Does the transmission mechanism of conventional monetary policy differ across central banks’ operational frameworks? The motivation of this question originates in the operational framework change in the U.S. in 2008 as a consequence of the unconventional policies implemented to mitigate the Great Financial Crisis (henceforth GFC). This change can be described as a switch from a “corridor” to a “floor” system. The main difference across the two frameworks is that under a corridor system the Federal Open Market Committee’s (FOMC’s) main monetary policy instrument is a market-determined rate, since the non-existence of excess reserves fosters overnight interbank trading, and requires an active management of reserve-supply by the monetary authority.
On the other hand, under a floor system the main instrument of monetary policy becomes the administrated rate at which central bank reserves are remunerated, because the presence of excess reserves lowers their market price (i.e., the interbank rate) until it is equal to the interest on reserves (henceforth IOR). This implies that as long as the aggregate banking-system’s demand for reserves is satiated, the central bank generates a “divorce” between monetary policy’s main instrument and the amount of reserves available in the system (see, e.g., Curdia and Woodford (2011); Keister et al. (2015) and Beckworth (2018)). New-style central banking refers to the floor system through which central banks in many advanced economies implement monetary policy ever since the Great Financial Crisis. Hall and Reis (2015) describe this new normal as central banks “with large balance sheets and payment of interest on reserves.”
Is there any reason to believe that these institutional changes could have affected the monetary transmission mechanism? The answer is yes. In fact, I show that the transmission mechanism of conventional monetary policy differs across operational frameworks. I do so by estimating a Hybrid-VAR model in the spirit of Bernanke and Blinder (1992) for both periods, i.e., before and after the operational framework switch. Under the old-style corridor system, I find that both loans and deposits fall after a contractionary monetary policy shock. These results provide evidence in favor of an active bank-lending channel before the operational system change. On the other hand, in the post-switch sample, i.e., under the new-style floor system, I find that the bank-lending channel breaks down since bank deposits and loans actually increase after a conventional tightening shock.
In order to understand the economic mechanisms at work, I develop a regime-switching two-agent New Keynesian (TANK) model. Taking Gerali et al. (2010) as a starting point the model features credit-supply frictions, as well as monetary policy that is implemented via open market operations (henceforth OMOs) and, thus, in this framework the target (or interbank) rate is deter- mined endogenously. These model features are generally outside the focus of NK models (see, e.g., Galí (2018)). They are, however, necessary to capture the bank-lending channel as first described in the seminal papers by Bernanke and Blinder (1988, 1992).
In line with the empirical estimates, my theoretical results show that in a dynamic TANK framework, as long as the reserve requirement constraint binds, the bank-lending channel works precisely as described in the simple static framework developed in Bernanke and Blinder (1988). However, in the presence of excess reserves, i.e., under a new-style floor system, the bank-lending chan- nel breaks down. In other words, when the reserve requirement constraint is slack an unexpected reserve reduction, conducted via OMOs, stimulates credit supply to the real economy.
Liquidity management costs give rise to this counterintuitive result. Concretely, this friction affects banks’ optimality condition by setting a wedge between the marginal costs of acquiring reserves (the interbank rate) and the marginal funding costs (the interest rate on deposits). Since the liquidity management costs are assumed to be increasing in the aggregate amount of central bank’s total reserves, a reserve contraction alleviates this friction by closing the wedge between the interbank and deposits rates, allowing banks to issue more deposits and extend more loans. This assumption is justified by several liquidity-related regulation changes linked to the operational framework switch, such as the introduction of payments of interest on reserves in 2008.
While this friction is present across both operational systems in the theoretical model, under the assumption of a permanently binding reserve requirement constraint, i.e., under the old-style corridor system, the original mechanism of the lending channel dominates the effects of a reduction in the wedge between the interbank and deposit rates. On the other hand, when the constraint becomes slack the effect of monetary policy via OMOs on banks’ credit supply is entirely determined by the liquidity management friction.
Furthermore, I also assess whether it is possible to have a similar experience as in the Great Financial Crisis under the new floor framework. More specifically, I investigate to what extent a run on non-banks, which are active in money markets, can propagate to the real economy under the new-style floor system. My results support the 2019 FOMC’s decision to maintain a floor framework, since a run on the non-depository financial institutions that are active in the Federal Funds market, similar to the one experienced during the GFC, has no consequences for aggregate demand. The effects of such a run are contained in the financial system since banks can accommodate the sudden rise in deposit holdings by transforming excess reserves into required reserves, leaving total reserves unchanged. In contrast, this purely transitory shock has highly persistent contractionary effects under the old-style corridor framework, because banks are not able to increase their deposits as quickly due to the slow reaction of the central bank in supplying higher reserve levels.
Let me relate these results to the literature. First, this paper is related to the empirical literature on the bank-lending channel. This strand of the literature was pioneered by the seminal contribution by Bernanke and Blinder (1992), where they estimate a simple structural VAR model and find that bank lending contracts after a tightening shock. To the best of my knowledge this is the first paper that estimates the effects of conventional monetary policy on credit supply using externally-identified shocks for both periods, i.e., before and after the operational framework switch.
