correction of the value of the debt. In order to overcome that, the operational deficit was adopted as the main deficit measure, which included only the nominal value of real interest payments. See Appendix C. 14 policy.30 Instead of freezing prices, its target was merely to keep inflation at 15 percent per month. The deficit was expected to reach 7 to 8 percent of GDP in 1988, and there was a temporary freeze of public-sector wages to reduce it. At first, this policy succeeded in avoiding an inflationary explosion, and the fiscal stance improved. The default on external debt was suspended, and the government started negotiations with external creditors. However, inflation started rising again, and the target of 15 percent per month was not achieved in the second quarter of 1988. In October 1988, a new constitution was enacted. The new constitution increased fiscal expenditures, reduced the flexibility of expenditure switching between fiscal accounts, and substantially increased labor costs. It did so by increasing expenditures and increasing the transfers from the central government to states without transferring the corresponding responsibilities, which induced an increase in the deficit of the central government. To put this into perspective, 92 percent of revenues were earmarked; that is, revenues from different sources were dedicated to specific programs or purposes (or both), reducing the flexibility of fiscal policies. In addition, the new constitution reduced the standard workweek from 48 to 44 hours and increased the cost of overtime. Summer Plan: The government implemented the Summer Plan in January 1989. Again, it was a hybrid plan, but the debate on the need for changes in fiscal and monetary policies was increasing. Like previous plans, it included a component of price freezing as well as the adoption of a nominal anchor. In this case, a fixed exchange rate (1 cruzado novo = 1,000 cruzados = US$1) was implemented for an indefinite time. Moreover, an attempt was made to end inflation indexation. On the fiscal and monetary side, the plan was to adopt a tight monetary policy and to fight inflation by controlling the public deficit. It intended to control expenditures and increase revenues through the privatization of publicly owned assets and a reduction in the wage bill of the public sector. Overall, the plan seemed to incorporate everything that was missing in the previous plans. Although it kept a heterodox flavor, it was mostly an orthodox plan aiming to reduce subsidies, close public firms, and fire public employees, with a deindexation plan that was sort of a small default. However, the government did not have the political power to carry it through. Without Congress, privatizations and other unpopular measures, such as the closing of public firms, were canceled. In the end, the reforms were not implemented. Moreover, the tight monetary policy put interest rates at high levels and increased the fiscal deficit of the government. With low credibility and a reform that did not go through, inflation 30The policy name reflects the meaning of black beans and rice in Brazilian culture. It is the dish that Brazilians eat every day. It is considered to be neither very interesting nor complicated, but it does the job of providing a healthy meal. 15 accelerated, and the Summer Plan also failed. The 1980s ended with inflation rates of about 70 percent a month and with almost 100 percent of the federal bond debt being rolled over in the form of zero-duration bonds.31 This state of affairs reflected not only the extremely high uncertainty regarding inflation and interest rates but also the fear of an explicit default of the public debt by the incoming administration, headed by President Fernando Collor de Mello. At the time, the credit risk of the public securities was clouded with widespread suspicion, which was indeed validated by the new administration’s actions. Collor de Mello was elected president of Brazil after 29 years of either indirect or undemocratic elections. The very day he took office, he launched the first Collor Plan. Collor Plan I: In March 1990, the government launched Collor Plan I. Prices and wages were frozen. The plan recognized that a reduction in deficits was necessary to end the hyperinflation, and it implemented both temporary and permanent fiscal policies. Among the temporary measures were the establishment of a tax on financial intermediation and the suspension of tax incentives. But the permanent policies were more important. An effort was made to reduce fiscal evasion (one of the president’s trademarks during the presidential campaign) and increase taxes. Other major components included privatizations and an administrative reform. However, that plan became famous for its (controversial) monetary policy. In an attempt to reduce the money supply, the government confiscated deposits in both transaction and savings accounts for a period of eighteen months.32 Those resources amounted to 80 percent of bank deposits and financial investments, which would be held at the Central Bank of Brazil and invested in federal government bonds. These resources were remunerated while they were kept at the central bank, but their rates of return were decided by the government itself and therefore were subject to partial defaults. Following the plan’s implementation, monetary aggregates fell sharply, especially the higher ones (Figure 20), and real GDP per capita contracted by 5.7 percent in 1990. This reduction in liquidity, however, was not sufficient to control inflation. Regarding the fiscal reform, the government?s threatening behavior toward public-sector employees made the reform very unpopular. The plan encountered extensive resistance, and in the end, it could not deliver on its promises. While some privatizations succeeded, most of its reforms were short-lived. Collor Plan II: In January 1991, the same government implemented the Collor Plan 31Zero-duration bonds are bonds that pay ex post the accrual of daily overnight interest rates. Therefore, the price of these bonds is insensitive to interest rate changes. It was a way to separate interest rate risk from maturity risk, thereby somewhat lengthening the very short-term public debt. 32While the government confiscated the amounts exceeding $50,000 cruzados novos, those resources actually became available before the end of the 18-month period, as Figure 20 shows. 16 II. Like its predecessor, this plan aimed to reduce government expenditures by firing civil servants and closing public services. It also proposed the privatization of state-owned enterprises. As usual, the plan included some price freezes. Wages were converted to a 12-month average, a new tablita was adopted based on the assumption that inflation would fall to zero, and the plan put an end to indexation.33 Not entirely related to the fight against inflation, this plan included as one of its motifs the need for Brazil to improve the quality of its products. In the words of the president, Brazil was producing horse-drawn coaches instead of cars. To achieve that goal, the government opened the Brazilian economy to foreign competition and privatized state-owned firms. Following the plan’s implementation, the country experienced a recession. The subsequent recovery, however, is usually attributed to enhanced competition in the economy. Inflation ended up rising again, but this plan did make two important permanent changes. First, it opened up the Brazilian economy and expanded trade, reversing the previous trend (Figure 19). Second, it increased productivity. In the beginning of 1992, when expectations of accelerated inflation did not materialize, the effects of the recovery on investors’ confidence started to show up in public debt markets. Those expectations had been based on the combination of price liberalization, corrections of public tariffs, and the devaluation that followed the floating of the exchange rate in October 1991, in face of the strong monetization of the hijacked assets during Collor Plan I.34 The return of investors’ confidence is also confirmed by the recovery of foreign exchange reserves after 1992. Following the high political turbulence that characterized the months preceding the impeachment of President Collor de Mello (October 2, 1992), the beginning of Itamar Franco’s presidency was marked by high uncertainty concerning economic policy. Proposals of another moratorium, and even repudiation of the public debt, were constantly in the press. It was only after the president nominated Fernando Henrique Cardoso, his fourth minister of finance in less than six months, that regained confidence led to higher external reserves. Real Plan: In February 1994, the government launched its last stabilization plan, the Real Plan, which would finally put an end to the hyperinflation. The plan that conquered Brazilian inflation did not have the blessing of the IMF, an always troubled relationship in previous decades. The plan’s concepts were different from those of its predecessors: it aimed to reduce deficits, modernize firms, and reduce the distortions that arose from previous price 33Tablita was the name for the interest rate conversion table when the currency changed. 