Working Paper Series
Stefan Gebauer, Falk Mazelis Macroprudential regulation and leakage to the shadow banking sector
No 2406 / May 2020
Disclaimer: This paper should not be reported as representing the views of the European Central Bank (ECB). The views expressed are those of the authors and do not necessarily reflect those of the ECB.
Abstract
Macroprudential policies are often aimed at the commercial banking sector, while a host of other non-bank nancial institutions, or shadow banks, may not fall under their jurisdiction. We study the eects of tightening commercial bank regulation on the shadow banking sector. We develop a DSGE model that dierentiates between regulated, monopolistic competitive commercial banks and a shadow banking system that relies on funding in a perfectly competitive market for investments. After estimating the model using euro area data from 1999 { 2014 including information on shadow banks, we nd that tighter capital requirements on commercial banks increase shadow bank lending, which may have adverse nancial stability eects. Coordinating macroprudential tightening with monetary easing can limit this leakage mechanism, while still bringing about the desired reduction in aggregate lending. In a counterfac- tual analysis, we compare how macroprudential policy implemented before the crisis would have dampened the business and lending cycles.
JEL: E32, E58, G23
Keywords: Macroprudential Policy, Monetary Policy, Policy Coordination, Non-Bank
Financial Institutions, Financial Frictions
ECB Working Paper Series No 2406 / May 2020 | 1 |
Non-technical summary
In the aftermath of the global nancial crisis of 2007/2008, a broad consensus has been reached among scholars and policy makers that a macroprudential approach towards - nancial regulation should focus on systemic developments in nancial markets { such as swings in aggregate credit or nancial market volatility { and on the role of nan- cial cycles for business cycle movements. Contemporaneously, the development of nan- cial intermediaries in canonical pre-crisis dynamic stochastic general equilibrium (DSGE) models has been central to the research program, with particular attention placed on banking-augmented macro models that allow the assessment of the e ectiveness of di er- ent macroprudential tools in the presence of nancial frictions. However, many models largely abstract from heterogeneity in nancial markets, e.g. assuming that regulators reach the entire nancial sector.
In this study we develop and estimate a quantitative DSGE model for the euro area that features credit intermediation by di erent nancial institutions: regulated commercial banks and unregulated shadow banks. On the aggregate level, the two intermediaries engage in similar activities, but di er along several dimensions on the micro-level. First, while commercial banks can directly be reached with macroprudential tools, shadow banks are unregulated in the model. In reality, the shadow banking sector comprises a multitude of diverse and highly specialized institutions, such that the implementation of universal macroprudential regulation towards the sector as a whole may prove dicult. Furthermore, while { in line with empirical evidence { commercial banks exert market power, shadow banks are assumed to be subject to market forces and modeled as price takers. Due to lack of regulation, intermediating funds via these institutions is risky, and investors will limit their exposure to shadow banks endogenously. The presence of shadow banking implies a trade-o for regulators, as tighter regulation on commercial banks induces credit leakage towards the unregulated part of the nancial system.
How much credit tightening induced by macroprudential regulation will therefore be counteracted by the additional credit take-up of shadow banks? Given the diverse nature of nancial rms involved in shadow bank credit intermediation, their regulation falls into the court of various regulatory authorities. This makes consistent and comprehensive regulation more dicult to attain. In the absence of macroprudential regulation of non- bank nancial institutions, how can monetary policy react to limit the side e ect of leakage from regulated to unregulated intermediaries after adjustments in capital requirements of commercial banks?
We employ ECB data on euro area shadow and commercial banks in a full-information Bayesian estimation exercise and employ the quantitative model for policy analyses. Overall credit uctuation in response to monetary policy and banking regulation shocks is
ECB Working Paper Series No 2406 / May 2020 | 2 |
dampened, as shadow bank credit movements counteract deviations of commercial bank lending to some degree, resulting from the credit leakage mechanism. Furthermore, monetary policy can mitigate unintended credit leakage to the shadow banking sector in response to unanticipated increases in macroprudential regulation, as changes in short-term interest rates aect both commercial and shadow bank credit. Finally, we show in a coun- terfactual analysis how the level of regulatory capital would have evolved if the regulator had imposed counter-cyclical requirements already since 1999.
ECB Working Paper Series No 2406 / May 2020 | 3 |
- Introduction
A macroprudential approach towards nancial regulation focuses on systemic developments in nancial markets like swings in aggregate credit or nancial market volatility, as well as on the role of nancial cycles for business cycle movements.1 While a lot of regulation has focused on the commercial banking sector2, comprehensive regulation of the non-bank nancial sector has proven to be more dicult to attain. Since non-bank nan- cial institutions might take up some of the lending that banks have been prohibited from extending due to macroprudential policies3, understanding the interaction between commercial bank regulation and non-bank nancial institutions is crucial for the assessment of macroprudential policies.
We analyze the implications of considering non-bank nancial intermediaries, or shadow banks, in the conduct of macroprudential regulation of the commercial banking sector. The shadow banking sector comprises a set of diverse institutions conducting highly specialized tasks in the nancial system. Observed from an an aggregate level, the shadow banking sector intermediates funds from savers to borrowers in a similar fashion to the commercial banking system, but without being subject to macroprudential policies. How much of the credit tightening induced by macroprudential regulation will therefore be counteracted by the additional credit take-up of shadow banks? Given the diverse nature of nancial rms involved in shadow bank credit intermediation, their regulation falls into the court of various regulatory authorities. This makes consistent and comprehensive regulation more dicult to attain. In the absence of macroprudential regulation of non-bank nancial institutions, how can monetary policy react to limit the side eects?
To answer these questions, we derive a dynamic stochastic general equilibrium (DSGE) model with savers and borrowers, and two types of nancial institutions that intermediate funds between these two groups: commercial banks and shadow banks. Both types of intermediaries are based on distinct microeconomic foundations that allow for structural dierences with respect to regulatory coverage and market structure in the two sectors. We then apply Bayesian techniques and rely on economic and nancial data for the euro area to estimate the parameters of our model.
We contribute to the literature in several ways. On the technical side, we derive a heterogeneous nancial system by combining elements of two canonical frameworks for
- See Borio and Shim (2007), or Borio (2009, 2011) for a detailed description of the macroprudential approach.
- Rules laid down in the latest round of Basel accords on banking regulation (Basel III ) strongly focus on supervisory and regulatory tools targeting macro developments in credit and risk-taking, such as rules on interbank lending, cyclical adjustments of capital requirements, and supervision on bank interconnect- edness.
- See Cizel et al. (2019).
ECB Working Paper Series No 2406 / May 2020 | 4 |
modeling nancial frictions in DSGE models: our commercial banking sector is based on the work by Gerali et al. (2010), whereas our shadow banking sector is modeled similar to the nancial sector in Gertler and Karadi (2011). This is the rst paper to model both approaches in one consistent setting with heterogeneous nancial markets.4 The second contribution is the inclusion of data on euro area shadow bank lending in the estimation procedure, which has so far been limited to other jurisdictions. Third, we contribute to the policy discussion around unintended consequences of macroprudential measures and interaction with monetary policy: the presence of shadow banks can aect the setting of macroprudential policy, even if not directly enforceable on all nancial intermediaries.
Eective and targeted macroprudential policy is often considered to be the rst-best response in reaction to nancial instability concerns (see for instance Rubio and Carrasco- Gallego, 2015). We challenge this view in the face of a growing non-bank nancial sector in the euro area (see Figure 1) that remains sparsely regulated, and a potential for macroprudential and monetary policy coordination. Our approach is therefore intended as a caveat to work that considers nancial stability issues of the aggregate nancial sector or commercial banking sector only instead of dierentiating between banking and non-bank nance. Changes in macroprudential regulation for banks can trigger credit leakage towards unregulated institutions, and neglecting such changes in the composition of credit can impede the ecacy of such policies.5
The explicit policy tool we consider in this study is capital requirement regulation, which under Basel III represents a key macroprudential tool regulators can apply to the commercial banking system to prevent banks from engaging in excessive leverage and risk-taking. Countercyclical capital requirements can be raised to avoid excessive credit growth in boom times, and lowered whenever credit developments are subdued. In a subsequent policy exercise we assess the ability of a coordinated monetary policy response in combination with macroprudential policy to limit the leakage mechanism while still bringing about the desired reduction in aggregate lending.
In the following section, we review the literature on both the current state of nan- cial market-augmented macroeconomic models and on studies evaluating macroprudential regulation and coordination with other policy areas. We then introduce the full DSGE
4The updated version of the ECB's New Area Wide Model (NAWM II) depicts another example of a model where elements of both the Gerali et al. (2010) and the Gertler and Karadi (2011) framework are combined, see Coenen et al. (2019). However, the wholesale" and retail" banks motivated by the two frameworks are part of a representative bank holding group in this model, whereas both bank types will be separate entities in our setup.
- Increasing intermediation by non-bank nancial institutions does not per se depict an undesirable development. In some circumstances, technological advances as well as lower dependency on regulatory and institutional provisions of non-bank nancial institutions can increase the eciency in the intermediation process and increase overall welfare in the economy (Buchak et al., 2018; Ordonez, 2018). However, higher shares of intermediation being conducted by unregulated shadow institutions, which may exploit regulatory arbitrage, potentially increases risks to nancial stability, which is our concern here.
ECB Working Paper Series No 2406 / May 2020 | 5 |
Figure 1: Commercial and Shadow Bank Loans to Non-Financial Corporates
Bn. EUR
6000 | 40 | ||||||||||
5000 | 35 | ||||||||||
30 | |||||||||||
4000 | 25 | ||||||||||
Percent | |||||||||||
3000 | 20 | ||||||||||
2000 | 15 | ||||||||||
10 | |||||||||||
1000 | 5 | ||||||||||
0 | 0 | ||||||||||
1999 | 2001 | 2003 | 2005 | 2007 | 2009 | 2011 | 2013 | 2015 | 2017 | 2019 | |
Year |
Other Financial Intermediaries (OFI) | Commercial Banks (CB) | OFI/CB (rhs) | |
Note: Outstanding amount of loans of commercial and shadow banks (OFI) to non-nancial coprorates
(billions of euro). Source: Euro Area Accounts and Monetary Statistics (ECB).
model in Section 3.6 Section 4 introduces the data we use and discusses the econometric procedure we employ to derive estimates of key parameters of the model. In Section 5, we use our model to simulate the eects of neglecting shadow bank intermediation in macroprudential policy, before we conclude in Section 6.
- Literature
The literature on DSGE models including nancial intermediaries and frictions has been growing rapidly in the past years. The approach developed in Gertler and Karadi (2011) and Gertler and Kiyotaki (2011) introduces nancial frictions based on an agency problem arising between banks and households, as banks are allowed to divert household funds away from investment projects for private benets. Given that households are aware of potential misconduct, the ability of banks to obtain funding via deposits is limited.
A second strand of macro models incorporating nancial frictions focuses on the role of collateral that borrowers have to place with lenders in return for funding. Iacoviello
6We highlight the key mechanism entailing credit leakage towards shadow banks in the wake of tighter commercial banking regulation in a simplied two-period version of the model available in the separate Online Appendix which can be found here.
ECB Working Paper Series No 2406 / May 2020 | 6 |
and Guerrieri (2017) and Iacoviello (2005) introduce housing as collateral and relate the amount of borrowing by impatient households to movements in the value of collateral. Extending the approach, Gerali et al. (2010) introduce a banking sector in a canonical New Keynesian model for the euro area and locate the collateral friction between borrowers and banks. By modeling the banking sector explicitly, they are able to incorporate speci c characteristics of the euro area banking sector, such as market power and sluggish adjustment of bank interest rates in response to changes in the monetary policy rate.
Focusing on macroprudential regulation7, Angelini et al. (2014) implement collateral constraints and capital requirements set according to a simple rule in an estimated euro area New Keynesian DSGE model. They nd that macroprudential policy is particularly e ective in times of nancial distress, i.e. when shocks a ecting credit supply hit the economy. By employing a modeling framework based on the Holstrom and Tirole (1997) intermediation setup with rule-based policy makers in place, Christensen et al. (2011) nd that countercyclical regulatory policies can stabilize the economy in response to shocks originating in the banking sector.
Some studies evaluate the optimal degree of coordination between macroprudential policy makers and monetary policy in banking-augmented macro models. Within a class of simple policy rules, Angeloni and Faia (2013) nd that the best combination includes countercyclical capital ratios and a response of monetary policy to asset prices or bank leverage. Gelain and Ilbas (2017) introduce a central bank and a macroprudential regulator in a Smets and Wouters (2007) New Keynesian DSGE model augmented by a Gertler and Karadi (2011) nancial intermediary. They nd that in a fully cooperative scenario, a higher weight placed by the regulator on output gap stabilization { which depicts a joint policy objective { is bene cial for reducing macroeconomic volatility. In a non-cooperative setup, a higher weight on credit growth stabilization by the macroprudential regulator is bene cial. In a similar approach, Bean et al. (2010) nd that a combination of monetary and macroprudential policies appears to be more e ective as a means of leaning against the wind than relying on traditional monetary policy alone. Evaluating di erent policy regimes, Beau et al. (2012) nd that, over the business cycle, conicts among both policy makers should be limited. By deriving jointly optimal Ramsey policies, Collard et al. (2017) focus on di erent types of lending and show that limited liability and deposit insurance can cause excessive risk-taking in the nancial sector. Silvo (2019) evaluates Ramsey-optimal policies in a New Keynesian framework augmented by Holstrom and Tirole (1997). In line with Angelini et al. (2014), she nds that macroprudential policies play a modest stabilizing role in response to aggregate supply shocks, but are highly
- Banking-augmented macro models are frequently employed in the evaluation of dierent aspects of nancial stability, such as bank runs (Gertler and Kiyotaki, 2015; Gertler et al., 2016) or the eectiveness of (un-)conventional scal and monetary policies in times of nancial distress (Gertler and Karadi, 2011; Curdia and Woodford, 2010a,b, 2011).
ECB Working Paper Series No 2406 / May 2020 | 7 |
e ective when the nancial sector is the source of uctuations.
In all of the above studies, nancial intermediaries are modeled as homogeneous representative agents. Gertler et al. (2016) augment the canonical Gertler and Karadi (2011) framework by introducing a heterogeneous banking system with wholesale as well as retail banks. Wholesale banks represent the shadow banking part of the nancial system engaged in interbank funding, whereas retail banks collect household deposits to lend to both the wholesale banks and the non- nancial sector.8 Meeks and Nelson (2017) use a calibrated model to show how the interaction between shadow banks and commercial banks can a ect credit dynamics and that securitization in combination with high leverage in the shadow banking sector can have adverse e ects on macroeconomic stability. Verona et al. (2013) develop a model where shadow banks directly engage in credit intermediation between households and rms. In contrast to our model, they assume that shadow banks act under monopolistic competition to derive a positive spread between the lending rate of shadow banks and the risk-free rate. They show that incorporating shadow banks increases the magnitude of boom-bust dynamics in response to an extended period of loose monetary policy. Mazelis (2016) develops a model including traditional banks, shadow banks, and investment funds and studies the relevance of di erent types of credit for macroeconomic volatility. He concludes that a more equity-based nancial system can mitigate the credit crunch during recessions when the economy is stuck at the e ective lower bound of nominal interest rates (ELB).
Closest to our study, Begenau and Landvoigt (2016) and Feve and Pierrard (2017) evaluate how the existence of shadow banks can alter the e ectiveness of capital requirements. Studying the US case, the former study shows that tightening regulation for commercial banks can result in a shift of intermediation away from safer commercial banks towards unregulated and more fragile shadow banks, such that the net bene t of raising capital requirements for commercial banks only depends on the initial level of fragility in the nan- cial system. Similarly, Feve and Pierrard (2017) estimate a real business cycle model with US data and identify a leaking of intermediation towards shadow banks. They conclude that the degree of stabilization due to higher capital requirements for commercial banks can be dampened when more funds are channeled via the shadow banking sector.9 Aikman et al. (2018) discuss the role of credit intermediation by market-based nancial institutions that do not represent traditional banks for optimal coordination between monetary and macroprudential policies.
8The notion of a wholesale banking sector has already been introduced in Gertler and Kiyotaki (2015). Furthermore, Gertler and Kiyotaki (2011) discuss interbank borrowing. However, no distinct separation between wholesale and retail banks has been undertaken in these studies.
- Alternative microfoundations for the leakage of credit towards shadow bank institutions in response to rising capital requirements were derived in the theoretical banking literature. See for instance Ordonez (2018), Farhi and Tirole (2017), Plantin (2015), or Harris et al. (2014).
ECB Working Paper Series No 2406 / May 2020 | 8 |
- A Sketch of the DSGE Model
In this Section, we introduce shadow banking into a DSGE model akin to the euro area banking model developed by Gerali et al. (2010) and Gambacorta and Signoretti (2014).10 We furthermore provide a detailed description of the implication of introducing shadow banks for the eectiveness of commercial bank regulation in a stylized two-period model presented in the Online Appendix of this study. In the DSGE model, patient households serve as savers and provide funds to impatient entrepreneurs that act as borrowers, implemented by dierent values for the discount factor used in the utility functions of both agents. Households cannot directly provide funds to borrowing rms, but have to place deposits in nancial intermediaries that then provide loans to rms, which use the funds for production.11 Households can allocate savings between two types of intermediaries: shadow banks and commercial banks. Commercial banks face regulatory capital require- ments, whereas shadow banks are not obliged by regulation to back a minimum of assets with equity. As in Gerali et al. (2010), commercial banks exert market power when setting interest rates on loans and deposits and adjust these rates incompletely in response to policy changes.
In contrast to commercial banks, shadow banks act under perfect competition. They are neither subject to macroprudential regulation, nor do they have recourse to government support schemes such as deposit insurance and central bank liquidity facilities. Conse- quently, saving in shadow banks is more risky from the household perspective. Default risk can thus result in a positive spread between the rates households demand from shadow banks compared to commercial banks. To capture the dependence of shadow banks on market funding, we draw on the incentive constraint in Gertler and Karadi (2011). We assume that the lack of regulation is akin to the risk that shadow bankers can divert a share of funds, defaulting on the remaining liabilities in the process. Whenever the benets from doing so exceed the returns from behaving honestly, shadow bankers face an incentive to disappear from the market and leave investors with losses on their investments. Households are aware of this risk and limit their funding to an amount that motivates the shadow banker of continuing operations rather than diverting a share and defaulting on the rest. The implicit default risk the household faces when placing funds in shadow banks thus results in a spread between shadow bank and commercial bank deposit rates, as households demand higher compensation when placing funds in these institutions.
