The role of leverage in quantitative easing decisions: Evidence from the UK

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

North American Journal of Economics and Finance journal homepage: www.elsevier.com/locate/najef

The role of leverage in quantitative easing decisions: Evidence from the UK ⁎

Dionisis Philippasa, , Stephanos Papadamoub, Iuliana Tomuleasac,d a

Department of Finance, ESSCA School of Management, 55 Quai Alphonse Le Gallo, 92513, Boulogne, Paris, France Laboratory of Economic Policy and Strategic Planning, Department of Economics, University of Thessaly, 28th October Street 78, 38333 Volos, Greece c Department of Finance, Faculty of Economics and Business Administration, Alexandru Ioan Cuza University of Iasi, 22 Blvd Carol I, 700505 Iasi, Romania d University of Clermont Auvergne, 41 Blvd. F. Mitterand, 63000 Clermont-Ferrand, France b

A R T IC LE I N F O

ABS TRA CT

JEL classification: G1 G21 G28 E52

The paper analyses the implications arising from the impulses and responses of the banking sector in the UK, through the banks’ portfolio balance sheet information, when determined by the quantitative easing implementation. In a panel vector autoregressive framework, we examine the effects of Bank of England's asset purchases on disaggregated leverage and bank profitability for different types of financial institutions, which reflect differences in the sequencing of the quantitative easing strategy. We find that the transmission channel of QE to the growth of economic activity depends on the degree of financial leverage, the holdings of securities and lending rates but with diverging magnitude for the different types of UK financial institutions.

Keywords: Quantitative easing Financial institutions Leverage Panel VAR

1. Introduction The global financial crisis that started in 2008 and its aftermath posed significant challenges for monetary authorities. Unconventional monetary policies remain one of the few levers available for policy makers to exercise, with the most common referring to the extension of their balance sheets by large-scale asset purchases (LSAPs), known as quantitative easing (hereafter QE). Traditionally, QE means focusing on buying longer-term government bonds from banks, allowing the sovereign yields to serve as a benchmark for the pricing of riskier privately issued securities (Eser & Schwaab, 2016; Krishnamurthy & Vissing-Jorgensen, 2011). As a result, the yields on privately issued securities, and consequently the bank lending rates, are expected to decline in parallel with those on government bonds, with the hope that this will stimulate longer-term investments and hence the aggregate demand, thereby supporting price stability (Bowman, Cai, Davies, & Kamin, 2015). Another important issue is that QE can have a significant impact on the profitability of banks. Lower interest rates reduce banks' profitability in traditional activities, pushing them into more dangerous activities, which can generate higher returns. The previous literature has observed that, in a low-interest environment, banks are trying to take on higher-risk projects, acting on the basis of the monetary policy risk channel (Borio & Haibin, 2012; Delis & Kouretas, 2011). This risk appetite may also be higher because central banks maintain low interest rates until economies recover (Delis, Iftekhar, & Mylonidis, 2017). Adrian and Hyun (2010) argue that a positive increase in asset values through a significant fall in interest rates increases collateral, making access to credit easier for

⁎

Corresponding author. E-mail addresses: [email protected] (D. Philippas), [email protected] (S. Papadamou), [email protected] (I. Tomuleasa).

https://doi.org/10.1016/j.najef.2018.04.014 Received 4 November 2017; Received in revised form 17 April 2018; Accepted 23 April 2018 1062-9408/ © 2018 Elsevier Inc. All rights reserved.

Please cite this article as: Philippas, D., North American Journal of Economics and Finance (2018), https://doi.org/10.1016/j.najef.2018.04.014

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

borrowers. However, when market participants expect a policy response from the central bank to overcome the negative economic crises, financial institutions also increase leverage levels (Diamond & Rajan, 2012; Farhi & Tirole, 2012). Therefore, higher aggregate leverage levels increase riskiness and, potentially, loan losses; which may have a negative impact on the profitability of banks. Following the recent financial crises globally, deleverage plays a key role due to increasing risk aversion undertaken from the financial institutions. The paper analyses the impulse and response between leverage undertaken by different types of banking institutions and, the asset purchases by the BoE as part of its QE strategy, oriented towards the UK financial institutions. We moreover provide useful insights on the understanding of lending behaviour and, cash and security holdings from different types of banking institutions, following a QE shock. Addressing this issue is a challenge, because it is of great interest to disentangle the implications of the effects of QE decisions on the UK financial sector, a major international financial hub. The setting of monetary policy is performed under several pressures that could force an abrupt change in the policy strategies being promoted within a wide variety of financial and macroeconomic signals. There is a considerable amount of empirical literature concerning the broader macroeconomic impact of QE via bond market rates. The main strand of the literature has focused on the two main transmission channels through which asset purchases can affect long-term interest rates by observing the policy signaling channel, which works through changing market expectations about future monetary policy and, portfolio balance channel, which arises from the reduction in the available supply of the assets purchased (Christensen & Rudebusch, 2012; D’Amico, English, López-Salido, & Nelson, 2012; Gilchrist & Zakrajšek, 2013; Hamilton & Wu, 2012; Neely, 2015; Steeley, 2015). A minor part of the literature has tried to estimate the macroeconomic effects of unconventional monetary policy measures via the linkages between the interest rate spreads and the real economy (Chen, Cúrdia, & Ferrero, 2012; Chung, Laforte, Reifschneider, & Williams, 2012). Nevertheless, few studies, to the best of our knowledge, have examined the impact of QE on the profitability and solidity of financial institutions, focusing mainly on US data (Lambert & Ueda, 2014; Mamatzakis & Bermpei, 2016; Mamatzakis, Matousek, & Vu, 2015). These studies argue that an unconventional monetary policy reinforces banks’ solidity by allowing them to reduce their leverage and extend the maturity of their debt. A handful of recent studies have attempted to highlight the role of financial institutions’ leverage decisions but for the case of conventional monetary policy, business cycles and real economic activity in the USA (Istiak & Serletis, 2016). In the light of the above discussion, it is important to make further considerations when discussing QE policy interactions. A crucial question is raised regarding the extent to which the critical role of the UK financial sector’s leverage can ensure the success of quantitative easing. In periods with high deleverage, QE is successful if it reduces the risks of a liquidity shortfall, encouraging banks to extend credit to higher interest-paying parties through the leverage decisions undertaken and thereby boost economic growth, even though the banks are forced to undertake more risks. Moreover, leverage ratios may be significant determinants when identifying the key roles of financial institutions in credit risk transfer in a VAR framework (Yang & Zhou, 2013). Nevertheless, given the level of leverage that the banking sector can experience, banks can stop intermediating loans and may hold onto funds to improve their viability rather than on-lending to the private sector and pass on the additional liquidity to the real economy, thereby making the QE policy ineffective (Joyce, Miles, Scott, & Vayanos, 2012). We address these issues from a different angle that innovates and contributes to filling some of the existing gaps in the literature in at least two dimensions. Firstly, we set up a panel vector autoregressive (panel VAR) framework, characterized by crosssectional heterogeneity and dynamic interdependencies. In this content, we treat QE as an endogenous policy response with respect to two modelling assumptions. In the first assumption, we employ different major types of UK financial institutions, in line with the study of Papadamou and Siriopoulos (2012), and we discuss the extent to which QE has exerted different impacts on their performance. This type of identification tries to shed light on a significant gap regarding the vital importance of different types of financial institutions in studying the implications of QE decisions, without being oriented narrowly towards a macroeconomic perspective. In the second assumption, we consider a decomposition of leverage into three main components, namely the gross loans to equity, liquid assets to equity and securities to equity components. This framework of decomposition may expose the financial sector’s access to asset liquidation, its resilience to short-term liquidity stress and its reliance on volatile short-term sources of funding (i.e. higher funding liquidity risk), respectively. In this content, we can assess whether loans would be granted to the real economy and would withstand adverse non-performing loans’ shocks and, moreover, measure the extent to which a financial institution should leverage in riskier market securities and financing sources and adverse market risks. Consequently, our study analyses the discrete role of leverage’s components in the QE policies implemented and real economic activity for the different types of UK financial institutions. These two modelling assumptions differentiate our paper from other studies employing similar empirical methodologies or addressing related topics. We draw policy implications based on both directions of impulse and response functions between the transmission of QE strategy and the performance of UK financial institutions’ balance sheets. On one hand, we examine the impulse analysis of QE on the balance sheets and the extent to which the financial variables of interest can play a key role in the GDP growth. On the other hand, we investigate the QE policy response to different shocks of lending rate and real economic activity, as well as on the profitability and the components of leverage. Our findings are of great importance to the existing literature, because they highlight the impulses and responses between the profitability and the discrete role of leverage’s components of the financial sector on the one hand, and the transmission of Bank of England’s QE policies for the real economy on the other. The first main finding is that the asset purchases by the BoE are a determining factor that provides financial institutions with the possibility to improve their profitability. A significant reduction in profitability is identified for almost all types of the UK financial institutions, with a diverging magnitude between these types, mainly due to the securities that are held and the diversification benefits of other institutions by their involvement in different sectors of 2