Additionally, there are empirical studies on the lending channel that employ bank-level data, and that also find supporting evidence for an active bank-lending channel in the U.S. pre-GFC period.5 On the other hand, during the new-style system period, recent contributions find that contractionary monetary policy increases banks’ credit lines (see Greenwald et al. (2020)) and that banks’ liquidity constraints actually become looser after monetary tightening (Salgado-Moreno (2021)). Similarly, and in line with my results, Diamond et al. (2021) estimate that each dollar of reserves created via large scale asset purchases crowds out 13 cents of bank lending. Moreover, recent studies analyzing data from Sweden and Denmark, where there is also abundant liquidity, find that the traditional bank-lending channel breaks down (see Eggertsson et al. (2019) and Cao and Dinger (2018)).
Second, even though the theoretical DSGE literature on financial frictions soared after the financial crisis, there is still only a small branch of this literature that either focuses on the supply side of credit, or analyzes alternative operational systems. To the best of my knowledge, there is only one other contribution, by Arce et al. (2020), that analyzes the interaction between monetary policy and the operational framework in the context of a NK model. However, while the focus of the current paper is the regime switch experienced in the U.S. in 2008, Arce et al. (2020) concentrate on the Euro-system. Thus, their interbank market is characterized by search and matching frictions. In the U.S. however, as will be shown below, unsecured overnight lending has been dominated by a few non-depository institutions, highlighting the importance of incorporating these into my analysis.
Further closely-related theoretical contributions that do not rely on NK models are, e.g., Andolfatto (2015) and Martin et al. (2016). The former develops a tractable OLG model to analyze the monetary transmission mechanism before and after the great recession. He finds that under the new system OMOs “have no real or nominal effects, apart from expanding the supply of excess reserves in the banking sector.” However, the simple structure in the model presented in Andolfatto (2015) does not include liquidity management costs that, as mentioned above, are behind the key results in this paper. This difference highlights the importance of incorporating explicitly an interbank market and the liquidity management expenses incurred by banks when analyzing the transmission mechanism of monetary policy.
The mechanism behind the counterintuitive relationship between monetary policy and banks’ credit supply under the new-style floor system is similar to the one in Martin et al. (2016). They develop a simple static model of the U.S. banking system under the new operational framework and find that introducing frictions in the form of increasing costs in the size of banks’ balance sheets implies a negative relation between the amount of reserves and banks’ credit supply. However, in contrast to this paper, Martin et al. (2016) do not explicitly model an interbank market and assume perfect competition and flexible prices in all markets. The stylized interbank setting developed in the present paper captures the key characteristics of the Federal Funds market (as will be discussed below), and it generates an endogenous target rate for the central bank, which implements monetary policy via OMOs. Thus, a further contribution of this paper is that in this richer model, it is possible to analyze the counteracting effects (arising from simultaneous changes in target rates and in the central bank’s liabilities) of conventional monetary policy.
Third, in recent years there has been an increasing interest in the relative merits of different monetary operational frameworks. For example, Keister (2016), among others, has argued that the new-style system yields a more effective implementation of monetary policy and concludes that the floor system “promote[s] a safer and more resilient payments system.” On the other hand, Selgin (2018) among others, argue in favor of a return to a corridor-style framework, since the new system “dramatically reduced the effectiveness of open market operations.” Based on the model developed in this paper, I find that both arguments are not mutually exclusive. Even though the monetary transmission mechanism differs across frameworks, the presence of excess reserves implies that a run on non-depository financial institutions, and a subsequent collapse in interbank funding, has no real consequences, supporting the FOMC’s decision to keep a floor system. Moreover, the main result of this paper, i.e., a change in the transmission mechanism of conventional monetary policy, is in line with the literature documenting changes in bank behavior (see, e.g., Dutkowsky and VanHoose (2017, 2018)), portfolio allocation (see, e.g., Beckworth (2018)) and credit supply to the real economy (see, e.g., Diamond et al. (2021)).
Finally, the presence of liquidity management costs in this paper is related to a growing branch of the literature focusing on liquidity management of financial institutions using different types of models (see, e.g., Schulhofer-Wohl and Clouse (2018) who use a sequential bargaining framework to model the Federal Funds market with excess reserves, or Bianchi and Bigio (Forthcoming) who introduce an over-the-counter interbank market into a general equilibrium model). Moreover, in this nascent strand of the literature there are recent studies that estimate liquidity costs or the effects of liquidity management on other balance sheet items. For example, Diamond et al. (2021) estimate that an increase in reserves leads to higher marginal cost of lending and to lower marginal cost of deposits. Additionally, Roberts et al. (2018) find that banks subject to a specific liquidity-related regulation introduced in Basel III lend less to the real economy compared to banks that remained unaffected by this new regulation. Similarly, using pre-GFC data Curfman and Kandrac (2021) find that “liquidity regulations impair credit supply as banks shift their balance sheets to favor more liquid assets.”
The remainder of the paper is organized as follows. First, section 2.1 provides several key stylized facts from the regime switch and its consequences for the interbank market and banks’ liquidity management. Second, the empirical evidence from the Hybrid-VAR model is presented in section 2.2. Afterwards, in section 3, I develop a model that is rich enough to capture the credit channel both under a corridor and under a floor system. Section 4 describes the model’s parametrization and the main results, followed by some applications and a robustness analysis. Finally, section 5 concludes.