On the loan market, regulation only applies with respect to commercial banks, as
10The complete nonlinear DSGE model is presented in Appendix A.
11As in Gambacorta and Signoretti (2014), we assume that all debt contracts are indexed to current ination. In this respect, we deviate from the framework in Gerali et al. (2010) and Iacoviello (2005) by eliminating the nominal debt channel from the model. This channel potentially a ects the redistribution of funds between borrowers and savers and thus macroeconomic developments in response to unexpected changes in the price level, which we do not consider here.
ECB Working Paper Series No 2406 / May 2020 | 9 |
Intermediaries | Regulator | ||||||||||||
Commercial Banks (CB) | |||||||||||||
Assets | Liabilities | ||||||||||||
Entrepreneurs (E) | Households (H) | ||||||||||||
Assets | Liabilities | Loans to E | Deposit Debt | ||||||||||
Assets | Liabilities | ||||||||||||
Loans | Bank Equity | ||||||||||||
from CB | CB | ||||||||||||
Deposits | |||||||||||||
Productive | Loans | ||||||||||||
Capital | from SB | Shadow Banks (SB) | |||||||||||
Assets | Liabilities | Risky | |||||||||||
Net Worth | Loans to E | Risky Debt | Investment | ||||||||||
in SB | |||||||||||||
SB Equity |
Note: Overview of model relationships between agents involved in nancial intermediation.
entrepreneurs have to full an externally set loan-to-value ratio when demanding funds from commercial banks. Consequently, entrepreneurs can only borrow up to a certain amount of their collateral value at hand, which is given by the stock of physical capital that they own and use for production purposes. However, they can use their remaining collateral for borrowing from shadow banks.12 An overview of the relationships between savers, borrowers and the two intermediaries is given in Figure 2.
Households provide labor to entrepreneurs and either consume or save in nancial in- termediaries. Entrepreneurs produce intermediate goods and sell them on a competitive market to retailers that dierentiate, repackage and sell them on in a monopolistically competitive market, resulting in a nal goods price that includes a mark-up on the marginal cost. Furthermore, capital goods producers are introduced to derive a market price for capital. The central bank conducts monetary policy by setting the nominal short-term interest rate according to a Taylor rule.13
In the baseline model, macroprudential regulation is determined exogenously, before we introduce a macroprudential regulator that follows a countercyclical policy rule for capital
- Details on the microfoundation of the entrepreneur's credit constraints and the superiority of commer- cial bank credit are provided in the Online Appendix.
- In this model, we abstract from any unconventional monetary policy and assume that the economy is not at the eective lower bound (ELB) of nominal interest rates.
ECB Working Paper Series No 2406 / May 2020 | 10 |
requirements in Section 5. In this respect, our baseline model used in the estimation procedure of Section 4 reects the regulatory framework in the euro area in place before the introduction of Basel III. Under the preceding Basel II regulations, countercyclical adjustments of capital requirements for commercial banks were not set systematically.
3.1 Households
The representative patient household i maximizes the expected utility
1 | "(1 | lP (i)1+P | # | ||||||||||||||
E0 | P | a | )"t log(ct (i) | a | ct1 ) | (1) | |||||||||||
P | P | P;C | P;S | 1 + P | |||||||||||||
ct | (i); lt | (i); dt | (i); dt | (i) | Xt | P | z | P | P | P | t | ||||||
max | t | ||||||||||||||||
=0 | |||||||||||||||||
subject to the budget constraint | |||||||||||||||||
ctP (i) + dtP;C (i) + dtP;S(i) wtltP (i) + (1 + rtdC1 | )dtP;C1 (i) + (1 + rtdS1 )dtP;S1 | (i) + ttP (i) | (2) |
where cPt (i) depicts current consumption and lagged aggregate consumption is given by cPt1 . Working hours are given by ltP and labor dis-utility is determined by P . Preferences are subject to a disturbance "zt a ecting consumption. The ow of expenses includes current consumption and real deposits to be made to both commercial and shadow banks, dP;Ct (i) and dP;St (i). Resources consist of wage earnings wtP ltP (i) (where wtP is the real wage rate for the labor input of each household), gross interest income on last period deposits (1 + rtdC1 )dP;Ct1 (i) and (1 + rtdS1 )dP;St1 (i), and lump-sum transfers tPt that include dividends from rms and banks (of which patient households are the ultimate owners).
3.2 Entrepreneurs
Entrepreneurs use labor provided by households as well as capital to produce intermediate goods that retailers purchase in a competitive market. Each entrepreneur i derives utility from consumption cEt (i), which it compares to the lagged aggregate consumption level of all entrepreneurs. They maximize expected utility
ct (i); lt (i); bt | (i); bt | (i) | 1 | |||||||||||||||
Xt | ||||||||||||||||||
E | P | max | E;S | E0 | Et log(ctE(i) aEctE1 ) | (3) | ||||||||||||
E;C | =0 | |||||||||||||||||
subject to the budget constraint | ||||||||||||||||||
cE(i) + wtlP (i) + (1 + rbc | )bE;C (i) + (1 + rbs | )bE;S(i) + qkkE(i) | ||||||||||||||||
t | t | t1 | t1 | t1 | t1 | t t | ||||||||||||
ytE(i) | E;C | E;S | k | k | E | |||||||||||||
= | + b | t | (i) + b | t | (i) + q | t | (1 | )k | t1 | (i) (4) | ||||||||
xt | ||||||||||||||||||
ECB Working Paper Series No 2406 / May 2020 | 11 |
with k depicting the depreciation rate of capital, qtk the market price for capital in terms of consumption, and xt determining the price markup in the retail sector.
Entrepreneurs face a constraint on the amount they can borrow from commercial banks which depends on the value of collateral the rm holds. The collateral value of the entrepreneurs is determined by their expected physical capital stock in the period of repayment (t + 1), which is given by Et[(1 k)ktEt+1].14 Whereas a regulatory loan-to- value (LTV) ratio mCt applies for funds borrowed from commercial banks, shadow bank funding is not prone to regulation. As outlined in detail in the Online Appendix, due to a positive spread between interest rates charged for shadow bank and commercial bank loans, entrepreneurs have an incentive to borrow from commercial banks rst and turn to shadow bank lending only whenever the possible amount of commercial bank funds, determined by mCt ktE(i), is reached. Further borrowing can be obtained from shadow banks by using capital holdings not reserved for commercial bank funds, (1 mCt )ktE(i). Thus, the two respective borrowing constraints are given by
(1 + rtbc)btE;C (i) mtC Et[qtk+1(1 k)ktE(i)]; | (5) |
(1 + rtbs)btE;S(i) (1 mtC )Et[qtk+1(1 k)ktE(i)] | (6) |
where the LTV ratio for commercial banks mCt is set exogenously by the regulator and follows an exogenous AR(1) process.
We follow Iacoviello (2005) and assume that the borrowing constraints bind around the steady state such that uncertainty is absent in the model.15 Thus, in equilibrium, entrepreneurs face binding borrowing constraints, such that Equations 5 and 6 both hold with equality.
3.3 Financial Intermediaries
Both commercial banks and shadow banks intermediate funds between households and rms. While they both engage in intermediation in a similar fashion, we assume the two types of agents to be structurally dierent along various dimensions.
First, we assume that commercial banks are covered by banking regulation, which implies that they have to fulll requirements on the amount of capital they have to hold
- In Iacoviello (2005), entrepreneurs use commercial real estate as collateral. However, we follow Gerali et al. (2010) by assuming that creditworthiness of a rm is judged by its overall balance sheet condition where real estate housing only depicts a sub-component of assets.
- Iacoviello (2005) discusses the deviation from the certainty equivalence case in Appendix C of his paper.
ECB Working Paper Series No 2406 / May 2020 | 12 |
compared to the size of their balance sheet. Second, they are eligible for central bank liquidity assistance and government guarantees such as deposit insurance schemes.16 Thus, for households and rms, commercial banks depict safe deposit institutions, given that they are both covered by regulation and have access to government support schemes. We furthermore assume market power in the loan and deposit markets for commercial banks,17 and model it using the same Dixit-Stiglitz framework as employed in Gerali et al. (2010). Thus, in both loan and deposit markets, commercial banks are able to charge some markup on loan rates and pay deposit rates conditional on a markdown. In line with Gerali et al. (2010), we model commercial banks by distinctively separating a single bank into three units: two retail branches responsible for retail lending and retail deposits, respectively, and one wholesale branch that manages the bank capital position. While the two retail branches operate under monopolistic competition, we assume lending and deposit taking between retail and wholesale units to operate perfectly competitively.
Shadow banks, in contrast, face no regulatory burden but in turn are not covered by structural support schemes. Consequently, the shadow banking sector increasingly depends on creditor trust, which is captured by a moral hazard problem that governs the degree of leverage of shadow bank institutions. Whereas commercial banks' charter values as well as their funding opportunities via central banks provide a bu er in case of illiquidity, shadow banks are exposed to funding pressures that can lead to instantaneous exit from participation in the market. Reduced regulatory burdens in the establishment of shadow banking operations supports regular inow to this market. As a consequence, while we assume commercial banks to be in nitely lived in our model, we allow for frequent entry to and exit from the shadow banking system.
3.3.1 Commercial Banks
In the following, we discuss the maximization problem of the wholesale unit of the commercial bank as the capital requirement set by regulators applies directly to this branch of the commercial bank.18 Due to space limitations, we will not discuss the maximization problems of the retail deposit and loan branches here as they are identical to the problems outlined in Gerali et al. (2010) and we refer to their study.
Wholesale Unit The wholesale branches of commercial banks operate under perfect competition and are responsible for the capital position of the respective commercial bank.
16Even though not explicitly modeled, the assumption of an existing insurance scheme lies behind the idea of shadow bank deposits being more risky than deposits placed with commercial banks.
17The existence of market power in the euro area was indicated in various empirical studies, see for instance Fungacova et al. (2014) or De Bandt and Davis (2000).
18Thus, the modeling of the wholesale unit closely resembles the commercial bank outlined in Section 1.3.1. in the Online Appendix.
ECB Working Paper Series No 2406 / May 2020 | 13 |
On the asset side, they hold funds they provide to the retail loan branch, bCt , and these retailers ultimately lend the funds to entrepreneurs as credit bE;Ct . As retailers act under monopolistic competition, the retail rate rtbc comprises a markup over the wholesale loan rate rtC . On the liability side, the wholesale unit combines commercial bank net worth, or capital, ktC , with wholesale deposits, dCt , that are provided by the retail deposit branch, but originally stem from deposits placed in the retail branch by patient households (dP;Ct ). Furthermore, the capital position of the wholesale branch is prone to a regulatory capital requirement tC . Moving away from the regulatory requirement imposes a quadratic cost to the bank, which is proportional to the outstanding amount of bank capital and parameterized by Ck .
The wholesale branch maximization problem can be expressed as
kC | ktC | 2 | ||||||
C C | dC C | C | C | |||||
bC ; dC | bt | rt dt | t | kt | ||||
2 | btC | |||||||
max rt |
t t
subject to the the balance sheet condition
bCt = ktC + dCt :
The rst-order conditions yield the following expression:
rt | = rt | k | kC | t | kC | 2 | |||
btC | btC | : | |||||||
C | dC | C | t | C | t |
(7)
(8)
(9)
As the commercial bank has access to central bank funding in the model, we assume that the rate paid on wholesale deposits gathered from the retail deposit unit of the commercial bank (and so originally from households and rms) has to be equal to the risk-free policy rate, rt, via arbitrage:
rtdC = rt
such that the spread between the loan and deposit rates on the wholesale level is given by
kC | kC | 2 | |||||||
rt = rt | k | t | : | (10) | |||||
btC | btC | ||||||||
C | C | t | C | t |
This expression indicates that the marginal benet from further lending, the spread earned on intermediation at the margin, has to be equal to the marginal costs from doing so in equilibrium. This marginal cost increases whenever the deviation of commercial bank capital holdings from the regulatory requirement increases.
Assuming symmetry between banks and reinvestment of prots in banks, aggregate
ECB Working Paper Series No 2406 / May 2020 | 14 |
bank capital KtC is accumulated from retained earnings only:
KtC = (1 C )KtC1 + JtC1 | (11) |
where JtC depicts aggregate commercial bank pro ts derived from the three branches of the bank, see Gerali et al. (2010). Capital management costs are captured by C .
3.3.2 Shadow Banks
In contrast to the commercial banking sector, shadow banks do not operate under monopolistic competition. The shadow banking sector is assumed to consist of a multitude of di erentiated and specialized business entities, which, taken together, engage in similar intermediation activity as commercial banks. Given the exibility and the heterogeneity of the shadow banking system, we assume shadow banks operate under perfect competition.
Instead of being constrained by regulation, as commercial banks are, shadow banks' ability to acquire external funds is constrained by a moral hazard problem that limits the creditors' willingness to provide external funds. To avoid excessive equity capital accumulation { and eventually exclusive nancing via equity rather than debt { shadow bankers are assumed to have a nite lifetime: Each shadow banker faces an i.i.d. survival probability S with which they will be operating in the next period. This exit probability functions in the maximization problem of the shadow banker as an additional discount factor, which ensures that they are always net debtors to the households. To make up for the outow, every period new shadow bankers enter with an initial endowment of wS they receive in the rst period of existence. The number of shadow bankers in the system remains constant.19
As long as the real return on lending (rbS | rdS) is positive, it is pro table for the |
t | t |
shadow bank to accumulate capital until they exit the intermediation sector. Thus, the shadow bank's objective to maximize expected terminal wealth, vt(j), is given by
1 | ||
vt(j) = max Et | (1 S)Si Si+1ktS+1+i(j): | (12) |
Xi | ||
=0 |
We introduce a moral hazard problem that leads to the possibility of positive spreads earned by shadow banks.20 We allow for the possibility that shadow banks divert a fraction of available funds, S, and use them for private bene ts at the beginning of each period. Households can consequently only recover the leftover share (1 S) afterwards. However,
19The complete derivation of the shadow bank problem and a deeper discussion of the approach used is presented in Section A.3.2 of Appendix A.
20See Section 2 in the Online Appendix.
ECB Working Paper Series No 2406 / May 2020 | 15 |
diverting funds and running away" is equivalent to declaring bankruptcy for the shadow bank, such that it will only do so if the return of declaring bankruptcy is larger than the discounted future return from continuing operations:
vt(j) SqtkbtE;S(j): | (13) |
Equation 13 depicts the incentive constraint the shadow banker faces when trying to acquire funds from households.21 As we assume some shadow bankers to exit each period and new bankers to enter the market, aggregate capital ktS is determined by the capital of continuing shadow bankers, ktS;c, and the capital of new bankers that enter, ktS;n
ktS = ktS;c + ktS;n | (14) |
and combining the expressions for ktS;c and ktS;n derived in Appendix A yields the following law of motion for shadow bank capital:
ktS = S[(rtbS1 | rtdS1 )tS1 + (1 + rtdS1 )]ktS1 + !SqtkbtE;S1 : | (15) |
The shadow bank balance sheet condition
qkbE;S | (j) = dP;S | (j) + kS(j) | (16) |
t t | t | t |
in combination with the demand for shadow bank credit by borrowers given by Equation 6 determines shadow bank lending bE;St (j).
Finally, we assume a non-negative spread between the interest rates earned on shadow bank deposits, rtdS, and on the deposits households can place with commercial banks, rtdC , which is determined by the parameter S, with 0 S 1. In Section 1.1. of the Online Appendix, we provided a microfoundation for the existence of a positive spread, and use the results to incorporate a relationship between the two deposit rates similar to the relation stated in Equation 4 in the two-period model:
dS | 1 + rdC | |||
t | : | |||
1 + rt | = | |||
S" | ||||
1 | ||||
t |
As in the two-period version of the model, the parameter S
(17)
determines the spread
21Compared to Equation 37 in the Online Appendix, the interest rate term on the right-hand side is missing here. In the DSGE model, we do not have xed shadow bank capital, but interest returns from the previous period are booked into shadow bank capital at the end of a respective period. In the innite- horizon case, the timing of events is such that at the beginning of any period t, shadow banks use net worth ktS(j) together with deposits dP;St (j) to lend out nancial claims bE;St (j). Afterwards, the shadow banker decides whether to run away or not. In case of behaving honestly, they receive net returns rtbS rtdS on intermediation at the end of period t, and these returns are then part of the capital stock in the next period, ktS+1(j).
ECB Working Paper Series No 2406 / May 2020 | 16 |
between the gross rates on both deposit types and is implicitly related to the default probability of shadow banks. As a shortcut, we calibrate S and assume the existence of a spread shock "t following an auto-regressive process to motivate exogenous swings in the spread on interest rates earned on the two deposit types.
- Estimation
4.1 Data
All economic variables used in the estimation exercise are drawn from the European System of Accounts (ESA 2010) quarterly nancial and non- nancial sector accounts, provided by the ECB and Eurostat.22 For the real and nominal economy, we include information on real gross domestic product, real consumption, real investment, and consumer price as well as wage ination. Information on commercial bank balance sheets { commercial bank deposits held by private households and commercial bank loans granted to the non- nancial corporate sector { is gathered from the data set in Monetary Financial Institutions" (MFIs) collected by the ECB. Data on commercial bank interest rates on household deposits and rm loans are drawn from di erent sources within the ECB Statistical Data Warehouse and harmonized in line with the procedure recommended by Gerali et al. (2010). We also use the short-term EONIA rate as a quarterly measure of the policy rate. For shadow bank variables, we use information provided in the ECB data base on di erent monetary and other nancial institutions, as discussed in detail in Appendix B.
