Chapter

Chapter 8. A Brazilian Perspective on Macroprudential and Monetary Policy Interaction

Author(s):
Luis I. Jacome H., Yan Carriere-Swallow, Hamid Faruqee, and Krishna Srinivasan
Published Date:
October 2016
Share
  • ShareShare
Show Summary Details
Author(s)
Fabia A. de Carvalho and Marcos R. de Castro 

This chapter examines the interaction between monetary and macroprudential policy in Brazil under both normative and positive perspectives. The chapter investigates optimal combinations of simple and implementable macroprudential and monetary policy rules that react to the financial cycle, based on a dynamic stochastic general equilibrium (DSGE) model built to reproduce Brazilian particularities and estimated with Bayesian techniques using data spanning the inflation-targeting period. The chapter also looks at whether recent macroprudential policy announcements in Brazil that targeted credit variables had important spillover effects on variables targeted by monetary policy. To this end, we use a daily panel of private inflation forecasts surveyed by the central bank. We also investigate the impact of announcements of macroprudential policy changes on the gap between inflation forecasts and the inflation target. Finally, the chapter presents an overview of the difficulties facing macroprudential policy in Brazil after the global financial crisis and discusses a few important future challenges.

The Latin American region sailed well through the years immediately following the collapse of Lehman Brothers in the United States. With several recurrent sovereign debt and systemic banking crisis episodes behind it, the region had shifted from procyclical to countercyclical fiscal and monetary policies by the late 1990s. This helped the region build solid macroeconomic fundamentals (Végh and Vuletin 2013) and sound and resilient financial systems, giving room to maneuver throughout the initial phases of the global financial crisis.

To alleviate risks of financial instability, owing to either a liquidity crunch or a worsening of external accounts, Latin American countries pursued different combinations of macroeconomic and macroprudential policy responses. This included measures to stimulate domestic demand by boosting bank loan origination, in some cases using public banks, to avoid a credit crunch. In addition, with abundant international liquidity associated with a balance of risks shifting favorably toward emerging markets, the region received significant inflows of foreign capital, which also helped fuel credit expansion.1

More recently, plummeting commodity prices and economic slowdown in major trading partners, such as China, have been a drag on the region’s growth, particularly in the context of fiscal buffers being practically exhausted. Increased household indebtedness, deteriorating labor market prospects, and worsening credit conditions have become important challenges from both macroeconomic and macroprudential policy perspectives. In some instances, financial and macro-economic cycles have been synchronous, and macroprudential policy has worked favorably towards meeting macroeconomic policy goals. However, at other times, macroprudential policy decisions might have had undesired spillovers from a macroeconomic policy perspective.

This chapter contributes to the ongoing debate by exploring two issues related to the interaction between macroprudential and monetary policy, taking Brazil as the central case. First, from a normative perspective, given the frequent and varied use of macroprudential instruments in the country even prior to the financial crisis, we seek an optimal combination of macroprudential and monetary policy, using the DSGE model in Carvalho and Castro (2015), which was carefully customized to represent essential features of the Brazilian banking system.2 Second, we take the analysis further by comparing the optimal set of policies with narrower subsets of policies that can be more easily implemented in a timely manner.

With respect to the first task, the optimal monetary and macroprudential policy mix for Brazil is still an unexplored issue. This chapter tries to answer the question by finding the optimal combination of sets of macroprudential and monetary policies that may react to a credit gap, given ample evidence that this indicator is a good, if not the best, early warning indicator of financial crisis (see Silva, Sales, and Gaglianone 2012 for the Brazilian case and Taylor 2015 and Drehman and Tsatsaronis 2014 for cross-country studies). Basel III also recommends the use of the credit gap to justify changes in the countercyclical capital buffer.

In order to find the optimal policy combinations, we follow the method proposed by Schmitt-Grohé and Uribe (2007), and focus on simple and implementable policy rules.3 We also investigate the properties of more easily implementable rules given the Brazilian regulatory framework.

Several studies have investigated the optimality of monetary policy reacting to financial conditions. Some have found that alternative monetary policy rules that react to financial variables have negligible stabilization gains when compared with strict inflation targeting or traditional Taylor rules (Bernanke and Gertler 2001; Faia and Monacelli 2007; Gilchrist and Leahy 2002; Iacoviello 2005). Other studies find that it can be welfare improving to let monetary policy react to financial variables (Angeloni and Faia 2013; Benigno and others 2011; Cúrdia and Woodford 2010; De Fiori and Tristani 2013; Fendoğlu 2014; Kannan, Rabanal, and Scott 2012). These studies are heterogeneous with respect to the model structure, financial frictions, financial targets,4 and parameterization. The conclusions might be model dependent, and, for a particular model, they can also be sensitive to the parameterization. They are also highly sensitive to the set of disturbances allowed in the model.5

This study distinguishes itself from others in a number of aspects. First, we include a varied and practical set of macroprudential policy instruments interacting with monetary policy, while most of the literature focuses only on monetary policy as the single instrument to stabilize multiple targets, including financial conditions.6 Second, our model is of a small and open economy with foreign trade and financial flows, while most of the literature focuses on closed economies.7 Third, our model has features that are necessary to reproduce the main aspects of the Brazilian credit market, such as heavy regulation of housing loans and savings deposits and a consumer credit segment in which credit originations are strongly based on households’ future labor income, yet facing significant default ratios. Conducting optimal policy analysis in models intended for practical use at central banks is a strikingly different approach compared to what has been usually adopted in the literature. The preferred choice of prototype models in this literature is most likely due to the dimension of practical models and the challenges faced in their estimation.

We find that certain combinations of reserve requirements and risk weights can result in losses that are very close to the optimum, involving a more complete combination of instruments, including the Basel III countercyclical capital buffer. The more restricted combination of instruments also results in dynamic responses very close to the optimal rules. Given the fact that changes to reserve requirements and risk weights are more easily implementable, the finding gives support to the Central Bank of Brazil’s extensive use of reserve requirements and risk weights to affect credit, and the stability of the overall capital requirement ratio to date.

With respect to the second task, although most of the literature is concerned with the issue of whether monetary policy should react to financial variables, the reverse argument has not been explored. Given the possible lack of synchronization of macroprudential and monetary policy in some recent episodes in Brazil, it is important to investigate whether macroprudential policy announcements can potentially affect the anchoring of inflation expectations. To this end, we use a panel of private inflation forecasts surveyed on a daily basis by the Central Bank of Brazil’s Investor Relations Office to estimate the impact of some macroprudential policy events—which explicitly targeted the credit market—on the formation of inflation expectations. We draw on the work of Carvalho and Minella (2012) to find a representative expectations formation rule, but we augment it with the investigated events in addition to some other necessary controls.

Among 14 events analyzed in our study, we find a subset of 6 events that suggest that macroprudential policy announcements affected the gap between inflation expectations and the inflation target. In 4 of these events, the impact was in the direction of widening the gap. When we group the events that were expected to increase credit into two different sets, one when monetary policy was contractionary and the other when monetary policy was expansionary, we find that the former had a positive significant impact on inflation expectations, while the latter was not significant. This can be interpreted as evidence that when macroprudential policy announcements are not synchronized with monetary policy, the anchoring of inflation expectations can be difficult.

The next section of this chapter describes how Brazil managed to build a solid financial system after a banking crisis. The chapter then reviews the main macroprudential measures implemented in Brazil in the aftermath of the financial crisis and during the postcrisis period. We present the optimal policy exercise using the DSGE model that was tailored to the Brazilian economy and then empirically investigate the impact of macroprudential policy announcements on monetary policy credibility before presenting the chapter’s conclusions.

The Brazilian Banking Crisis and Bank Regulatory Reform

The last episode of a banking crisis in Brazil immediately followed the implementation of the inflation stabilization plan—the Real Plan—in 1994 (Reinhart and Rogoff 2011). At that time, inflation stabilization had eliminated an important source of bank revenues and exposed banks’ practices to vulnerabilities that could undermine financial stability. To address these risks, in the first years of the inflation stabilization period the government implemented two major bank restructuring programs: the Program of Incentives to Restructure and Strengthen the National Financial System (PROER) and the Program of Incentives to Reduce the State-Level Public Sector in Bank Activity (PROES).

Local government banks had had a long history of impaired credit portfolios, with high default rates, posing systemic risks to the financial system and feeding fiscal imbalances. PROES addressed these problems through either the privatization of public banks or the transformation of public commercial banks into development banks, which were prohibited from extending loans to their public controllers. The Fiscal Responsibility Law, enacted in 2001, outlawed credit operations between public banks and their public controllers, further enhancing fiscal discipline.

PROER was a milestone in the regulatory framework of the Brazilian financial system. One of the pillars of this program was the enhanced framework under which the central bank—which is also the regulatory and supervisory authority—was authorized to intervene in troubled financial institutions. The program also included a number of other important measures, including a deposit-insurance facility.

In addition to these major restructuring programs, Brazil adopted best practices with respect to its bank regulatory and supervisory framework by adhering to the first Basel Accord in 1994 and adopting a strict regulatory and supervisory stance thereafter.8 Basel III capital regulations were first published in March 2013 and their phasing in started within a few months, in October. The Basel Committee on Banking Supervision performed its last assessment of Basel III regulations in Brazil in December 2013 and found that the country was compliant with the terms of the agreement. Brazil’s financial regulatory and supervisory framework ranks favorably among those evaluated by the Financial Stability Board (FSB), while it was ranked first in the IMF’s 2012 assessment of countries’ compliance with Basel principles. The Brazilian financial system is well capitalized and shows comfortable levels of liquidity indicators, with a Basel index of 16.7, a provisions-to-capital ratio, net of delinquencies, of 11 percent, and a net-assets-to-short-term-liabilities ratio of about 200.9

Overview of Monetary and Macroprudential Policy in Brazil After the Financial Crisis

Brazilian banks were not exposed to subprime loans or troubled assets during the global financial crisis. Hence, the crisis affected the Brazilian financial system mostly through the liquidity channel.10 However, uncertainties with respect to the viability of small banks that were negatively affected by the shortage of foreign credit lines caused a temporary disruption in interbank liquidity provision, particularly for smaller banks.