34The recovery of the stock of public debt in the portfolio of the private sector was a clear demonstration that asset holders were willing to return to business as usual in spite of the disruptions of repeated interventions that had been made in the rules of indexation and the liquidity of public securities during the previous 12 years. One should bear in mind that the majority of economic analysts at the time were forecasting that the government would never again be able to place new debt. 17 freezes. Moreover, unlike previous plans it was planned in advance, with several measures being taken before its official announcement. Its first stage started in June 1993, when the government launched the Programa de A¸c˜ao Imediata (Program for Immediate Action), designed to focus on fiscal imbalances that would arise when the seigniorage revenues fell.35 It included an increase in existing tax rates, such as income tax, the creation of new taxes, such as the tax on financial intermediation, and the renegotiation of subnational government debt in an attempt to control the deficits of subnational governments.36 Another fiscal adjustment came in the beginning of 1994, with the Fundo Social de Emergˆencia (Emergency Social Fund), a way to suspend part of the earmarked revenues of states and municipalities, providing more flexibility in the government budget. On the monetary side, a clearly stated intention to limit issuances of the new currency led to the adoption of a high interest rate policy and high reserve requirement ratios (100 percent reserve requirements on new deposits after July 1, 1994). In addition, the plan included changes in the institutional framework in which the central bank operated, such as the transfer of management of the external debt to the National Treasury Secretariat and the reduction of the size and duties of the National Monetary Council. Section 4discusses those changes in more detail. On top of all these measures, the government also reached an agreement on its external debt renegotiations under the Brady Plan, and in March 1994 its defaulted debt was securitized and the country regained access to international capital markets.37 The Real Plan did not involve price freezes itself, but it was able to solve the problems of staggered wages and prices. Actually, this was considered the most controversial aspect of the plan but ended up being very successful. The creation of a new unit of account, the URV— Unidade Real de Valor (Unit of Real Value)—aimed at establishing a parallel unit of value to the cruzeiro real, the inflated currency. The idea was to make the unit temporary. Prices were quoted in both URVs and cruzeiros reais, but payments had to be made exclusively in cruzeiros reais. The URV worked like a shadow currency that had its parity to cruzeiro real constantly adjusted, since it was one-to-one with the dollar. Therefore, a conversion rate of the URV/cruzeiro novo (the old currency) was set every day, and many conversions were left to free negotiation between economic agents, while the government interfered largely in oligopolized prices. This approach would allow agents to observe the low inflation of the parallel currency, therefore breaking the expectations of high inflation once the currency was changed. The URV was created in February 1994 when the Real Plan was officially launched, 35The program was announced in June 1993, but many of the reforms were implemented later in that year or even in 1994. 36See Portugal (2017) for a discussion of the fiscal adjustments that were adopted in that period. 37The agreement was signed in November 1993. 18 and in May 1994 the government announced that the real would become the new currency in July 1994. The government kept its plan, and the URV was eliminated on July 1, 1994, when it was converted to the new currency, the real, at the rate 1 dollar = 1 real = 1 URV = 2,750 cruzeiros reais, followed by the adoption of a crawling-peg regime. After the currency conversion was implemented, inflation rates fell significantly, and the hyperinflation period in Brazil came to an end. Figure 21 illustrates how the increasing primary surplus since 1993 allowed the government to reduce its seigniorage revenues after the Real Plan was implemented. The drop in seigniorage revenues was associated with a reduction in both inflation and money growth rates. In Table 3we use monthly data to compute the government budget constraint around the time the Real Plan was implemented. Note that while the primary surplus increased in May/93–May/94, the government was able to increase foreign reserves by 6.3 percent of GDP (which explains the negative 6.3 percent of GDP relative to net external debt in Table 3). Then, between May/93–May/94 and May/94–May/95, seigniorage revenues fell from an average of 2.7 percent to only 0.7 percent of GDP, and that was possible because of the increase in primary surplus, from 2.8 to 4.0 percent of GDP, as a result of the fiscal measures described above. However, in May/95– May/96, after inflation was under control and seigniorage revenues fell, the government showed a deterioration in its primary balance that would be reversed in subsequent years. That fiscal deterioration was financed through domestic debt issuance, which shows that the credibility of the reforms played an important role, as it allowed the government to keep seigniorage revenues at low levels. In addition, the increase in real money balances following the Real Plan also contributed to increasing the financing options of the government (Figure 22). Lastly, Brazil started to accumulate external debt again after 1994, but that was done by the private sector (or by public entities that were not included in the fiscal and debt statistics), which borrowed from abroad and concurrently financed the government. That accounts for the current account deficits after 1994 (Figure 14), explained by lower trade balances as a result of increasing imports (Figure 19). Here, it is important to mention that the analysis above used the official primary deficits plus transfers as the benchmark measure of government primary deficits. We did so because there are large discrepancies in fiscal statistics around the time the Real Plan was implemented.38 If instead we considered only the primary deficits from government accounts, without the transfers, we would observe a transition from large primary deficits to large primary surplus upon the implementation of the Real Plan, without the subsequent deterioration in primary balances mentioned above. We discuss that in Appendix C. We were not 38One advantage is that it allows us to make the analysis using monthly data because the official fiscal statistics covering the national public sector are reported at an annual frequency. 