We apply full-information Bayesian techniques to estimate some of the model parame- ters. Our baseline sample covers the period between 1999:Q1 and 2013:Q4, as we assume that the e ective lower bound (ELB) on nominal interest rates in the euro area was reached in 2014.23 Furthermore, as the e ective implementation of the Basel III framework under the Capital Requirements Directive IV (CRD IV) of the European Union took place from 2014:Q1 onward, we estimate our baseline model reecting the regulatory landscape under Basel II for the period before the implementation of the new framework.24
In total, we use eleven time series,25 and we apply the Metropolis-Hastings algorithm
- See Appendix B for a detailed description of the data set.
- See for instance Coeur (2015) for a discussion of the beginning of the ELB period in the euro area. We conduct a robustness analysis using a dierent sample period in Appendix C.
- In the euro area, the implementation of Basel III is governed by the Capital Requirements Directives IV (CRD IV) and the subsequent Regulation on Prudential Requirements for Credit Institutions and Investment Firms (CRR), which came into force on January 1, 2014. Thus, as euro area countries did not implement the policy measures put forward under Basel III before the beginning of 2014, we are eectively covering the pre-Basel III era of banking regulation in the euro area with our sample for the baseline estimation.
- See charts in Figure 10 of Appendix B.
ECB Working Paper Series No 2406 / May 2020 | 17 |
to derive draws from the posterior distribution, by running ve chains with 500,000 draws each in the baseline estimation. We evaluate convergence in the estimation by considering the approach of Brooks and Gelman (1998). We furthermore check for the identication of parameters following Ratto and Iskrev (2011).26
4.2 Calibration and Prior Distributions
Table 1 depicts calibrated parameters. In most cases, we apply the calibration used by Gerali et al. (2010). We adjust parameters on the loan (deposit) rate markup (markdown) for commercial bank lending "bE ("d) to match the mean spreads in our extended sample.
bE | " | d | |||||
As the loan rate markup (deposit rate markdown) is given by | " | ( | ), we set param- | ||||
" | bE | d | |||||
1 | " 1 |
eters to match the average annualized loan rate spread (deposit rate spread) with respect to the EONIA of 240 basis points (35 basis points) in our extended sample.27
In addition, by incorporating shadow banks and macroprudential regulation in the model, we introduce three new parameters: S, S, and S. Given our broad denition of shadow banks, nding empirical equivalents to shadow bank deposit returns is not straightforward. The shadow bank aggregate we consider covers institutions with highly diverse investment portfolios, dierent types of investors placing funds, and ultimately highly varying returns on the specic activity they are engaged in. We calibrate S such that the implied default probability of shadow banks is approximately ve percent per quarter and the resulting annualized spread between shadow bank and commercial bank deposit rates is approximately two percentage points in steady state.
Furthermore, we ensure in the calibration that the share of shadow bank intermediation in total intermediation is approximately one-third in steady state and that the size of the average shadow bank loan portfolio is three times the size of shadow bank capital. These values are comparable to statistical gures derived in empirical studies on the euro area shadow banking sector based on similar data (Bakk-Simonet al., 2012; Malatesta et al., 2016) and resemble average values in our data set. The latter calibration allows us to treat S as a transformed parameter in the estimation, and the resulting post-estimation value is given by 0.944. Our value of S, the share of divertible funds, turns out to be lower than the calibrated value in Gertler and Karadi (2011), where the authors settled on a value of 0.381 in the calibration of the US model.28 Furthermore, we set the steady-
state commercial bank capital requirement, C , equal to 8 percent, which resembles the
- Details on convergence statistics and identication tests are available upon request.
- In Gerali et al. (2010), the retail deposit rate spread is stated to be 125 basis points. However, we include the period after 2008 in our sample, where bank market power was adversely aected by the global nancial crisis and the debt crisis in Europe and thus lending and deposit margins for commercial banks were reduced signicantly.
28An economic interpretation of the lower share that intermediaries can divert in the euro area could be given by a higher degree of creditor protection.
ECB Working Paper Series No 2406 / May 2020 | 18 |
Table 1: Calibrated Parameters | |||||||||||||||
Parameter | Description | Value | |||||||||||||
S | Deposit Rate Spread Parameter | 0.05 | |||||||||||||
S | SB Share of Divertible Funds | 0.2 | |||||||||||||
S | SB Survival Probability | 0.944 | |||||||||||||
C | Steady-State Capital Requirement | 0.08 | |||||||||||||
P | Inverse Frisch Elasticity of Labor Supply | 1 | |||||||||||||
P | Discount Factor Households | 0.9943 | |||||||||||||
E | Discount Factor Entrepreneurs | 0.975 | |||||||||||||
mC | Steady-StateLTV Ratio vs. Commercial Banks | 0.3 | |||||||||||||
Capital Share in Production Function | 0.2 | ||||||||||||||
"d | Deposit Rate Markdown Given by | "d | -0.9 | ||||||||||||
d | |||||||||||||||
bE | " 1 | ||||||||||||||
"bE | Loan Rate Markup Given by | " | 2 | ||||||||||||
" | bE | 1 | "y | ||||||||||||
"y | |||||||||||||||
Goods Market Markup Given by | 6 | ||||||||||||||
y | |||||||||||||||
" | 1 | ||||||||||||||
"l | Labor Market Markup Given by | "l | 5 | ||||||||||||
l | |||||||||||||||
k | " 1 | ||||||||||||||
Depreciation Rate Physical Capital | 0.025 | ||||||||||||||
b | Bank Capital Management Cost | 0.1049 | |||||||||||||
overall level of capital-to-asset holdings demanded from commercial banks under Basel
- The steady-state LTV ratio for commercial banks mC is calibrated following Gerali et al. (2010), implying relatively strict regulation on collateral and a signi cant scope for shadow bank lending based on collateral criteria.29
For the prior distributions, we mostly follow Gerali et al. (2010) for the parameters originally estimated in their study. As we apply a Calvo (1983) pricing framework instead of Rotemberg (1982), we rely on a prior distribution similar to those introduced by Smets and Wouters (2003, 2007) for the Calvo parameter p. We choose a slightly tighter prior distribution for the Taylor-rule parameter on ination and change the distribution on the respective output parameter to a Beta distribution compared to the Normal distribution used in the original study. We thus give preference to a prior distribution that does not, even theoretically, allow for negative values of the parameter. Table 2 reports prior and posterior distributions for structural parameters as well as parameters describing exogenous processes. As in Gerali et al. (2010), we take the posterior medians as parameter estimates, and report estimates derived in their study for comparability.
For the parameter that governs the cost of deviation from the capital requirement, C , we assume a uniform distribution ranging from 0 to 25. Since this parameter is dicult to identify in an observable empirical counterpart, the non-informative nature of this prior in principle allows sucient exibility for the posterior to assume a broad range of values depending on the highest likelihood of the entire model and parameter set.
We nally use the same priors for all exogenous process parameters, including the
29Changing the LTV ratio to higher levels does not change the estimation results dramatically.
ECB Working Paper Series No 2406 / May 2020 | 19 |
parameters related to the two newly introduced shocks to the spreads between shadow bank and commercial bank returns ("t ), as can be seen in Table 2.
4.3 Posterior Distributions
In Table 2 we also report summary statistics of the posterior distributions for the model parameters. We furthermore provide marginal densities of the prior and posterior distributions for the structural parameter estimates in Figure 11 in Appendix B.
Even though the mode of the posterior for the Calvo parameter turns out to be slightly lower than the estimate derived in Smets and Wouters (2003), price stickiness is a sig- ni cant feature in the model. The posterior mode for the investment adjustment cost parameter i turns out to be of similar magnitude as the parameter derived in Smets and Wouters (2003), whereas Gerali et al. (2010) report a larger value for this parameter.
Sluggish interest rate adjustment appears particularly strong in the market for commercial bank deposits, indicated by high posterior mode and median values for the deposit rate adjustment cost parameter d. Furthermore, loan rates adjust more rapidly to changes in the policy rate compared to commercial bank deposit rates. Commercial banks therefore appear to react to changes in monetary policy by a more exible adjustment of loan rates in response to competition from shadow banks which operate under perfect interest-rate pass through, compared to a situation where shadow banking is absent.
As indicated in the previous section, the uniform prior for the commercial bank capital requirement adjustment cost parameter Ck was selected due to a weak identi cation prob- lem, and the resulting estimated median turns out to be slightly lower than in Gerali et al. (2010).30 Parameters related to monetary policy are broadly in line with results derived for instance in Gerali et al. (2010) and Smets and Wouters (2003), with our estimated posterior modes for the Taylor rule parameters , y and r taking on values in-between the estimated parameters derived in these studies. Finally, household habit formation is slightly weaker than in Gerali et al. (2010). For all shock processes, persistence turns out to be relatively high, with the processes for commercial bank deposit rate markdown shocks and price markup shocks depicting exceptions.
30We conducted a sensitivity analysis to evaluate the robustness of the estimation and found that model dynamics are unchanged when this parameter is varied.
ECB Working Paper Series No 2406 / May 2020 | 20 |
2020 May / 2406 No Series Paper Working ECB
21
Table 2: Prior and Posterior Distributions
Prior Distribution | Posterior Distribution | GNSS (2010) | ||||||||
(Median) | ||||||||||
Distribution | Mean | Std.Dev. | 5 Perc. | Median | 95 Perc. | Mode | ||||
Structural Parameters | ||||||||||
p | Calvo Parameter | Beta | 0.5 | 0.10 | 0.83 | 0.87 | 0.90 | 0.86 | - | |
i | Investment Adjustment Cost | Gamma | 2.5 | 1.0 | 2.98 | 3.98 | 5.14 | 3.67 | 10.18 | |
d | Deposit Rate Adjustment Cost | Gamma | 10.0 | 2.5 | 10.00 | 13.26 | 16.72 | 12.62 | 3.50 | |
bE | Loan Rate Adjustment Cost | Gamma | 3.0 | 2.5 | 4.84 | 8.34 | 14.23 | 7.56 | 9.36 | |
C | CCR Deviation Cost | Uniform | 0.0 | 25.0 | 0.01 | 10.05 | 21.32 | 24.71 | 11.07 | |
k | ||||||||||
TR cientCoe | Gamma | 1.5 | 0.25 | 1.44 | 1.87 | 2.30 | 1.75 | 1.98 | ||
y | TR cientCoe y | Gamma | 0.20 | 0.05 | 0.14 | 0.24 | 0.34 | 0.20 | 0.35 | |
r | Interest Rate Smoothing | Beta | 0.75 | 0.10 | 0.84 | 0.88 | 0.91 | 0.88 | 0.77 | |
aP | HH Habit Formation | Beta | 0.50 | 0.10 | 0.70 | 0.77 | 0.84 | 0.77 | 0.86 | |
Exogenous Processes (AR cients)Coe | ||||||||||
Deposit Rate Spread | Beta | 0.8 | 0.1 | 0.62 | 0.81 | 0.95 | 0.85 | - | ||
z | Consumer Preference | Beta | 0.8 | 0.1 | 0.82 | 0.87 | 0.92 | 0.87 | 0.39 | |
a | Technology | Beta | 0.8 | 0.1 | 0.31 | 0.42 | 0.52 | 0.42 | 0.94 | |
mE | Entrepreneur LTV | Beta | 0.8 | 0.1 | 0.91 | 0.94 | 0.97 | 0.95 | 0.89 | |
d | Deposit Rate Markdown | Beta | 0.8 | 0.1 | 0.27 | 0.36 | 0.46 | 0.36 | 0.84 | |
bE | Loan Rate Markup | Beta | 0.8 | 0.1 | 0.51 | 0.63 | 0.75 | 0.64 | 0.83 | |
qk | Investment ciencyE | Beta | 0.8 | 0.1 | 0.33 | 0.46 | 0.58 | 0.49 | 0.55 | |
y | Price Markup | Beta | 0.8 | 0.1 | 0.25 | 0.36 | 0.47 | 0.37 | 0.31 | |
l | Wage Markup | Beta | 0.8 | 0.1 | 0.64 | 0.71 | 0.77 | 0.71 | 0.64 | |
C | Commercial Bank Capital | Beta | 0.8 | 0.1 | 0.93 | 0.96 | 0.99 | 0.97 | 0.81 | |
k | ||||||||||
Exogenous Processes (Standard Deviations) | ||||||||||
Deposit Rate Spread | Inverse Gamma | 0.01 | 0.05 | 0.002 | 0.007 | 0.016 | 0.005 | - | ||
z | Consumer Preference | Inverse Gamma | 0.01 | 0.05 | 0.008 | 0.011 | 0.014 | 0.011 | 0.027 | |
a | Technology | Inverse Gamma | 0.01 | 0.05 | 0.025 | 0.029 | 0.033 | 0.028 | 0.006 | |
mE | Entrepreneur LTV | Inverse Gamma | 0.01 | 0.05 | 0.006 | 0.008 | 0.009 | 0.007 | 0.007 | |
d | Deposit Rate Markdown | Inverse Gamma | 0.01 | 0.05 | 0.002 | 0.002 | 0.002 | 0.002 | 0.032 | |
bE | Loan Rate Markup | Inverse Gamma | 0.01 | 0.05 | 0.002 | 0.002 | 0.003 | 0.002 | 0.063 | |
qk | Investment ciencyE | Inverse Gamma | 0.01 | 0.05 | 0.001 | 0.002 | 0.002 | 0.002 | 0.019 | |
r | Monetary Policy | Inverse Gamma | 0.01 | 0.05 | 0.001 | 0.001 | 0.002 | 0.001 | 0.002 | |
y | Price Markup | Inverse Gamma | 0.01 | 0.05 | 0.001 | 0.002 | 0.002 | 0.001 | 0.598 | |
l | Wage Markup | Inverse Gamma | 0.01 | 0.05 | 0.035 | 0.041 | 0.047 | 0.040 | 0.561 | |
C | Commercial Bank Capital | Inverse Gamma | 0.01 | 0.05 | 0.003 | 0.003 | 0.004 | 0.003 | 0.031 | |
k | ||||||||||
Note: Results are based on 5 chains with 500,000 draws each based on the Metropolis-Hastings algorithm. GNSS (2010) refers to the results reported in Gerali et al. (2010).
- Policy Analyses
We use our estimated model to evaluate whether disregarding credit intermediation via the shadow banking sector in macroprudential policy decisions has quantitative implications for policy decisions and the macroeconomy. In this context, we discuss potential implications for policy coordination between central banks and macroprudential regula- tors. Furthermore, in a counterfactual analysis, we assess how regulators would have set capital requirements under a countercyclical policy rule in the fashion of the Basel III regulatory framework had it been in place throughout the existence of the common currency. To do so, we introduce a policymaker following a countercyclical rule in the pre-Basel III model and simulate the development of capital requirements over the course of the monetary union. We thereby evaluate the implications from policy rules with different target variables. Furthermore, we discuss to what extent the level of implied capital requirements would have changed if regulators took not only commercial bank credit, but overall credit into account.
5.1 Macroprudential Regulation
In the following analyses, we discuss dierent regulatory regimes, depending on the degree of shadow bank consideration. We therefore implement, in the estimated Basel II model, dierent types of regulators that follow countercyclical rules for adjusting commercial bank capital requirements. We take key elements of the Basel III framework into account: countercyclical adjustment of capital requirements in response to swings in the credit cycle and the primary focus on commercial banking in the application of macroprudential policy. As indicated above, before the implementation of Basel III, the requirement on total capital holdings was 8 percent, and no countercyclical adjustment of requirements was intended. We raise the steady-state capital requirement for all regulator types from 8 percent to 10.5 percent31 and change the capital requirement equation in the model from an exogenous AR(1) process to a regulation-specic countercyclical rule described in more detail below. We leave the rest of the calibration and estimated parameters unchanged, as they were derived from the estimation using the true regulatory setup and economic data before the implementation of Basel III.
We discuss two dierent versions of the Basel III macroprudential regulator { in addition to the case without countercyclical capital regulation as under Basel II { that can apply capital requirements only to commercial banks, but cannot enforce regulation on the shadow banking system. The dierence between these types emerges from the degree
31The Basel III capital requirement consist of dierent types of buers banks have to hold: 8 percent (minimum Tier 1+2 capital) plus 2.5 percent (capital conservation buer), yielding 10.5 percent for total capital.
ECB Working Paper Series No 2406 / May 2020 | 22 |
to which shadow banking is considered when setting capital requirements for commercial banks. We dene a moderate regulator that only takes variation in commercial bank credit into account when setting capital requirements for commercial banks. In comparison, a prudent regulator considers overall credit, which includes both commercial and shadow bank credit.32
The macroprudential policy rules we consider resemble the countercyclical capital requirement on commercial bank balance sheets introduced with the Basel III accords. The macroprudential authority raises the requirement on the capital-to-asset ratio whenever a certain credit measure rises above the level perceived as stable, and lowers the requirement whenever the credit gap is negative. We distinguish between four target variables that indicate the credit cycle: credit levels, credit growth, as well as the level and the growth rate of the credit-to-GDP ratio.33 The regulator thus raises the capital-to-asset ratio tC
above the steady-state level of capital requirements C whenever the respective measure deviates positively from its steady-state value, and vice versa.
5.1.1 The Moderate Regulator
We rst evaluate the policy setting of a moderate regulator that only focuses on developments in commercial bank credit when setting capital requirements for commercial banks. The policy rules the moderate regulator follows in each scenario resemble the rule derived in Angelini et al. (2014) where deviations of the respective credit measure from steady state are targeted:
Credit Growth Rule: | tC |
Credit/GDP Growth Rule: | tC |
Credit Level Rule: | tC |
Credit/GDP Level Rule: | |
where |
BM | = bE;C | |||||||
bE;C ; | ||||||||
bM | btE;C | bE;C | ||||||
t | t | |||||||
Zbt | = | |||||||
Yt | ||||||||
Y | ||||||||
32Under Basel III, the specic credit measure that should be applied is not stated explicitly in the regulatory statutes, and the primary focus of regulators lies on credit intermediated by commercial banks.
33The choice of target variables is inspired by the common measures employed in the DSGE literature on capital requirements. See for instance Rubio and Carrasco-Gallego (2016), Bekiros et al. (2018), Angelini et al. (2014), Angeloni and Faia (2013), and Christensen et al. (2011).