To even out liquidity positions in the interbank market, the Brazilian central bank implemented unconventional changes in reserve requirement regulations. These instruments were important to give the central bank room to maneuver in moments of distress.

On another front, policy measures were adopted to reduce the volatility caused by strong international liquidity inflows into the country, a response to quantitative-easing programs and unconventional policies adopted by central banks in major advanced economies. Some of those measures aimed to reduce the incentives for foreign investors to invest in short-term assets, while others called for stricter requirements on banks’ foreign exchange exposure.

Signs of a possible credit crunch led to a set of regulatory relief measures, triggering intervention by public banks in the credit market mainly through looser credit origination conditions. The strong response of public banks changed the composition of credit in the system and fueled an important acceleration of consumer indebtedness, to a point where the country ranked sixth in the world in terms of household debt service and principal payments to income.11 More recently, household indebtedness with respect to housing loans shows signs of moderation, while indebtedness with respect to other credit segments is clearly decelerating (Figure 8.1).

Figure 8.1.Household Indebtedness and Income Commitment in Brazil

(Percent)

Source: Central Bank of Brazil.

To date, the evolution of consumer indebtedness has not posed a large threat to financial stability, given relatively low levels of credit to GDP, but looser terms of credit origination have contributed to vulnerabilities in some market segments.

The regulatory policy response to this was either specific to identified vulnerabilities in certain markets or of a more general nature. For instance, the implementation of risk-weight factors directly related to the maturity and loan-to-value of credit operations proved effective in constraining their impact on specific targets. Martins and Schechtman (2013) and Afanasieff and others (2015) provide evidence supporting the precision of the measures adopted in 2010 for auto loans. In some instances, the direction of macroprudential measures was in line with the monetary stance of the economy.

An important challenge facing both the regulatory and the monetary authorities has been the fast and intense financial deepening process that Brazil has undergone over recent years. Figure 8.2 shows the evolution of banks’ regional presence in Brazil, and Figure 8.3 shows the migration of social classes over the years. Financial inclusion has been the result of technological and regulatory improvements in the financial system, income distribution policies, public banks’ credit origination policies, and a stable macroeconomic environment.

Figure 8.2.Evolution of Banks’ Regional Presence in Brazil

Source: Central Bank of Brazil.

Note: Points refer to bank counters, branches, or correspondents.

Figure 8.3.Social Mobility in Brazil

Source: Central Bank of Brazil, Fundação Getulio Vargas (FGV).

Sahay and others (2015) find a positive relationship between the pace of financial deepening and the risk of crisis and macroeconomic instability, conditional on the efficiency of a financial system’s regulation and supervision. To avoid these risks, the Brazilian central bank has closely monitored the financial deepening process and the quality of credit origination so that credit growth and income commitment are kept within sustainable boundaries. Essential to this task is the Credit Bureau, created in 1997 and restructured in 2008, which collects detailed information on each and every credit origination in the banking system above R$1,000 (about US$300). Such transactions currently account for 99 percent of the entire credit portfolio of the Brazilian financial system. The Credit Bureau is managed by the central bank, and the information available is processed and analyzed on a daily basis not only by the supervisory and regulatory departments of the central bank, but also by economic departments, constituting an important input for a broad set of policy decisions.

Other important risk-mitigating measures have been put in place, including: the approval of the Credit Default Law in 2005;12 improvements in the Deposit Guarantee Fund (including the introduction of a similar fund for cooperative credit unions that target low-income borrowers); enactment of a law that approved the creation of a positive borrowers’ record (in addition to credit registries); and a derivatives exposure registry (CED).

The central bank’s capacity to monitor the Brazilian financial system is in many respects unique in the world. Not only is it comprehensive in terms of banks’ operations, portfolios, and exposures, it also responds in a timely manner and is appropriately designed to monitor and detect inconsistencies in the wide range of information given to the central bank. Table 8.1 offers a glimpse at the dimension of database monitoring at the Central Bank of Brazil.

Table 8.1Database Monitoring at the Central Bank
Assets and securities markets
  • Data sources: SELIC, CETIP, BM&F BOVESPA, Brazilian Payments System, and all financial institutions

  • Processes 40 million registers per day

  • Processes over 900 documents per month

  • Produces daily macro- and microprudential analysis on liquidity and market risks of the financial system

  • Monitors the market for public bonds and the behavior of bank funding daily

  • Releases information on the Central Bank of Brazil’s website

Credit operations
  • Data sources: Monthly information from financial institutions

    • 480 million operations

    • Credit operations outstanding of 75 million clients

    • Each operation has 36 information fields

  • Produces monthly micro- and macroprudential analysis on credit risks of the financial system

  • Manages the Credit Bureau System and publishes information for the public and for financial institutions on credit operations

  • Releases information on the Central Bank of Brazil’s website

Purchasing consortium groups
  • Information on 13.7 million quotas, distributed among 21,000 groups

  • Data on 9 million quotas with past-due earnings

  • 880 million data registers received on a quarterly basis and 60,000 received on a monthly basis

  • Produces individual and aggregate quarterly analysis on the purchasing consortium segment

  • Releases information on the Central Bank of Brazil’s website

Foreign exchange operations
  • Foreign exchange system:

    • 207 authorized financial institutions

    • 7.8 million operations per year

  • 31,000 operations per day

  • Central bank receives additional 25.5 million operations per year via monthly files

  • Produces daily microprudential analysis on foreign exchange operations carried out by financial institutions

  • Monitors foreign inflows and the foreign exchange flow daily

  • Releases information on the Central Bank of Brazil’s website

Accounting information
  • Receives 1,136 bank financial statements on a monthly basis and 2,267 limit statements (600,000 monthly registers)

  • Receives over 7,300 documents on a quarterly basis (2.2 million quarterly registers)

  • Produces monthly macro- and microprudential analysis on the financial-economic situation of the financial institutions

  • Monitors the adherence of financial institutions to regulatory operational limits monthly

  • Releases information on the Central Bank of Brazil’s website

Others
  • Other sources of information:

    • Regulators

    • Deposit guarantee fund

    • Custody chambers

    • Registry chambers

    • External auditors

    • Rating agencies

    • International organizations—Financial Stability Board

    • Government databases

    • Private databases—SERASA

    • Institutions that are not regulated by the Central Bank of Brazil

Source: Central Bank of Brazil.
Source: Central Bank of Brazil.

Another factor that has limited the impact of higher default rates of low-income borrowers to the rest of the financial system is the fact that theseloans have been mostly originated by a public bank (Caixa Econômica Federal) as part of a wider policy of social inclusion.13

Interaction Between Monetary and Macroprudential Policy

Monetary and financial stability are the core missions of the Central Bank of Brazil. The Monetary Policy Committee (COPOM) was created by the central bank in 1996 with the purpose of setting the monetary policy stance,14 and since 1999 its decisions must be directed toward achieving the inflation targets set by the National Monetary Council.15

The Financial Stability Committee (COMEF) was created by the central bank in 2011 to set directives and guidelines for central bank conduct in order to preserve financial stability, assess systemic risk, and carry out macroprudential over-sight.16 Although COMEF’s guidelines are enforced, the board of governors is not constrained by the meeting days of COMEF or COPOM to set the central bank’s policy instruments (with the exception of the monetary policy interest rate). In addition, both COPOM and COMEF comprise the same members, the Central Bank of Brazil’s board of governors. Board members have restated that each committee’s objectives and decisions are independent. However, communication, which is seen as essential to avoid misperceptions that could undermine policy effectiveness, remains a challenge.

There are pros and cons in having supervision and financial regulation within the central bank. For instance, the IMF (2013) considers this double assignment a vulnerability of the overall Brazilian regulatory and supervisory framework, as it brings time consistency issues and communication challenges.

However, as also stated in the IMF report, there are several benefits from this arrangement, such as having macroprudential policy decisions drawing on central bank expertise in financial and macroeconomic analyses and data availability, which facilitates the analysis of the side effects of each policy. The report also mentions the benefits of better shielding macroprudential policy from political influence compared to when this function is assigned to a separate regulatory body.

In addition to continuously improving the regulatory and supervisory stance of the Brazilian financial system, the central bank has actively used a variety of instruments to try to influence the financial cycle, with either narrow or broader purposes. Important policy choices for these purposes have been risk-weight factors, reserve requirements, and taxation of foreign capital inflows; overall capital requirement ratios have remained unchanged since the implementation of Basel I. In most instances, communication of the targeted impact of policy decisions has not been detailed and the macroprudential policy decision framework remains highly discretionary.

During the inflation-targeting period, monetary policy has followed the traditional inflation-targeting framework, with the (SELIC) interest rate being the central policy instrument. Very rarely were reserve requirements explicitly used to reinforce the monetary policy stance.17 Their use has been associated mostly with macroprudential purposes, and only occasionally have they been used to attain other goals, such as draining liquidity from the large inflows of foreign capital, or in times of distress in government bonds issuances.

National Monetary Council Resolution #4193 of March 1, 2013 instituted the additional conservation and countercyclical bank capital requirement to come into effect in 2016. According to Central Bank of Brazil Comunicado #20615 of February 17, 2011, the countercyclical capital requirement will be activated in case of excessive credit growth that potentially builds up systemic risk. Any changes in the countercyclical capital requirement should be announced one year in advance. So far, the decision framework for the activation of this instrument is still work in progress.

Given that currently available policy instruments have been used to affect the financial cycle, and a new instrument is soon to be implemented, several questions arise. How should these instruments interact? How strongly should they respond to the financial cycle? Should all of them be used for the same purpose? In addition, given the unsettled debate on whether monetary policy should be concerned with financial stability, what would be the recommendation for Brazil?