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

activity. However, we recognize that the significant reduction in profitability for real estate banks presents significant benefits for the economic activity in the UK. The transmission channel of QE to GDP growth based on financial institutions’ leverage has a significantly positive effect through securities holding for commercial banks and bank holding companies. This second main finding complements the previous studies about the effect of an unconventional monetary policy on the GDP via leverage in the USA (Istiak & Serletis, 2016; Lambert & Ueda, 2014). The contribution of commercial banks in liquidity and leverage responses to the QE shock is of considerable importance and, consequently, the increase in leverage seems to be attributed mainly to risk-taking behaviour by commercial banks. The evidence shows that a negative shock to the economic activity leads the majority of UK financial institutions to increase their leverage by undertaking significantly high risks, indicating countercyclical effects. The third main finding is the evidence that QE is also transmitted to the real economy via the significant reduction in the retail banking rates, in comparison with other studies focusing only on the transmission via bond rates (Pesaran & Smith, 2016; Weale & Wieladek, 2016). We evidence that the BoE reduces its asset purchases when lending rates are dropped, economic activity is augmented and the leverage of commercial banks is increased. In line with the studies of Putnam (2013) and Yang and Zhou (2017), we therefore argue that the role of QE transmission and its exit strategies by central banks is particularly dominant and challenging to implement and, moreover, have the potential to suspend a return to the normal conduct of monetary policy to the detriment of longer-term economic growth, rational leverage and potential future inflation. The remainder of the paper is organized as follows. Section 2 discusses the panel VAR framework, including the modelling assumptions. Section 3 provides a detailed description of the data sources and draws some initial insights from a preliminary analysis. Section 4 illustrates the empirical findings together with a discussion of the results and their policy implications. Finally, Section 5 concludes. 2. Model setup The panel VAR framework is a coherent approach to estimating interdependencies by treating all the variables as endogenous and allowing time lags across the variables. Recent relevant studies have used empirical panel VAR modelling frameworks with different structural identification approaches to address a variety of issues, such as the transmission of shocks across units, countries and time (Beetsma & Giuliodori, 2011; Ciccarelli, Maddaloni, & Peydró, 2013). In a panel VAR framework, a cross-sectional dimension is added to the common VAR representation that may reveal additional information about interdependencies. Without loss of generality, we illustrate the specification of our panel VAR framework, assuming one lag. Let yi,t be the ki × 1 vector of endogenous variables for each unit i , i = 1, …, N . The ki × 1 vector of endogenous variables takes the form Yi,t = [y1,′ t … yN′ ,t ]′. The panel VAR is written as:

Yi,t = Ai,0 + Ai (l) Yi,t − 1 + ui,t

(1)

where Ai,0 is the vector of all the deterministic common components (e.g., constants, seasonal dummies and deterministic polynomial in time) of the data for all units i , t denotes the time parameter, where t = 1, …, T , coefficients Ai (l) , and ui,t is the G × 1 vector of contemporaneously correlated random disturbances with zero mean and the non-singular variance–covariance matrix Σu . Assuming that the data-generating process features dynamic homogeneity, the pooled estimation approach with fixed effects can be used to estimate the parameters of the model by potentially capturing idiosyncratic but constant heterogeneities across variables and units. However, if different assumptions are imposed in the model specification (e.g., for large number of units denoted as N and small number of time periods, denoted as T ), the pooled estimation approach is biased. One way to overcome this difficulty is to employ the generalized method of moments (GMM) approach initially proposed by Arellano and Bond (1991). According to these authors, when the cross-sectional size of units (N ) is large, and T is fixed and small, given the fact that lagged regressors are used as instruments, the first assumption is derived by estimating the model parameters with the GMM procedure, which is consistent when T is small. In this paper we impose two assumptions to obtain plausible results. The first assumption of the panel VAR framework derived herein is that cross-sectional heterogeneity and dynamic interdependencies are assumed by introducing fixed effects, thus allowing for time-variant individual characteristics.1 Therefore, the panel VAR is characterized by dynamic interdependencies in which the lags of all endogenous variables of all units enter the model for every unit i , cross-sectional heterogeneity whereby innovations are correlated contemporaneously and where the intercept, slope and variance of the shocks ui,t may be unit-specific. In this setting we impose a block structure on the matrix of contemporaneous coefficients (i.e., short-run restrictions) to compute the structural parameters prior to generating impulse response functions, based on the study by Frame, Hancock, and Passmore (2012). Under the first assumption and a common set of L ⩾ k + l instruments, recall Eq. (1) in a compact form: (2)