19 able to explain the differences between both series. We chose the primary deficits plus transfers as our benchmark measure because it is closer to the primary deficit series reported by the Central Bank of Brazil, and it has been the measure preferred by economists analyzing the fiscal policy in Brazil at that time (e.g., Giambiagi and Alem 2011 and Portugal 2017). Under our benchmark measure of primary deficits, the fiscal deterioration from May/94– May/95 to May/95–May/96 is usually explained by the increase in wages that resulted from wage negotiations in 1994, and by the Bacha effect, which worked as the reverse of the Olivera-Tanzi effect.39 The reason is that fiscal revenues in Brazil were very well indexed to inflation, but fiscal expenditures were not.40 So the executive branch could, and indeed did so, cut the real value of expenditures just by disbursing the originally planned nominal amounts with some delay, as higher inflation rates would rapidly erode the real value of those expenditures. Of course, this had the collateral effect of creating large problems, since public hospitals would run out of money at the end of the year, several bridges or roads would stay unfinished for many years, and so on. Guardia (1992) studied the budget for 1990 and 1991 in detail and reported significant differences between total expenditures in the (federal) budget and the actual expenditures. In 1990 and 1991, total expenditures were, respectively, approximately 63 and 60 percent of approved expenditures. Therefore, after inflation was under control, the government had to deal with the large discrepancies between nominal expenditures and revenues in the budget, which could explain the temporary deterioration in fiscal balances in the subsequent years. The success of the Real Plan in conquering hyperinflation is indisputable, but many discussions have taken place regarding which were the most important points in accounting for it. As is evident, many important changes were taking place around the implementation of the Real Plan, so it is hard to answer that question. One important condition was the availability of foreign financing, as foreign capital inflows resumed after the government reached an agreement with its foreign creditors under the Brady Plan.41 That was the end of a long process of foreign debt rescheduling. Another important factor was the fiscal reform that increased primary surpluses (Figure 21). That reform also included other important fiscal measures that, most likely, did not have an immediate impact on government fiscal statistics. Among those measures was the imposition of fiscal constraints on subnational governments, seen as an important achievement of the Real Plan. The deficits of subnational 39See Bacha (2003) and Tanzi (1977). 40A daily index, the UFIR (Fiscal Reference Unit), was computed based on inflation. Taxes would be denominated in this indexed unit of account and then translated to the nominal hyperinflated currency on the very day that taxes were paid to the banking system. 41Capital inflows were a main factor in the expansion of the interest-bearing public debt, as the Central Bank of Brazil conducted massive sterilized purchases of foreign exchange. In 1993, so much capital was flowing into Brazil that the government implemented controls on capital inflows (Carvalho and Garcia 2008). 20 governments became a big issue in the 1980s and 1990s, and many attempts were made, often including bailouts, to solve that issue. In 1989, for example, a debt renegotiation with the state governments took place, and only two years previously the government had renegotiated the debt of 10 state banks. The state banks, in particular, were constantly used to finance subnational government deficits. Many of them actually operated with negative reserves, which ultimately pressured the central bank to expand the monetary base (see Section 4.1). In 1993, the reforms enabled the federal government to use the fiscal revenues of subnational governments as debt guarantees and also forbid the state banks from making new loans to their respective state governments. That was the beginning of a sequence of reforms that would culminate in the implementation of the Fiscal Responsibility Law in 2000. Finally, the tightness of monetary policy was an important characteristic of the Real Plan, and it still characterizes monetary policy to this day. 3.3 1995–2016: Moderate Growth with Higher Stability The last subperiod of our analysis represents the period of lowest inflation in Brazilian history. Inflation rates averaged only 8 percent per year, accompanied by the adoption of active fiscal and monetary policy rules. Economic growth resumed, but at moderate rates. Real GDP per capita grew 1.2 percent per year on average. We also observed the process of fiscal consolidation, with primary surpluses in 1995–2002 and 2003–2011 averaging 2.3 and 2.8 percent of GDP, respectively (row 9, Table 1). Despite all those advancements, the country experienced a big shift in its economic policy at the onset of the international financial crisis in 2008–2009, which eventually culminated in a rapid deterioration of its fiscal balances and a deep recession in recent years. These events have raised concerns about the capability of the government to maintain a low-inflation regime in the future.42 3.3.1 1995–2002 The years following the implementation of the Real Plan represented a consolidation of the reforms that had begun in the previous subperiod. The government continued the privatization process and promoted both fiscal and banking reforms. Part of these reforms were possible only because of the success of the Real Plan in conquering hyperinflation, which gave the government political support for pushing its reform agenda. The public?s assessment of the new low-inflation scenario became clear in the following presidential elections. Fernando Henrique Cardoso, the finance minister during the formulaiton of the Real Plan, was elected 42For thorough analyses of that period, we refer to Werneck (2014a,b). 21 president of Brazil in the first round, not only in the presidential elections of 1994 but again in the 1998 elections.43 The low-inflation regime, however, also brought some challenges. For example, a banking crisis followed the Real Plan, during which some private and state-owned banks failed. One of the reasons for the failure was that the fall in inflation led to a fall in seigniorage-like revenues (the float) that were partially captured by these banks. Here is an example of how this mechanism works. The Central Bank of Brazil increases the monetary base by $1,000 reais by depositing that amount in the account that holds the bank reserves. Assuming that the banks have no incentives to hold any voluntary reserves, the banking system lends the $1,000 reais to the public. The public borrows that amount, and after spending it, the $1,000 reais return to the banks as deposits. Assuming that the reserve requirement ratio is, for example, 10 percent, the banks keep $100 reais as reserves and now have $900 reais left to lend to the public again. Then the public borrows that amount, and after spending it, the $900 reais return to the banking system as deposits again. The banking system holds $90 reais as reserves and lends the rest. That process continues indefinitely, and the increase in the amount of deposits converges to $10,000 reais (= $1,000 + (1 −10%) ×$1,000 + (1−10%)2×$1,000 + . . .), which represents ten times (the inverse of the reserve requirement ratio) the initial increase in the monetary base. The ratio of deposits to bank reserves in Brazil is illustrated in Figure 23. Finally, given that banks charge interest when lending money to the public, and that deposits are usually not remunerated, that process represents an increase in the revenues of the banking system. So, overall, in order for inflation to fall, the Central Bank of Brazil must make fewer such increases to the monetary base and thus must collect fewer revenues, as described immediately above. This decrease in revenues hurt the balance sheets of banks and, all else equal, contributed to the banking crisis. Besides the banking crisis, the government also faced turbulence in international capital markets. The first episode was the 1997 Asian financial crisis, which was immediately followed by the 1998 Russian financial crisis. After the latter, there was a speculative attack on the real, and the Central Bank of Brazil experienced a fast deterioration of its international reserves (Figure 24). The IMF stepped in, but the situation was such that the central bank could no longer hold the crawling-peg regime, and in January 1999, a floating exchange rate regime was adopted. This change also culminated in the replacement of the governor of the 43According to the constitution of 1988, the president is elected by a majority voting in a two-round system. If a presidential candidate receives more than 50 percent of the valid votes in the first round, that is, after the exclusion of blank and null votes, then the candidate is elected president without a second round. So far, no other presidential candidate besides Fernando Henrique Cardoso has been elected in the first round after the constitution of 1988 was enacted. Voting is mandatory in Brazil. 