Our contribution to the normative perspective of macroprudential regulation in Brazil is to use a model that was built and estimated for Brazil to find an optimal combination of macroprudential and monetary policies that are allowed to react to the financial cycle, which in this study is represented by the credit gap. We focus on the (wide) set of macroprudential instruments that have been more intensely used in Brazil to influence credit markets, especially after the financial crisis, namely reserve requirements on time, savings, and demand deposits and risk-weight factors on consumer, commercial, and housing loans, in addition to the new countercyclical capital buffer and monetary policy.

Optimal Policy

To search for the optimal monetary and macroprudential policy combination, we use the DSGE model with financial frictions developed by Carvalho and Castro (2015), which incorporates the main features of the Brazilian credit market, including the heavily regulated housing loan market. This model was estimated with Brazilian data spanning the inflation-targeting regime. The model was carefully built to allow for relevant policy analysis at the Central Bank of Brazil and was shown to fit the empirical behavior of several key policy variables.

Consumer loan origination in the model is dependent on lenders’ expectations with respect to borrowers’ capacity to pay loans with future labor income, with endogenous default. This has been an important feature of the Brazilian credit market. Housing loans use houses as collateral, but indebtedness in this market affects borrowers’ available income, affecting their decisions with respect to consumer credit. Housing loan payments have seniority over consumer loans, which explains the very low default rates compared with the consumer credit segment observed in the data. Commercial credit takes capital as collateral and also faces endogenous default.

In addition to financial frictions that aim to represent the Brazilian credit market, the model also incorporates important features regarding Brazil’s connection with the rest of the world. The model includes all major balance of payment accounts, with special attention to foreign direct investment (FDI), which has been the most important source through which foreign capital has accumulated in the country. The interaction of FDI with the financial system is indirect. The recipients of FDI flows are entrepreneurs who also fund their projects with bank loans.

The real sector of the economy is modeled in line with the standard DSGE literature. Households are distributed in groups of savers and borrowers, both supplying labor to a continuum of labor unions that operate under monopolistic competition, and consuming goods and housing. Savers have a wider array of possible investment opportunities and are more patient than borrowers, who take out risky loans for consumption and housing. Entrepreneurs manage productive capital. Domestic producers combine capital and labor to produce intermediate goods that will be combined with imported intermediate goods to produce final goods for consumption (private and public), investment, and exports. Price frictions are introduced in both domestic and imported intermediate goods. The model also incorporates capital and housing investment producers. Exporting firms face adjustment costs on quantum changes and take working capital loans from domestic banks. Figure 8.4 shows the structure of the real economy.

Figure 8.4.The Real Sector of the Carvalho and Castro (2015) Open-Economy Dynamic Stochastic General Equilibrium Model with Financial Frictions for Brazil

The financial sector is composed of a retail money market fund that takes deposits from savers and issues foreign debt to invest in banks’ time deposits and government bonds. The banking conglomerate is composed of a continuum of competitive banks that get funding from deposit branches and extend credit to households, entrepreneurs, and export firms through their lending branches. They optimally choose their balance sheet composition, subject to regulatory requirements and several frictions intended to replicate banks’ incentives to deal with regulatory constraints. They can accumulate capital by retaining profits, the choice variable in the intertemporal dynamic optimization problem of the bank. Figure 8.5 shows the financial structure of the model.

Figure 8.5.The Financial Flows in the Carvalho and Castro (2015) Open-Economy Dynamic Stochastic General Equilibrium Model with Financial Frictions for Brazil

The model has the following macroprudential instruments: reserve requirements on demand, savings and time deposits, risk weights on consumer, commercial and housing loans, tax on credit, and standard minimum capital requirement ratios. Reserve requirements on demand deposits are not remunerated, whereas the other types of reserve requirements are remunerated at exactly the same rate that accrues on bank deposits. In the benchmark (estimated) model, neither macroprudential policies nor monetary policy respond to the credit cycle.

One advantage of analyzing the interaction between monetary and macroprudential policies using DSGE models is that these models can account for the side effects that the use of one policy tool has on the targets of the others.

We use the model to seek an optimal combination of macroprudential and monetary policy rules that may react to the credit gap. For each of the exercises we perform, the optimal policies are obtained by minimizing a loss function composed of the volatility of output, inflation, the policy interest rate, and total credit.18 The weights on output, inflation, and interest rate volatility in this loss function are obtained in such a way that the minimization of a loss function composed of only these three variables would result in an optimal monetary policy rule in the benchmark model, in which the policies do not directly respond to the credit gap. The weight attributed to credit is arbitrary.19

Optimization takes into account all sources of fluctuation in the model, an approach that is also adopted by Lambertini, Mendicino, and Punzi (2013). Since the model is estimated, the influence of each shock on the optimal solution will rely on realistic values of the stochastic processes governing the shocks. Several studies address the optimal responses to a few selected shocks, but given the fact that in practice considerable judgment is required to assess the real-time source of a shock driving economic variables, it is equally important to find an optimal rule that could be transparent to and predictable for the public.

The optimal simple monetary and macroprudential policies are allowed to react to the credit gap. Monetary policy follows an augmented, forward looking, Taylor-type rule:

where π¯t is the nonzero inflation target, yt is GDP detrended by permanent technology and population growth shocks, bE,t, bC,t, and bH,t are commercial, consumer, and housing credit gaps from the stationary trend driven by permanent technology and population growth shocks, all variables indexed as “ss” represent steady-state values, y is steady-state detrended GDP, R is the steady-state interest rate, and εR,t is a white noise shock.

The capital requirement ratio is augmented with the countercyclical capital buffer:

where the traditional component is centered on the current required ratio (11 percent):

and the countercyclical capital buffer follows:

where the steady state value of ΓCC,t is 1.

Reserve requirement ratios on demand, savings, and time deposits react to the total credit gap according to the following policy rules:

where τD,t, τS,t, and τT,t are reserve requirement ratios on demand, savings, and time deposits, εD,t, εS,t, and εT,t are white noise shocks, and all variables indexed as “ss” represent steady-state values.

Actual capital adequacy is calculated as the ratio of bank capital to risk-weighted assets:

and risk-weighted assets are computed according to

where ςE,t, ςC,t, and ςH,t are risk-weight factors on commercial, consumer, and housing loans, ςB,t is the risk-weight factor on banks’ portfolios of liquid assets, which in the model is composed of risk-free public bonds, and hence ςB,t = 0. The last term, vt, is an AR(1) process to account for the share of Brazilian financial system assets that are not formally included in the model.

Risk-weight factors are allowed to react to their specific credit segments, since Carvalho and Castro (2015) show that these instruments have a primary impact on their specific credit segments.20 They can be expressed according to the following policy rules:

where εE,t, εC,t, and εH,t are white noise shocks.

We follow Schmitt-Grohé and Uribe (2007) and focus on simple and implementable policy rules. We find the optimal coefficients {ρ, ρCC, ρD, ρS, ρT, ρE, ρC, ρH, γπ, γy, χ, γCC, γD, γS, γT, γE, γB, γH} of the policy rules in equations (8.1), (8.4), (8.5), (8.6), (8.7), (8.10), (8.11), and (8.12) that minimize the loss function,21 where all autoregressive parameters are restricted to the (0,1) interval and the policy parameters are unconstrained.

Table 8.2 shows the optimization results for three possible weights for the credit gap in the loss function, considering all policy instruments operating simultaneously. For each weight, we proceed with two types of optimization: one in which we do not constrain the support of the policy parameters and another in which we constrain the reaction of monetary policy to the credit gap to the nonnegative support. Constraining the set of possible solutions for optimal simple rules is common in the literature.22

Table 8.2Optimal Simple Rules: Comparing Constrained and Unconstrained Optima at Different Credit Gap Weights in the Loss Function
Credit Gap Weight = 0.001Credit Gap Weight = 0.01Credit Gap Weight = 0.5
RulesReaction Parameters of the RulesUnconstrained OptimumConstrained OptimumUnconstrained OptimumConstrained OptimumUnconstrained OptimumConstrained Optimum
Monetary PolicyCoefficient of reaction to inflation4.444.414.044.041.814.39
Coefficient of reaction to output0.900.891.720.852.695.95
Coefficient of reaction to total credit gap−0.060.00−0.110.00−0.350.75
Autoregressive coefficient0.930.930.930.940.910.93
Risk-Weight Factors (RWFs)RWF of consumer loans: Reaction to consumer credit0.100.120.090.090.270.33
RWF of consumer loans:0.990.990.990.990.840.96
Autoregressive coefficient
RWF of commercial loans:0.830.840.840.84−0.930.00
Reaction to commercial credit
RWF of commercial loans:0.290.300.310.310.560.56
Autoregressive coefficient
RWF of housing loans:0.630.620.610.610.490.49
Reaction to housing credit
RWF of housing loans:0.980.980.980.970.000.00
Autoregressive coefficient
Reserve Requirements (RRs)RR on time deposits: Reaction to total credit gap20.9620.9620.9620.9620.9121.07
RR on time deposits:0.030.020.000.000.670.98
Autoregressive coefficient
RR on demand deposits:5.025.025.025.025.025.16
Reaction to total credit gap
RR on demand deposits:0.000.000.000.000.760.84
Autoregressive coefficient
RR on savings deposits:10.9710.9610.9610.9610.8711.26
Reaction to total credit gap
RR on savings deposits:0.890.950.900.910.270.48
Autoregressive coefficient
Countercyclical Capital Buffer (CCB)Autoregressive coefficient0.460.450.440.440.460.48
Reaction to total credit gap3.113.113.113.113.1010.21
Value of the objective function0.001070.001080.001890.001940.004040.00462
Variation coefficient0.0130.0130.0150.0150.0250.024
Inflation0.0130.0150.0250.024
Interest rate0.0120.0120.0140.0140.0230.024
Output0.0570.0560.0520.0500.0470.038
Credit-to-GDP0.0750.0760.0690.0700.0460.054

In the constrained solution, we find that, in general, increasing the weight of the credit gap in the loss function increases the volatility of inflation and of the interest rate in the optimal solution, but reduces the volatility of credit and output. The relative magnitude of the former is substantially higher than the latter, implying that increasing the importance attributed to the financial cycle in monetary policy comes with an important cost in terms of the inflation target. In addition, the optimized constrained rules require a very aggressive response of monetary policy to inflation. As the weight of the credit gap in the loss function increases, so does the optimal monetary policy reaction to the output gap, which is an indirect channel to affect borrowers’ incentives to get loans. Only with very large weights of the credit gap is it optimal for monetary policy to react to the credit gap within the constrained solutions. For low values of the weight associated with the credit gap, the constrained solutions achieve losses that are very close to the unconstrained solutions, suggesting some flatness in this region of constrained objective function. With respect to unconstrained solutions, in all of them the optimal reaction of monetary policy to the credit gap is found to be negative, somehow counterbalanced by an increase in the reaction to output. Although this result is not unprecedented in the literature (Faia and Monacelli 2007),23 it is unlikely that the monetary policymaker will implement such a response. In addition, as mentioned earlier, the gains from adopting these policy combinations are only relevant when the weight of the credit gap in the loss function is very high. Hence, we shall restrict our analysis to the solutions where the monetary policy reaction to the credit gap is nonnegative.