Yt = Zt A + Ut

where Yt is the vector of the endogenous variables, Zt = ING × (A0 yi′,t1) , which contains all the remaining deterministic common components of the data for all units i , A = (Ai (l))′ = (ai′)′ with Gk × 1 vectors, and Ut is the GN × 1 vector of innovations serially correlated contemporaneously with zero mean and variance–covariance matrix Σu . The individual heterogeneity is endorsed in the 1 One way to address the implicit selectivity bias in our accounting data is to use fixed effects to ensure robustness in the empirical analysis in relation to nonrandom selectivity rather than the random-effects estimator.

3

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

levels of the variables.2 Subtracting the means of each variable calculated for each firm-year and introducing fixed effects eliminate any bank-specific time dummies that capture aggregate and global shocks that may affect all firms in the same way and preserve the orthogonality between the transformed variables. Since A varies with cross-sectional units, it depends on a lower dimension vector that prevents any meaningful unconstrained estimation. For a structural interpretation, we use the following standard linear accounting identity:

Yt =

∑

Zt γj ϑj + Ut + Zt et

(3)

j

where Zt γj can capture any potential common, unit-specific, variable-specific and lag-specific information in the regressors, ϑj are factors that capture the determinants of A and et is the error term of the linearization. The decomposition allows us to measure the common and unit-specific influences for endogenous Yt . Finally, the equation-by-equation GMM estimation yields consistent estimates of panel VAR, in which the joint estimation of the system of equations makes cross-equation hypothesis testing straightforward (Holtz-Eakin, Newey, & Rosen, 1988). To check the robustness of the GMM estimator, we test the optimal lag order in both the panel VAR specification and the moment condition using the moment and model selection criteria (MMSC) for GMM models based on the J statistic of over-identifying restrictions proposed by Andrews and Lu (2001). The dynamics of the model can be investigated by impulse response analysis (IRF). The IRFs are informative for the shocks and interactions arising between the endogenous variables of the system. The standard errors of the impulse response functions and confidence intervals are generated using Monte Carlo simulations. The impulse response function is derived to one standard deviation shock to equation j corresponding to variable k at time t on the expected values of Y at time horizon t + h . The second model assumption is identified as a restricted version of the panel VAR framework and examines the dynamic heterogeneity in the responses to shocks that may arise for different consistent formulations of the cross-sectional panel. Suppose that we run the model for one type of financial institution, denoted as d , from the full-panel sample. Comparing the impulse response functions obtained for the d-type financial institutions each time allows us to assess roughly the contribution of the d-type institutions. Therefore, the restricted vector to be estimated in Eq. (3) is now specified as:

Yt∗ = [y1,′ d,t … yN′ ,d,t ]

(4)

where Yt∗ is the ki,d × 1 vector of endogenous variables for unit i , i = 1, …, N and d denotes the type of financial institutions examined for the restricted model. In addition, suppose that we run the model excluding one of the variables in the full endogenous vector, denoted as (ki−1) × 1. This form of the restricted model is obtained by the exclusion of the k-variable, and it can reveal the contribution of the omitted variable to the impulse response functions of the d-type restricted model. The restricted vector to be estimated in Eq. (4) is given as:

Y k∗,t = [y1,′ d(k,ti− 1) … yN′ (,kdi,−t 1) ]

(5)

where Y k∗,t is now the (ki−1) × 1 vector of endogenous variables included in the restricted model setup for unit i , i = 1, …, N and (ki−1) . A panel VAR is similar to large-scale VARs (e.g. spatial VARs or global VARs) where dynamic and static interdependencies are allowed for, but no consideration is given to the existence of panel dimensionality. Thus, all variables are treated symmetrically regardless of whether they belong to a unit or not, and of whether they measure the same quantity in different units or not. When the existence of a cross sectional dimension is omitted, it may not be damaging in terms of forecasting since dimensionality shrinkage is more important than the way of how it is implemented. However, cross sectional heterogeneity imposes a structure on the covariance matrix of the error terms which is generally disregarded in the policy exercises (see Canova & Ciccarelli, 2013). Using a cousin model of spatial VARs, it is strictly assumed that a shock originated in one unit can be transmitted after one period to another unit if these units are neighbors; if they are not neighbors, delayed effects are longer depending on how many neighbors are between them (Anselin, 2010). On the other hand, the use of other cousin models, such as global VAR (GVAR) which includes global variables (Canova & Ciccarelli, 2013; Dragomirescu-Gaina & Philippas, 2015), is a restricted large-scale VAR, where variables of different units in an equation are weighted. Since the weights are unit-specific and a-priori determined, a GVAR imposes a particular structure on the interdependencies. In this context, there are two main advantages using the panel VAR approach when addressing the research question associated with this study. First, we obtain UK financial institutions’ dynamic responses to shocks because of the model’s ability to approximate complicated, interdependent adjustment paths with the time-series information. Panel VARs are particularly suited to analyzing the transmission of idiosyncratic shocks across units and time (Canova, Ciccarelli, & Ortega, 2012; Love & Zicchino, 2006). In addition, we can control for individual heterogeneity and specify the time-varying relationships between dependent and independent variables. Panel VARs have also been frequently used to construct time-varying effects across heterogeneous groups of units (see the survey of Canova and Ciccarelli (2013)).

2 Within this context, if the data-generating process features dynamic heterogeneity, both a within and a between estimator will give inconsistent estimates of the parameters, even when N is large and T is small, since the error term is also likely to be correlated with the endogenous regressors.