ECB Working Paper Series No 2406 / May 2020 | 23 |
and indicates the dierence of a variable compared to its one-period lag. The reaction parameter determines the degree of policy sensitivity, and we calibrate it to a value of 7, which is broadly in line with the parameter values derived in Angelini et al. (2014). Furthermore, we allow for exogenous shocks "t to the capital requirement, and assume an auto-regressive shock process and smoothing in the adjustment of capital requirements, governed by parameter which we calibrate at a value of 0.9.
5.1.2 The Prudent Regulator
In addition, a prudent regulator is introduced that takes lending by the shadow banking sector into account when setting capital requirements for commercial banks. Despite the lack of a unifying regulatory framework for shadow banks, we assume that the prudent regulator is able to derive estimates of non-bank credit intermediation. The regulator therefore considers not only commercial bank credit, but movements in overall credit.34 The policy rules stated in Equations 18 to 21 are thus altered for the prudent regulator such that:
Credit Growth Rule: | tC = (1 | ) | C | + (1 | ) BtP + tC | 1 + "t ; | (22) | |||||||||||||||||
t = (1 | ) | bt | t 1 | t | ||||||||||||||||||||
Credit/GDP Growth Rule: | )C + (1 | |||||||||||||||||||||||
C | ZP + | C | + " ; (23) | |||||||||||||||||||||
Credit Level Rule: | tC = (1 | ) Btb | + | t | 1 | + "t ; | (24) | |||||||||||||||||
) | C | + (1 | ||||||||||||||||||||||
P | C | |||||||||||||||||||||||
t = (1 | ) | bt | t 1 | t | ||||||||||||||||||||
Credit/GDP Level Rule: | )C + (1 | + | (25) | |||||||||||||||||||||
C | ZP | C | + " ; | |||||||||||||||||||||
where | b | |||||||||||||||||||||||
BP = (bE;C + bE;S) | ||||||||||||||||||||||||
(bE;C + bE;S); | ||||||||||||||||||||||||
bP | btE;C + btE;S | bE;C + bE;S | ||||||||||||||||||||||
t | t | t | ||||||||||||||||||||||
Zbt = | : | |||||||||||||||||||||||
Yt | ||||||||||||||||||||||||
Y |
5.2 Impulse Response Analysis
In the following, we derive impulse responses for two policy shocks: a standard monetary policy shock and a shock to commercial bank capital requirements. We analyze the rst shock to evaluate whether our model is able to replicate stylized facts from the large literature on monetary policy shocks, and to study potentially diering reactions of commercial bank and shadow bank intermediation. We then evaluate the impact of
34The ECB has stressed the importance to consider both commercial bank and overall credit in their scoreboard approach" for macroprudential regulation. See for instance Const^ancio et al. (2019) for a review of the ECB's approach towards macroprudential policy and the role of market-based nance in regulatory statutes.
ECB Working Paper Series No 2406 / May 2020 | 24 |
an unanticipated increase of capital requirements to shed light on credit leakage towards shadow bank intermediation in response to tighter commercial bank regulation. We nally discuss the potential of coordination between monetary and macroprudential policies to avoid potentially unintended side eects of tighter commercial bank regulation, i.e. credit leakage towards the shadow banking sector.
5.2.1 Monetary Policy Shock
Figure 3: Impulse Responses: Monetary Policy Shock { Moderate Regulator
0.02 | Lending | 0 |
CB Lending | SB Lending |
Dev. | 0 | Dev. | ||||||
-0.02 | -0.1 | |||||||
% | -0.04 | % | ||||||
-0.06 | -0.2 | |||||||
5 | 10 | 15 | 20 | |||||
Quarters | ||||||||
0 | Output | 0 | ||||||
% Dev. | -0.02 | Abs. Dev. | ||||||
-0.04 | -0.02 | |||||||
-0.06 | -0.04 | |||||||
-0.08 | ||||||||
5 | 10 | 15 | 20 | |||||
Quarters | ||||||||
0.2 | Investment | 0.2 | ||||||
% Dev. | 0 | Abs. Dev. | 0.1 | |||||
-0.2 | 0 | |||||||
-0.4 | 5 | 10 | 15 | 20 | -0.1 | |||
Quarters | ||||||||
CB Spread | 0 | |||||||
Dev. | 0.05 | Dev. | ||||||
-0.1 | ||||||||
Abs. | 0 | Abs. | -0.2 | |||||
-0.05 | 5 | 10 | 15 | 20 | -0.3 | |||
Quarters |
Dev. | 0.4 | |||||||
0.2 | ||||||||
% | ||||||||
0 | ||||||||
5 | 10 | 15 | 20 | |||||
Quarters | ||||||||
Inflation | 0 | |||||||
Dev. | -0.02 | |||||||
-0.04 | ||||||||
% | ||||||||
-0.06 | ||||||||
5 | 10 | 15 | 20 | |||||
Quarters | ||||||||
Policy Rate | 0 | |||||||
Dev. | -1 | |||||||
% | ||||||||
5 | 10 | 15 | 20 | -2 | ||||
Quarters | ||||||||
SB Spread | 0.0004 | |||||||
Dev. | 0.0003 | |||||||
Abs. | 0.0002 | |||||||
5 | 10 | 15 | 20 | 0.0001 | ||||
Quarters |
5 | 10 | 15 | 20 |
Quarters | |||
Consumption |
5 | 10 | 15 | 20 |
Quarters | |||
Capital Requirements |
5 | 10 | 15 | 20 |
Quarters | |||
Deposit Spread |
5 | 10 | 15 | 20 |
Quarters |
Credit Growth Credit/GDP Growth Credit Level Credit/GDP Level No Regulator
Note: Impulse responses to a one-standard-deviation monetary policy shock. Rates in absolute deviations from steady state, all other variables as percentage deviations from steady state.
ECB Working Paper Series No 2406 / May 2020 | 25 |
Figure 4: Impulse Responses: Monetary Policy Shock { Prudent Regulator
0.02 | Lending | 0 | ||||||
Dev. | 0 | Dev. | ||||||
-0.02 | -0.1 | |||||||
% | -0.04 | % | ||||||
-0.06 | -0.2 | |||||||
5 | 10 | 15 | 20 | |||||
Quarters | ||||||||
0 | Output | 0 | ||||||
% Dev. | -0.02 | Abs. Dev. | ||||||
-0.04 | -0.02 | |||||||
-0.06 | -0.04 | |||||||
-0.08 | ||||||||
5 | 10 | 15 | 20 | |||||
Quarters | ||||||||
0.2 | Investment | 0.2 | ||||||
% Dev. | 0 | Abs. Dev. | 0.1 | |||||
-0.2 | 0 | |||||||
-0.4 | 5 | 10 | 15 | 20 | -0.1 | |||
Quarters | ||||||||
CB Spread | 0 | |||||||
Dev. | 0.05 | Dev. | ||||||
-0.1 | ||||||||
Abs. | 0 | Abs. | -0.2 | |||||
-0.05 | 5 | 10 | 15 | 20 | -0.3 | |||
Quarters |
CB Lending | SB Lending | |||||||||||
Dev. | 0.4 | |||||||||||
0.2 | ||||||||||||
% | ||||||||||||
0 | ||||||||||||
5 | 10 | 15 | 20 | 5 | 10 | 15 | 20 | |||||
Quarters | Quarters | |||||||||||
Inflation | 0 | Consumption | ||||||||||
Dev. | -0.02 | |||||||||||
-0.04 | ||||||||||||
% | ||||||||||||
-0.06 | ||||||||||||
5 | 10 | 15 | 20 | 5 | 10 | 15 | 20 | |||||
Quarters | Quarters | |||||||||||
Policy Rate | 1 Capital Requirements | |||||||||||
% Dev. | 0 | |||||||||||
-1 | ||||||||||||
5 | 10 | 15 | 20 | 5 | 10 | 15 | 20 | |||||
Quarters | Quarters | |||||||||||
SB Spread | 0.0004 | Deposit Spread | ||||||||||
Dev. | 0.0003 | |||||||||||
Abs. | 0.0002 | |||||||||||
5 | 10 | 15 | 20 | 0.0001 | 5 | 10 | 15 | 20 | ||||
Quarters | Quarters |
Credit Growth Credit/GDP Growth Credit Level Credit/GDP Level No Regulator
Note: Impulse responses to a one-standard-deviation monetary policy shock. Rates in absolute deviations from steady state, all other variables as percentage deviations from steady state.
Several empirical studies have identied dierent reactions in credit intermediated within and outside the regular banking system in response to monetary policy shocks. Igan et al. (2017) nd that some institutions (money market mutual funds, security broker- dealers) increase their asset holdings after monetary policy tightening, whereas issuers of asset-backed securities (ABS) decrease their balance sheets. Pescatori and Sole (2016) use a VAR framework including data on commercial banks, ABS issuers, and other nance companies as well as government-sponsored entities (GSEs). They nd, inter alia, that
ECB Working Paper Series No 2406 / May 2020 | 26 |
monetary policy tightening decreases aggregate lending activity, even though the size of the non-bank intermediary sector increases. Similarly,Den Haan and Sterk (2011), using US ow-of-funds data, nd that non-bank asset holdings increase in response to monetary tightening, even though overall credit declines or stays relatively at. Mazelis (2016) distinguishes between commercial banks depending on deposit liabilities, highly levered shadow banks which depend on funding from other intermediaries, and investment funds that draw funding from real economic agents directly. He nds that, whereas commercial bank credit remains relatively at after monetary tightening and is reduced only in the medium term, shadow banks and investment funds increase lending in response to monetary policy tightening in the short term. Nelson et al. (2018) nd similar results when looking at aggregate balance sheets, even though their de nition of shadow banks di ers from that in Mazelis (2016). For European banks, Altunbas et al. (2009) show that institutions engaged to a large extent in non-bank activities, such as securitization, are less a ected by monetary policy shocks.
We report the reaction of model variables to an unanticipated increase in the policy rate by 12.5 basis points in Figures 3 and 4. In line with standard ndings, output and its subcomponents { consumption and investment { decline in a hump-shaped manner and ination falls in response to tighter monetary policy. Total lending is reduced as credit costs increase due to higher interest rates, while aggregate demand deteriorates. Bank intermediation spreads increase, as monopolistic competitive banks raise lending rates to generate pro ts, which compensates for the decline in lending volume.35 However, higher commercial bank loan rates increase lending costs for borrowing rms. The collateral constraint 5 indicates that the amount of borrowing rms can obtain from commercial banks is limited by the LTV ratio mCt and the value of non-depreciated physical capital qtk+1(1 k)ktE(i). Due to this borrowing limitation, an increase in the borrowing cost rms face when acquiring loans from commercial banks 1 + rtbc causes a decline in the quantity of commercial bank loans bE;Ct (i).
On the margin, borrowers will nd it pro table to switch to alternative sources of funding. Shadow banks face higher re nancing costs due to an increasing risk premium, reected in the widening of the deposit rate spread. They are extending loan supply and accept a decline in the intermediation spread on impact to generate pro ts. Thus, the decline in commercial bank lending is partly counteracted by an increased intermediation and leverage of shadow banks. Whereas commercial bank credit falls by approximately
0.18 percent in response to higher interest rates in all scenarios, shadow bank credit increases by approximately 0.35 percent. We therefore con rm empirical evidence on the presence of credit leakage towards shadow banks in response to tighter monetary policy
35This nding is consistent with studies relying on a homogeneous description of the nancial sector, see for instance Gerali et al. (2010).
ECB Working Paper Series No 2406 / May 2020 | 27 |
in our model.
For the moderate regulator, the decline in credit and real activity is larger for rules based on growth rates (solid blue and dashed red lines in Figures 3 and 4) than for level rules (dashed green and gray lines). Policy makers relying on level rules lower capital requirements more aggressively in response to a monetary tightening, and thus the overall decline in lending is mitigated. For all types of rules, a moderate policymaker only concerned with developments in the commercial banking sector lowers capital requirements to a larger extent than the prudent counterpart when confronted with higher interest rates. Quantitatively, a moderate regulator cuts capital requirements by half a percent on im- pact, which depicts a reduction of roughly ve basis points compared to the steady-state level of capital requirements. While the requirements are quickly readjusted back to the steady-state level under growth-based rules, capital requirements are eased by up to 1.8 percent, or 19 basis points under the credit level-based rule.
The easing of capital regulation is significantly less pronounced under the prudent regulator (Figure 3). Strikingly, a regulator considering both commercial bank and shadow bank credit would actually increase capital requirements in our stylized simulation exercise whenever the rule is based on the credit-to-GDP (growth) gap. The significant increase in shadow bank credit, and a relatively strong decline in GDP { the denominator of the credit-to-GDP ratio { in response to higher interest rates is sufcient to trigger a slight increase in capital requirements. In turn, the reduction in overall credit is even more pronounced in the prudent regulation case with credit-to-GDP as in the moderate regulator case, and compared to the benchmark situation of no countercyclical policy maker. Also, the reduction in aggregate demand and output is slightly more pronounced in the case of a prudent macroprudential regulator.
We take this nding as indication that a dierent treatment of shadow banks in policy considerations can lead to dierent policy prescriptions in response to macroeconomic shocks, with respective consequences for macroeconomic developments and nancial sta- bility. Ultimately, the response by regulators depends on the primary objective of macro- prudential policy. As we do not explicitly take the eect of shadow banking on nancial stability into account, the results here indicate that a regulator concerned with excessive lending by unregulated shadow banks { and a potentially resulting increase in nancial instability { would prescribe a dierent policy for commercial banks when developments in the shadow banking sector are taken into considerations.36
ECB Working Paper Series No 2406 / May 2020 | 28 |
Figure 5: Impulse Responses: Capital Requirement Shock { Moderate Regulator
0.02 | Lending | 0 | |||||||
% Dev. | 0 | % Dev. | -0.05 | ||||||
-0.02 | |||||||||
-0.04 | -0.1 | ||||||||
-0.06 | -0.15 | ||||||||
5 | 10 | 15 | 20 | ||||||
Quarters | |||||||||
0 | Output | ||||||||
% Dev. | -0.01 | Abs. Dev. | 0 | ||||||
-0.02 | -0.002 | ||||||||
-0.03 | -0.004 | ||||||||
5 | 10 | 15 | 20 | ||||||
Quarters | |||||||||
0.2 | Investment | 0.005 | |||||||
% Dev. | 0 | Abs. Dev. | 0 | ||||||
-0.2 | |||||||||
5 | 10 | 15 | 20 | -0.005 | |||||
Quarters | |||||||||
0.1 | CB Spread | 0 | |||||||
Dev. | 0.05 | Dev. | -0.05 | ||||||
Abs. | Abs. | ||||||||
-0.1 | |||||||||
0 | -0.15 | ||||||||
5 | 10 | 15 | 20 | ||||||
Quarters |
CB Lending | 0.2 | SB Lending | ||||||||
Dev.% | ||||||||||
0.1 | ||||||||||
0 | ||||||||||
5 | 10 | 15 | 20 | 5 | 10 | 15 | 20 | |||
Quarters | Quarters | |||||||||
Inflation | 0 | Consumption | ||||||||
% Dev. | -0.005 | |||||||||
-0.01 | ||||||||||
5 | 10 | 15 | 20 | -0.015 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters | |||||||||
Policy Rate | Capital Requirements |
10 | ||||||||||
Dev. | 5 | |||||||||
% | ||||||||||
5 | 10 | 15 | 20 | 0 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters | |||||||||
SB Spread | 0.00002 | Deposit Spread | ||||||||
Dev. | 0 | |||||||||
Abs. | ||||||||||
5 | 10 | 15 | 20 | -0.00002 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters |
Credit Growth Credit/GDP Growth Credit Level Credit/GDP Level No Regulator
Note: Impulse responses to a one-standard-deviation capital requirement shock. Rates in absolute deviations from steady state, all other variables as percentage deviations from steady state.
5.2.2 Capital Requirement Shock
In the previous section we verify that our model can generate dynamics of commercial and shadow bank credit in response to tighter monetary policy which are qualitatively in line with empirical evidence. Similar evidence on the eects of regulatory changes on credit
36As we do not derive welfare implications of shadow banking here, no discussion about the optimality or desirability of dierent described policy responses can be drawn from the presented results. Gebauer (2020) introduces such a welfare analysis in a simplied version of our model.
ECB Working Paper Series No 2406 / May 2020 | 29 |
Figure 6: Impulse Responses: Capital Requirement Shock { Prudent Regulator
0.02 | Lending | 0 | |||||||
% Dev. | 0 | % Dev. | -0.05 | ||||||
-0.02 | |||||||||
-0.04 | -0.1 | ||||||||
-0.06 | -0.15 | ||||||||
5 | 10 | 15 | 20 | ||||||
Quarters | |||||||||
0 | Output | ||||||||
% Dev. | -0.01 | Abs. Dev. | 0 | ||||||
-0.02 | -0.002 | ||||||||
-0.03 | -0.004 | ||||||||
5 | 10 | 15 | 20 | ||||||
Quarters | |||||||||
0.2 | Investment | 0.005 | |||||||
% Dev. | 0 | Abs. Dev. | 0 | ||||||
-0.2 | |||||||||
5 | 10 | 15 | 20 | -0.005 | |||||
Quarters | |||||||||
0.1 | CB Spread | 0 | |||||||
Dev. | 0.05 | Dev. | -0.05 | ||||||
Abs. | Abs. | ||||||||
-0.1 | |||||||||
0 | -0.15 | ||||||||
5 | 10 | 15 | 20 | ||||||
Quarters |
CB Lending | 0.2 | SB Lending | ||||||||
Dev.% | ||||||||||
0.1 | ||||||||||
0 | ||||||||||
5 | 10 | 15 | 20 | 5 | 10 | 15 | 20 | |||
Quarters | Quarters | |||||||||
Inflation | 0 | Consumption | ||||||||
% Dev. | -0.005 | |||||||||
-0.01 | ||||||||||
5 | 10 | 15 | 20 | -0.015 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters | |||||||||
Policy Rate | Capital Requirements |
10 | ||||||||||
Dev. | 5 | |||||||||
% | ||||||||||
5 | 10 | 15 | 20 | 0 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters | |||||||||
SB Spread | 0.00002 | Deposit Spread | ||||||||
Dev. | 0 | |||||||||
Abs. | ||||||||||
5 | 10 | 15 | 20 | -0.00002 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters |
Credit Growth Credit/GDP Growth Credit Level Credit/GDP Level No Regulator
Note: Impulse responses to a one-standard-deviation capital requirement shock. Rates in absolute deviations from steady state, all other variables as percentage deviations from steady state.
intermediation is still relatively scarce, primarily due to issues of identifying unanticipated shocks to capital regulation on the macro level.37 Irani et al. (2018) use detailed US corpo-
37Compared to well-established procedures to identify monetary policy shocks, the empirical identication of macroprudential policy shocks is less straightforward. First, policy decisions are taken in a process that has only started to mature since the nancial crisis, while monetary policy has a long history of regular meetings of the monetary policy committees that announce their decisions in a public manner, at least in many developing countries over several decades. Second, as many of the macroprudential tools discussed now were only implemented over the last ten years, time series for respective measures are still short.