In Brazil, changes in minimum capital requirement ratios have to be authorized by the National Monetary Council (CMN), which is composed of not only the Central Bank of Brazil’s governor, but also the Minister of Finance and the Minister of Budget and Planning. With the implementation of Basel III, the countercyclical capital buffer can be set by the central bank, but needs to be announced 12 months in advance of its implementation. This constraint does not exist for reserve requirements or risk-weight factors. Hence, timely policy reactions to imbalances in the financial system are easier to be implemented through alternative policy instruments than with minimum capital requirements or even the countercyclical buffer.

Given the fact that a number of macroprudential instruments are available to the Brazilian central bank and can be more easily and immediately changed than capital requirements, we investigate whether optimal simple rules that make up only subsets of the available macroprudential tools can perform as well as the entire optimal set of macroprudential policy rules that react to the credit gap.

The following subsets are analyzed: (1) monetary policy with all macroprudential instruments; (2) monetary policy and the countercyclical capital buffer; (3) monetary policy, risk-weight factors, and reserve requirements; (4) monetary policy, the countercyclical capital buffer, and risk-weight factors; (5) monetary policy and risk-weight factors; (6) monetary policy and reserve requirements; and (7) all of the former combinations except for monetary policy.

Table 8.3 shows the optimum for each subset of optimal policy rules that include monetary policy. In most optimized combinations, the solution is pushed towards a very aggressive response of monetary policy to inflation and to output. The most important result in this exercise is that the subset of policies that includes monetary policy, risk-weight factors, and reserve requirements achieves almost the same loss as that obtained from the complete set of rules.

Table 8.3Optimal Simple Rules: Comparing Constrained Optima Using Different Subsets of Policy Rules for a Loss Function Credit-Gap Weight of 0.001
RulesReaction Parameters of the RulesComplete Set: Monetary Policy & CCB & RR & RWF1Monetary Policy & CC & RWFMonetary Policy & CCMonetary Policy & CC & RRMonetary Policy & RWF & RRMonetary Policy & RWFMonetary Policy & RR
Monetary PolicyCoefficient of reaction to inflation4.412.864.404.875.426.736.73
Coefficient of reaction to output0.890.400.881.631.300.421.15
Coefficient of reaction to total credit gap0.001.000.000.100.010.000.08
Autoregressive coefficient0.930.900.930.950.950.940.95
Risk-Weight FactorsReaction to consumer0.121.241.201.51
(RWFs)RWF of consumer loans: Autoregressive coefficient0.990.970.920.96
RWF of commercial loans:0.840.900.567.64
Reaction to commercial credit
RWF of commercial loans:0.300.960.930.72
Autoregressive coefficient
RWF of housing loans:0.620.220.070.18
Reaction to housing credit
RWF of housing loans:0.980.950.980.89
Autoregressive coefficient
Reserve RequirementsReaction to total credit gap20.9616.497.3822.01
(RRs)RR on time deposits:0.020.990.990.00
Autoregressive coefficient
RR on demand deposits:5.021.340.640.21
Reaction to total credit gap
RR on demand deposits:0.000.450.890.94
Autoregressive coefficient
RR on savings deposits:10.967.803.4233.83
Reaction to total credit gap
RR on savings deposits:0.950.990.000.99
Autoregressive coefficient
Countercyclical CapitalAutoregressive coefficient0.450.500.000.51
Buffer (CCB)Reaction to total credit gap3.110.1613.760.29
Objective0.001080.001490.001260.001140.001080.001400.00116
Variation coefficient0.0130.0130.0130.0140.0130.0110.012
Inflation
Interest rate0.0120.0130.0120.0130.0120.0120.012
Output0.0560.0620.0610.0550.0550.0640.062
Credit-to-GDP0.0760.1470.1040.0760.0810.1430.082

However, the optimal responses of monetary policy to inflation and output obtained in this exercise are very far from the values usually obtained in Taylor rule estimations using actual data. Hence, we proceed with a search for optimal simple rules that take these traditional parameters of the monetary policy as given, setting them according to the mode of the posterior distribution of the parameters estimated in Carvalho and Castro (2015). In other words, we seek to find simple and optimal macroprudential rules (reserve requirements, risk-weight factors, and countercyclical capital buffers) that can react to the credit gap, also allowing for the reaction coefficient of monetary policy to the credit gap to be obtained optimally.

Table 8.4 shows the results of this exercise. For the constrained optimal simple rules, we find that some subsets of macroprudential policy can perform almost as well as the complete set. The following subsets yield losses that are merely about 2 percent higher than the one with the complete set: (1) reserve requirements and the countercyclical capital buffer; (2) monetary policy reaction to the credit gap together with reserve requirements; and (3) monetary policy reaction to the credit gap together with risk-weight factors and reserve requirements. The combination of monetary policy, reserve requirements, and risk-weight factors reacting together to the credit gap requires a milder countercyclical response of each instrument. If only reserve requirements are allowed to help monetary policy react to the financial cycle, the optimal response of each of these instruments to the credit gap becomes very aggressive. Instead, if the countercyclical capital buffer is used together with reserve requirements and monetary policy, the latter loses its importance to directly target the credit cycle.

Table 8.4Optimal Simple Rules: Comparing Constrained Optima Using Different Subsets of Policy Rules for a Loss Function Credit-Gap Weight of 0.001 and for a Given Monetary Policy Reaction to Inflation and Output
RulesReaction Parameters of the RulesMonetary Policy & RR & RWF & CCBMonetary Policy & RWF & CCBMonetary Policy & RWF & RRMonetary Policy & RR & CCBMonetary Policy & RRMonetary Policy & RWF
Monetary PolicyCoefficient of reaction to total credit gap0.000.000.010.000.200.00
Risk-Weight Factors (RWFs)RWF of consumer loans: Reaction to consumer credit0.652.320.398.06
RWF of consumer loans: Autoregressive coefficient0.960.930.990.80
RWF of commercial loans: Reaction to commercial credit0.303.730.2316.75
RWF of commercial loans: Autoregressive coefficient0.680.990.730.90
RWF of housing loans: Reaction to housing credit0.070.260.0490.30
RWF of housing loans: Autoregressive coefficient0.710.950.990.13
Reserve RequirementsRR on time deposits: Reaction to total credit gap4.392.3517.4266.32
(RRs)RR on time deposits: Autoregressive coefficient0.980.730.94
RR on demand deposits: Reaction to total credit gap0.360.180.26
RR on demand deposits: Autoregressive coefficient0.850.730.87
RR on savings deposits: Reaction to total credit gap1.991.111.66
RR on savings deposits: Autoregressive coefficient0.120.730.59
Countercyclical Capital BufferCCB: Autoregressive coefficient0.500.500.50
(CCB)CCB: Reaction to total credit gap0.110.290.05
Objective0.001380.001670.001410.001410.001410.00152
Variation coefficient
Inflation0.0150.0150.0150.0150.0150.015
Interest rate0.0150.0150.0150.0150.0150.015
Output0.0600.0640.0610.0620.0630.061
Credit-to-GDP0.0870.1420.0920.0810.0800.122
Note: In this exercise, monetary policy reaction to inflation and output is set according to the model estimated in Carvalho and Castro (2015), that is, ρ = 0.829, γπ = 1.961, and γy = 0.185.
Note: In this exercise, monetary policy reaction to inflation and output is set according to the model estimated in Carvalho and Castro (2015), that is, ρ = 0.829, γπ = 1.961, and γy = 0.185.

Table 8.5 shows the subsets of optimal simple macroprudential rules obtained when we do not include the possibility that monetary policy reacts to the credit gap. In this case, most subsets achieve very similar losses. However, the loss is the highest when only risk-weight factors are allowed to react to the credit gap.