4

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

3. Data selection A major part of our sample comes from the Bankscope database and covers the annual accounting data of the financial institutions in the UK for a period spanning from 2005 to 2013. However, to ensure potential uniformity, which can be affected by the presence of missing data in Bankscope, in some cases we use the annual reports of the financial institutions for the variables of interest as data sources. Higher frequency (e.g. quarterly frequency) could, in principle, give a better insight into the link between the accounting ratios and the QE rounds. However, quarterly data are not available for most financial institutions. We follow Gambacorta (2005) who noted that the bias in the results obtained using annual data instead of quarterly data appears not to be significant. The time span structure was chosen to capture QE rounds and transformations observed in the UK financial sector in recent years. In the period preceding the crisis, UK financial institutions increasingly came to depend on wholesale funding rather than their customers’ deposits, an element that placed greater pressure on their structure. Expansionary monetary policies in the UK started on October 8th 2008, when the Monetary Policy Committee (MPC) lowered the bank rate to 4.5% with further reductions taking place in the subsequent months (e.g. on October 6th 2008 it reached 3%, on December 4th 2008 it fell to 2%, on January 8th 2009 it reached 1.5% and, on February 5th 2009 it fell to 1%). The last reduction was to 0.5%, being decided on March 5th 2009, the date when unconventional monetary policies were officially implemented in the UK. During the first QE program, a value of £75 billion in assets were purchased from the BoE, splitting them between two auctions: (a) 5–10 years and (b) 10–25 years. On March 19th 2009, the Asset Purchase Facility (APF) conducted its first purchases of corporate bonds. On May 7th 2009, it was announced that QE purchases would be extended by £50 billion to £175 billion. The buying range was extended to all conventional gilts with a residual maturity greater than three years. The second QE program was launched on October 6th 2011. The MPC announced the extension of QE asset purchases by £75 billion to £275 billion that would be financed by reserve issuance, whereas the ceiling on private assets held would remain to £50 billion. On July 5th 2012, the BoE announced an asset purchase up to £375 billion in total. Finally, on July 13th 2012, the Funding for Lending Scheme (FLS) was launched, which meant that the BoE would subsidize bank funding to increase lending. During the first and second QE programs spanning from March 2009 to November 2012, the BoE purchased £375 billion of medium- and long-term government gilts (representing approximately 24% of the domestic GDP). As a result, the balance sheet of the UK financial institutions was significantly expanded due to the liquidity support. On the brink of the financial crisis in the UK, financial institutions ended up having less capital and fewer liquid assets than in the past, given the fluctuations in the UK’s financial environment. Thus, our time span structure can evaluate the overall impact of QE on the UK financial sector without segregating the impact of different QE rounds. We draw on the accounting quantities, which are associated with the present research study. The first quantity straightforwardly derived from Bankscope is the returns on assets (hereafter ROA), used as a key ratio for the evaluation of profitability and as a measurement of the overall performance of a financial institution regarding its efficiency in utilizing assets to generate profits, given the structure of liabilities and equity (Athanasoglou, Brissimis, & Delis, 2008). Next, we drawn on three quantities derived from Bankscope, namely the liquid assets, defined as the sum of cash and cash equivalents, public securities and secured short-term loans, the gross loans as the total amount of issued credits and, the sum of securities, defined as the sum of investments of banks that include bonds, equity derivatives and any other type of securities. We divide all the three quantities by the total shareholders’ equity to derive a leverage decomposition into three components, denoted as liquid assets to equity, loans to equity and securities to equity, which reflect the extent to which the financial institutions are (de)leveraging within the QE framework effect.3 In the standard quantitative easing framework, it is common to assume that the central bank sets its policy interest rate taking into account real-economy variables, for example the real GDP, the output gap, the inflation deviation from the target and so on, when deciding on the amount of QE in which it will engage. In this context we draw on the real GDP derived from the BoE database4 and examine the extent to which it may have an impact on bank performance due to the fact that the demand for lending increases during cyclical upswings (Athanasoglou et al., 2008). Moreover, we use the lending rate, as the average long-term rate from the BoE, to examine the extent to which the lending between banks is decreasing. This choice of lending rate relies on the hypothesis that certain bank-specific characteristics (e.g., size, liquidity, short-term funding, cost-to-income proportion and capitalization) only influence the loan supply. Finally, and to nail down an indicator of the QE, we derive from the BoE database the average of the 12-months asset purchases made by the BoE, over its total assets which is commonly used in the empirical literature (Chen et al., 2012).5 Using the Bankscope database, the types of financial institutions are not always mutually exclusive (Bhattacharya, 2003). Consequently, we restrict our sample to five main types of financial institutions in the UK, which are mutually exclusive. Even though the analysis is implemented on a total sample of more than 300 financial institutions, the contribution of each type of financial institution to the QE responses is investigated further, given that each type may reveal significant information. However, due to the data availability and the low relevance of some financial institutions to QE practices, the empirical analysis is focused on five major types. Table 1 presents the types and the number of financial institutions included over the period studied.

3 The decomposition of leverage is in line with the multiplier leverage, which measures the risk associated with non-capital funding of overall balance sheets, defined as total assets to total shareholders’ equity and subordinated debt (Bordeleau, Crawford, & Graham, 2009) and it is not subject to the model and measurement errors associated with asset risk calculations. This definition is also similar to the regulatory leverage ratio used by the Office of the Superintendent of Financial Institutions (OSFI) and it is based on total regulatory capital as defined in Basel II. 4 Source: http://www.bankofengland.co.uk/boeapps/iadb/. 5 The Bank of England is also engaged in other types of QE activities, e.g. the MBS purchases to MBS issuance and also MBS outstanding, addressed by Hancock and Passmore (2011), which is beyond the scope of this paper.

5

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Table 1 UK financial institutions. Type of financial institution

Number of financial institutions

Commercial banks Private banking and asset management companies Real estate and mortgage banks Investment banks Bank holding companies Total

76 32 43 42 20 213

Note: The table presents the types and the number of UK financial institutions included over the period studied. Bankscope divides financial institutions by specialization as follows: commercial banks, savings banks, investment banks, real estate and mortgage banks, cooperative banks, credit banks, Islamic banks, non-banking credit institutions, bank holding companies, central banks, specialized governmental credit institutions and multilateral government banks. In terms of the distinctions between the five different types presented in the table, commercial banks are regarded as financial institutions that are owned by stockholders pursuing various lending activities to increase their profits. Real estate and mortgage banks specialize in real estate lending. Investment banks are underwriters that serve as intermediaries between issuers of securities and the investing public. Private banking and asset management companies focus on the management of clients’ current investments. Finally, bank holding companies own or control one or more banks.

Table 2 Summary statistics of ratios for UK financial institutions.

Loans to equity Liquid assets to equity Securities to equity ROA

Commercial banks

Investment banks

Bank holding companies

Private banking companies

Real estate banks

Mean

St.dev.

Mean

St.dev.

Mean

St.dev.

Mean

St.dev.

Mean

St.dev.