22 ultimately used. These operations were mostly done off-budget, so they must be taken into account when computing public deficits in Brazil, especially during the high-inflation periods. We use the transfers from the central bank to the Bank of Brazil as an approximation of those deficits, and we add those transfers to the deficits data.49 However, the evidence above suggests that Brazil effectively had multiple monetary authorities during the high-inflation period, so those transfers might not represent all of the deficit that was financed through money issuance. The reason is that given the size of the public financial sector, part of the seigniorage-like revenues (the float) described in Section 3.3.1 was captured by public banks, so the total revenue of the public financial sector was probably somewhere between the variation of the monetary base over GDP and the variation of M1 over GDP, both illustrated in Figure 29.50 However, we do not have enough information on those banks, so we restrict our analysis to the transfers between the central bank and the Bank of Brazil. Initially, the Central Bank of Brazil did not transfer its profits to the government Treasury, so the use of Conta de Movimento was a direct way in which the government could access seigniorage revenues. Interestingly, when that account was frozen in 1986, the government started to use another similar account between the Bank of Brazil and the central bank, Conta de Suprimentos Especiais, until its elimination in 1988.51 When both accounts became unavailable, the government established the transfer of the profits of the central bank to the Treasury.52 In fact, after 1988, there were two ways in which the government could access seigniorage revenues: through the transfer of profits and through the remuneration of its deposits at the Central Bank of Brazil. The new constitution in 1988 established that the government could not have accounts with commercial banks, but rather only one account, and that account should be at the central bank, the Conta ´ Unica do Tesouro (Single Government Account), which should be used for all its transactions. A particular feature of this account is that the Central Bank of Brazil became responsible for paying interest on its balances, which was based on the average remuneration of the government debt securities in its portfolio. Since deposits are usually not remunerated, those transfers can also be interpreted as part of the seigniorage revenues. Figure 28 demonstrates that they match well with the series of seigniorage revenues. The discussion above shows that the persistence and magnitude of the inflation process 49See Appendix C. 50Salviano Junior (2004) also makes that point in his analysis of the process that led to the privatization of public banks (PROES) following the Real Plan. 51Between 1965 and 1987, the average variation in the monetary base over GDP was 2.6 percent, while the average variation in the Bank of Brazil’s accounts over GDP was 2.8 percent. Those figures show that seigniorage revenues were used mostly to finance the operations of the Bank of Brazil. 52See Carvalho (2017) for a description of the evolution of the institutional framework regarding the relationship between the Central Bank of Brazil and the government Treasury. 29 in Brazil are closely related to the persistence and magnitude of the degree of passiveness of its monetary policy and to the government’s fiscal deficits, and that important structural changes took place in both the institutional arrangements and fiscal balances in the transition to the low-inflation period. In addition, it also provides an explanation of why the government increased seigniorage revenues in the 1970s despite its access to debt financing. Direct access to the central bank’s funds through Conta de Movimento provided incentives for some groups to use that account instead of going through negotiations with Congress. In this case, the “provinces” effect might have played an important role in accounting for the high and persistent inflation in Brazil. That was the term used by Calvo and Vegh (1999) as a possible explanation for chronic inflation episodes, in which different entities of the government choose their deficit level without taking into account its effect on the aggregate deficit, which would lead, in equilibrium, to higher deficits that are financed through higher seigniorage revenues, therefore leading to higher inflation.53 4.2 Inflation Indexation and Passive Monetary Policy Another factor that contributed to the Brazilian inflation process is indexation, which was often mandatory and became a widespread phenomenon in Brazil after its introduction in the 1960s. Most prices, wages, taxes, and the exchange rate, as well as asset prices, were indexed to past inflation. That, together with a peculiar form of monetary passiveness, would create the so-called inflation inertia. The main idea is that the widespread indexation would induce agents to expect even higher inflation rates in the future and to demand higher money balances, which would then be satisfied by the (passive) monetary policy. Thus agents’ expectations of higher inflation rates would translate into higher growth rates of money balances and higher inflation rates, leading to a vicious circle. In such an environment, inflationary shocks, like the maxi-devaluations undertaken during external crises, would permanently increase the inflation rate. As the monopsonist in the money market, the central bank could, in theory, opt not to validate such an increase in money demand by not increasing money supply, that is, by substantially increasing the interest rate. However, this was not the monetary policy that was followed during the pre– Real Plan years. Monetary policy at the time was more geared to keeping the ex-ante real interest rate at a low positive level (Pastore 1994, 1996). The adoption of a more active monetary policy by the Real Plan represented a major change in the way monetary policy was conducted, and that could explain part of its success. This hypothesis regarding inflation inertia gave many economists the illusion that the 53That issue was formally analyzed in Chari and Kehoe (2007) in the context of a monetary union. 30 hyperinflation could be cured without major improvements in the fiscal stance, as discussed in our description of the stabilization plans. This mechanism has been studied by many economists in Brazil (Pastore 2014; Cati, Garcia, and Perron 1999; Garcia 1996), but it has not been completely formalized yet, so we still cannot fully evaluate its importance.54 5 Final Remarks and Conclusion This chapter has shown that a simple fiscal story goes a long way in accounting for the inflation dynamics in Brazil, in which inflation rates are highly correlated with both money growth rates and seigniorage revenues. In 1960–1973, debt financing was still very restricted, and the government relied mostly on seigniorage revenues to finance its deficits. Then, in 1973–1988, even though the government had some access to debt financing, the institutional arrangement was such that seigniorage revenues were being channeled to finance deficits at the Bank of Brazil. During the period with the highest inflation rates, 1989– 1994, the connection between government financing needs and seigniorage revenues became more complex. The government obtained higher seigniorage revenues, but that was done through the profit transfers from the central bank to the Treasury and also through the remuneration of its deposits at the central bank, which were less direct mechanisms than before. Financial repression during Collor Plan I also guaranteed that part of household savings was used to finance government deficits and allowed the government to partially default on interest payments given that the remuneration of the confiscated savings was chosen by the government itself. Finally, in the low-inflation period (after 1994), Brazil