Table 8.5Optimal Simple Rules: Comparing Constrained Optima of Subsets of Policy Rules for a Given Monetary Policy Rule and a Loss Function Credit-Gap Weight of 0.001
RulesReaction Parameters of the RulesRR & RWF & CCBRWF & CCBRWF & RRRR & CCBRRRWF
Risk-Weight FactorsRWF of consumer loans: Reaction to consumer credit0.110.170.123.31
(RWFs)RWF of consumer loans: Autoregressive coefficient0.990.990.990.91
RWF of commercial loans: Reaction to commercial credit0.73−0.040.385.28
RWF of commercial loans: Autoregressive coefficient0.710.990.850.99
RWF of housing loans: Reaction to housing credit0.370.150.030.36
RWF of housing loans: Autoregressive coefficient0.990.990.990.95
ReserveRR on time deposits:5.232.6820.6817.42
Requirements (RRs)Reaction to total credit gap
RR on time deposits:0.980.990.960.94
Autoregressive coefficient
RR on demand deposits:0.520.272.030.26
Reaction to total credit gap
RR on demand deposits:0.840.880.450.60
Autoregressive coefficient
RR on savings deposits:2.641.3310.111.64
Reaction to total credit gap
RR on savings deposits:0.380.890.330.49
Autoregressive coefficient
CountercyclicalAutoregressive coefficient0.500.500.51
Capital Buffer (CCB)Reaction to total credit gap0.0525.580.23
Objective0.001360.001360.001380.001400.001410.00167
Variation coefficient
Inflation0.0150.0150.0150.0150.0150.015
Interest rate0.0150.0150.0150.0150.0150.015
Output0.0590.0600.0590.0620.0620.064
Credit-to-GDP0.0850.0840.0890.0800.0810.141
Note: In this exercise, we set the monetary policy parameters according to the model estimated in Carvalho and Castro (2015), that is, ρ = 0.829, γπ = 1.961, γy = 0.185, and χ = 0.
Note: In this exercise, we set the monetary policy parameters according to the model estimated in Carvalho and Castro (2015), that is, ρ = 0.829, γπ = 1.961, γy = 0.185, and χ = 0.

Reserve requirements and risk-weight factors have actually been used on a number of occasions for macroprudential purposes in Brazil, especially after the financial crisis, and some of them countercyclically. Our results corroborate the perception that the direction of these policies can be used to help correct the buildup of risks in the Brazilian financial system with an efficiency to minimize the volatility of targeted macroeconomic variables similar to that of the combination with countercyclical capital. In addition, the other macroprudential instruments are easier to implement and have targeted impact on specific variables.

Impulse Responses Under Different Rules

Next, we compare the dynamic responses of the model under four different combinations of policy rules. In the first (“Benchmark”), the policy rules do not react to credit, and the model is exactly the one estimated in Carvalho and Castro (2015). The second combination is composed of optimal simple rules for the countercyclical capital buffer, reserve requirements, risk-weight factors, and monetary policy (all parameters in the augmented Taylor rule are included in the optimization but are constrained to the nonnegative support, and the autoregressive parameters are constrained to the 0–1 interval). The third combination refers to optimal simple rules for the countercyclical capital buffer, risk-weight factors, and reserve requirements, taking all monetary policy parameters as given and set at the mode of the posterior estimated in Carvalho and Castro (2015). The last combination is composed of optimal simple rules for risk-weight factors and reserve requirements, also taking all monetary policy parameters as given, and set at the mode of the posterior estimated in Carvalho and Castro (2015). As mentioned earlier, the complete set of optimal simple rules that includes monetary policy requires a very aggressive response to inflation.

Figures 8.6 and 8.7 focus on exogenous shocks originating in the banking sector. In Figure 8.6, the model is perturbed by a negative shock to bank capital, which is close in meaning to what Gertler, Kiyotaki, and Queralto (2012) refer to as a “crisis shock.” This shock simulates, for instance, the impact of a drop in bank capital due to losses that negatively affect banks’ net worth. In Figure 8.7, the model is shocked with a drop in banks’ preference for liquidity, which simulates a situation in which banks reduce their risk aversion and try to increase their exposure to the credit risk. This would result in a relaxation in credit origination conditions. For both shocks, the responses under countercyclical macroprudential policy rules are strikingly different from those obtained from the benchmark model. The optimal policies sharply reduce the volatility of total credit, and the dynamics of the main economic variables under subsets of active policy rules are very close to the complete set of active rules. The main difference stands in the banks’ decisions concerning balance sheet locations and dividend distribution. For the bank capital shock, the subsets of optimal policies usually generate stronger responses in bank variables. With respect to the shock on bank liquidity preferences, since the complete policy combination requires a very strong response of reserve requirements to credit conditions, bank liquidity is more severely affected than in the case of subsets of optimal policy rules. Dividend distribution is also more strongly impacted in the case of the complete set.

Figure 8.6.Comparing Combinations of Optimal Simple Macroprudential and Monetary Policy Rules: Negative Shock to Bank Capital—10 Percent Drop on Impact

Note: bp = basis points; CapitalReq = capital requirements; MoP = monetary policy; OSR = optimal simple rules; pp = percentage points; RR = reserve requirements; RWF = risk-weight factors; ss dev = % change from the steady trend.

Figure 8.7.Comparing Combinations of Optimal Simple Macroprudential and Monetary Policy Rules: Bank Liquidity Preference Shock—65 Percent Drop on Impact

Note: bp = basis points; CapitalReq = capital requirements; MoP = monetary policy; OSR = optimal simple rules; pp = percentage points; RR = reserve requirements; RWF = risk-weight factors; ss dev = % change from the steady trend.

The monetary policy shock makes it more clear how aggressive the complete set of optimal simple rules is in terms of its impact on the real economy (Figure 8.8). To stabilize credit under a monetary policy shock, the complete set of optimal simple rules requires a more sluggish response of interest rates, substantially affecting output, consumption, labor market conditions, and housing investment. The subset of optimal policies does almost as well as the complete set in stabilizing credit, but it also has the potential to stabilize real economy variables. In fact, the subsets of optimal simple rules improve the stabilization of the real economy compared to the benchmark model, a feature that cannot be observed in the complete set.

Figure 8.8.Comparing Combinations of Optimal Simple Macroprudential and Monetary Policy Rules: Monetary Policy Shock—100 Basis Point Increase on Impact

Note: bp = basis points; CapitalReq = capital requirements; MoP = monetary policy; OSR = optimal simple rules; pp = percentage points; RR = reserve requirements; RWF = risk-weight factors; ss dev = % change from the steady trend

Figures 8.9 to 8.12 compare the model dynamics after external shocks. In the estimated benchmark model, a drop in world output has a recessionary impact on the domestic economy, with a significant reduction in investment and consumption. An increase in foreign direct investment inflows has an expansionary impact on domestic credit, but the impact on inflation and output is contractionary given the appreciation of the domestic currency. An increase in the world interest rate leads to a depreciation of the exchange rate, which calls for a response of monetary policy. A commodity price boom, which in the model is represented by a shock to export prices,24 has an expansionary impact on credit, through the increase in available income and a surge in investment. In all cases, the optimal simple rules stabilize credit, and the impact of each subset of rules on real economy variables depends on the credit segment that is most significantly affected by each policy combination. The main difference in the dynamic responses of the model to different subsets of optimal policies occurs for banks’ balance sheet variables, given that each subset requires a different reaction from each macroprudential instrument, affecting banks’ incentives distinctly.

Figure 8.9.Comparing Combinations of Optimal Simple Macroprudential and Monetary Policy Rules: Shock to World Output—1 Percent Drop

Note: bp = basis points; CapitalReq = capital requirements; MoP = monetary policy; OSR = optimal simple rules; pp = percentage points; RR = reserve requirements; RWF = risk-weight factors; ss dev = % change from the steady trend.

Figure 8.10.Comparing Combinations of Optimal Simple Macroprudential and Monetary Policy Rules: Shock to Foreign Direct Investment Flows—1 Percentage Point Increase

Note: bp = basis points; CapitalReq = capital requirements; MoP = monetary policy; OSR = optimal simple rules; pp = percentage points; RR = reserve requirements; RWF = risk-weight factors; ss dev = % change from the steady trend.

Figure 8.11.Comparing Combinations of Optimal Simple Macroprudential and Monetary Policy Rules: Shock to Foreign Interest Rates—100 Basis Point Drop

Note: bp = basis points; CapitalReq = capital requirements; MoP = monetary policy; OSR = optimal simple rules; pp = percentage points; RR = reserve requirements; RWF = risk-weight factors; ss dev = % change from the steady trend.

Figure 8.12.Comparing Combinations of Optimal Simple Macroprudential and Monetary Policy Rules: Shock to Export Prices—20 Percent Increase

Note: bp = basis points; CapitalReq = capital requirements; MoP = monetary policy; OSR = optimal simple rules; pp = percentage points; RR = reserve requirements; RWF = risk-weight factors; ss dev = % change from the steady trend.

In the benchmark model of Carvalho and Castro (2015), macroprudential policy instruments that do not react to economic or financial cycles are more effective in stabilizing the credit-to-GDP ratio when shocks originate in the financial system. In fact, each instrument will have a potential niche where its impact is more pronounced. In general, macroprudential policies have a greater impact on financial variables, whereas monetary policy has a stronger effect on real variables, except for the case of housing loans, which are strongly influenced by monetary policy. Capital requirements have a strong impact on capital investment given the sensitivity of the value of capital to available funding from bank loans. Risk-weight factors have a substantial impact on actual capital adequacy ratios. Reserve requirements have the most important impact on bank liquidity.

The Impact of Macroprudential Policy Announcements on Inflation Expectations

A strand in the literature advocates that monetary policy can exacerbate risks to financial stability. In the case of Brazil, very few studies have investigated this risk-taking channel. Tavares, Montes, and Guillén (2013) study the impact of monetary policy on bank risk perception, associating the stance of monetary policy with lending spreads and on insurance engaged by borrowers against credit default. They find that contractionary monetary policy induces banks to use more insurance (the reverse is true in the case of expansionary monetary policy). The same is observed for reserve requirements, which also affect the risk-taking behavior of banks through insurance. Montes and Peixoto (2012) also find a positive relation between bank risk perception and the stance of monetary policy in Brazil.

To the best of our knowledge, the reverse channel has not been explored, especially for Brazil. Macroprudential policy announcements can have an impact on variables that are targeted by the monetary authority, and, depending on the coordination of business and financial cycles, macroprudential policy announcements can have an impact on the anchoring of inflation expectations.

According to the IMF (2013), monetary and macroprudential policies were complementary in Brazil during the postcrisis period. Both were used countercyclically, leaning against the business and the financial cycle, which were synchronized during the period analyzed. In more recent times, however, this synchronization has been challenged, and while monetary policy became more contractionary given inflationary pressures, some macroprudential measures were implemented with the purpose of easing credit conditions in specific segments.