5.89 4.99 2.62 0.68

6.01 4.39 3.41 1.99

3.72 3.34 3.80 1.13

3.95 4.58 4.79 5.15

6.92 4.55 3.56 1.94

6.91 4.49 3.83 3.66

4.90 11.87 2.70 0.83

5.85 11.48 3.38 2.16

13.56 2.28 1.95 0.39

4.52 1.58 1.58 0.37

Note: The table presents the summary statistics of the ratios of interest for UK financial institutions, namely loans to equity, liquid assets to equity and securities to equity, and the ROA. The table illustrates the mean and the standard deviation of the UK financial institutions by type.

Next, we rely on some statistical analysis to provide insights that can further motivate our empirical analysis. The findings here are not decisive for the main conclusions of the paper, but they offer a preliminary perspective of the data. Table 2 illustrates the mean and the standard deviation for the variables of interest by the type of UK financial institution. The idea behind this table is to examine whether the types of UK financial institutions with comparable averages have heterogeneous deviations from the mean. Comparing the results suggested in Table 2, we obtain some interesting findings. Firstly, there is a comparable (or close to) mean value between the types of financial institutions, although their deviations are highly heterogeneous. This suggests that distinguishing financial institutions by type and examining their partial contribution to the QE programme can play a key role, because they are all quite sensitive to unconventional shocks but differ in their degree of sensitivity. Moreover, the heterogeneity of the leverage’s component of liquidity across the types of financial institutions indicates short-term liquidity stress. To provide further insights into the contribution of the leverage’s components that can motivate the distinguishing of financial institutions by type, we derive histograms of the three components of leverage for the five types of UK financial institutions, as shown in Fig. 1. The findings indicate strong evidence of heterogeneity between the different types of financial institutions across the components, indicating the handling of different processes for each type of financial institution, an element that is robust to our hypothesis not to consider all financial institutions within the same modelling framework. This adjustment is in line with the UK HM Treasury’s report in 2012, even though these measures are planned to enter into force in 2019 and therefore the effects will only become visible later on. When reviewing loans to equity, the majority of financial institutions have values below 10%, while there are outliers in all the types with values that exceed 20%. This implies that they promote a very aggressive growth strategy accompanied by a correspondingly increased insolvency risk. In the case of liquid assets to equity, there is evidence of a high value crossing the 40% level for a few cases of investment banks, real estate banks and bank holding companies, indicating that they have high-quality liquid assets that can be converted easily and immediately into cash. This fact can be confirmed by the results obtained for securities to equity where, for these institutions, a high value might have been registered, meaning that they deal with creditworthy securities with short-term maturities. The issues of liquidity and leverage suggest of adopting a tighter regulation with the purpose of withstanding new stress scenarios, which make the financial system more resilient to the major risks that placed pressure on the performance of UK financial institutions, such as the economic downturn, borrower defaults, pressures in funding markets, credit conditions and sovereign risk. At the 6

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Fig. 1. Histograms by type of financial institution. Note: The figure provides histograms of loans to equity, liquid assets to equity and securities to equity, for all the types of UK financial institutions.

minimum, the conditions for achieving this objective are higher spreads on lending activities and reduced leverage. Achieving these goals would imply a rebalancing of the financial institutions’ funding profiles and a more focused approach on the activities that exploit their comparative advantage. In reality the transition determined a trade-off between deleveraging and revenue generation. Though, as shown in Fig. 1, this regulatory framework had an impact, particularly on commercial banks and bank holding companies, a large number of the institutions ensured a minimal level of liquidity. 4. Empirical findings under the model setup In this section, we present the empirical results with a relevance significance and discuss the implications, derived from the panel VAR model framework. We start by selecting the optimal lag length for the panel VAR framework, using MMSC for the GMM models based on the J statistic of over-identifying restrictions (Andrews & Lu, 2001). The first-order lag specification is chosen to ensure no serial correlation of residuals in the VARX models after estimating the model. Finally, we bear in mind that, when computing the bootstrapped error bands by simulating the model, we use the sample covariance matrix, since the number of endogenous variables in our model is lower than the dimension of the time series included. In this setting, our panel VAR framework is repetitively estimated for all types6 and the d-type of UK financial institutions, as shown in Table 3. Our results indicate some important findings. Looking on the group of macroeconomic variables, we observe that GDP growth is mainly driven by lending rate and the QE variable, along with its own lag values. For the cases of commercial banks and bank holding companies, the security holdings variable has a significant effect on GDP growth, while there is a negative relationship between real estate banks’ profitability and economic growth. A significant reduction in QE is presented in case of a significant positive shock on real GDP growth and security holdings of commercial banks and bank holding companies. Lending rate is reduced significantly due to QE policies; however, it increases when economic activity and security holdings of commercial banks and bank holding companies are boost. Loan to equity ratio presents the higher persistence in case of real estate banks and commercial banks. The positive effect of lagged profitability on loan holdings is presented mainly on commercial banks, and to a lower degree on private banking companies. Cash holdings variable for investment banks is only affected from the QE policy while the lagged lending rates, security holdings and 6

Others studies focusing on bank network used LASSO methodology for high dimensional VARs (see for instance Demirer, Diebold, Liu, & Yilmaz, 2018).

7

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Table 3 Panel VAR – Results. All types of banks

Commercial banks

Investment banks

Real estate banks

Bank holding companies

Private banking companies

GDP growth (GDP_growth)t−1 (QE)t−1 (Lending Rate)t−1 (Loan_Equity)t−1 (Cash_Equity)t−1 (Sec_Equity)t−1 (ROA)t−1

0.526*** −0.039*** −1.125*** 0.009 −0.011 0.060* −0.003

0.537*** −0.037*** −1.083*** −0.024 0.022 0.134*** −0.046

0.587*** −0.062*** −1.545*** 0.048 −0.055 −0.042 −0.028

0.490*** −0.031** −0.925*** 0.029 −0.335 −0.255 −1.602***

0.490*** −0.024 −0.798 0.010 −0.169 0.181** 0.032

0.492*** −0.032 −1.023** 0.044 −0.013 0.062 0.246*

QE (GDP_growth)t−1 (QE)t−1 (Lending Rate)t−1 (Loan_Equity)t−1 (Cash_Equity)t−1 (Sec_Equity)t−1 (ROA)t−1

