Macroprudential policy and intra-group dynamics: The effects of reserve requirements in Brazil

Highlights

• This paper studies how reserve requirements transmit within a banking group.

• The sample covers the period from 2008 to 2014 and the Brazilian banking system.

• The exposure of bank headquarters to the policy affects branches’ credit supply.

• Results are driven by the global financial crisis period and state-owned banks.

• Freed-up liquidity after a loosening benefits weaker and liquidity scarce branches.

Abstract

We examine whether liquidity dynamics within banking groups matter for the transmission of macroprudential policy. Using matched bank headquarters-branch data for identification, we find a lending channel of reserve requirements for municipal branches whose headquarters are more exposed to the policy tool. The result is driven by the 2008–2009 crisis and is stronger for state-owned branches, especially when being less profitable and liquidity constrained. These findings suggest the presence of cross-regional distributional effects of macroprudential policies operating via internal capital markets.

Introduction

The global financial crisis showed that disruptions in the financial sector can have large negative effects on the real economy. To reduce systemic risk in financial markets, several changes in the regulatory framework of the banking system have been made. This policy consensus about the reform of banking regulation was characterized by the introduction of macroprudential policies, a combination of policy tools aimed at reducing the risk of systemic imbalances by steering the cycle of banks’ credit supply (Duffie, 2018). One necessary condition for macroprudential policies to be effective is in such a context their capacity to tighten or loosen banks’ funding constraints (Aiyar et al., 2014).

Although the literature suggests that bank funding structures and internal capital markets are important for the transmission of monetary policy to credit supply (e.g., Campello, 2002, Kishan and Opiela, 2000, Holod and Peek, 2010), there is limited evidence of whether a similar rationale applies to macroprudential policies. We draw on a rich data set on Brazilian banking groups to explore the effect of reserve requirements targeting headquarters’ deposit ratio on credit supply by their municipal bank branches. From a theoretical perspective, an increase in reserve requirements should constrain branches’ credit supply because liquidity has to be placed as reserves at the central bank. The requirement is proportional to the headquarters’ deposit ratio such that a banking group's exposure to the policy depends on its funding structure.¹

The matched bank headquarters-branch data set we use comes with two key advantages. First, we can look at a novel mechanism, namely the transmission of reserve requirements to credit supply within a banking group. Second, the data structure allows accounting for reverse causality and credit demand side effects as explained in the next paragraph. Our results show evidence for a lending channel of reserve requirements within a banking group with potential cross-regional distributional effects. Branches whose headquarters are more exposed to targeted deposits are more likely to increase credit supply when reserve requirements are loosened. Importantly, it is not only banks’ funding structure that matters for transmitting the policy but also intra-group liquidity allocation: the lending channel is stronger for weaker branches in terms of having, ex-ante, a low profitability and a large internal liquidity reliance, particularly when they are state-owned. This result is in line with evidence suggesting a cross-subsidization role of internal capital markets (see, e.g., Cremers et al., 2011). While the result bears important implications in itself, it would be more difficult to capture it based on an alternative data structure.²

We exploit variation across headquarters, branches and municipalities to follow an identification strategy based on three main building blocks. First, to reduce concerns about reverse causality, we conduct our analysis at the level of individual branches belonging to a banking group. We can thus separate the corporate level at which reserve requirements are imposed from the level at which loans are granted. This point is strengthened considering that reserve requirements are actively used by the Brazilian Central Bank to steer the local credit cycle when foreign capital shocks hit. Hence, changes in reserve requirements are likely to be exogenous from the perspective of regional bank branches.

Second, we identify the effect of reserve requirements on branches’ credit supply by exploiting the fact that bank headquarters vary in their reliance on targeted deposits. The heterogeneous effect of reserve requirements along the distribution of headquarters’ deposit ratios provides a proper identification of changes in credit supply triggered by reserve requirements. This argument follows the established approach of analyzing the transmission of policy shocks conditional on banks’ balance sheet exposures. While this approach was originally used to analyze monetary policy shocks (see, e.g., Kashyap and Stein, 1995, Kashyap and Stein, 2000), more recent contributions have extended it to explore the transmission of macroprudential policies to bank lending (see, e.g., Gambacorta and Mistrulli, 2004, Danisewicz et al., 2017).

Third, to isolate credit supply from demand and to account for different investment opportunities across regions, we exploit that several branches of different headquarters are active in one municipality. We follow the literature (Carlson et al., 2013, Dursun-de Neef, 2019) by including quarter-municipality fixed effects in a panel model that absorb time-varying and municipality-specific changes in credit demand to which branches in a given region are commonly exposed.

We implement this research design on granular data for the Brazilian banking system covering balance-sheet information for every active bank in the country between 2008 and 2014. These data allow us to link individual headquarters with their regional branches aggregated at the level of Brazilian municipalities. The high reporting frequency of the data (compared to alternative data sources such as Orbis Bank Focus) allows us to properly track changes in banks’ credit induced by adjustments in reserve requirements. Finally, we exploit the large presence of foreign and state-owned banks in Brazil to explore whether results differ depending on banks’ ownership structure, similar to Aiyar et al. (2014) and Coleman and Feler (2015).