showed primary surpluses and low seigniorage revenues. The fiscal story is also consistent with the fiscal adjustments that were implemented by PAEG and the Real Plan, the two successful stabilization plans that managed to reduce inflation in Brazil. In both cases, the drop in seigniorage revenues that followed their implementation was compensated by higher primary surpluses. The credibility of these stabilization plans also played an important role, given that they paved the way for increased access to debt financing by the government, which allowed the adjustments to be imple54Garcia (1996) highlights a distinct aspect of the Brazilian hyperinflation with respect to the causal relation between seigniorage and inflation. He points out that although seigniorage revenues were used to finance fiscal deficits, the causal relation was not a direct one from higher deficits to higher inflation and higher seigniorage. According to his partial equilibrium model, seigniorage was endogenously generated by the interaction between indexation and money demand in the case in which the government provided a currency substitute, that is, inflation-indexed accounts offered by the banks that invested in government bills with daily liquidity. To keep this system from collapsing, the central bank had to match the increase in nominal money demand, thereby generating seigniorage. The central bank, however, could generate neither more nor less. Garcia (1996) calls this the “non-controllability” of seigniorage revenues. 31 mented more gradually. In the case of the Real Plan, the increased access to debt financing even allowed for a temporary deterioration in fiscal balances in subsequent years, in which increased access to debt financing enabled the government to keep low seigniorage revenues. On the other hand, the connection between government fiscal deficits and growth performance is much less clear. Even though many of Brazil’s recessions were associated with government fiscal imbalances, such as in the early 1960s, early 1980s, and the most recent one, the large reduction in macroeconomic instability that was achieved after the Real Plan in 1994 did not result in much higher growth rates of real GDP per capita. Nevertheless, the periods of fast economic growth in 1968–1973 and most recently in 2004–2008 were associated with improvements in fiscal balances, which leads us to conclude that fiscal stability is a necessary but not sufficient condition for sustained growth. That has shown to be a more difficult task and would have to include, for example, reforms to Brazil’s education, tax, political, and judicial systems. Another lesson from the Brazilian case is that the government often used off-budget transactions, mostly through public banks and state-owned enterprises, to “hide” its fiscal deficits, especially in the high-inflation period (pre-1994). Some recent developments in Brazilian public finances, as the pedaladas fiscais, accounting maneuvers that eventually led to the impeachment of former president Dilma Rousseff, remind us that we should always be aware that there are constant attempts to “violate” the intertemporal budget constraint. Finally, we observed that Brazil’s transition to its low-inflation regime beginning in 1994 was associated with the institution of active monetary policy in a context in which fiscal dominance has been avoided with several initiatives to improve the fiscal stance, most notably the Fiscal Responsibility Law. Unfortunately, after this great achievement, and the very substantial improvement in the fiscal stance, especially after the floating of the currency in 1999, Brazil’s primary surplus decreased rapidly in the last few years. To make matters worse, the public-debt-to-GDP ratio accelerated again and now hovers at record levels. With the tax burden at also very high levels, Brazil must finally confront the political challenges required to rein in public expenditures. This major challenge facing the next president of the country will have to be tackled, or the Real Plan will pass to history as a long noninflationary interregnum. References Abreu, Marcelo de Paiva, and Rog´erio L. F. Werneck. 2014. “Estabiliza¸c˜ao, abertura e privatiza¸c˜ao, 1990-1994.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. 32 Bacha, Edmar L. 2003. “Brazil’s Plano Real: A view from the Inside.” In Development Economics and Structuralist Macroeconomics: Essays in Honor of Lance Taylor. Cheltenham, UK: Edward Elgar. Calvo, Guillermo A., and Carlos A. Vegh. 1999. “Inflation Stabilization and BOP Crises in Developing Countries.” In Handbook of Macroeconomics, vol. 1C, edited by John Taylor and Michael Woodford, 1531–1614. Amsterdam: North Holland. Carneiro, Dion´ısio D. 2014. “Crise e esperan¸ca, 1980–1984.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. Carneiro, Dion´ısio D., and Eduardo M. Modiano. 2014. “Ajuste externo e desequil´ıbrio interno, 1980–1984.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. Carvalho, Antonio C. C. D’´ Avila Jr. 2017. “BC e Tesouro: um estudo sobre a constitui¸c˜ao, leis complementares, leis ordin´arias e medidas provis´orias.” In A crise fiscal e monet´aria Brasileira, 2nd ed., edited by Edmar L. Bacha. Rio de Janeiro: Civiliza¸c˜ao Brasileira. Carvalho, Bernardo S. de M., and M´arcio G. P. Garcia. 2008. “Ineffective Controls on Capital Inflows Under Sophisticated Financial Markets: Brazil in the Nineties.” In Financial Markets Volatility and Performance in Emerging Markets, edited by Sebastian Edwards and M´arcio G.P. Garcia. Chicago: University of Chicago Press. Cati, Regina Celia, M´arcio G. P. Garcia, and Pierre Perron. 1999. “Unit Roots in the Presence of Abrupt Governmental Interventions with an Application to Brazilian Data.” Journal of Applied Econometrics 14 (1): 27–56. Cerqueira, Ceres Aires. 2003. D´ıvida externa brasileira. 2nd ed. Banco Central do Brasil. Chari, V. V. and Patrick J. Kehoe. 2007. “On the Need for Fiscal Constraints in a Monetary Union.” Journal of Monetary Economics 54 (8): 2399–2408. Costa Neto, Yttrio Corrˆea da. 2004. Bancos oficiais no Brasil: origem e aspectos de seu desenvolvimento. Banco Central do Brasil. Franco, Gustavo H. B. 2017. “O conselho monet´ario nacional como autoridade monet´aria: das origens aos dias actuais.” In A crise fiscal e monet´aria brasileira, 2nd ed., edited by Edmar L. Bacha. Rio de Janeiro: Civiliza¸c˜ao Brasileira. 33 Garcia, M´arcio G. P. 1996. “Avoiding Some Costs of Inflation and Crawling Toward Hyperinflation: The Case of the Brazilian Domestic Currency Substitute.” Journal of Development Economics 51 (1): 139–159. Giambiagi, Fabio, and Ana Claudia Alem. 2011. Finan¸cas p´ublicas - teoria e pr´atica no Brasil. 4th ed. Amsterdam: Elsevier. Guardia, Eduardo R. 1992. “Or¸camento p´ublico e pol´ıtica fiscal: aspectos institucionais e a experiˆencia recente–1985/1991.” Master’s thesis, Universidade Estadual de Campinas, Instituto de Economia. Lago, Luiz Aranha Corrˆea. 2014. “A retomada do crescimento e as distor¸c˜oes do milagre, 1967–1974.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. Mesquita, M´ario M. C. 2014. “Infla¸c˜ao, estagna¸c˜ao e ruptura, 1961–1964.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. Modiano, Eduardo M. 2014. “A ´opera dos trˆes cruzados, 1985–1990.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. N´obrega, Ma´ılson da. 2005. O futuro chegou: institui¸c˜oes e desenvolvimento no Brasil. Rio de Janeiro: Globo. Orenstein, Luiz, and Antonio Claudio Sochaczewski. 2014. “Democracia com desenvolvimento, 1956–1961.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. Pastore, Affonso Celso. 1994. “D´eficit p´ublico, a sustentabilidade do crescimento das d´ıvidas interna e externa, senhoriagem e infla¸c˜ao: uma an´alise do regime monet´ario brasileiro.” Brazilian Review of Econometrics 14 (2): 177–234. Pastore, Affonso Celso. 1996. “Por que a pol´ıtica monet´aria perde efic´acia?” Revista Brasileira de Economia 50 (3): 281–311. Pastore, Affonso Celso. 2014. Infla¸c˜ao e crises: o papel da moeda. Amsterdam: Elsevier. Pedras, Guillermo B. V. 2009. “Hist´oria da d´ıvida p´ublica no Brasil: de 1964 at´e os dias atuais.” In D´ıvida p´ublica: a experiˆencia brasileira, edited by Lena Oliveira de Carvalho, Otavio Ladeira de Medeiros, and Anderson Caputo Silva. Brasilia: Secretaria 34 do Tesouro Nacional, Banco Mundial. Portugal, Murilo. 