−6.299*** 0.902*** 1.186 −0.095 0.015 −0.459** 0.044

−6.339*** 0.882*** 0.792 0.0803862 −0.338 −0.863** 0.405

−6.563*** 0.98*** 3.506 −0.201 0.264 0.127 0.118

−6.096*** 0.849*** −0.220 −0.204 2.341 1.325 11.09***

−6.181*** 0.819*** −0.521 −0.111 0.780 −1.171** −0.069

−6.166*** 0.851*** 0.363*** −0.331 0.130 −0.574 −1.704

Lending rate (GDP_growth)t−1 (QE)t−1 (Lending Rate)t−1 (Loan_Equity)t−1 (Cash_Equity)t−1 (Sec_Equity)t−1 (ROA)t−1

0.420*** −0.028*** 0.159** 0.005 −0.002 0.042** −0.008

0.425*** −0.026*** 0.187 −0.001 0.023 0.080*** −0.011

0.430*** −0.037*** −0.007 0.010 −0.029 −0.017 −0.019

0.402*** −0.024*** 0.257 0.012 −0.174 −0.144 −0.898***

0.410*** −0.020 0.311 0.001 −0.064 0.117** 0.005

0.411*** −0.023** 0.224 0.026 −0.006 0.058 0.122

Loan to equity (GDP_growth)t−1 (QE)t−1 (Lending Rate)t−1 (Loan_Equity)t−1 (Cash_Equity)t−1 (Sec_Equity)t−1 (ROA)t−1

−0.047 0.012 0.371** 0.595*** −0.005 0.025 0.032**

−0.001 −0.008 0.001 0.623*** −0.001 −0.012 0.178**

−0.033 0.018 0.444 0.407** −0.088 0.143 0.025

−0.068 0.018 0.463 0.885*** 0.412 −0.167 0.350

−0.011 −0.004 0.172 0.480* 0.091 −0.223 −0.009

−0.155 0.025 0.513 0.399*** 0.032 0.001 0.322*

Cash to equity (GDP_growth)t−1 (QE)t−1 (Lending Rate)t−1 (Loan_Equity)t−1 (Cash_Equity)t−1 (Sec_Equity)t−1 (ROA)t−1

0.048 0.009 0.191 0.066 0.585*** 0.206 0.040*

0.002 0.007 0.246 0.022 0.699*** 0.106 0.034

−0.132 0.085*** 1.461** −0.141 0.411*** 0.555** 0.041

−0.030 0.008 0.134 0.075 0.453*** −0.150 −0.280

0.087 0.014 0.263 0.069 0.648*** −0.178 −0.009

0.404 −0.092 −1.932** 0.289** 0.551*** −0.224 0.081

Securities to equity (GDP_growth)t−1 (QE)t−1 (Lending Rate)t−1 (Loan_Equity)t−1 (Cash_Equity)t−1 (Sec_Equity)t−1 (ROA)t−1

−0.042 −0.004 0.088 −0.002 −0.006 0.434*** −0.011

−0.089 0.006 0.239 −0.013 0.002 0.383*** 0.050

−0.055 0.001 0.375 0.068 −0.149** 0.386 −0.034

0.013 −0.007 −0.042 −0.029 0.151* 0.890*** 0.285

−0.134 −0.016 −0.087 −0.070 0.119 0.117 −0.001

−0.003 −0.034* −0.432 −0.001 0.054 0.551*** 0.128

ROA (GDP_growth)t−1 (QE)t−1 (Lending Rate)t−1 (Loan_Equity)t−1 (Cash_Equity)t−1 (Sec_Equity)t−1 (ROA)t−1

0.060 0.001 0.014 0.036 0.038 −0.079 0.276***

0.132* −0.017 −0.374 −0.010 0.036 0.007 0.683***

−0.069 0.058 0.922 0.188 0.101 −0.137 0.264*

0.030* −0.004 −0.089 −0.003 0.016 0.017 0.664***

0.140 −0.001 0.226 −0.047 −0.073 0.106 0.157*

0.064 0.001 0.023 0.060* 0.014 −0.110 0.746***

Robust 1167 183 Yes

Robust 458 69 Yes

Robust 180 32 Yes

Robust 275 41 Yes

Robust 98 16 Yes

Robust 156 25 Yes

Diagnostics GMM weight matrix: No. of obs No. of panels PVAR satisfies stability conditions

(continued on next page)

8

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Table 3 (continued)

J-Statistic

All types of banks

Commercial banks

Investment banks

Real estate banks

Bank holding companies

Private banking companies

21.33

17.69

14.46

12.46

13.41

21.21

Note: The table presents the estimated coefficients of the panel VAR framework for all the types of the UK financial institutions. The first column presents the variable of interest in the panel VAR along with the lag explanatory variables. The panel in the bottom of the table of results presents the diagnostic tests (similar to the study of Holtz-Eakin et al. (1988)). In particular, we include the J statistic that provides evidence for the validity of the over-identifying restrictions of the model, the number of observation and panels included, and the eigenvalues for stability condition. The eigenvalues lie inside the unit root circle; therefore, the stability condition is satisfied for all the cases. One star (*), two stars (**) and three stars (***) denote that the coefficients are significant at the 10%, 5% and 1% levels, respectively.

cash holdings also play a significant role on cash holding management. The higher persistence on security holdings over time is shown in real estate banks and private banking companies. Theses two types of financial institutions, together with commercial banks, show the highest persistence on profitability. 4.1. Impulse-response analysis Under the model assumptions, our panel VAR framework is repetitively estimated for all types and the d-type of UK financial institutions with the analysis focusing on IRFs (one standard deviation). The cross-sectional interactions within the different types of financial institutions each time can reflect the extent to which the institutions are subject to QE imposed by the Bank of England. Analyzing the response of the financial sector to shocks resulting from the QE policy, it is implicitly assumed that the variables of interest respond within the period to the BoE QE policy. Finally, we simulate the model 5000 times to obtain confidence intervals and median estimates for the impulse responses. 4.1.1. Quantitative easing impulses and transmission to GDP growth We start the empirical analysis by setting the QE effect impulses and the transmission to the UK real GDP growth, as shown in Fig. 2 (panels A, B and C). In panels A and B, there is strong evidence that, a positive shock of QE leads to a significant reduction in the lending rate with beneficial effects on real GDP growth for all the cases of financial institutions. This finding is also in line with the study of Joyce and Spaltro (2014) who found that QE has small but significant impact on bank lending in the UK, especially for smaller banks. The second important finding is the evidence that, during the period of the positive shock of QE, the profitability (indicated by ROA) of commercial banks and real estate banks is reduced significantly, highlighting their role, compared with the others (see panel C, Fig. 2). This finding is in line with Mamatzakis and Bermpei (2016), who identified a reduction in the profitability of US banks during quantitative easing implementation by the Fed. However, we indicate that the financial institutions affected most in the UK are the commercial and real estate banks. The reduction in their net interest margin may be attributed to the reduction in the lending rate via QE policies. Interest income is a significant source of their total income; nevertheless, lower interest margins in accordance with low rates can be beneficial for the real economy when considering the effect of a positive shock of ROA on real GDP growth after

