How can macroprudential policies transmit within a banking group?

The unexpected shock represented by the COVID-19 pandemic illustrates the importance of building robust macroprudential frameworks to increase countries’ resilience against sudden disruptions in financial markets. By now, a widespread opinion among commentators and policy makers is that the macroprudential frameworks that were implemented over the past decades were effective in moderating market stress, a view supported by ample evidence on the effectiveness of macroprudential policies.

In a new research project, we put the functioning of macroprudential policies under the magnifying glass, focusing on how intra-group links between banks’ headquarters and their regional branches affect the pass-through of reserve requirements to credit supply in Brazil. Shifting the focus to banks’ group structures when evaluating the effectiveness of reserve requirements is important as adjustments within a group can create relevant blind spots for policy makers. On the one side, intra-group liquidity dynamics can intervene with the desired effect of steering banks’ credit by managing banks’ available funding. On the other side, banks may have incentives to reallocate funds heterogeneously across regions and industries as a response to the policy, making use of their group structure. The latter may open the scope for unintended distributional effects.

To explore these questions, we construct a research design based on matched bank-branch data for Brazil covering 2008 to 2014. The sample period includes several episodes in which reserve requirements on demand deposits — a macroprudential tool frequently used by the Brazilian Central Bank — were adjusted to steer banks’ credit supply. Given that policy decisions can be a function of credit market dynamics, it is not straightforward to find a setting that allows for a clean identification of supply-side effects. Our setting addresses identification concerns as follows:

• First, we reduce reverse causality concerns by focusing on credit by individual municipal branches, separating the corporate level at which reserve requirements are imposed (that is, headquarters) from the level at which loans are granted (that is, branches).

• Second, we identify the effect of reserve requirements by exploiting the fact that bank headquarters vary in their reliance on targeted demand deposits.

• Third, to isolate credit supply, we exploit that several branches of different headquarters are active per municipality, allowing us to control for local credit demand.

This empirical design allows us to draw conclusions from a rich sample of 6,081 individual bank branches traced on a quarterly frequency and belonging to 56 banks operating in 1,678 Brazilian municipalities.

Figure 1 shows the evolution of reserve requirements from 2006 to 2014, illustrating the multiple changes in reserve requirements during this period. The figure contrasts this evolution with trends in cross-border banking flows from and to Brazil. This comparison highlights a relevant dimension of our identification: historically, reserve requirements have responded to shifts in capital flows, tending to ease banks’ funding when global financial conditions tighten. This dynamic, in which reserve requirements react to global factors, further mitigates reverse causality concerns.

Figure 1: Quarterly evolution of reserve requirements and cross-border banking claims in Brazil

Note: This graph describes the pattern of the reserve requirements for demand deposits (in %, solid line - left axis) as provided by the Central Bank of Brazil. The dashed (dotted) line describes the evolution of quarterly cross-border liabilities (assets) of the Brazilian banking system (in billions of USD), as obtained from the Locational Banking Statistics of the Bank for International Settlements.

Our empirical approach delivers robust results, providing new evidence on the functioning of macroprudential policies such as reserve requirements within a banking group. Our main results suggest that…

• … changes in reserve requirements targeting banks’ headquarters significantly affect bank branches’ credit supply.  The effect is larger for branches linked to headquarters relying more on targeted demand deposits.

• However, this result depends crucially on the stage of the economic cycle and on banks’ ownership structure. In particular, the effect shows up in periods when reserve requirements are reduced (for example, the global financial crisis), while branches of state-owned banks react the most to the policy tool.

How do group structures and characteristics of individual units drive these results? To shed light on this question, we use detailed data on branches’ balance sheets including their reliance on internal funding. Focusing on the subsample of state-owned banks for the period when reserve requirements were most effective (that is, during the 2008–09 global financial crisis), we find that branches with a low ex-ante profitability and a high dependence on internal funds were most responsive to reserve requirements and responded more to a loosening of the policy by expanding loans.  Two important conclusions are drawn from this finding:

• First, branches’ own liquidity constraints seem to be one factor in explaining the heterogeneous response to reserve requirements.

• Second, our results suggest the presence of a within-bank reallocation of funds toward weaker branches, highlighting the importance of corporate political considerations for the transmission of macroprudential policies. 

Our findings have the following implications for policy makers concerned with the functioning of reserve requirements in emerging market economies. On the one hand, intragroup structures lead to a heterogeneous effect of macroprudential policies on banks’ credit supply, opening the scope for possible distributional effects at the macro level, for instance across regions and industries. On the other hand, the importance of corporate political considerations for the transmission of macroprudential policies raises questions regarding the emergence of market distortions. These two conclusions offer an interesting venue for future research.