ECB Working Paper Series No 2406 / May 2020 | 30 |
rate loan data to evaluate the e ect of capital requirement changes on the development of non-bank nancial intermediation. Relying on data derived from a supervisory register on syndicated loans, they nd that shadow bank credit increases in response to commercial bank capital constraints. Similarly, Buchak et al. (2018) examine data on Fintech lenders in residential mortgage markets. They nd that commercial banking contracted due to a higher regulatory burden { such as higher bank capital requirements as well as mortgage market-related regulatory changes { and was partly replaced by unregulated shadow bank intermediation. To evaluate the e ect of credit leakage towards shadow banks in response to tighter regulation, we simulate an unanticipated increase in commercial bank capital requirements by one percentage point (resembling a positive ten-percent deviation from steady state) and provide impulse responses in Figures 5 and 6.
Whereas overall lending is reduced by increased bank capital requirements, lower credit intermediation by commercial banks is partly o set by increased shadow bank activity in all scenarios. Due to the leakage mechanism laid out in detail in the Online Appendix, higher capital requirements result in a deviation of actual capital-to-asset ratios held by commercial banks from the regulatory requirement and increase the intermediation cost for commercial banks (Equation 10). In response, both the wholesale and retail loan rates rtC and rtbc increase. As in the case of a monetary policy tightening, due to the collateral constraint 5, the quantity of loans bE;Ct (i) declines. Monopolistically competitive banks raise loan rates and generate pro ts via retained earnings, and borrowing from shadow banks becomes relatively more attractive. The initially una ected demand for credit by entrepreneurs eases the shadow bank leverage constraint (Equation 13). In response, the shadow bank lending rate spread declines and shadow bank intermediation and leverage increases.38 Thus, in response to a macroprudential tightening, borrowers increase the share of relatively costly shadow bank loans which raises the overall cost of borrowing. Finally, tighter macroprudential regulation reduces overall lending activity and ultimately, due to lower credit supply, dampens economic activity. Lower aggregate demand reduces ination, and monetary policy consequently responds by lowering interest rates.
The di erent degrees to which macroprudential policymakers take shadow banks into consideration has implications for the development of both credit and macroeconomic variables in the model. Following an unanticipated rise in capital requirements, regulators following level-based rules pursue a path of relative rapid policy normalization compared to the case of growth-based rules, both under moderate and prudent regulation. In return, even though the drop in commercial bank lending is equally pronounced on impact, credit returns to its steady-state level more quickly under level rules. Therefore, the reduction in overall credit is relatively smaller in the case of the level-based regulators, and the de-
38A similar rationale for credit leakage eects in response to tighter regulation is provided in Aikman et al. (2018).
ECB Working Paper Series No 2406 / May 2020 | 31 |
scribed losses in aggregate demand are weaker. Ination is reduced to a lesser extent and monetary policy reacts less aggressively in the case of level-based rules.
5.2.3 Policy Coordination
In the two preceding policy exercises, unexpected tightening by one policymaker triggered counteractive measures implemented by the other to mitigate adverse e ects on price stability and output. Furthermore, tighter regulation and monetary policy caused leakage towards the shadow banking sector. Both observations indicate a potential role for coordination among policymakers to mitigate dampening macroeconomic implications of tighter regulation and, in particular, to limit the unintended leakage of credit towards the shadow banking sector.
In the following exercise, we evaluate to what extent a coordinated reaction using monetary policy could limit the increase in shadow bank lending in response to tighter bank capital regulation in our model setup. Higher capital requirements indeed reduce lending activity by commercial banks, as shown in the analysis of the previous section. However, the contemporaneous increase in shadow bank lending depicts a limitation of macropru- dential policy eciency. First, the intended reduction in lending activity is partly counteracted by an increase in shadow bank intermediation, resulting in a smaller reduction in overall lending compared to a situation without shadow bank intermediation. Second, an increase in shadow bank lending potentially increases nancial instability as a relatively larger share of intermediation is now conducted by unregulated nancial institutions.
To discuss bene ts from policy coordination in light of the limits on macroprudential regulation emerging from the existence of shadow banks, we evaluate whether monetary policy can be employed to avoid leakage of credit intermediation towards shadow banks. Whereas capital requirements only a ect commercial banks, interest rates depict a universal tool that reaches through all the cracks in the economy" (Stein, 2013). To this end, we apply the monetary policy reaction that is necessary to keep shadow bank intermediation at its steady-state level in response to the capital requirement shock discussed in the previous Section.
Figures 7 and 8 depict impulse responses to an unanticipated increase in commercial bank capital requirements by one percentage point in combination with a contemporaneous response by monetary policy that mitigates the reaction of shadow bank intermediation. In the simulations, the reaction in shadow bank lending is negligible as the central bank lowers interest rates by approximately 5 6 basis points in response to tighter capital requirements. As indicated in Section 5.2.1, commercial bank credit increases in reaction to monetary policy easing, and therefore the decrease in commercial bank lending is less pronounced in Figures 7 and 8 compared to the respective reductions in Figures 5 and 6.
ECB Working Paper Series No 2406 / May 2020 | 32 |
Combination
Figure 7: Impulse Responses: Policy Coordination { Moderate Regulator
0 | Lending | 0 | ||||
Dev. | -0.02 | Dev. | -0.02 | |||
-0.04 | ||||||
% | % | |||||
-0.04 | -0.06 | |||||
5 | 10 | 15 | 20 | |||
Quarters | ||||||
Output | 0.02 | |||||
0.02 | Abs. Dev. | 0.015 | ||||
% Dev. | ||||||
0.01 | 0.01 | |||||
0.005 | ||||||
0 | 5 | 10 | 15 | 20 | 0 | |
Quarters | ||||||
Investment | 0.05 | |||||
0 | ||||||
% Dev. | Abs. Dev. | 0 | ||||
-0.05 | -0.05 | |||||
-0.1 | 5 | 10 | 15 | 20 | -0.1 | |
Quarters | ||||||
0.04 | CB Spread | |||||
0 | ||||||
Dev. | 0.03 | Dev. | ||||
0.02 | -0.01 | |||||
Abs. | Abs. | |||||
0.01 | -0.02 | |||||
0 | 5 | 10 | 15 | 20 | -0.03 | |
Quarters |
CB Lending
0.15 | |||||
Dev. | 0.1 | ||||
0.05 | |||||
% | |||||
0 | |||||
5 | 10 | 15 | 20 | -0.05 | |
Quarters | |||||
Inflation | |||||
0.02 | |||||
Dev. | 0.01 | ||||
% | |||||
5 | 10 | 15 | 20 | 0 | |
Quarters | |||||
Policy Rate |
10 | |||||
Dev. | 5 | ||||
% | |||||
5 | 10 | 15 | 20 | 0 | |
Quarters | |||||
SB Spread | 0 | ||||
Dev. | -0.00005 | ||||
-0.0001 | |||||
Abs. | |||||
-0.00015 | |||||
5 | 10 | 15 | 20 | -0.0002 | |
Quarters
SB Lending
5 | 10 | 15 | 20 |
Quarters | |||
Consumption |
5 | 10 | 15 | 20 |
Quarters | |||
Capital Requirements |
5 | 10 | 15 | 20 |
Quarters | |||
Deposit Spread |
5 10 15 20
Quarters
Credit Growth Credit/GDP Growth Credit Level Credit/GDP Level No Regulator
Note: Impulse responses to a combination of capital requirement and monetary policy shocks. Rates in absolute deviations from steady state, all other variables as percentage deviations from steady state.
Therefore, even though monetary policy easing partly counteracts the intended reduction in overall lending stemming from an increase in capital requirements, it can help to mitigate potentially undesired leakage towards shadow banks as a side eect of tighter commercial bank regulation.
ECB Working Paper Series No 2406 / May 2020 | 33 |
Combination
Figure 8: Impulse Responses: Policy Coordination { Prudent Regulator
0 | Lending | 0 | ||||
Dev. | -0.02 | Dev. | -0.02 | |||
-0.04 | ||||||
% | % | |||||
-0.04 | -0.06 | |||||
5 | 10 | 15 | 20 | |||
Quarters | ||||||
Output | 0.02 | |||||
% Dev. | 0.02 | Abs. Dev. | ||||
0.01 | 0.01 | |||||
0 | 5 | 10 | 15 | 20 | 0 | |
Quarters | ||||||
Investment | 0.05 | |||||
0 | ||||||
% Dev. | Abs. Dev. | 0 | ||||
-0.05 | -0.05 | |||||
-0.1 | 5 | 10 | 15 | 20 | -0.1 | |
Quarters | ||||||
0.04 | CB Spread | |||||
0 | ||||||
Dev. | Dev. | |||||
0.02 | -0.01 | |||||
Abs. | Abs. | |||||
-0.02 | ||||||
0 | 5 | 10 | 15 | 20 | -0.03 | |
Quarters |
CB Lending | 0.2 | SB Lending | ||||||||
Dev.% | ||||||||||
0.1 | ||||||||||
0 | ||||||||||
5 | 10 | 15 | 20 | 5 | 10 | 15 | 20 | |||
Quarters | Quarters | |||||||||
Inflation | Consumption | |||||||||
Dev. | 0.02 | |||||||||
0.01 | ||||||||||
% | ||||||||||
5 | 10 | 15 | 20 | 0 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters | |||||||||
Policy Rate | Capital Requirements |
10 | ||||||||||
Dev. | 5 | |||||||||
% | ||||||||||
5 | 10 | 15 | 20 | 0 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters | |||||||||
SB Spread | 0 | Deposit Spread | ||||||||
Dev. | -0.0001 | |||||||||
Abs. | ||||||||||
5 | 10 | 15 | 20 | -0.0002 | 5 | 10 | 15 | 20 | ||
Quarters | Quarters |
Credit Growth Credit/GDP Growth Credit Level Credit/GDP Level No Regulator
Note: Impulse responses to a combination of capital requirement and monetary policy shocks. Rates in absolute deviations from steady state, all other variables as percentage deviations from steady state.
5.3 Counterfactual Simulation
Finally, we evaluate how euro area regulators would have set capital requirements under Basel III, if the framework would have been in place already in 1999 and throughout the existence of the common currency. For all regulatory regimes, we use the estimated baseline model (Section 4) to lter the data and simulate the evolution of endogenous model variables over the period 1999 { 2014 in a counterfactual analysis. In the counterfactual
ECB Working Paper Series No 2406 / May 2020 | 34 |
simulation we allow for endogenous feedback between capital requirements and macroeconomic and nancial variables. We focus on hypothetical capital requirements that the regulator would have set in response to macroeconomic and nancial shocks under the growth-based rules of Equations 18{19 and 22{23, reported in Figure 9.39
Figure 9: Counterfactual Analysis: Dierent Regulatory Regimes
Capital Requirement
16 | ||||
14 | ||||
Percent | 12 | |||
10 | ||||
8 | ||||
6 | ||||
1999 | 2004 | 2009 | 2014 |
Moderate Credit Level Moderate Credit/GDP Level Prudent Credit Level Prudent Credit/GDP Level No Regulator
Note: Simulated path of capital requirements based on shock series identied in estimation of Section 4
and for dierent regulators of Section 5.1.
Independent of the rule type, both the moderate and the prudent regulator would have applied some form of countercyclical regulation by reducing capital requirements in times of nancial distress and by raising requirement in times of excessive lending. All rules would have prescribed a sharp tightening of credit standards from the mid-2000s onward, in response to massive credit growth in the European nancial sector. Over the course of the global nancial crisis starting in 2008 and the European debt crisis, both regulators would have prescribed a reduction in capital requirements due to subdued lending activity in the euro area.
Thereby, the moderate regulators only concerned with commercial bank credit would have set lower capital requirements relative to their prudent counterparts throughout the sample period. Also, they would have raised capital requirements more aggressively prior to the 2008 nancial crisis, as commercial bank credit grew rapidly in these years (Figure 1). Prudent regulators on the other hand would have eased requirements later and less strongly in response to the nancial crisis: In the years 2008-2009, shadow bank credit continued to grow further while the increase in commercial bank lending stalled. If the leakage channel would have been taken into account by the regulator, the migration of credit to the shadow banking system at a time when nancial stability considerations take
39We focus on the growth-based rules to avoid taking a stance on the steady-state level of credit, which would be required to construct the level-based rules, Equations 20{21 and 24{25.
ECB Working Paper Series No 2406 / May 2020 | 35 |
center stage may have led to yet different dynamics.40
These ndings indicate that the existence of shadow banking and the resulting credit leakage requires a detailed understanding of the exact transmission mechanism of nan- cial regulation. Considering non-bank nancial intermediation in regulation for commercial banks as depicted by the prudent regulator can, on the one hand, result in lower capital requirements and a resulting lower leakage of credit to unregulated intermediaries. On the other hand, giving a stronger weight to the developments in overall credit relative to the leakage motive, can, as in our simulation, lead to higher requirements for commercial banks. Furthermore, as shown in Section 5.2.3, monetary policy can play an active role in mitigating shadow bank intermediation. Thus, it can play a crucial part whenever leakage concerns limit the scope for tighter commercial banking regulation.
Table 3: Di erences in Variation Under Di erent Policy Regimes
Moderate Regulator | Prudent Regulator | ||||
Credit | Credit/GDP | Credit | Credit/GDP | ||
GDP | -5.85 | -5.16 | -5.74 | -4.72 | |
Consumption | -3.20 | -2.46 | -2.81 | -1.70 | |
Investment | -25.14 | -23.77 | -25.77 | -23.70 | |
Ination | -1.98 | -2.00 | -2.20 | -2.24 | |
Policy Rate | -2.12 | -2.19 | -2.35 | -2.47 | |
Total Lending | -4.12 | -4.23 | -4.56 | -4.74 | |
CB Lending | -5.43 | -5.35 | -5.67 | -5.56 | |
SB Lending | -5.10 | -4.39 | -4.37 | -3.29 |
Note: Percentage dierence in the variance of macroeconomic and nancial variables, compared to the variation under the baseline scenario without cycli-
cal regulation. For variable X, the percentage dierence X is dened as X = [V ar(XReg) V ar(XNoReg)]=V ar(XNoReg).
In Table 3, we report the percentage di erence in the variance of simulated variables when considering macroprudential rules compared to the baseline scenario without cyclical capital requirements. We nd that the growth-based rules in Figure 9 would have been e ective at reducing macroeconomic volatility. Moderate regulators are particularly successful in reducing volatility in real macroeconomic variables, as the variance reduction in GDP and consumption is largest under this set of regulators. Investment volatility also declines slightly more under moderate regulation if growth in the credit-to-GDP gap is used as the target variable. In turn, prudent regulators are better equipped at stabilizing nominal variables and overall lending.
40Implications of the leakage channel for optimal capital regulation are further explored in Gebauer (2020). He shows that under optimal policy, credit leakage provides a motive for mitigating the regulatory response to commercial bank credit whenever commercial and shadow bank credit move in the same direction in response to macroeconomic disturbances.
ECB Working Paper Series No 2406 / May 2020 | 36 |
- Conclusion
We develop a DSGE model featuring two dierent types of nancial intermediaries: regulated commercial banks and unregulated shadow banks. Methodologically, we combine two seminal strands of the literature for modeling nancial frictions that were independently developed in recent years. In doing so, we exploit dierences with respect to market power and regulatory coverage in the two frameworks and argue that they can be applied to structurally dierent nancial institutions.
We highlight the key mechanism of bank capital requirements and evaluate how tighter regulation of commercial bank credit intermediation can result in higher intermediation activity by unregulated shadow banks. We estimate the structural parameters of the model via Bayesian methods using euro area data on both commercial and shadow banks.
We use our estimated model to evaluate quantitative responses of macroeconomic variables to unexpected changes in macroprudential and monetary policy. We nd that macroprudential tightening leads to a reduction in commercial bank credit, but increases intermediation by shadow banks. If a macroprudential rule is employed, this leakage mechanism can be reduced, but not eliminated.
Whereas capital requirements can only be employed with respect to commercial banks, interest rates depict a universal tool to reach though all the cracks in the economy" (Stein, 2013). We evaluate whether monetary policy can be employed to counteract the leakage mechanism in a coordinated macroprudential and monetary policy interaction scenario. Even though monetary easing partly counteracts the intended reduction in overall lending stemming from an increase in capital requirements, it can help to mitigate potentially undesired leakage towards shadow banks as a side eect of tighter bank regulation.
We furthermore evaluate in a counterfactual analysis how regulation would have been set had it followed Basel III rules, and how this would have aected macro indicators through the global nancial crisis and the sovereign debt crisis. We nd macroprudential tightening during episodes of credit increases, and easing during credit crunches. Further- more, regulators only concerned with commercial bank credit would have raised capital requirements more strongly in the years preceding the global nancial crisis, when growth in commercial bank lending was particularly pronounced. However, more 'prudent' regulators taking both commercial and shadow bank credit into consideration would have generally applied higher levels of capital requirements. Also, these prudent regulators would have eased requirements only later and less strongly after the outbreak of the - nancial crisis, as shadow bank credit continued to grow thereafter.