Fig. 2. Impulses of QE transmission to GDP. Note: Panel A and B presents the statistically significant responses of lending rates to a quantitative easing shock, and the responses of GDP growth to QE and lending rate shocks respectively for each type of UK financial institutions. The black line represents the median estimate of the response. The shadow grey area around the median estimate line of the response represents the statistically significant 95% confidence bands generated from 5000 Monte Carlo bootstrap resampling. The interested reader can find analytical results at the Supplementary file. Panel C presents the statistically significant responses of securities to equity, liquid to equity and ROA variables of interest to a quantitative easing shock. Moreover, the effect of these variables on real GDP growth is also, identified. The black line represents the median estimate of the response. The shadow grey area around the median estimate line of the response represents the statistically significant 95% confidence bands generated from 5000 Monte Carlo bootstrap resampling. We denote the leverage components, namely securities to equity and liquid assets to equity, as “Securities to equity” and “Liquid to equity”, respectively. The interested reader can find analytical results at the Supplementary file. 9

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Fig. 2. (continued)

one period for real estate banks. Another significant finding derived from panel C in Fig. 2 is the evidence that a positive shock of QE coexists with the significant increase in the securities to equity ratio, for commercial banks. The real GDP growth responds positively and significantly to a QE positive shock after one period. Therefore, the significant activity of this type of banks in terms of asset leverage may contribute to the UK real GDP growth. Moreover, the drop in liquid assets to equity ratio for private banking companies may contribute to the increase in the real GDP growth, given the response of the latter to a positive shock to the liquid assets to equity for these types of banks and bank holding companies. 4.1.2. The role of financial institutions’ variables in the GDP growth response and QE shock We further explore the role of financial institutions’ variables in the effects of quantitative easing on the real GDP growth and on profitability. To do so, we repetitively estimate our panel VAR framework, by excluding each time one relevant variable of interest and comparing the responses with the ones from the full baseline framework. Fig. 3 presents the results associated with this question. The main finding is that the leverage component securities to equity amplifies the effect of a QE shock on the real economic activity for bank holding companies and to a lesser degree for private banking companies and commercial banks. In the case of real estate banks, the effect of a QE shock on GDP growth is much less positive with the exclusion of the profitability (ROA) variable, underlying the importance of interest rate drop for these financial institutions with a crucial role on economic growth. In the majority of the cases, the ROA responses to a positive QE shock are negative with the exception of the private banking companies. However, when the securities to equity ratio is omitted, the ROA also responds in the same manner for this type of financial institution, following the others’ ROA response to QE. Therefore, the leverage component securities to equity is of great importance, providing a tool to the private banking companies to avoid experiencing a significant reduction in their profitability. In the case of the bank holding companies, the same leverage component has a beneficial effect by reducing the negative effect of QE on ROA. These results have significant implications for bank managers when facing a significant easing monetary policy. A well-diversified bank strategy to interest and non-interest income activities may reduce the negative effect of a QE strategy on bank profitability. 4.1.3. Does the QE policy respond to shocks of leverage and profitability? Fig. 4 shows the responses of the BoE QE transmission to leverage and profitability. The findings illustrate that the BoE reduces asset purchases when a positive shock in growth occurs and increases asset purchases when a positive lending rate shock takes place. Looking into the financial institutions’ variables, we observe a significant reduction of asset purchases as evidence after a positive shock to the leverage component securities to equity for commercial banks. The same finding holds for bank holding companies but with a lower level of statistical significance. Our results show that the response of the QE variable is positive after a positive shock to profitability for the real estate banks to reduce the lending rates and boost the economy, given the significant role of real estate banks in housing lending. 4.1.4. Does economic activity affect leverage and profitability? We address this question by testing the impulses of real GDP growth to leverage components and ROA. Fig. 5 illustrates the results of the IRFs for all types of financial institutions. The first main finding is that real GDP growth has a significant positive effect on real estate banks’ and bank holding companies’ profitability and to lesser extent on the profitability of commercial banks. This finding may be attributed to different business model followed by different types of banks. The second main finding is that a negative shock to the real GDP growth may increase the securities to equity for three out of five types of financial institutions, namely commercial banks, real estate banks and bank holding companies. Moreover, the leverage component loans to equity increases after a negative GDP growth shock for the real estate and commercial banks, adding to their leverage. The liquid assets to equity are reduced in the case 10

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Fig. 2. (continued)

of a negative GDP growth shock for commercial banks and bank holding companies, adding more to their risk profile, while for real estate banks it is increased, lowering their risk profile. These results imply that risks are undertaken when the economic conditions are worse. This is particularly the case for commercial banks and bank holding companies. By increasing their leverage, these institutions hope to resist a potential reduction in their profitability due to low economic activity. However, this may increase their risk significantly, given that poor conditions in the economic environment lead to losses. Even though the monetary authorities are afraid of deleverage over weak economic growth, they should take measures for bank capital adequacy due to a possible worsening of the economic conditions. 4.2. Forecast error variance decomposition analysis Table 4 presents the results of variance decomposition when including a specific type of financial institutions or including all together. The findings underline some important aspects of our analysis. We stress evidence that after five periods almost 5.4% of GDP growth variability, 8.9% of lending rate variability and 8.7% of QE variability explained by cash and security holdings of bank holding companies. In 5-period horizon, the commercial banks’ security holdings explain 2.8% of QE and 2.9% of lending rate variability respectively, highlighting the importance of these two types of banks on real economy. Another interesting finding is that the lending rate’s contribution to real GDP growth and QE variability is the highest for the model corresponding to investment banks (8.8% and 7.7%, respectively). Therefore, our results confirm that investment banks can also play a crucial role on BoE unconventional policies. We also document that after five period horizon, almost 10% of loan to equity ratio variability attributed to cash and security holdings management for real estate banks and investment banks, while GDP growth variability explains 8.2% of loan to equity ratio variability and 2.1% of security holdings variability for real estate banks. This later finding underlines the importance of economic activity on the functioning of this type of banks. Loans to equity ratio explains a significant percentage (23.9%) of cash holdings variability for real estate banks and this may be attributed to strict regulation that they face. For commercial banks and bank holding companies, GDP growth together with loans holdings is responsible for 9.4% and 26.4% of cash holdings variability, respectively. Cash management seems to be of crucial importance on security holdings for investment banks (32%–35% in Table 4) and to a lower degree on real estate banks (9.8%–16%). Moreover, it is worth to mention the 21.2% of variability on cash holdings and 12.9% of variability on profitability attributed to GDP growth conditions for bank holding companies. Finally, the 27% of real estate banks profitability attributed to macroeconomic policy factors and loan holdings, as it was expected based on their business model. 4.3. Discussion Up until to 2013, banking regulation in the UK involved three organizations, namely the Financial Services Authority (FSA) the Bank of England and the Treasury. The Financial Services and Markets Act 2000 introduce a substantial reform to the regulation of the UK financial services sector, by introducing a consolidation of responsibility for regulation and supervision of banks, building societies, insurance companies, and investment firms within a single body – the Financial Services Authority, which is based on a single statutory process with broadly harmonized prudential and regulatory requirements. However, real estate banks faced liquidity and funding restriction compared to other types of banks, given their significant contribution to the real economy and their specialized character. Moreover, our results indicate the significant effect of QE policy on their profitability (see IRFs and FEVD analysis) which is attributed to their restricted activity, mainly on lending. In 2007, the Butterfill Act reform enables real estate banks to borrow a greater proportion (up to 75%) of their funding from the wholesale markets and modified the relevant law so that, in the event of a building society insolvency, members' shares would rank equally with liabilities to creditors. The Banking Act of 2009 and the Review of Lord Turner (chairman of the FSA) lead banks to hold more assets and improve regulation of liquidity. The former was 11