2017. “Pol´ıtica fiscal na primeira fase do Plano Real, 1993–1997.” In A crise fiscal e monet´aria brasileira, 2nd ed., edited by Edmar L. Bacha. Rio de Janeiro: Civiliza¸c˜ao Brasileira. Salviano Junior, Cleofas. 2004. Bancos estaduais: dos problemas crˆonicos ao PROES. Banco Central do Brasil. Silva, Anderson Caputo. 2009. “Origem e hist´oria da d´ıvida p´ublica no Brasil at´e 1963.” In D´ıvida p´ublica: a experiˆencia brasileira, edited by Lena Oliveira de Carvalho, Otavio Ladeira de Medeiros, and Anderson Caputo Silva. Brasilia: Secretaria do Tesouro Nacional, Banco Mundial. Tanzi, Vito. 1977. “Inflation, Lags in Collection, and the Real Value of Tax Revenue.” IMF Economic Review 24 (1): 154–167. Werneck, Rog´erio L. F. 1991. “Public sector adjustment to external shocks and domestic pressures, 1970-85.” In The Public Sector and the Latin American Crisis, 1970-1985, edited by Felipe Larrain and Marcelo Selowsky. San Francisco: International Center for Economic Growth. Werneck, Rog´erio L. F. 2014a. “Alternˆancia pol´ıtica, redistribui¸c˜aao e crescimento, 2003– 2010.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. Werneck, Rog´erio L. F. 2014b. “Consolida¸c˜ao da estabiliza¸c˜ao e reconstru¸c˜ao institucional, 1995–2002.” In A ordem do progreso: dois s´eculos de pol´ıtica econˆomica no Brasil, 2nd ed., edited by Marcelo de Paiva Abreu. Rio de Janeiro: Elsevier. 35 A Tables Table 1: Government Budget Accounting (% of GDP) subperiod 60–64 65–72 73–80 81–94 95-02 03-11 12-16 Sources (1) domestic debt 0.0 0.8 -0.2 -0.3 -0.6 1.1 3.5 (2) external debt 0.0 0.0 1.7 -1.8 0.0 -2.3 0.0 (3) real monetary base -0.4 -0.2 -0.2 -0.1 0.1 0.1 -0.1 (4) seigniorage 4.1 1.9 2.4 3.2 0.4 0.4 0.3 Uses (5) interest on domestic debt 0.3 0.4 1.3 0.4 2.0 1.7 3.2 (6) interest on external debt 0.0 0.0 0.7 1.9 0.3 0.3 0.1 (7) primary deficits 2.9 1.0 0.2 3.1 -3.0 -2.7 -1.1 (8) transfers (residual) 0.5 1.1 1.6 -4.5 0.7 -0.1 1.5 Other measures (9) primary deficits + transfers 3.4 2.2 1.8 -1.4 -2.3 -2.8 0.4 (10) fiscal deficits 3.2 1.5 2.1 5.5 -0.8 -0.6 2.1 (11) fiscal deficits + transfers 3.7 2.6 3.7 1.0 0.0 -0.8 3.6 Sources: Brazilian Institute of Geography and Statistics (IBGE) and Central Bank of Brazil (CBB). Notes: See Appendix Cfor a description of the budget constraint. Transfers are computed as (8)=(1)+(2)+(3)+(4)-(5)-(6)-(7). The fiscal deficit is computed as (10)=(5)+(6)+(7). 36 Table 2: Investment by State-Owned Enterprises (% of GDP) sub-period 60–72 73–80 81–94 95–00 All sectors 2.2 4.7 2.7 1.3 Manufacturing 1.0 1.5 0.6 0.3 Energy 0.4 1.0 0.8 0.2 Transportation 0.3 0.9 0.3 0.0 Communication 0.1 0.8 0.6 0.5 Source: Brazilian Institute of Geography and Statistics (IBGE), Estat´ısticas do S´eculo XX (https://seculoxx.ibge.gov.br/economicas.html). Table 3: Government Budget Accounting: Real Plan (% of GDP) subperiod May/92– May/93 May/93– May/94 May/94– May/95 May/95– May/96 Sources (1) domestic debt -1.4 -2.6 -4.1 7.8 (2) external debt -2.1 -6.3 0.6 -4.1 (3) real monetary base -0.3 -0.1 1.4 0.1 (4) seigniorage 2.8 2.7 0.7 0.3 Uses (5) interest on domestic debt 0.8 -3.4 2.7 3.3 (6) interest on external debt -0.1 -0.2 -0.1 0.2 (7) primary deficits + transfers -1.8 -2.8 -4.0 0.5 Sources: Brazilian Institute of Geography and Statistics (IBGE) and Central Bank of Brazil (CBB). Notes: See Appendix Cfor a description of the budget constraint. Transfers are computed as (8)=(1)+(2)+(3)+(4)-(5)-(6)-(7). 37 B Figures 38 Figure 7: Sectoral composition of the public debt, 1960–2016 (a) External debt (b) Domestic debt Sources: IPEADATA, Central Bank of Brazil (CBB), and Brazilian Institute of Geography and Statistics. Notes: The series of net debt, both external and domestic, of the federal government, state governments, and SOEs are from IPEADATA. We add the balances of Bank of Brazil accounts on the balance sheet of the central bank to the series of net domestic debt of the federal government. See Appendix Cfor a description of how we compute the nominal GDP series. 45 Figure 8: Bank of Brazil (BB) and the central bank Sources: Brazilian Institute of Geography and Statistics (IBGE) and Central Bank of Brazil (CBB). Notes: The variation in the balance of BB accounts denotes the variation in the balance of the accounts of the Bank of Brazil that are reported on the balance sheet of the Central Bank of Brazil, which comprises the Conta de Movimento and the Conta de Suprimentos Especiais. See Appendix Cfor a description of how we compute the nominal GDP series. 46 Figure 9: Deficit and transfers Sources: Brazilian Institute of Geography and Statistics (IBGE) and Central Bank of Brazil (CBB). Notes: Our definition of “deficit” is the primary deficit plus real interest payments on debt discounting for real GDP growth, and transfers represent the residual (the variation in debt that cannot be explained by official deficit statistics). See Appendix Cfor a description of how we compute the series of deficits, transfers, and nominal GDP. 47 Figure 10: Average maturity of debt Sources: Central Bank of Brazil (CBB) and National Treasury Secretariat (STN). Notes: We use the series of average maturity of the registered external debt (code 3688) from 1971 to 2012, and the series of the average maturity of the gross external debt (code 3689) from 2005 to 2016, both from the central bank. Regarding the maturity of the domestic debt, from 1964 to 2008, we used the average maturity reported in the statistical appendix, table A.4.5, of the book D´ıvida P´ublica: A Experiˆencia Brasileira published by the National Treasury Secretariat in 2009. After 2009, we use the series of average maturity of the federal government debt securities (code 10621) from the central bank. The series codes correspond to the codes used on the website of the CBB. 48 Figure 11: Indexation of federal government debt securities Source: IPEADATA. Notes: Debt securities indexed to inflation correspond to those indexed to either IGP-DI or IGP-M. Those indexed to IPCA or other price indexes are included in others. Debt securities that are not indexed correspond to nominal bonds. Finally, debt securities indexed to the interest rate denote those that are indexed to the interest rate that is used by the central bank as its policy instrument, the SELIC. These securities are sometimes referred to as zero-duration bonds. 49 Figure 12: Nominal interest rate and inflation Source: IPEADATA. Notes: Interest rates correspond to the daily averages of the interest rates charged in the overnight operations between the central bank and other financial institutions. Inflation rates are computed using the General Price Index, IGP-DI, from Getulio Vargas Foundation. 50 Figure 13: Inflation and stabilization plans Source: IPEADATA. Note: Inflation rates are computed using the General Price Index, IGP-DI, from Getulio Vargas Foundation. 51 Figure 14: International accounts Source: Central Bank of Brazil (CBB). Notes: We use the exchange rate series with average purchase values in the year (code 3693) from the central bank to convert dollar values to reais. The current account, trade balance, and net interest income values correspond to the series with codes 2301, 2301, 2398, respectively. The series codes correspond to the codes used on the website of the CBB. See Appendix Cfor a description of how we compute the series of nominal GDP. 52 Figure 15: Real exchange rate Sources: Central Bank of Brazil (CBB), IPEADATA, and Bureau of Economic Analysis (BEA). Notes: We consider the bilateral real exchange rate between the United States and Brazil. The real exchange rate is computed as the US price index times the exchange rate divided by the price index in Brazil. We use the US CPI as the price index for the United States, the exchange rate series (R$/US$) with average purchase values from the central bank, and the General Price Index, IGP-DI, from Getulio Vargas Foundation as the price index for Brazil. 