We therefore highlight the need for understanding credit leakage and the emerging trade-o for regulators taking lending by both regulated and unregulated intermediaries into account when lending by both intermediaries increases. On the one hand, an increase
ECB Working Paper Series No 2406 / May 2020 | 37 |
in overall credit might indicate the need for tighter regulation. On the other hand, tighter regulation on the regulated entity only fuels credit leakage to the unregulated entity, with potential implications for nancial stability. This paper develops a framework for nancial regulators to think of such trade-os, and take them into account when making macroprudential decisions, potentially by including a role for policy coordination with the monetary authority, which requires further investigation.
ECB Working Paper Series No 2406 / May 2020 | 38 |
References
Adrian, T. (2014). Financial Stability Policies for Shadow Banking. Federal Reserve Bank of New York Sta Report, (664).
Adrian, T. and Liang, N. (2014). Monetary Policy, Financial Conditions, and Financial Stability. Federal Reserve Bank of New York Sta Report, (690).
Aikman, D., Giese, J., Kapadia, S., and McLeay, M. (2018). Targeting Financial Stability: Macroprudential or Monetary Policy? Bank of England Sta Working Paper, (734).
Altunbas, Y., Gambacorta, L., and Marques-Ibanez, D. (2009). Securitisation and the bank lending channel. European Economic Review, 53(8):996{1009.
Angelini, P., Neri, S., and Panetta, F. (2014). The Interaction between Capital Requirements and Monetary Policy. Journal of Money, Credit and Banking, 46(6):1073{1112.
Angeloni, I. and Faia, E. (2013). Capital Regulation and Monetary Policy with Fragile Banks. Journal of Monetary Economics, 60(2013):311{324.
Bakk-Simon, K., Borgioli, S., Giron, C., Hempell, H., Maddaloni, A., Recine, F., and Rosati, S. (2012). Shadow Banking in the Euro Area - An Overview. ECB Occasional Paper Series, (133).
Bean, C., Paustian, M., Penalver, A., and Taylor, T. (2010). Monetary Policy After the Fall. Bank of England Working Paper.
Beau, D., Clerc, L., and Mojon, B. (2012). Macro-Prudential Policy and the Conduct of Monetary Policy. Banque de France Working Paper, (390).
Begenau, J. and Landvoigt, T. (2016). Financial Regulation in a Quantitative Model of the Modern Banking System. Working Paper.
Bekiros, S., Nilavongse, R., and Uddin, G. S. (2018). Bank Capital Shocks and Coun- tercyclical Requirements: Implications for Banking Stability and Welfare. Journal of Economic Dynamics and Control, 93:315{331.
Borio, C. (2009). Implementing the Macroprudential Approach to Financial Regulation and Supervision. Financial Stability Review, 13:31{41.
Borio, C. (2011). Rediscovering the Macroeconomic Roots of Financial Stability Policy: Journey, Challenges, and a Way Forward. Annual Review of Financial Economics, 3(1):87{117.
Borio, C. and Shim, I. (2007). What Can (Macro)-Prudential Policy Do to Support Monetary Policy? BIS Working Papers, 242.
ECB Working Paper Series No 2406 / May 2020 | 39 |
Brooks, S. P. and Gelman, A. (1998). General Methods for Monitoring Convergence of Iterative Simulations. Journal of Computational and Graphical Statistics, 7(4):434{455.
Buchak, G., Matvos, G., Piskorski, T., and Seru, A. (2018). Fintech, Regulatory Arbitrage, and the Rise of Shadow Banks. Journal of Financial Economics, 130(3):453{483.
Calvo, G. A. (1983). Staggered Prices in a Utility-Maximizing Framework. Journal of Monetary Economics, 12(3):383{398.
Christensen, I., Meh, C., and Moran, K. (2011). Bank Leverage Regulation and Macroeconomic Dynamics. Bank of Canada Working Paper, (2011-32).
Cizel, J., Frost, J., Houben, A. G. F. J., and Wierts, P. (2019). Eective Macroprudential Policy; Cross-Sector Substitution from Price and Quantity Measures. Journal of Money, Credit and Banking, 51(5):1209{1235.
Coenen, G., Karadi, P., Schmidt, S., and Warne, A. (2019). The New Area-Wide Model
- An Extended Version of the ECBs Micro-Founded Model for Forecasting and Policy Analysis with a Financial Sector. ECB Working Paper Series, 2200.
Coeure, B. (2015). How Binding is the Zero Lower Bound? Speech at the conference "Removing the Zero Lower Bound on Interest Rates", London, 18 May 2015.
Collard, F., Dellas, H., Diba, B., and Loisel, O. (2017). Optimal Monetary and Prudential Policies. American Economic Journal: Macroeconomics, 9(1):40{87.
Const^ancio, V., Cabral, I., Detken, C., Fell, J., Henry, J., Hiebert, P., Kapadia, S., Al- timari, S. N., Pires, F., and Salleo, C. (2019). Macroprudential Policy at the ECB: Institutional Framework, Strategy, Analytical Tools and Policies. ECB Occasional Paper Series, (227).
Curdia, V. and Woodford, M. (2010a). Conventional and Unconventional Monetary Policy. Federal Reserve Bank of St. Louis Review, 92:229{264.
Curdia, V. and Woodford, M. (2010b). Credit Spreads and Monetary Policy. Journal of Money, Credit and Banking, 42(s1):3{35.
Curdia, V. and Woodford, M. (2011). The Central-Bank Balance Sheet as an Instrument of Monetary Policy. Journal of Monetary Economics, 58(1):54{79.
De Bandt, O. and Davis, E. P. (2000). Competition, Contestability and Market Structure in European Banking Sectors on the Eve of EMU. Journal of Banking and Finance, 24(6):1045{1066.
Den Haan, W. J. and Sterk, V. (2011). The Myth of Financial Innovation and the Great Moderation. The Economic Journal, 121(553):707{739.
ECB Working Paper Series No 2406 / May 2020 | 40 |
Doyle, N., Hermans, L., Molitor, P., and Weistro er, C. (2016). Shadow Banking in the Euro Area: Risks and Vulnerabilities in the Investment Fund Sector. ECB Occasional Paper Series, (174).
Farhi, E. and Tirole, J. (2017). Shadow Banking and the Four Pillars of Traditional Financial Intermediation. NBER Working Paper, (23930).
Feve, P. and Pierrard, O. (2017). Financial Regulation and Shadow Banking: A Small- Scale DSGE Perspective. TSE Working Paper, 17-829.
FSB (2011). Shadow Banking: Scoping the Issues. A Background Note of the Financial Stability Board, (12).
FSB (2017). Assessment of Shadow Banking Activities, Risks and the Adequacy of Post- Crisis Policy Tools to Address Financial Stability Concerns. Report to the G20, (July).
Fungacova, Z., Solanko, L., and Weill, L. (2014). Does Competition Inuence the Bank Lending Channel in the Euro Area? Journal of Banking and Finance, 49:356{366.
Gambacorta, L. and Signoretti, F. M. (2014). Should Monetary Policy Lean Against the Wind?: An Analysis Based on a DSGE Model With Banking. Journal of Economic Dynamics and Control, 43(June):146{174.
Gebauer, S. (2020). Welfare-Based Optimal Macroprudential Policy with Shadow Banks.
mimeo.
Gelain, P. and Ilbas, P. (2017). Monetary and Macroprudential Policies in an Estimated Model with Financial Intermediation. Journal of Economic Dynamics and Control, 78:164{189.
Gerali, A., Neri, S., Sessa, L., and Signoretti, F. M. (2010). Credit and Banking in a DSGE Model for the Euro Area. Journal of Money, Credit and Banking, 42(1):107{141.
Gertler, M. and Karadi, P. (2011). A Model of Unconventional Monetary Policy. Journal of Monetary Economics, 58(1):17{34.
Gertler, M. and Kiyotaki, N. (2011). Financial Intermediation and Credit Policy in Business Cycle Analysis. In Friedman, B. M. and Woodford, M., editors, Handbook of Monetary Economics. Elsevier.
Gertler, M. and Kiyotaki, N. (2015). Banking, Liquidity, and Bank Runs in an In nite Horizon Economy. American Economic Review, 105(7):2011{2043.
Gertler, M., Kiyotaki, N., and Prestipino, A. (2016). Wholesale Banking and Bank Runs in Macroeconomic Modelling of Financial Crises. In Taylor, J. B. and Uhlig, H., editors, Handbook of Macroeconomics. Elsevier.
ECB Working Paper Series No 2406 / May 2020 | 41 |
Harris, M., Opp, C. C., and Opp, M. M. (2014). Higher Capital Requirements, Safer Banks? - Macroprudential Regulation in a Competitive Financial System. Chicago Booth Working Paper.
Holstrom, B. and Tirole, J. (1997). Financial Intermediation, Loanable Funds, and the Real Sector. The Quarterly Journal of Economics, 112(3):663{691.
Iacoviello, M. (2005). House Prices, Borrowing Constraints, and Monetary Policy in the Business Cycle. American Economic Review, 95(3):739{764.
Iacoviello, M. and Guerrieri, L. (2017). Collateral Constraints and Macroeconomic Asym- metries. Journal of Monetary Economics, 90:28{49.
Igan, D., Kabundi, A., Nadal de Simone, F., and Tamirisa, N. (2017). Monetary Policy and Balance Sheets. Journal of Policy Modeling, 39(1):169{184.
Irani, R. M., Iyer, R., Meisenzahl, R. R., and Peydro, J.-L. (2018). The Rise of Shadow Banking: Evidence from Capital Regulation. Federal Reserve Board of Governors Finance and Economics Discussion Series, (2018-039).
Malatesta, F., Masciantonio, S., and Zaghini, A. (2016). The Shadow Banking System in the Euro Area: Denitions, Key Features and the Funding of Firms. Italian Economic Journal, pages 1{21.
Mazelis, F. (2016). Implications of Shadow Bank Regulation for Monetary Policy at the Zero Lower Bound. Working Paper.
Meeks, R. and Nelson, B. (2017). Shadow Banks and Macroeconomic Instability. Journal of Money, Credit and Banking, 49(1):1483{1516.
Nelson, B., Pinter, G., and Theodoridis, K. (2018). Do Contractionary Monetary Policy Shocks Expand Shadow Banking? Journal of Applied Econometrics, 33(2):198{211.
Ordonez, G. (2018). Sustainable Shadow Banking. American Economic Journal: Macroeconomics , 10(1):33{56.
Pescatori, A. and Sole, J. (2016). Credit, Securitization and Monetary Policy: Watch Out for Unintended Consequences. IMF Working Paper, (16/76).
Plantin, G. (2015). Shadow Banking and Bank Capital Regulation. Review of Financial Studies, 28(1):146{175.
Pozsar, Z., Adrian, T., Ashcraft, A., and Boesky, H. (2010). Shadow Banking. Federal Reserve Bank of New York Sta Report, (458).
ECB Working Paper Series No 2406 / May 2020 | 42 |
Ratto, M. and Iskrev, N. (2011). Identication Analysis of DSGE Models with DYNARE. MONFISPOL, 225149.
Rotemberg, J. J. (1982). Monopolistic Price Adjustment and Aggregate Output. Review of Economic Studies, 49(4):517{531.
Rubio, M. and Carrasco-Gallego, J. A. (2015). Macroprudential and Monetary Policy Rules: A Welfare Analysis. The Manchester School, 83(2):127{152.
Rubio, M. and Carrasco-Gallego, J. A. (2016). The New Financial Regulation in Basel
- and Monetary Policy: A Macroprudential Approach. Journal of Financial Stability, 26:294{305.
Silvo, A. (2019). The Interaction of Monetary and Macroprudential Policies. Journal of Money, Credit and Banking, 51(4):859{894.
Smets, F. and Wouters, R. (2003). An Estimated Dynamic Stochastic General Equilibrium Model of the Euro Area. Journal of the European Economic Association, 1(5):1123{1175.
Smets, F. and Wouters, R. (2007). Shocks and Frictions in US Business Cycles: A Bayesian DSGE Approach. American Economic Review, 97(3):586{606.
Stein, J. C. (2013). Overheating in Credit Markets: Origins, Measurement and Policy Responses. Remarks at a Research Symposium sponsored by the Federal Reserve Bank of St. Louis.
Verona, F., Martins, M. M., and Drumond, I. (2013). (Un)anticipated Monetary Policy in a DSGE Model with a Shadow Banking System. International Journal of Central Banking, 9(3):78{124.
ECB Working Paper Series No 2406 / May 2020 | 43 |
- Appendix: The Full Nonlinear DSGE Model
A.1 Households
The representative patient household i maximizes the expected utility
1 | "(1 aP )"tzlog(ctP (i) aP ctP1 ) | lP (i)1+P | # | ||
E0 = | Pt | t | (A.1) | ||
1 + P | |||||
Xt | |||||
=0 |
which depends on current individual consumption cPt (i) as well as lagged aggregate consumption cPt and working hours ltP . Labor dis-utility is parameterized by P . Preferences are subject to a disturbance a ecting consumption, "zt . Household choices are undertaken subject to the budget constraint:
cPt (i) + dP;Ct (i) + dP;St (i) wtltP (i) + (1 + rtdC1 )dP;Ct1 (i) + (1 + rtdS1 )dP;St1 (i) + tPt (i) (A.2)
The ow of expenses includes current consumption and real deposits to be made to both commercial and shadow banks, dP;Ct (i) and dP;St (i). Due to the di erence in the discount factor for households ( P ) and entrepreneurs ( E), households only place deposits, but do not borrow any funds from nancial market agents. Resources consist of wage earnings wtP ltP (i) (where wtP is the real wage rate for the labor input of each household), gross interest income on last period deposits (1 + rtdC1 )dP;Ct1 (i) and (1 + rtdS1 )dP;St1 (i), and lump- sum transfers tPt that include dividends from rms and banks (of which patient households are the ultimate owners). First-order conditions yield the consumption Euler equation and labor-supply condition:
ctP (i) | " ctP+1(i) | # | ||||
"z | t | 1 + rdC | ||||
t | = | Et | (A.3) | |||
P | ||||||
P | P | wt | ||||
lt | (i) | = | (A.4) | |||
cP (i) | ||||||
t |
A.2 Entrepreneurs
Entrepreneurs use labor provided by households as well as capital to produce intermediate goods that retailers purchase in a competitive market. Each entrepreneur i derives utility from consumption cEt (i), which it compares to the lagged aggregate consumption level of all entrepreneurs. He maximizes expected utility
ECB Working Paper Series No 2406 / May 2020 | 44 |
1 | |||
Xt | |||
E0 | Et log(ctE(i) aEctE1 | ) | (A.5) |
=0 |
by choosing consumption, the use of physical capital ktE, loans from both commercial and shadow banks (bE;Ct , bE;St ), and labor input from households. He faces the following budget constraint:
cE(i) + wtlP (i) + (1 + rb | )bE;C | (i) + (1 + rb | )bE;S | (i) + qkkE(i) | |||||||||||
t | t | t1 | t1 | t1 | t1 | t t | |||||||||
ytE(i) | E;C | E;S | k | k | E | ||||||||||
= | + bt | (i) + bt | (i) + qt | (1 | )kt1 | (i) (A.6) | |||||||||
xt | |||||||||||||||
with k depicting the depreciation rate of capital and qtk the market price for capital in terms of consumption. As we assume that intermediate goods are sold on a wholesale market at price Ptw and are transformed by retailers in a composite nal good whose price index is Pt, we dene xt as the price markup of the nal over the intermediate good. We thus express output ytE produced by the entrepreneur in terms of the relative
competitive price of the wholesale good, given by 1 and which is produced according to
xt
the Cobb-Douglas technology
ytE(i) = AtEktE1 (i) ltE(i)1 | (A.7) |
where the (stochastic) total factor productivity (TFP) is given by AEt .
Entrepreneurs face constraints on the amount they can borrow from commercial and shadow banks as discussed in Section 3:
(1 + rtbc)btE;C (i) mtC Et[qtk+1(1 | k)ktE(i)] | (A.8) |
(1 + rtbs)btE;S(i) (1 mtC )Et[qtk+1(1 | k)ktE(i)] | (A.9) |
where the LTV ratio for commercial banks mCt is set exogenously by the regulator and follows an exogenous AR(1) process.
A.3 Financial Intermediaries
In our model, we have two nancial market agents that intermediate funds between households and rms: commercial banks and shadow banks. While they both engage in intermediation in a similar fashion, we assume the two types of agents to be structurally dierent
ECB Working Paper Series No 2406 / May 2020 | 45 |
along various dimensions, as discussed in Section 3.
A.3.1 Commercial Banks
In the following, we discuss the maximization problem of the wholesale unit of the commercial bank as the capital requirement set by regulators applies directly to this branch of the commercial bank. Due to space limitations, we will not discuss the maximization problems of the retail deposit and loan branches here as they are identical to the problems outlined in Gerali et al. (2010) and refer to their study.