To investigate whether macroprudential policy announcements had a significant impact on the anchoring of inflation expectations in Brazil, we select events when macroprudential policy was changed by explicitly targeting credit-related variables and assess their impact on the gap of inflation expectations from the inflation target.25Table 8.6 lists the events that classify under this category.

Table 8.6Macroprudential Policy Events
Event NumberEvent WindowEvent DescriptionAuthors’ Interpretation of the Expected Impact on CreditCurrent Nominal Policy Interest Rate CycleTime Span since Last Change in Policy DirectionPrevious Nominal Policy Interest Rate Cycle
119-aug-2014 to 22-aug-2014Reduces RWF of long-term retail credit operations (Circular 3714)IncreaseStability4 monthsIncrease
224-jul-2014 to 28-jul-2014Changes the compliance terms of reserve requirements on time deposits by introducing optional compliance with credit origination. Increases the set of institutions that can partially comply with reserve requirements on demand deposits with credit origination related to a specific development program (Circular 3712)Increase
323-jun-2014 to 25-jun-2014Changes the calculation of risk-weight factors for retail loans (Circular 3711)IncreaseStability3 monthsIncrease
Postpones the implementation of a stricter mandatory allocation of funds to rural credit (Resolução 4336) and gives more flexibility to compliance with mandatory rural credit originations (Resolução 4348)IncreaseStability2 monthsIncrease
430-sep-2013 to 2-oct-2013Increases the maximum value of real estate authorized to be financed with lower interest rates. Sets loan-to-value caps on housing loans (Resolução 4271)AmbiguousIncrease5 monthsStability
59-aug-2013 to 13-aug-2013Sets risk-weight factors on rural credit inversely related to lending rates (Resolução 4259)IncreaseIncrease4 monthsStability
Changes the time window for computing the incidence base of mandatory allocation of demand deposits on microcredit originations (Resolução 4242)Neutral
62-jul-2013 to 4-jul-2013Speeds up the schedule for normalization of the remuneration of reserve requirements on time deposits (the previous regulation reduced the remuneration of required reserves on time deposits if banks did not purchase credit portfolios of small financial institutions) (Circular 3660)ReductionIncrease3 monthsStability
719-jun-2013 to 21-jun-2013Changes several regulatory pieces concerning mandatory rural credit origination (Resoluções 4233, 4234 and 4235)NeutralIncrease2 monthsStability
828-feb-2013 to 4-mar-2013Implements Basle 3NeutralStability4 monthsReduction
927-dec-2012 to 2-jan-2013Changes the compliance terms of reserve requirements on demand deposits by introducing optional compliance with credit origination, with a potential impact of R$15 billion in new credit origination (Circular 3622)IncreaseStability2 monthsReduction
1014-sep-2012 to 17-sep-2012Cancels additional reserve requirements on time deposits. Reduces the reserve requirement ratio on time deposits (Circular 3609)IncreaseReduction12 monthsIncrease
1123-aug-2012 to 27-aug-2012Changes required allocation of funds to rural loans, giving incentives for credit originations at low lending rates (Resolução 4127). Adds flexibility to the requirements for issuing long-term bank instruments (Letra Financeira) (Resolução 4123)IncreaseReduction11 monthsIncrease
1228-jun-2012 to 2-jul-2012Increases the set of institutions allowed to obtain export credit (Circular 3604). Increases mandatory allocation of demand deposits on rural credit (Resolução 4096) and reduces additional resere requirements on demand deposits (Circular 3603)IncreaseReduction9 monthsIncrease
1321-may-2012 to 23-may-2012Increase the set of credit operations allowed to be used as compliance with resere requirements on time deposits (Circular 3594). Requires the registry of collateral on housing and vehicle loans in authorized asset exchange systems.IncreaseReduction8 monthsIncrease
1410-feb-2012 to 14-feb-2012Increases the set of financial institutions allowed to partly comply with traditional and additional reserve requirements on time deposits with purchases of credit portfolios from other institutions and other operations. Increases the set of institutions exempted from these reserve requirements (Circular 3576)IncreaseReduction5 monthsIncrease

To assess the impact of the events on inflation expectations in Brazil, we use a panel of 12-month-ahead private inflation forecasts, surveyed on a daily basis by the Central Bank of Brazil’s Investor Relations Office from 2011 to 2014. To control for other factors influencing inflation expectations, we follow Carvalho and Minella (2012) by estimating an expectations-formation-type rule. The rule is further augmented to account for the events under investigation here and includes the addition of dummy controls for the week of monetary policy meetings (and the preceding week) and for times when the consensus forecast was above the upper bound of the inflation target.26 The estimated equation is:

where t corresponds to each day in the sample, and the variables are described as follows:

  • π^i,te,12m is the gap between the 12-month-ahead inflation forecast for each participant i and the center of the inflation target band;

  • π^median,te,12m is the gap between the median of 12-month-ahead inflation forecasts and the center of the inflation target band;

  • Stdte,12m is the standard deviation of 12-month-ahead inflation forecasts surveyed on a daily basis by the Central Bank of Brazil’s Investor Relations Office;

  • Δ20FXt is the change in the BRL/US$ daily quote over the past 20 days;

  • Δ20Embit is the change in JP Morgan’s EMBI Brazil over the past 20 days;

  • Δ20Selict is the change in the annualized monetary policy (SELIC) rate over the past 20 days;

  • Dcopom,t takes the value 1 on the days included in the following interval: the Friday immediately preceding a monetary policy meeting and the Monday immediately following it. For all other days, it takes the value 0;

  • Dmedian,t takes the value 1 on the days when the median of 12-month-ahead inflation forecasts are above the upper bound of the inflation target. For all other days, it takes the value 0;

  • Devent,i,t takes the value 1 in the event of window days according to Table 8.2. For all other days, it takes the value 0.

We perform a fixed effects panel regression applying a covariance matrix that is robust to heteroscedasticity, autocorrelation with moving average (MA)–type errors, and cross-sectionally dependent errors27. Since forecasts are made for 12 months ahead, the MA structure duly considers this time span. Table 8.7 shows the regression results, where “Forecast Gap” corresponds to the variable πi,te,12m in equation (8.13), “Median Gap” corresponds to the variable π^median,te,12m, and “Panel std” corresponds to the variable Stdte,12m.

Table 8.7Panel Regression Results: Single Events
Forecast GapRobust Coeff.Std. Err.zP > z[95% Conf. Interval]Forecast Gap
Forecast gap (−20)0.6570.01253.900.000***0.6330.681
Median gap (−20)0.1090.0821.320.187−0.0530.270
Panel std (−5)0.4580.2471.860.064*−0.0260.942
Δ FX (−5)0.4960.2142.310.021**0.0760.916
Δ Π (−5)0.8760.1207.290.000***0.6401.112
Δ Embi (−5)0.1090.1290.850.398−0.1440.362
Δ R (−5)0.5770.4531.270.203−0.3111.464
Dummy: COPOM week−0.0040.018−0.200.839−0.0380.031
Dummy: Median above target0.2500.0813.100.002***0.0920.408
Dummy: event 10.2120.0277.840.000***0.1590.265
Dummy: event 2−0.0320.038−0.860.392−0.1060.042
Dummy: event 30.0050.0430.120.908−0.0780.088
Dummy: event 40.1770.0682.600.009***0.0430.311
Dummy: event 50.2030.0484.240.000***0.1090.298
Dummy: event 6−0.0390.046−0.850.395−0.1290.051
Dummy: event 7−0.0400.047−0.840.399−0.1330.053
Dummy: event 8−0.2530.037−6.770.000***−0.326−0.180
Dummy: event 90.2260.0445.100.000***0.1390.313
Dummy: event 10−0.0000.026−0.010.994−0.0510.051
Dummy: event 110.0530.0361.460.144−0.0180.123
Dummy: event 12−0.1500.016−9.140.000***−0.182−0.118
Dummy: event 13−0.0330.033−0.990.324−0.0980.032
Dummy: event 140.0250.0400.610.544−0.0550.104
Total (centered) SS = 18,196.85Total (uncentered) SS = 18,196.85Residual SS = 5,006.06Number of observations = 68,722Number of groups = 138F(23,1020) = 7,659.44Prob > F = 0.0000Centered R2 = 0.725Uncentered R2 = 0.725Root MSE = 0.27Note: Fixed-effects estimation. Statistics robust to heteroskedasticity and time clustering and kernel-robust to common correlated disturbances (Driscoll and Kraay 1998). Kernel = Bartlett; bandwidth = 242 days. SS = steady state.*p < .05; **p < .01; ***p < .001.
Total (centered) SS = 18,196.85Total (uncentered) SS = 18,196.85Residual SS = 5,006.06Number of observations = 68,722Number of groups = 138F(23,1020) = 7,659.44Prob > F = 0.0000Centered R2 = 0.725Uncentered R2 = 0.725Root MSE = 0.27Note: Fixed-effects estimation. Statistics robust to heteroskedasticity and time clustering and kernel-robust to common correlated disturbances (Driscoll and Kraay 1998). Kernel = Bartlett; bandwidth = 242 days. SS = steady state.*p < .05; **p < .01; ***p < .001.

We find that on six different occasions, macroprudential policy announcements had an impact on the gap between inflation expectations and the inflation target. Event numbers 1, 4, 5, and 9 contributed to increase the gap. Event number 1 was not particularly intended to increase credit, but the movement was in the direction of relaxing credit constraints. In event number 4, while the increase in the maximum value of the real estate that could be financed with more favorable rates would contribute to expanding credit, the implementation of a loan-to-value cap could have the opposite effect on credit. In any case, market participants seem to have interpreted it as inflationary. Event number 9 was specifically intended to stimulate credit origination through changes in the way banks could comply with reserve requirements on demand deposits. Event numbers 8 and 12 contributed to reducing the gap between inflation expectations and the inflation target. Event number 8 corresponded to the announcement of the implementation of Basel III. Event number 12 did not have an intention of reducing credit. Hence, the negative sign obtained in the estimation seems at odds with the intention of the event.