53 Figure 16: Interest payments refinanced and in arrears Sources: Central Bank of Brazil (CBB) and IPEADATA. Notes: We use the monthly series for interest actually paid (code 2833), in arrears (code 2835), and refinanced (code 2834) relative to the external debt from the central bank. We use the exchange rate series with average purchase values in the month from IPEADATA to convert dollar values to reais. We use accumulated values in three months to smooth the series and report annualized values by multiplying the series by four. The series codes correspond to the codes used on the website of the CBB. See Appendix Cfor a description of how we compute the series of nominal GDP. 54 Figure 24: Exchange rate and foreign reserves Source: Central Bank of Brazil (CBB). Figure 25: Size of National Monetary Council (CMN) and inflation Sources: IPEADATA and Franco (2017). 61 Figure 26: Composition of the balance sheet of the Central Bank of Brazil (a) Assets (b) Liabilities Source: Central Bank of Brazil (CBB). 62 Figure 27: Government debt securities held by the Central Bank of Brazil Source: IPEADATA. Figure 28: Transfers from the central bank to Bank of Brazil and to the treasury Source: Central Bank of Brazil (CBB). 63 Figure 29: Variation in M0 and M1 over GDP Source: IPEADATA. 64 C Appendix This appendix describes the data and methodology that were used to compute the government budget constraint that is used in Table 1. Also discussed are the different primary deficit series that are available for Brazil. We consider the following consolidated budget constraint of the government in units of the domestic currency (in our case, reais): Ptbt+EtP∗ tb∗ t+Mt=Pt(Dt+Tt) + Ptrt−1bt−1+EtP∗ tr∗ t−1b∗ t−1+Mt−1,(1) where Pis the price level of the domestic good in local currency, bis the real stock of domestic debt in units of the domestic good, Eis the nominal exchange rate between the foreign (dollar) and local currencies, P∗is the price level of the foreign good in foreign currency, b∗is the real stock of external debt in units of the foreign good, Mis the monetary base in units of the local currency, Dis the real primary deficit in units of the domestic good, ris the gross real return on domestic debt, r∗is the gross real return on the external debt, and Tdenotes the transfers in units of the domestic good. The transfers, T, account for the residual in equation (1). We follow the same methodology as in chapter 2, with two simplifying assumptions: (1) we assume that the share of the domestic good in the domestic price level is one (α= 1), and we express the stock of external debt, b∗ t, in units of the foreign good, so the value of the external debt in units of the foreign currency is P∗ tb∗ t; and (2) we assume that all domestic debt is real because most domestic debt in Brazil is indexed to inflation or nominal interest rates (or both), and we do not have data on interest payments that separate real from nominal debt. Dividing equation (1) by PtYt, in which Ytis real GDP of the domestic country, and after some algebra, we can express the consolidated government budget constraint as (θt−θt−1) | {z } domestic debt +ξt(θ∗ t−θ∗ t−1) | {z } external debt + (mt−mt−1) | {z } monetary base +πtgt−1 πtgt mt−1 | {z } seigniorage (2) =rt−1−gt gt θt−1 |{z } interest on domestic debt +ξt r∗ t−1−gt gt θ∗ t−1 |{z } interest on external debt + (dt+τt) | {z } primary deficit + transfers , where we used the following definitions: θt≡bt Yt,θ∗ t≡b∗ t Yt,ξt≡EtP∗ t Pt,mt≡Mt PtYt,gt≡Yt Yt−1, and πt≡Pt Pt−1. Next, we describe how we use the available data to compute the government budget constraint in equation (2), whose averages are reported in Table 1. The fiscal deficit, 65 as in Figure 1c, corresponds to interest on domestic debt +interest on external debt + primary deficit. We start by describing how we compute the price level, real GDP, and exchange rate series.55 An important issue in our case is that stock values are expressed in units of the domestic currency at the end of the period, while the official price statistics report average prices within the period. In periods of high inflation, the difference between end-of-period and average prices is substantial. To overcome that, we deflate stock values by an estimate of end-of-period prices. In addition, for the most recent period (after 1990), we are able to compute the government budget constraint for each month, so we use the twelve-month sum of the budget constraint to compute the annual sequence of government budget constraints. To do that, we also construct a monthly series of real GDP. Price index We use the General Price Index (IGP) from Getulio Vargas Foundation. It is composed of three price indexes: 60 percent Producer Price Index (IPA), 30 percent Consumer Price Index (IPC), and 10 percent Construction Price Index (INCC). Prices are collected from day one to day thirty of the reference month. Data are available at the IPEADATA website, IGP-DI - geral - ´ındice (ago. 1994 = 100). End-of-period values are computed as the geometric average between tand t+ 1. We normalize the IGP-DI such that its average is equal to the GDP deflator in 1995. The GDP deflator in 1995 is equal to the nominal GDP in 1995 divided by 100 (we normalize real GDP to 100 in 1995). Data on nominal GDP are from IPEADATA, Produto interno bruto (PIB) a pre¸cos de mercado - referˆencia 2000). Regarding the price index of the foreign good, we use the GDP deflator for the United States for the annual series and the US CPI for the monthly series. Real GDP Data are from the Brazilian Institute of Geography and Statistics (IBGE), series SCN53. It is annual, from 1947 to 2014. The series show the percentage variation in volume. We normalize real GDP in 1995 to 100 and use the variation to construct the series of real GDP. We use the annual variation in real GDP (reference year 2010) between 2015 and 2016 to update the series to 2016. It is available at the IPEADATA website. To construct a monthly series, we assume that the annual values correspond to July values and use linear interpolation to compute real GDP for the other months. Exchange rate We use the exchange rate (buy price) between the Brazilian currency and US$. Both average and end-of-period series are available at the IPEADATA website. 55Nominal GDP is computed as the price level multiplied by real GDP. 66 Subperiod 1960–1981 Most series are available at an annual frequency for this subperiod, so we only compute the budget constraint at an annual frequency. We use the sum of three series to compute the stock of domestic debt Ptbt: federal government debt securities out of the central bank, debt securities of states and municipalities, and d´ıvida p´ublica fundada. Data are from IBGE, Estat´ısticas do S´eculo XX. We use the series of monetary base, M0 (end-of-period) from IPEADATA, as our measure of Mt. We use the series of real interest payments on domestic debt divided by nominal GDP as our measure of interest payments on domestic debt rt−1−gt gtθt−1. We avoid manipulating these series in the first subperiods because the series of interest payments and stock of debt come from different sources. The series of real interest payments is computed as the series of nominal interest payments on domestic debt minus the series of monetary correction, multiplied by the fraction of federal debt securities out of the central bank, all from IBGE, Estat´ısticas do S´eculo XX. Regarding the net external debt, we assume that b∗ tis zero up to 1972. After that, we use the series of registered public external debt (code 3564) minus foreign reserves (code 3566) as the measure of P∗ tb∗ t, both from the Central Bank of Brazil.56 For the series of interest payments on external debt, we only observe the total payments of interest on gross external debt, which includes both public and private debt. We observe both the series of total gross external debt (code 3682) and public gross external debt, so we assume that interest payments are proportional to the stocks, which gives us a series of interest payments on public external debt, P∗ t(r∗ t−1−1)b∗ t−1. The series of primary deficit is computed as expenditures minus tax revenues plus the transfers from the central bank to the Bank of Brazil. Tax revenues include direct and indirect taxes. Expenditures correspond to government consumption, subsidies, transfers, and investment, net of other current net revenues. Data are from IBGE, Estat´ısticas do S´eculo XX. The transfers to Bank of Brazil are computed as the variation in the balance of the Bank of Brazil accounts at the central bank: Conta de Movimento and Conta de Suprimentos Especias. The transfers τtare computed as the residual