Wholesale Unit The wholesale branches of commercial banks operate under perfect competition and are responsible for the capital position of the respective commercial bank. On the asset side, they hold funds they provide to the retail loan branch, bCt , which ultimately lends these funds to entrepreneurs at a markup in the form of loans, bE;Ct . On the liability side, it combines commercial bank net worth, or capital, ktC , with wholesale deposits, dCt , that are provided by the retail deposit branch, but originally stem from deposits placed in the retail branch by patient households (dP;Ct ). The wholesale bank balance sheet is thus given by
btC = ktC + dtC : | (A.10) |
Furthermore, the capital position of the wholesale branch is prone to a regulatory capital requirement, tC . Moving away from the regulatory requirement imposes a quadratic cost cCt to the bank, which is proportional to the outstanding amount of bank capital and parameterized by Ck :
C | kc | 2 | ||||||||
C | 2 | btC | C | C | ||||||
k | t | |||||||||
ct | = | t | kt | : | (A.11) |
The wholesale branch thus maximizes the discounted sum of real cash ows:
bt | ; dt | Xt | |||||||||
Lw = max E0 | 0P;t | (1 + rtC )btC | btC+1t+1 + dtC+1t+1 | (1 + rtdC )dtC + | |||||||
C | C | =0 | |||||||||
C | kC | 2 | |||||||||
2 | btC + btS | ||||||||||
+ (kC | t+1 | kC ) | k | t | C | kC | (A.12) | ||||
t+1 | t | t | t |
where we assume the net wholesale loan rate rtC and the deposit rate rtdC to be given from the perspective of the maximizing bank. We can use the objective together with the balance sheet constraint A.10 to get:
ECB Working Paper Series No 2406 / May 2020 | 46 |
rt bt | rt dt | 2 | btC | 2 | |||
kt . | |||||||
C C | dC C | kC | ktC | C | C | ||
We can thus express the maximization problem as:
C | kC | 2 | ||||||||||||
L | bC ; dC | |||||||||||||
w | b | dC | C | 2 | btC | C | C | |||||||
C | k | t | ||||||||||||
= | max rt | bt | rt | dt | t | kt | : | |||||||
t t |
The rst-order conditions yield the following expression:
rt | = rt | k | kC | t | kC | 2 | |||
btC | btC | ||||||||
b | dC | C | t | C | t |
(A.13)
(A.14)
As the commercial bank has access to central bank funding in the model, we assume that the rate paid on wholesale deposits gathered from the retail deposit unit of the commercial bank (and so originally from households and rms) has to be equal to the risk-free policy rate, rt, by arbitrage:
rtdC = rt
such that the spread between the loan and deposit rates on the wholesale level is given
by
kC | kC | 2 | |||||||
rt | rt = k | t | : | (A.15) | |||||
btC | btC | ||||||||
b | C | t | C | t |
Assuming symmetry between banks and reinvestment of prots in banks, aggregate bank capital KtC is accumulated from retained earnings only:
KtC = (1 C )KtC1 + JtC1 | (A.16) |
where JtC depicts aggregate commercial bank prots derived from the three branches of the bank, see Gerali et al. (2010). Capital management costs are captured by C .
A.3.2 Shadow Banks
The balance sheet of each shadow bank j in each period is given by
qkbE;S | (j) = dP;S | (j) + kS(j) | (A.17) |
t t | t | t |
where the asset side is given by the funds lend to entrepreneurs, bE;St (j), multiplied with the relative price for these claims, qtk. Shadow banks' liabilities consist of household deposits dP;St (j) and net worth, or shadow bank capital ktS(j).
ECB Working Paper Series No 2406 / May 2020 | 47 |
Shadow bankers earn an interest rate on their claims rtbS. The net prots of shadow banks, i.e. the dierence between real earnings on nancial claims and real interest payments to depositors, determine the evolution of shadow bank capital:
ktS+1(j) = (1 + rtbS)qtkbtE;S(j) | (1 + rtdS)dtP;S(j) | (A.18) | |||
or | |||||
kS | (j) = (rbS | rdS)qkbE;S(j) + (1 + rdS)kS(j): | (A.19) | ||
t+1 | t | t t t | t | t | |
For the shadow banker, as long as the real return on lending, (rbS | rdS) is positive, | ||||
t | t |
it is protable to accumulate capital until it exits the shadow banking sector. Thus, the shadow bank's objective to maximize expected terminal wealth, vt(j), is given by
vt(j) = maxEt | 1 | (1 | S)SiSi+1ktS+1+i(j) | (A.20) | |||||||||||
Xi | |||||||||||||||
=0 | |||||||||||||||
or | |||||||||||||||
1 | |||||||||||||||
Xi | S | Si Si+1 | dS | k | E;S | dS | S | ||||||||
(1 ) | [(r | r | )q | b | (j) + (1 + r | )k | (j)]: (A.21) | ||||||||
vt(j) = maxEt | t+i | t+i | t+i | t+i | t+i | ||||||||||
t+i | |||||||||||||||
=0 |
We introduce a moral hazard problem discussed in Section 3. Diverting funds and running away" is equivalent to declaring bankruptcy for the shadow bank, such that it will only do so if the return of declaring bankruptcy is larger than the discounted future return from continuing and behaving honestly:
vt(j) SqtkbtE;S(j): | (A.22) |
Equation A.22 is the innite-horizon incentive constraint in the two-period model the shadow banker faces when demanding funds from households. Following Gertler and Karadi (2011), we can rewrite it as:
vt(j) = SqkbE;S | (j) + SkS(j) | (A.23) |
t t t | t t |
with
tS = Et[(1 S)S(rtbS rtdS) + SSt;tS | +1tS+1] | (A.24) |
ECB Working Paper Series No 2406 / May 2020 | 48 |
and | ||||||||||
S = Et[(1 | S) + SSzS | S | ] | (A.25) | ||||||
t | t;t+1 | t+1 | ||||||||
S | qtk+ibtE;S+i (j) | |||||||||
where t;t+i | depicts the gross growth rate in nancial claims between t | |||||||||
qtkbtE;S(j) | ||||||||||
S | ktS+i(j) | |||||||||
and t+i, whereas zt;t+i | determines the gross growth rate of shadow bank capital. | |||||||||
ktS(j) | ||||||||||
With these denitions, we can express the incentive constraint as | ||||||||||
tSktS(j) + tSqtkbtE;S(j) SqtkbtE;S(j): | (A.26) |
With constraint A.26 being binding, bank capital determines the amount that the shadow banker can lend out:
k E;S | tS | S | S | S | |||||
qt bt | (j) = | kt | (j) = t | kt | (j) | (A.27) | |||
S | S | ||||||||
t |
where St is the asset-to-capital ratio, or the shadow bank leverage ratio. As shadow banks' incentive to divert funds increases with leverage, Equation A.27 limits the shadow bank's leverage ratio to the point where costs and benets of cheating are exactly leveled. Thus, due to the nancial friction, shadow banks, even not facing an externally set capital requirement that limits their leverage, are prone to an endogenous capital constraint that limits their ability to increase leverage.41
Rewriting bank capital as
kS | (j) = [(rbS | rdS)S + (1 + rdS)]kS(j) | (A.28) | |||
t+1 | t | t | t | t | t |
we get
S | ktS+1(j) | bS | dS | S | dS | ||||
zt;t+1 | = | = (rt | rt | )t | + (1 + rt | ) | (A.29) | ||
kS(j) | |||||||||
t |
and
S | qk | bE;S | (j) | S | S | ||||||
t+1 | t+1 | t+1 | |||||||||
t;t+1 | = | = | zt;t+1: | (A.30) | |||||||
qtkbtE;S(j) | tS | ||||||||||
41We assume that in the simulations, parameters are set such that the constraint always binds within a local region around steady state in equilibrium. Similarly to condition 41 in the Online Appendix, an equilibrium with a binding incentive constraint is characterized by 0 < tS< S, which can be shown with Equation A.27.
ECB Working Paper Series No 2406 / May 2020 | 49 |
As none of the components of St depends on rm-specic factors, we can drop the subscript j by summing across individual shadow bankers to get for total shadow bank lending:
qtkbtE;S = tSktS | (A.31) |
with bE;St depicting aggregate lending/nancial claims the shadow banking sector provides and ktS being the aggregate capital held by shadow banks in period t.
As we assume some shadow bankers to exit each period and new bankers to enter the market, we know that aggregate capital ktS is determined by capital of continuing shadow bankers, ktS;c, and capital of new bankers that enter, ktS;n:
kS = kS;c | + kS;n | : | (A.32) | |
t | t | t |
As a fraction S of existing shadow bankers survives each period, we know that at period t, we have for ktS;c
ktS;c = S[(rtbS1 | rtdS1 )tS1 + (1 + rtdS1 )]ktS1 : | (A.33) |
For new shadow bankers, we assume that they get some start-up capital from the household the shadow banker belongs to. This start-up value is assumed to be proportional to the amount of claims exiting shadow bankers had intermediated in their nal period. With i.i.d. exit probability S, total nal period claims of exiting shadow bankers at t are given by (1 S)qtkbE;St1 . We assume that each period the household transfers a fraction
!SS of this value to entering bankers, such that in the aggregate, we get:
1
ktS;n = !SqtkbtE;S1 : | (A.34) |
Combining Equations A.32, A.33 and A.34, we get the following law of motion for shadow bank capital:
ktS = S[(rtbS1 | rtdS1 )tS1 + (1 + rtdS1 )]ktS1 + !SqtkbtE;S1 : | (A.35) |
Finally, we assume a non-negative spread between the interest rates earned on shadow bank deposits, rtdS, and on the deposits households can place with commercial banks, rtdC , which is again determined by the parameter S, with 0 S 1. In the Online Appendix, we provide a microfoundation for the existence of a positive spread, and use the results to incorporate a relationship between the two deposit rates similar to the relation stated in the two-period model:
ECB Working Paper Series No 2406 / May 2020 | 50 |
1 + rdC | |||||
: | (A.36) | ||||
S" | |||||
1 | |||||
t |
As in the two-period version of the model, the parameter S determines the spread between the gross rates on both deposit types and is implicitly related to the default probability of shadow banks. As a shortcut, we will calibrate S and assume the existence of a spread shock "t following an auto-regressive process to motivate exogenous swings in the spread on interest rates earned on the two deposit types.
A.4 Capital Goods Producers and Retailers
Following Gerali et al. (2010), the rst-order condition for capital goods producers is given by
1 = qtk"1 | i It"qk | 1! | 2 | i | It"qk | 1! | It"qk | #+ | ||||||||||||
t | t | t | ||||||||||||||||||
2 | It1 | It1 | It1 | |||||||||||||||||
" tE | It | k | ! | k | 2 | # | ||||||||||||||
It | ! | |||||||||||||||||||
tE+1 | k | i It+1"tq+1 | It+1"tq+1 | qk | ||||||||||||||||
+ EEt | qt+1 | 1 | "t+1 | (A.37) | ||||||||||||||||
and capital accumulation is given by | ||||||||||||||||||||
i I "qk | 2 | |||||||||||||||||||
Kt = (1 | k)Kt1 | + "1 | t t | ! #: | (A.38) | |||||||||||||||
2 | It1 |
We assume price stickiness a la Calvo (1983) in the retail sector.42 Thus, only a share or retailers indicated by p is able to adjust prices in a given period. Retailers' marginal costs are given by
mctE = | 1 | : | (A.39) |
xt |
A.5 Monetary Policy and Market Clearing
The central bank sets the policy rate according to a Taylor-type rule given by
42In the studies by Gerali et al. (2010) and Gambacorta and Signoretti (2014), price stickiness was modeled using Rotemberg (1982) pricing. However, we decided to use the more convenient Calvo pricing approach in the model.
ECB Working Paper Series No 2406 / May 2020 | 51 |
(1 + rt) = (1 + r)(1 | )(1 + rt1 | ) | (1 r) | yt1 | ! | y(1 r) | (A.40) | ||||
! | "tr | ||||||||||
r | r | t | yt | ||||||||
where the weights on ination and output growth are given by and y, respectively. The steady-state policy rate is given by r and "rt de nes a white noise monetary policy shock.
The market clearing condition is given by the aggregate resource constraint
k | k | kKb | ||||||
t1 | ||||||||
Yt = Ct + qt | (Kt | (1 | )Kt1 | ) + | + ACt | (A.41) | ||
t | ||||||||
with ACt determining the overall adjustment costs and composite consumption given by Ct = cPt + cEt .
ECB Working Paper Series No 2406 / May 2020 | 52 |
- Appendix: Data and Estimation
We derive our data set from the European System of Accounts (ESA 2010) quarterly nancial and non- nancial sector accounts, provided by the ECB and Eurostat. Commercial bank balance sheet data is gathered from the data set on Monetary Financial Institutions" (MFIs), whereas shadow bank data is based on statistics on Other Financial Institutions" (OFIs) as well as on data on investment funds and money market funds (MMFs) provided by the ECB. Commercial bank interest rate data is combined from different sources, as indicated below. All variables except for interest rates are seasonally and working day adjusted and expressed in real terms. We furthermore detrend macroeconomic variables (real GDP, real consumption, real investment) and intermediary loans and deposits by applying log-di erences. We then subtract the sample means from the data after log-di erentiation to arrive at average growth rates of zero for these variables. Interest rates and price and wage ination variables are also demeaned. A detailed description of each variable is given below, and the nal time series used in the estimations are plotted in Figure 10.
B.1 Real Economic and Commercial Bank Data
For the real economy, we include information on real gross domestic product, real con- sumption, real investment, and consumer price as well as wage ination. We furthermore use data on commercial bank deposits held by private households, commercial bank loans granted to the non- nancial corporate sector, the short-term EONIA rate as a quarterly measure of the policy rate, and measures for interest rates on household deposits and rm loans. We detrend non-stationary seasonally adjusted data (real consumption, real investment, bank deposits and loans) by using demeaned log-di erenced data and demean all interest and ination rates.
Real GDP: Real gross domestic product, euro area 19 ( xed composition), deated using the GDP deator (index), calendar and seasonally adjusted data (national accounts, main aggregates (Eurostat ESA2010)).
Real consumption: Real consumption expenditure of households and non-pro t institutions serving households (NPISH), euro area 19 ( xed composition), deated using Consumption deator (index), calendar and seasonally adjusted data (national accounts, main aggregates (Eurostat ESA2010)).
Real investment: Real gross xed capital formation (GFCF), euro area 19 ( xed com- position), deated using GFCF deator (index), calendar and seasonally adjusted data (national accounts, main aggregates (Eurostat ESA2010)).
ECB Working Paper Series No 2406 / May 2020 | 53 |
Ination: Harmonized index of consumer prices (HICP) overall index, quarterly changes, euro area (changing composition), net ination rate, calendar and seasonally adjusted data.
Wage ination: Labor cost index, OECD data, euro area 19 ( xed composition), wages and salaries, business economy, net wage ination, calendar and seasonally adjusted data.
Nominal interest rate (policy rate): EONIA rate, ECB money market data.
Commercial bank loans: Real outstanding amounts of commercial bank (MFIs excluding ESCB) loans to non- nancial corporations, euro area (changing composition), deated using HICP, calendar and seasonally adjusted data.
Commercial bank deposits: Real deposits placed by euro area households (overnight deposits, with agreed maturity up to two years, redeemable with notice up to 3 months), outstanding amounts, euro area (changing composition), deated using HICP, calendar and seasonally adjusted data.
Interest rate on commercial bank loans: Annualized agreed rate (AAR) on commercial bank loans to non- nancial corporations with maturity over one year, euro area (changing composition), new business coverage. Before 2003: Retail interest Rates Statistics (RIR), not harmonized data. Starting 2003:Q1: MFI Interest Rate Statistics (MIR), harmonized data.
Interest rate on commercial bank deposits: Commercial bank interest rates on household deposits, weighted rate from rates on overnight deposits, with agreed maturity up to two years, redeemable at short notice (up to three months), euro area (changing composition), new business coverage. Before 2003: Retail interest Rates Statistics (RIR), not harmonized data. Starting 2003:Q1: MFI Interest Rate Statistics (MIR), harmonized data.
B.2 Shadow Bank Data
In addition to the variables on commercial bank and real activity, we include data on shadow banks in the euro area in our sample. In comparison to lending provided by commercial banks, we derive a time series on shadow bank lending to non- nancial corporates. In doing so, we are able to include an empirical measure of shadow bank credit. Deriving information on the European shadow banking system is challenging since 1) a wide variety of shadow bank de nitions are used among scholars and practitioners and 2) euro area data on nancial institutions that could be classi ed as shadow banks is available at a much lower level of detail and in a less structured manner than information on commercial banks. Therefore, one has to compromise between the conceptional de nition of shadow banks used and the empirical counterparts that can be analyzed with available data.
ECB Working Paper Series No 2406 / May 2020 | 54 |
In practice, the shadow banking system consist of a multitude of nancial institutions partly fullling highly specialized task in a prolonged chain of credit intermediation (Adrian, 2014; Adrian and Liang, 2014; Pozsar et al., 2010). Given the diverse nature of non-bank nancial institutions, a variety of denitions of shadow banks have been pro- posed, covering either a particular set of institutions (institutional approach) or a range of activities dierent entities are jointly engaged in (activity approach). We base our empirical measures of shadow banks on the broad" denition of the shadow banking system provided by the Financial Stability Board (FSB, 2017, 2011), which states that the shadow banking system is the system of credit intermediation that involves entities and activities outside the regular banking system" (FSB, 2011, p.2) and that ...this implies focusing on credit intermediation that takes place in an environment where prudential regulatory standards and supervisory oversight are either not applied or are applied to a materially lesser or dierent degree than is the case for regular banks engaged in similar activities" (FSB, 2011, p.3).
More precisely, we follow the institutional approach employed by ECB sta to apply the FSB broad denition to available euro area data (Malatesta et al., 2016; Doyle et al., 2016; Bakk-Simonet al., 2012). The core of this approach depicts the use of the Other Financial Intermediaries" (OFIs) aggregate in the Eurosystem's nancial accounts data. Within the aggregate, all activities of nancial intermediaries not classied as Monetary Financial Institutions" (MFIs) are captured. Thus, the OFI aggregate depicts a residual component and not only includes institutions universally accepted as shadow banks. For instance, the insurance corporations and pension funds sector (ICPFs) is mainly engaged in activities that are not related to shadow bank intermediation, and we therefore exclude balance sheet items of these institutions from our shadow bank aggregates. Furthermore, the OFI aggregate is lacking information on money market funds (MMFs), which are classied as MFIs. However, there is a broad consensus in the literature that MMFs engage in activities that could possibly be counted as shadow bank intermediation,43 and we therefore include MMF information in the shadow bank aggregate. Our benchmark shadow bank denition (1) therefore closely resembles the broad shadow bank denition by the FSB and covers the whole range of OFIs except for ICPFs, plus MMFs (Scenario 1 in Table 4).44
The OFI sector, in line with the broad denition of shadow banks given by the FSB,
- See for instance Adrian (2014), Adrian and Liang (2014), Pozsar et al. (2010), or FSB (2017, 2011).
- Detailed information on the composition of the OFI sector has only recently been provided by the ECB. For instance, the collection of detailed balance sheet data on investment funds and nancial vehicle corporations (FVCs) was only initiated in 2008 and 2009, respectively. Also, harmonized data on MMFs is available only from 2006 onward in the MFI statistics, but can be gathered from other sources for earlier years (see Appendix B). Balance sheet information on these institutions accounts for approximately 50 percent of the total OFI sector, with the rest being characterized by smaller and more heterogeneous entities. As highlighted by Doyle et al. (2016), one should therefore be aware of the fact that not all institutions in the remaining half of the OFI aggregate could unambiguously be declared as shadow banks.