Our results are threefold and can be summarized as follows. First, we find robust evidence that reserve requirements targeting bank headquarters’ funding side are transmitted into their affiliated branches’ credit supply, whereas the effect of reserve requirements becomes stronger if headquarters have a larger exposure to demand deposits. This baseline result remains robust when controlling for demand side effects, monetary policy and a large range of other confounding factors. Branches’ lending sensitivity to reserve requirements pertains at the aggregate level and is not netted out by borrowers’ substituting credit between banks. Second, the result depends crucially on the stage of the economic cycle and bank ownership. It is driven by periods of economic downturns when reserve requirements are loosened and by branches belonging to state-owned banks. Third, for the sample of state-owned banking groups and the crisis period, we find that the loosening of the policy contributed to the maintenance of credit supply by branches with low ex-ante profitability. Our analysis relates this finding to branch-level liquidity constraints, as the effect is stronger for state-owned branches with low liquid assets and low net intra-group assets. This finding suggests a within-bank reallocation towards weaker branches, highlighting that the transmission of a macroprudential tool is not only driven by different investment opportunities across regions, but also by corporate politics operating via internal capital markets.

As the study focuses on a single macroprudential instrument in Brazil, external validity might be limited to similar countries (e.g., as concerns the share of state-owned or foreign banks) and comparable tools. It is to note, however, that the Brazilian setting provides a case study that resembles the characteristics of other emerging countries using reserve requirements, especially in Latin America. For example, Cordella et al. (2014) highlight that most Latin American countries actively use reserve requirements (65% or 11 out of 17 countries in their sample), motivated by banks’ deposit-intensive funding structures and credit cycles fueled by external capital flows.

This paper contributes to three main strands of literature. First, there is an evolving literature on the effectiveness of macroprudential policies (see e.g., Claessens et al., 2013, Haldane et al., 2014), studying the heterogeneous effects of macroprudential policy by relying on bank-level data (Acharya et al., 2020, Barbone Gonzalez et al., 2018, Buch and Goldberg, 2017, Epure et al., 2017). For instance, Aiyar et al. (2014) use a sample of domestically-owned banks and foreign-owned branches and subsidiaries in the UK from 1998 to 2007 and find that stricter bank-specific capital regulation of domestic banks and foreign subsidiaries leaks to unregulated foreign branches, which increase their lending. The differential responses to home regulation of foreign branches versus subsidiaries located in the UK are found by Danisewicz et al. (2017). Three main contributions differentiate our paper from these studies. First, we look at a different instrument of macroprudential policy – reserve requirements for demand deposits – in the context of an emerging country that uses this tool to steer the transmission of cycles of capital flows from abroad. Second, we analyze how the characteristics of banks’ funding structures drive the effectiveness of reserve requirements. Third, we show that macroprudential instruments may have cross-regional distributional effects driven by liquidity reallocation within a banking group.

Second, our focus on banks’ intra-group dynamics adds to the literature on the transmission of liquidity shocks via internal capital markets. Early literature on internal capital markets discussed the role of banking groups’ strength for affiliates’ lending and the internal transmission of monetary policy (see, e.g., Ashcraft, 2008, Campello, 2002, Dahl et al., 2002, Houston et al., 1997, Houston and James, 1998). More recent studies have analyzed the cross-border transmission of liquidity or regulatory shocks within international bank holding companies (e.g., Aiyar et al., 2014, Buch and Goldberg, 2015, Cetorelli and Goldberg, 2012a, Cetorelli and Goldberg, 2012b, Danisewicz et al., 2017, De Haas and van Lelyveld, 2010, Frey and Kerl, 2015). Using also Brazilian branch-level data, Coleman et al. (2017) show that banks make use of internal liquidity management after liquidity shocks to support their branches’ lending. Our study is closer to Cremers et al. (2011) who investigate how internal capital markets are used by banks to provide intertemporal insurance against deposit shortfalls.

Finally, other studies have relied on country-level data or descriptive analysis to evaluate the functioning of reserve requirements in Latin America (Montoro and Moreno (2011), Da Silva and Harris (2012)). Tovar Mora et al. (2012) and Glocker and Towbin (2015) estimate structural VAR models to analyze the effect of monetary policy and reserve requirement on aggregate credit growth in Latin America and Brazil, respectively. Dassatti Camors et al. (2019) study one increase in reserve requirements in Uruguay using credit register data and find evidence for a contraction of the loan supply. In contrast, our paper analyzes the effect of reserve requirements on intra-group dynamics for several changes in reserve requirements and using bank-level data. We contribute to this literature by showing how internal capital markets affect the transmission of a macroprudential policy.

The paper is structured as follows. Section 2 discusses the use of reserve requirements in Brazil as a macroprudential tool. Section 3 describes the data and shows descriptive statistics. Section 4 explains the empirical estimation approach, discusses our identification scheme, and presents regression results. Section 5 concludes.