An alternative specification was tested, including shorter lags of the controlling variables, and event numbers 4, 5 and 9 remained significant,28 suggesting that these events had an important impact on the anchoring of inflation expectations.

To test whether the cycle of monetary policy matters for the impact of macroprudential announcements on inflation expectations, we perform the same regression, except that, instead of using individual events, we separate those that would likely have an expansionary impact on credit into two groups: Group A is composed of events that happened when the cycle of monetary policy was contractionary, and group B is composed of events that happened in expansionary monetary policy cycles. Our monetary policy cycle classification is described below. If the change in the policy rate that immediately preceded the event was in the direction of increasing it, the monetary policy stance was considered to be contractionary. If the policy rate was stable, but the previous cycle had an increase in interest rates, the monetary policy stance was also considered to be contractionary. If the change was to reduce policy rates or if the current cycle was one of stable rates immediately following a reduction cycle, the monetary policy stance was considered to be expansionary. Hence, group A is composed of event numbers 1, 2, 3, 4, and 5, and group B is composed of event numbers 9, 10, 11, 12, 13, and 14. Event number 7 was not included in either of these groups, since the expected impact would be either neutral or of a reduction in credit. They were treated separately in the estimation.

Table 8.8 shows the estimation results. We find a significantly positive coefficient for the events in group A, but the coefficient for group B is not significant. The strict interpretation of this result is that macroprudential policy announcements that are interpreted to increase credit in moments when monetary policy is contractionary have unwanted effects on the anchoring of inflation expectations. This certainly creates challenges for monetary policy conduct and for central bank communication.

Table 8.8Panel Regression Results: Grouped Events

Event group A: events expected to increase credit when the monetary policy stance was contractionary

Event group B: events expected to increase credit when the monetary policy stance was expansionary

Forecast GapRobust CoefficientStandard ErrorzP > z[95% Confidence Interval]
Forecast gap (−20)0.6570.01253.1400.000***0.6330.681
Median gap (−20)0.1050.0811.300.192−0.0530.263
Panel std (−5)0.4330.2431.780.074*−0.0430.910
ΔFX (−5)0.5030.2112.390.017**0.0900.916
Δ Π (−5)0.8760.1207.300.000***0.6411.112
Δ Embi (−5)0.0970.1290.760.450−0.1550.350
Δ R (−5)0.6560.4371.500.134−0.2011.513
Dummy: COPOM week−0.0030.018−0.150.877−0.0370.032
Dummy: Median above target0.2520.0803.170.002***0.0960.409
Dummy: events group A0.1210.0542.220.026**0.0140.227
Dummy: events group B0.0150.0360.410.683−0.0550.084
Dummy: event 7−0.0410.047−0.870.382−0.1320.051
Dummy: event 8−0.2510.038−6.630.000***−0.326−0.177
Total (centered) SS = 18,196.85Total (uncentered) SS = 18,196.85Residual SS = 5,030.21Number of observations = 68,722Number of groups = 38F(13,1018) = 1,748.76Prob > F = 0.0000Centered R2 = 0.72Uncentered R2 = 0.72Root MSE = 0.27Note: Fixed-effects estimation. Statistics robust to heteroskedasticity and time clustering and kernel-robust to common correlated disturbances (Driscoll and Kraay 1998). Kernel = Bartlett; bandwidth = 242 days. SS: steady state.*p < .05; **p < .01; ***p < .001.
Total (centered) SS = 18,196.85Total (uncentered) SS = 18,196.85Residual SS = 5,030.21Number of observations = 68,722Number of groups = 38F(13,1018) = 1,748.76Prob > F = 0.0000Centered R2 = 0.72Uncentered R2 = 0.72Root MSE = 0.27Note: Fixed-effects estimation. Statistics robust to heteroskedasticity and time clustering and kernel-robust to common correlated disturbances (Driscoll and Kraay 1998). Kernel = Bartlett; bandwidth = 242 days. SS: steady state.*p < .05; **p < .01; ***p < .001.

Conclusions

This chapter has discussed the interaction between monetary and macroprudential policy in Brazil from both normative and positive perspectives. From the normative perspective, we used an estimated DSGE model built to reproduce Brazilian particularities to investigate optimal combinations of simple and implementable macroprudential and monetary policy rules that react to the financial cycle. We find combinations of reserve requirements, risk-weight factors, and monetary policy that can achieve good results in terms of the central bank’s loss function. The results are very close to those of a more comprehensive optimal combination of macroprudential policy, which includes the countercyclical capital buffer together with all other macroprudential policy instruments considered in this study. This chapter has argued that the smaller sets of optimal policy rules are also easier to implement in Brazil. Reserve requirements and risk-weight factors have actually been used on a number of occasions for macroprudential purposes in Brazil, especially after the financial crisis, and some of them countercyclically. Our results corroborate the perception that the direction of these policies can be used to help correct the buildup of risks in the Brazilian financial system with an efficiency similar to the combination with countercyclical capital.

From the positive perspective, we investigate whether recent macroprudential policy announcements that targeted credit variables had important spillover effects on the inflation target pursued by monetary policy in Brazil. To this end, we used a rich survey panel of private inflation forecasts collected by the Central Bank of Brazil’s Investor Relations Office on a daily basis and investigated the impact of announcements of macroprudential policy changes on inflation forecasts. We find that some events increased the gap between inflation forecasts and inflation targets. When we group the events that were expected to increase credit into two different sets, one when monetary policy was contractionary and the other when monetary policy was expansionary, we find that the former had a positive and significant impact on inflation expectations, while the latter did not have a significant effect. This can be interpreted as evidence that when macroprudential policy announcements are desynchronized from monetary policy, the anchoring of inflation expectations can be challenged. This stresses the importance of improving communication of the central bank’s policy intentions.

This chapter has also presented an overview of the macroprudential policy challenges in Brazil since the global financial crisis and offered a glimpse at a few important future challenges. Financial deepening, foreign capital flows, and the impact of fiscal policy on the credit cycle have been particularly relevant challenges that warrant further analysis. Financial deepening has resulted from financial inclusion, following technological improvements in the financial system, income distribution policies, public banks’ credit origination policies, and a long period of macroeconomic stability. Household indebtedness increased substantially and credit accelerated, but several risk-mitigating measures have been put in place to strengthen the resilience of the financial system, in addition to a tight supervisory and regulatory policy stance.

However, since a substantial part of the risk inherent to the financial deepening process has been taken by public banks, and on some occasions transferred to the National Treasury, the impact of the fiscal policy stance on the Brazilian credit cycle should be constantly monitored and anticipated.

References

    AfanasieffT.F.CarvalhoE.CastroR.Coelho and J.Gregório.2015. “Implementing Loan-to-Value Ratios: The Case of Auto Loans in Brazil (2010–11).Working Paper 380Central Bank of BrazilBrasilia.

    AgénorP. and L.Silva.2013. “Inflation Targeting and Financial Stability: A Perspective from the Developing World.Working Paper 324Central Bank of BrazilBrasilia.

    AngeloniI. and E.Faia.2013. “Capital Regulation and Monetary Policy with Fragile Banks.Journal of Monetary Economics60(3): 31124.

    BenignoG.H.ChenC.OtrokA.Rebucci and E.Young.2011. “Monetary and Macroprudential Policies: An Integrated Analysis.Paper prepared for the IMF Twelfth Jacques Polak Annual Research ConferenceWashington, DCNovember 10–11.

    BernankeB. and M.Gertler.2001. “Should Central Banks Respond to Movements in Asset Prices?American Economic Review91(2): 25357.

    Brzoza-BrzezinaM. and M.Kolasa.2013. “Bayesian Evaluation of DSGE Models with Financial Frictions.Journal of Money Credit and Banking45(8): 145176.

    CarvalhoF. and C.Azevedo.2008. “The Incidence of Reserve Requirements in Brazil: Do Bank Stockholders Share the Burden?Journal of Applied Economics11(1): 6190.

    CarvalhoF. and M.Castro.2015. “Foreign Capital Flows, Credit Growth and Macroprudential Policy in a DSGE Model with Traditional and Matter-of-Fact Financial Frictions.Working Paper 387Central Bank of BrazilBrasilia.

    CarvalhoF. and A.Minella.2012. “Survey Forecasts in Brazil: A Prismatic Assessment of Epidemiology, Performance, and Determinants.Journal of International Money and Finance31(6): 137191.

    Cesa-BianchiA. and A.Rebucci.2015. “Does Easing Monetary Policy Increase Financial Instability?IMF Working Paper 15/139International Monetary FundWashington, DC.

    CúrdiaV. and M.Woodford.2010. “Credit Frictions and Optimal Monetary Policy.Journal of Money Credit and Banking42: 335.

    De FioreF. and O.Tristani.2013. “Optimal Monetary Policy in a Model of the Credit Channel.Economic Journal Royal Economic Society123(571): 90631.

    DrehmanM. and K.Tsatsaronis.2014. “The Credit-to-GDP Gap and Countercyclical Capital Buffers: Questions and Answers.BIS Quarterly Review (March).

    DriscollJ. and A. C.Kraay.1998. “Consistent Covariance Matrix Estimation with Spatially Dependent Data.Review of Economics and Statistics80: 54960.

    FaiaE. and T.Monacelli.2007. “Optimal Interest Rate Rules, Asset Prices, and Credit Frictions.Journal of Economic Dynamics and Control31(10): 322854.

    FendoğluS.2014. “Optimal Monetary Policy Rules, Financial Amplification, and Uncertain Business Cycles.Journal of Economic Dynamics & Control46: 271305.

    GertlerM.N.Kiyotaki and A.Queralto.2012. “Financial Crises, Bank Risk Exposure and Government Financial Policy.Journal of Monetary Economics59: S17S34.

    GilchristS. and J.Leahy.2002. “Monetary Policy and Asset Prices.Journal of Monetary Economics49: 7597.