ECB Working Paper Series No 2406 / May 2020 | 55 |
Table 4: Di erent De nitions of Shadow Banks Based on the OFI Aggregate
Including | Including | Lending | |
Scenario | Investment Funds | Money-Market Funds | Counterparties |
1 | X | X | NFCs |
2 | X | Total economy | |
covers non-MMF investment funds. Whereas some studies highlight the increasing role of direct investment fund lending to the non- nancial private sector in the euro area since the recent global nancial crisis (Doyle et al., 2016), other studies discuss the special role investment funds play in the nancial system and question the adequacy of considering these institutions as intermediaries between real economy borrowers and lenders. For instance, Bakk-Simonet al. (2012) argue that investment funds are indeed covered by regulation, even though substantially di erent than commercial banks. They therefore question whether the de nition of shadow banks being intermediaries outside the regulatory system given by the FSB applies to investment funds. Consequently, we use as a robustness check an alternative measure of shadow bank loans that excludes investment funds (Scenario 2 in Table 4). However, we are not able to gather counterparty information for investment fund lending before 2008, and therefore use total lending of the OFI sector less investment fund lending in this second estimation, instead of lending to non- nancial corporations only.
Shadow bank loans (including investment funds): Loans of other nancial intermediaries (OFI) to non-nancialcorporations, excluding insurance corporations and pension funds, including investment funds, euro area 19 (xed composition), deated using HICP, calendar and seasonally adjusted data.
Shadow bank loans (excluding investment funds): Loans of other nancial intermediaries (OFI) to total economy, excluding insurance corporations and pension funds, excluding investment fund assets (deposits, loans, and nancial derivatives), euro area 19 (xed composition), deated using HICP, calendar and seasonally adjusted data.
ECB Working Paper Series No 2406 / May 2020 | 56 |
Figure 10: Euro Area Observable Time Series Used in Estimation | |||||||||||
GDP | Consumption | Investment | |||||||||
0.002 | 0.01 | ||||||||||
0 | 0 | ||||||||||
-0.005 | 0 | -0.01 | |||||||||
-0.01 | -0.002 | -0.02 | |||||||||
1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 |
CB Deposits | CB Deposit Rate | CB Firm Loans |
0.015 | 0.002 | 0.01 | |||||||||
0.01 | 0 | 0 | |||||||||
0.005 | |||||||||||
0 | -0.01 | ||||||||||
-0.005 | -0.002 | ||||||||||
1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 |
Price Inflation | Wage Inflation | CB Firm Loan Rate | |||||||||
0.002 | 0.008 | 0.004 | |||||||||
0.006 | |||||||||||
0.002 | |||||||||||
0 | 0.004 | ||||||||||
-0.002 | 0.002 | 0 | |||||||||
0 | |||||||||||
-0.002 | |||||||||||
-0.004 | -0.002 | ||||||||||
1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 |
Policy Rate | SB Firm Loans (Broad) | SB Firm Loans (Narrow) | |||||||||
0.005 | 0.04 | 0.02 | |||||||||
0 | 0.02 | 0 | |||||||||
0 | |||||||||||
-0.005 | -0.02 | -0.02 | |||||||||
1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 | 1999 | 2004 | 2009 | 2014 |
Note: Real stock and volume data (real GDP, real consumption, real investment, loans and deposits by commercial and shadow banks) are expressed as demeaned log-di erences. Wage and price ination and interest rates are quarterly net rates and expressed in absolute deviations from sample means.
B.3 Prior and posterior distributions
Figure 11 reports the prior and posterior distributions for the baseline estimation reported
in Table 2.
ECB Working Paper Series No 2406 / May 2020 | 57 |
Figure 11: Prior and Posterior Distributions: Baseline Estimation
p | i | d | bE | C | ||||||||||||||
k | ||||||||||||||||||
15 | 0.6 | 0.15 | 0.2 | 1.5 | ||||||||||||||
0.04 | ||||||||||||||||||
0.4 | 1 | |||||||||||||||||
10 | ||||||||||||||||||
0.1 | ||||||||||||||||||
0.1 | ||||||||||||||||||
0.02 | ||||||||||||||||||
5 | 0.2 | 0.05 | 0.5 | |||||||||||||||
0 | 0 | 0 | 0 | 0 | 0 | |||||||||||||
0.5 | 1 | 2 4 6 8 | 5 | 15 25 | 0 | 10 | 20 | 0 | 20 | 1 | 2 | 3 | ||||||
y | r | aHH | z | a | ||||||||||||||
8 | 15 | 8 | 4 | 10 | 6 | |||||||||||||
6 | ||||||||||||||||||
10 | 6 | 4 | ||||||||||||||||
4 | 2 | |||||||||||||||||
4 | ||||||||||||||||||
5 | ||||||||||||||||||
5 | 2 | |||||||||||||||||
2 | 2 | |||||||||||||||||
0 | 0 | 0 | 0 | 0 | 0 | |||||||||||||
0 | 0.4 | 0.5 | 1 | 0.5 | 1 | 0.5 | 1 | 0.5 | 1 | 0.2 0.6 | ||||||||
mE | d | bE | qk | y | l | |||||||||||||
20 | 6 | 6 | 10 | |||||||||||||||
4 | ||||||||||||||||||
4 | 4 | |||||||||||||||||
4 | ||||||||||||||||||
5 | ||||||||||||||||||
10 | ||||||||||||||||||
2 | 2 | |||||||||||||||||
2 | 2 | |||||||||||||||||
0 | 0 | 0 | 0 | 0 | 0 | |||||||||||||
0.5 | 1 | 0 | 0.4 0.8 | 0 | 0.5 | 1 | 0.2 0.6 | 0 | 0.4 0.8 | 0.5 0.7 0.9 | ||||||||
C | z | a | mE | d | ||||||||||||||
k | ||||||||||||||||||
20 | 200 | 150 | 400 | |||||||||||||||
100 | 1000 | |||||||||||||||||
100 | ||||||||||||||||||
10 | ||||||||||||||||||
50 | 100 | 200 | 500 | |||||||||||||||
50 | ||||||||||||||||||
0 | 0 | 0 | 0 | 0 | 0 | |||||||||||||
0.5 | 1 | 0 | 0.15 | 0 | 0.04 | 0.01 | 0.04 | 0.01 0.04 | 0.01 0.04 | |||||||||
bE | qk | r | y | l | C | |||||||||||||
k | ||||||||||||||||||
600 | 2000 | 3000 | ||||||||||||||||
1000 | ||||||||||||||||||
2000 | 2000 | 100 | ||||||||||||||||
400 | ||||||||||||||||||
1000 | 500 | |||||||||||||||||
1000 | 1000 | 50 | ||||||||||||||||
200 | ||||||||||||||||||
0 | 0 | 0 | 0 | 0 | 0 | |||||||||||||
0 | 0.04 | 0.01 0.04 | 0.01 0.04 | 0.01 0.04 | 0.02 | 0.06 | 0.01 0.04 | |||||||||||
Posterior | Prior |
Note: Prior and posterior distributions from the baseline estimation reported in Table 2.
ECB Working Paper Series No 2406 / May 2020 | 58 |
- Appendix: Robustness Checks
In the following, we estimate our baseline model on two dierent specications of the sample. First, to account for uncertainty around the exact date of the beginning of the ELB phase in the euro area, we provide evidence on estimated parameters when using an earlier end date as in the baseline specication. We are also aware of structural changes in the nancial system after the 2007/2008 nancial crisis and over the course of the subsequent European debt crisis which potentially altered the role and eectiveness of shadow banking in the euro area. To take these considerations into account, we re-estimate our model for the period of 1999:Q1 to 2008:Q4, thereby excluding both the post-nancial crisis and ELB period from the estimation. In addition, excluding the period after 2008 allows for a straightforward comparison of results to Gerali et al. (2010), who used the same period in the estimation. Estimation results are reported in the middle columns of Table 5. In addition, we restate our baseline estimation results for comparison.
Whereas result from the pre-crisis period estimation are qualitatively comparable to the baseline estimates, some slight quantitative dierences in parameter estimates can be observed. The mode estimates for the parameter governing investment adjustment costs turns out to be lower in the estimation using the pre-crisis sample. By including the years after 2008 - a period characterized by the aftermath of the global nancial crisis and by the subsequent European sovereign debt crisis - the rise in investment adjustment costs could be driven by higher investment volatility - due to a signicant fall in investment activity in the post-crisis years and the more moderate growth thereafter - in the post- 2008 period. Furthermore, estimates for interest rate adjustment costs are higher in the pre-crisis sample compared to the full sample.
Second, we re-estimate our model by applying a dierent denition of shadow banks, i.e. by excluding investment funds from the shadow bank aggregate, as discussed in Section 4 and appendix B (Scenario 2 in Table 4). We report parameter estimates in the right columns of Table 5, again in comparison to our baseline estimation. Our baseline results are not substantially aected when investment fund information is excluded. Commercial bank loan rate adjustment costs turn out to be slightly lower when investment fund information is excluded from the estimation, whereas other structural parameters - based on the comparison of posterior modes - do not dier from baseline results.
ECB Working Paper Series No 2406 / May 2020 | 59 |
2020 May / 2406 No Series Paper Working ECB
Table 5: Posterior Distributions: Robustness and Evaluation
Posterior Distribution | Pre-Crisis | Excluding Investment Funds | |||||||||||||
Parameter | |||||||||||||||
5 Perc. | Median | 95 Perc. | Mode | 5 Perc. | Median | 95 Perc. | Mode | 5 Perc. | Median | 95 Perc. | Mode | ||||
Structural Parameters | |||||||||||||||
p | Calvo Parameter | 0.83 | 0.87 | 0.90 | 0.86 | 0.76 | 0.81 | 0.86 | 0.81 | 0.83 | 0.86 | 0.90 | 0.86 | ||
i | Investment Adjustment Cost | 2.98 | 3.98 | 5.14 | 3.67 | 2.60 | 3.64 | 4.83 | 3.50 | 2.89 | 3.79 | 4.76 | 3.66 | ||
d | Deposit Rate Adjustment Cost | 10.00 | 13.26 | 16.72 | 12.62 | 11.37 | 15.06 | 18.99 | 14.74 | 9.89 | 12.94 | 16.25 | 12.62 | ||
bE | Loan Rate Adjustment Cost | 4.84 | 8.34 | 14.23 | 7.56 | 5.11 | 9.72 | 15.13 | 8.81 | 4.75 | 7.74 | 11.00 | 6.91 | ||
C | CCR Deviation Cost | 0.01 | 10.05 | 21.32 | 24.71 | 3.32 | 14.01 | 25.00 | 25.00 | 3.08 | 14.27 | 24.99 | 25.00 | ||
k | |||||||||||||||
TR Coecient | 1.44 | 1.87 | 2.30 | 1.75 | 1.23 | 1.62 | 2.03 | 1.58 | 1.43 | 1.87 | 2.31 | 1.80 | |||
y | TR Coecient y | 0.14 | 0.24 | 0.34 | 0.20 | 0.12 | 0.20 | 0.28 | 0.19 | 0.12 | 0.20 | 0.29 | 0.20 | ||
r | Interest Rate Smoothing | 0.84 | 0.88 | 0.91 | 0.88 | 0.79 | 0.86 | 0.91 | 0.86 | 0.84 | 0.88 | 0.91 | 0.88 | ||
aHH | HH Habit Formation | 0.70 | 0.77 | 0.84 | 0.77 | 0.53 | 0.66 | 0.78 | 0.64 | 0.71 | 0.78 | 0.85 | 0.77 | ||
Exogenous Processes (AR Coecients) | |||||||||||||||
Deposit Rate Spread | 0.62 | 0.81 | 0.95 | 0.85 | 0.65 | 0.81 | 0.96 | 0.85 | 0.65 | 0.82 | 0.96 | 0.85 | |||
z | Consumer Preference | 0.82 | 0.87 | 0.92 | 0.87 | 0.77 | 0.87 | 0.95 | 0.88 | 0.81 | 0.87 | 0.92 | 0.87 | ||
a | Technology | 0.31 | 0.42 | 0.52 | 0.42 | 0.31 | 0.42 | 0.52 | 0.42 | 0.32 | 0.42 | 0.53 | 0.42 | ||
mE | Entrepreneur LTV | 0.91 | 0.94 | 0.97 | 0.95 | 0.90 | 0.94 | 0.98 | 0.95 | 0.91 | 0.94 | 0.97 | 0.94 | ||
d | Deposit Rate Markdown | 0.27 | 0.36 | 0.46 | 0.36 | 0.28 | 0.38 | 0.49 | 0.38 | 0.27 | 0.37 | 0.47 | 0.37 | ||
bE | Loan Rate Markup | 0.51 | 0.63 | 0.75 | 0.64 | 0.52 | 0.67 | 0.82 | 0.67 | 0.50 | 0.63 | 0.76 | 0.64 | ||
qk | Investment Eciency | 0.33 | 0.46 | 0.58 | 0.49 | 0.27 | 0.40 | 0.54 | 0.41 | 0.34 | 0.46 | 0.58 | 0.47 | ||
y | Price Markup | 0.25 | 0.36 | 0.47 | 0.37 | 0.25 | 0.38 | 0.51 | 0.36 | 0.24 | 0.35 | 0.46 | 0.34 | ||
l | Wage Markup | 0.64 | 0.71 | 0.77 | 0.71 | 0.62 | 0.70 | 0.77 | 0.69 | 0.65 | 0.71 | 0.77 | 0.71 | ||
C | Commercial Bank Capital | 0.93 | 0.96 | 0.99 | 0.97 | 0.86 | 0.92 | 0.97 | 0.93 | 0.93 | 0.96 | 0.99 | 0.96 | ||
k | |||||||||||||||
Exogenous Processes (Standard Deviations) | |||||||||||||||
Deposit Rate Spread | 0.002 | 0.007 | 0.016 | 0.005 | 0.002 | 0.007 | 0.018 | 0.005 | 0.002 | 0.007 | 0.023 | 0.005 | |||
z | Consumer Preference | 0.008 | 0.011 | 0.014 | 0.011 | 0.005 | 0.008 | 0.011 | 0.007 | 0.008 | 0.011 | 0.014 | 0.011 | ||
a | Technology | 0.025 | 0.029 | 0.033 | 0.028 | 0.021 | 0.025 | 0.030 | 0.025 | 0.024 | 0.028 | 0.033 | 0.028 | ||
mE | Entrepreneur LTV | 0.006 | 0.008 | 0.009 | 0.007 | 0.004 | 0.006 | 0.007 | 0.005 | 0.006 | 0.008 | 0.009 | 0.007 | ||
d | Deposit Rate Markdown | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | 0.002 | ||
bE | Loan Rate Markup | 0.002 | 0.002 | 0.003 | 0.002 | 0.002 | 0.003 | 0.004 | 0.003 | 0.002 | 0.003 | 0.003 | 0.002 | ||
qk | Investment Eciency | 0.001 | 0.002 | 0.002 | 0.002 | 0.001 | 0.002 | 0.002 | 0.002 | 0.001 | 0.002 | 0.002 | 0.002 | ||
r | Monetary Policy | 0.001 | 0.001 | 0.002 | 0.001 | 0.001 | 0.002 | 0.002 | 0.002 | 0.001 | 0.001 | 0.002 | 0.001 | ||
y | Price Markup | 0.001 | 0.002 | 0.002 | 0.001 | 0.001 | 0.002 | 0.002 | 0.002 | 0.001 | 0.002 | 0.002 | 0.001 | ||
l | Wage Markup | 0.035 | 0.041 | 0.047 | 0.040 | 0.029 | 0.035 | 0.042 | 0.035 | 0.035 | 0.041 | 0.047 | 0.040 | ||
C | Commercial Bank Capital | 0.003 | 0.003 | 0.004 | 0.003 | 0.003 | 0.003 | 0.004 | 0.003 | 0.003 | 0.003 | 0.004 | 0.003 | ||
k |
Note: Results are based on 5 chains with 500,000 draws each based on the Metropolis-Hastings algorithm. Columns 3 to 6 report the posterior moments from the baseline estimation in Table 2. Columns 7 to 10 report results from the estimation using the sample 1999:Q1 to 2008:Q4, and columns 11 to 14 report results from the estimation using shadow banking data excluding information on investment funds.
60
Acknowledgements
The views in this paper are those of the authors and do not necessarily reflect the views of the European Central Bank or DIW Berlin. We thank Mathias Trabandt, Marius Clemens, Marcel Fratzscher, Wouter den Haan, Peter Karadi and participants at the 15th Dynare Conference, Lausanne, the Third Research Conference of the CEPR Network on Macroeconomic Modelling and Model Comparison (MMCN), Frankfurt am Main, the 2019 ASSA Annual Meeting, Atlanta, the 24th Annual Conference on Computing in Economics and Finance (CEF), Milan, the 2018 Verein für Socialpolitik Annual Conference, Freiburg and the FU Berlin Empirical Macroeconomics Seminar for useful comments.
Stefan Gebauer
DIW Berlin, Freie Universität Berlin, Berlin, Germany; email: sgebauer@diw.de
Falk Mazelis
European Central Bank, Frankfurt am Main, Germany; email: falk.mazelis@ecb.europa.eu
© European Central Bank, 2020
Postal address 60640 Frankfurt am Main, Germany
Telephone +49 69 1344 0
Website www.ecb.europa.eu
All rights reserved. Any reproduction, publication and reprint in the form of a different publication, whether printed or produced electronically, in whole or in part, is permitted only with the explicit written authorisation of the ECB or the authors.
This paper can be downloaded without charge from www.ecb.europa.eu, from the Social Science Research Network electronic library or from RePEc: Research Papers in Economics. Information on all of the papers published in the ECB Working Paper Series can be found on the ECB's website.
ISBN 978-92-899-4049-8 | ISSN 1725-2806 | doi:10.2866/722519 | QB-AR-20-058-EN-N |
Attachments
- Original document
- Permalink
Disclaimer
ECB - European Central Bank published this content on 13 May 2020 and is solely responsible for the information contained therein. Distributed by Public, unedited and unaltered, on 13 May 2020 09:14:05 UTC