of equation (2). Subperiod 1982–1990 In 1981, the Central Bank of Brazil started to publish the series of the public-sector borrowing requirement, which requires the computation of both net domestic and net external debt series of the nonfinancial public sector. We rely mostly on these series to construct the 56Codes denote the series codes on the website of the central bank. 67 budget constraint after 1981. As our measure of domestic debt, Ptbt, we use the series of net domestic public debt, available at IPEADATA. However, this series includes the monetary base as liabilities and the accounts of the Bank of Brazil at the central bank as assets, so we adjust the series for that. We subtract the monetary base and add the balance of the Conta de Movimento and Conta de Suprimentos Especias on the balance sheet of the central bank to the series of net domestic debt. Regarding the series of net public external debt, the central bank does not distinguish between exchange rate adjustments and inventory adjustments; it reports only the sum. We construct the series of exchange rate adjustments and use the difference as the proxy for inventory adjustments on the external debt.We assume that, for the 1981–1990 period, all the adjustments that are made to compute the nominal deficit based on the variation in total net public debt are from adjustments on the net external debt series. The reason is that this period is characterized by a sequence of external debt renegotiations, which could lead to variations in the net external debt figures without being related to nominal deficits. In order to compute the series of interest payments on external debt, we proceed in a similar fashion as in the previous subperiod, but now we adjust the value of interest payments on external debt for default. We compute the fraction of total interest paid over total interest due relative to total gross external debt and use this fraction to adjust for interest payments that were actually paid. The source of these data is the same as in Figure 16. Regarding the series of interest payments on domestic debt, we use the same method as before up to 1984. After that, the central bank started to publish the series of nominal interest payments. We then subtract the series of interest payments on external debt described above and use the result as our measure of interest payments on domestic debt. Subperiod 1990–2018 After 1990, the central bank began to publish its fiscal statistics on a monthly frequency. From that point on, we have all the information that is needed to compute the government budget constraint in (2), except for the primary deficit. At this point, we can only compute the sum of primary deficit plus transfers, corresponding to the residual of (2). This is similar to the primary deficit series that the central bank uses. As mentioned above, we proceed by summing the monthly budget constraints to have a annual series for the government budget constraint. We then use the primary deficit series from IBGE and subtract it from (primary deficit+transfers) to compute the transfers. Finally, note that the series of primary deficits from IBGE is available up to 2000. After that, we use the publications of the consolidated national public-sector accounts available on the website of the National Treasury Secretariat. 68 C.1 Primary Deficit Series in Brazil In Brazil, there are two main sources of data on the public-sector primary deficit: the Brazilian Institute of Geography and Statistics (IBGE) and the Central Bank of Brazil (CBB). The data from IBGE are annual and cover the period from 1947 to the present.57 The IBGE data include the federal government, states, and municipalities, and are published in the national accounts—public sector.58 The data are based on the executed budget of the government. The data on primary deficits from CBB, on the other hand, are available on an annual frequency since 1985 and on a monthly frequency since January 1991. These data include the federal government, states, and municipalities, as well as the central bank and state-owned enterprises (SOEs). We illustrate both (original) series in Figure 30. There are important differences in the methodology used to compute each of these series. The data from IBGE use the actual data reported on the books of government authorities and can be considered the traditional measure of primary deficit.59 The CBB, on the other hand, estimates its data based on the public-sector borrowing requirement. That is, the CBB computes the variation of the stock of net debt and money supply from the federal government, central bank, states and municipalities, and state-owned enterprises. That variation gives a proxy for the fiscal deficit of the government, which includes both interest payments and the primary deficit. CBB then estimates the interest payments based on the characteristics of assets and liabilities and computes the primary deficit as a residual. In Brazil, there is a preference among economists for using the deficit series from the CBB after 1985.60 A few factors explain that preference. First, at the onset of the external debt crisis in the early 1980s, authorities needed to work with recent data, and the statistics on the primary deficit took a long time to be released. The CBB, on the other hand, had the ability to compile debt information for both domestic and external debt in a timely fashion, so it decided to compute its own fiscal statistics based on the public-sector borrowing requirements. The second factor is the inclusion of SOEs. The debt series used by the CBB covered SOEs, whereas the other deficit series did not. Since SOEs were constantly used by the government to implement its economic policies (see main text), that was considered a significant advantage. Third, there was lack of confidence in the capability of the government to accurately report its finances. The National Treasury Secretariat, for example, was only 57From 1947 to 2000, the series can be downloaded from the Estat´ısticas do S´eculo XX from IBGE. We extrapolate it using the annual publications of the public-sector accounts on the website of the National Treasury Secretariat. 58Data on the federal government deficit and central government are also readily available on the website of the Department of the Treasury for the most recent period. 59For example, IBGE uses data from the Balan¸co Geral da Uni˜ao (Union General Budget) for the federal government. 60See Giambiagi and Alem (2011). 69 Figure 30: Primary deficit: CBB versus IBGE Sources: Brazilian Institute of Geography and Statistics (IBGE) and Central Bank of Brazil (CBB). created in 1986. Next, we discuss a few points regarding the primary deficit series of the central bank. First, the inclusion (or not) of SOEs in the primary deficit series does not make a big difference after 1985, especially after the 1990s when most privatizations took place (Figure 30). Second, the CBB did not take into account privatizations when estimating primary deficits. In other words, if the government sold some of its assets to finance current expenditures, that would not be captured in the deficit series from the CBB. Figure 31a compares both series, with privatization and no privatization, for the period after 1996, which is the year when the data on privatizations used by the CBB become available. Third, the CBB does not account for defaults. It estimates interest payments on an accrual basis. So if the government defaulted on its interest payments, the CBB would underestimate the primary deficit. The reason is that the CBB assumes that the government paid all the interest that was due. Therefore, it would conclude that the government had the resources to pay for the interest due, which must have come from lower primary deficits (or higher surpluses) according to CBB’s methodology. Remember that the primary deficit is computed as a residual. But if the government defaulted on the interest payments, then it must be the case that deficits were actually higher than the ones reported by the CBB. As the main text mentions, the country accumulated arrears on interest payments on the external debt for many years, especially in the 1990–1994 period. We use the information on the fraction of interest on external debt that was actually paid to correct for both the 70