    IacovielloM.2005. “House Prices, Borrowing Constraints and Monetary Policy in the Business Cycle.American Economic Review95(3): 73964.

    IacovielloM. and S.Neri.2010. “Housing Market Spillovers: Evidence from an Estimated DSGE Model.American Economy Journal: Macroeconomics2(2): 12664.

    International Monetary Fund (IMF). 2013. “The Interaction of Monetary and Macroprudential Policies.January29. Policy PaperWashington, DC. http://www.imf.org/external/np/pp/eng/2013/012913.pdf.

    JácomeL. I.E.Nier and P.Imam.2012. “Building Blocks for Effective Macroprudential Policies in Latin America: Institutional Considerations.IMF Working Paper 12/183International Monetary FundWashington, DC.

    KannanP.P.Rabanal and A.Scott.2012. “Monetary and Macroprudential Policy Rules in a Model with House Price Booms.B.E. Journal of Macroeconomics12(1) Article 16.

    LambertiniL.C.Mendicino and M.Punzi.2013. “Leaning against Boom-Bust Cycles in Credit and Housing Prices.Journal of Economic Dynamics and Control37: 150022.

    MartinsB. and R.Schechtman.2013. “Loan Pricing Following a Macro Prudential Within-Sector Capital Measure.Working Paper 323Central Bank of BrazilBrasilia.

    MontesG. and G.Peixoto.2012. “Risk-Taking Channel, Bank Lending Channel and the ‘Paradox of Credibility’: Evidence for Brazil.Encontro Nacional de Economia–ANPEC.

    PonticelliJ. and L.Alencar.2013. “Celeridade do Sistema Judiciário e Créditos Bancários para as Indústrias de Transformação.Working Paper 327Central Bank of BrazilBrasilia.

    ReinhartC. and K.Rogoff.2011. “From Financial Crash to Debt Crisis.American Economic Review101: 1676706.

    SahayR.M.ČihákP.N’DiayeA.BarajasR.BiD.AyalaY.GaoA.KyobeL.NguyenC.SaborowskiK. Svirydzenka and S.Yousefi.2015. “Rethinking Financial Deepening: Stability and Growth in Emerging Markets.IMF Staff Discussion Note 15/08International Monetary FundWashington, DC.

    Schmitt-GrohéS. and M.Uribe.2007. “Optimal Simple and Implementable Monetary and Fiscal Rules.Journal of Monetary Economics54: 170225.

    SilvaL. and R.Harris.2012. “Sailing through the Global Financial Storm: Brazil’s Recent Experience with Monetary and Macroprudential Policies to Lean against the Financial Cycle and Deal with Systemic Risks.Working Paper 290Central Bank of BrazilBrasilia.

    SilvaL.A.Sales and W.Gaglianone.2012. “Financial Stability in Brazil.Working Paper 289Central Bank of BrazilBrasilia.

    TavaresD.G.Montes and O.Guillén.2013. “Transmissão da Política Monetária pelos Canais de Tomada de Risco e de Crédito: uma análise considerando os seguros contratados pelos bancos e o spread de crédito no Brasil.Working Paper 308Central Bank of BrazilBrasilia.

    TaylorA.2015. “Credit, Financial Stability, and the Macroeconomy.NBER Working Paper 21039National Bureau of Economic ResearchCambridge, Massachusetts.

    VéghC. and G.Vuletin.2013. “The Road to Redemption: Policy Response to Crises in Latin America.Paper prepared for the IMF 14th Jacques Polak Annual Research ConferenceWashington, DCNovember 7–8.

The authors are grateful to Eduardo Lima, Laura Kodres, Solange Gouvea, Hamid Faruqee, Alfredo Cuevas, Jorge Roldós, Marcello Estevão, Alberto Torres, Troy Matheson, David Einhorn and the participants at the IMF’s midpoint seminar on “The Future of Central Banking in Latin America” for helpful comments and discussions.

Jácome, Nier, and Imam (2012) thoroughly discuss the measures implemented in the region.

Our approach of using a realistic model of the Brazilian economy, estimated with actual data, adds robustness to the results. De Fiore and Tristani (2013), for instance, recognize that their numerical findings of optimal rules are illustrative and the quantitative features derived from them should be validated through more complex models.

Ramsey-type optimal policy analysis requires an arbitrary weight of each class of agents in the model. Lambertini, Mendicino, and Punzi (2013) find an important role for heterogeneity with respect to classes of agents in determining welfare implications. They cannot find a uniform ranking of policy frameworks for both classes of agents in their model. In addition, rules that deviate from the optimal in individual terms have important welfare effects for only one class of agent, the borrower, which is more directly affected by the financial constraint.

In Faia and Monacelli (2007) monetary policy faces a trade-off between stabilizing consumer inflation or asset prices. In Angeloni and Faia (2013), financial targets are asset prices or bank leverage. In Fendoğlu (2014), financial targets are asset prices or credit spreads. They study optimal policy with costly-state verification–type financial frictions, but focus on monetary policy rules. In Kannan, Rabanal, and Scott (2012), the financial friction occurs in housing loans, but the external finance premium is assumed, rather than obtained from first-order conditions. Monetary policy is allowed to react to credit growth.

Brzoza-Brzezina and Kolasa (2013) provide an extensive analysis of model-implied differences in responses of the main economic variables by examining credit constraint and external finance premium financial accelerators vis-à-vis a standard New Keynesian model. For a detailed description of the impact of the set of disturbances allowed in a particular model on optimal policy rules, see Lambertini, Mendicino, and Punzi (2013).

Some exceptions that introduce a second policy instrument are Benigno and others (2011) and Cesa-Bianchi and Rebucci (2015), who study the interaction of monetary policy with macroprudential policy when borrowing constraints bind; Angeloni and Faia (2013), who introduce a countercyclical capital rule that interacts with monetary policy; and Lambertini, Mendicino, and Punzi (2013), who study the optimality of countercyclical loan-to-value ratio caps in a model based on Iacoviello and Neri (2010), focused on the mortgage market.

An exception is Benigno and others (2011), but the financial frictions they incorporate are significantly different from ours. They assume that there are eventually binding collateral constraints with a reduced set of nominal rigidities and that borrowing occurs in foreign currency, while our financial frictions in the borrowing side of the model come from costly-state verification, with bank borrowing carried out in domestic currency. Notice that our model has other important frictions that constrain banks’ balance sheet locations and have real effects.

Agénor and Silva (2013) have qualified Brazil’s bank supervisory environment as “strong, sophisticated and intrusive” with a “robust regulatory environment,” which differentiates the country from other middle-income countries.

Data from the IMF’s Financial Soundness Indicator database (http://fsi.imf.org/Default.aspx).

Silva and Harris (2012) provide an extensive report on the measures adopted in Brazil to fight the global financial crisis.

IMF, Financial Soundness Indicators database.

Ponticelli and Alencar (2013) show that the Credit Default Law allowed for a significant increase in the probability of collateral recovery in the case of liquidation of a borrowing firm. It also had significantly positive effects on loan originations to companies in the transformation industry (which was the only industry examined in the study). The law generated an overall impact in the form of lower lending rates, longer maturities, and lower collateral requirements. The effects were more noticeable in regions where judges are quicker to analyze these cases.

This could increase the procyclicality of housing loans because the fiscal stance of the economy could play an important role in the capacity to originate new loans.

Central Bank of Brazil Circular #2698 of June 20, 1996 created both the COPOM and its monetary policy instrument, the rediscount (TBC) rate, which was the official monetary policy instrument until 1999, when it was informally replaced by the base (SELIC) rate. Circular #2966 of February 8, 2000 formalized the SELIC rate as the central bank’s monetary policy instrument.

The National Monetary Council is composed of the Minister of Finance, Minister of Planning and Budget, and the Governor of the Central Bank of Brazil.

Central Bank of Brazil Portaria #65180 of May 18, 2011 created COMEF. The central bank’s portarias are legal instruments issued by the central bank governor. Circulares must be approved by the central bank’s board of directors.

For a more detailed overview of reserve requirements in the pre-global-crisis period, see Carvalho and Azevedo (2008).

Credit in the model is composed of consumer, housing, and commercial credit. For all of these credit segments, the model allows for endogenous default due to imperfect monitoring. Consumer credit is extended based on borrowers’ future labor income, net of payments related to housing loans. Housing loans are subject to loan-to-value (LTV) constraints that constrain borrowers’ available income. Commercial loans are taken by the entrepreneurs and are subject to LTV constraints.

We show below some sensitivity analysis on the impact of different weights on the credit gap in the loss function.

The volume of housing loans is not very sensitive to its corresponding risk-weight factor. The reason for this is that this market is heavily regulated with respect to both interest rates and funding sources.

We use the Optimal Simple Rule routine in Dynare, which is based on Sims’ minimization algorithm. The results that we report here are obtained after testing different initial points and comparing the value of the objective function obtained in each of these trials.

We find the optimal rules given all sources of disturbance estimated in the model of Carvalho and Castro (2015). The rule obtained from the setup that we adopt can be more easily compared with actual rules estimated with all shocks activated.

Schmitt-Grohé and Uribe (2007) is such an example. They constrain the optimal parameter search to a particular set of so called “realistic” values.

Faia and Monacelli (2007) find, in a model with agency costs and nominal frictions, that monetary policy should react to increases in the asset price by lowering the nominal interest rate.

The export sector is modeled in accordance with a commodity-based economy.

We use the upper bound of the inflation target interval and not the midpoint target.

The dummy for times when inflation expectations exceeded the upper inflation target controls for possible regime changes in the dynamics of inflation expectations.

The routine is implemented in Stata through the “xtivreg2” command, which applies the covariance matrix estimator of Driscoll and Kraay (1998).

Since the panel used in the estimation is highly unbalanced, changing the lag structure of the regressors can have important implications for the number of observations actually used in the estimation.

    Other Resources Citing This Publication