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

(caption on next page)

12

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Fig. 3. Responses of the real GDP and ROA to a positive QE shock – Identifying the role of omitted variables for the types of UK financial institutions. Note: The figure presents the responses of the real GDP and ROA to a positive QE shock for 10 period-horizons ahead. The blue line with rhombuses represents the sample containing all financial institutions, the red line with squares represents the sample when securities to equity (denoted as Sec/ Equity) are excluded, the green line with triangles represents the sample when loans to equity (denoted as Loans/Equity) are excluded, the purple line with two-ray asterisks represents the sample when liquid assets to equity (denoted as Liquid/Equity) are excluded and the light blue line with three-ray asterisks represents the sample when the ROA is excluded. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4. The BoE QE policy response to leverage and profitability. Note: The figure presents the response functions of QE to all (statistically significant) types of macroeconomic and financial shocks. The black line represents the median estimate of the response. The grey shadow area around the median estimate line of the response represents the statistically significant 95% confidence bands generated from 5000 Monte Carlo bootstrap resamplings. We denote the leverage components, namely securities to equity, loans to equity and liquid assets to equity, as “Securities to equity”, “Loans to equity” and “Liquid to equity”, respectively. The interested reader can find analytical results at the Supplementary file.

clearly presented in our FEVD analysis for commercial banks and bank holding companies, where a significant part of security holdings variation is explained from the QE variable. Therefore, these two types of banks increase security holdings after a positive QE shock, as it can also be seen from the impulse response functions. The priority of the Lord Turner’s review was to ensure that in future financial activities are always regulated according to their economic substance and not because of their legal form. More complex and risky investment banking activities require more liquidity holding. The finding of increased variability of cash holdings explained by the QE variable for private investment banks is on the above direction. According to the Financial Services Act 2012, the financial institutions which typically demonstrate a high degree of leverage, liquidity or maturity mismatch, or financial interconnectedness can transmit, and often amplify, shocks arising elsewhere in the financial system. However, not all types of financial institutions will have the same capacity to either transmit contagion or offer 13

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Fig. 5. Effect of economic activity impulses on leverage and profitability. Note: The figure presents the profitability (ROA) responses (statistically significant) to a shock from the three components of leverage, across the different types of UK financial institutions, for 10 period-horizons ahead. The black line represents the median estimate of the response. The grey shadow area around the median estimate line of the response represents the statistically significant 95% confidence bands generated from 5000 Monte Carlo bootstrap resamplings. We denote the leverage components, namely securities to equity, loans to equity and liquid assets to equity, as “Securities to equity”, “Loans to equity” and “Liquid to equity”, respectively. The interested reader can find analytical results at the Supplementary file.

the same opportunity for arbitrage. Our findings support this argument. Nevertheless, the global financial crisis has shown us how swiftly and harshly credit risk and liquidity risk can crystallize, while some sources of funding can vanish, spawning concerns regarding the valuation of assets and capital adequacy. The issues of liquidity and leverage were among the reasons for regulators to adopt a tighter regulation and supervisory practice with the purpose of withstanding new stress scenarios. The regulatory steps on liquidity have been taken at the European level through the Capital Requirements Regulation (CRR) and the Capital Requirements Directive (CRDIV). These new regulations aim to make the financial system more resilient to the major risks that put pressure on the performance of the UK financial institutions such as the economic downturn, borrower defaults, pressures in funding markets, credit conditions and sovereign risk. Moreover, the new regulation framework implies a rebalancing of the financial institutions’ funding profiles and a more focused approach on the activities that exploit their comparative advantage (e.g. ring-fencing banks). By providing banks with sufficient regulatory buffers, this approach counters the risk of swiftly winding down positions during contractionary phases, therefore lessening fire-sale spirals, which could eventually determine a credit crunch via depleted capital levels. The regulatory instruments, such as the capital-based or liquidity instruments, can also be useful to address externalities referring to systemic risk, arising from banks’ interconnectedness. The low interest rate environment observed in the last years has potentially amplified the value of liquid assets held by financial institutions. Generally, a shift in central bank’s monetary policy is deliberately targeting an increase in the overall interest rates (effect that might be further amplified by an increase in the refinancing rates). Thus, an increase in interest rates might further impact the liquidity buffer via revaluation of the assets. However, our analysis highlighted that there might be negative externalities of QE policies in a new regulatory framework on bank profitability especially for commercial banks and real estate banks. While the regulatory framework took initially a universal perspective, applicable to all financial institutions, its impact differed based on banks’ specific business models. More specifically, this new regulatory trend has a particular importance for commercial banks and bank holding companies. As such, more diversified business models inclined to be more adapted to capital and liquidity requirements compared to specialized banks. Moreover, based on our findings the focus of the BoE on these two types of banks may signal the reduction of such unconventional 14

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Table 4 Variance decomposition. Response variable

GDP growth All types of banks Commercial banks Investment banks Real estate banks Bank holding companies Private banking companies QE All types of banks Commercial banks Investment banks Real estate banks Bank holding companies Private banking companies Lending rate All types of banks Commercial banks Investment banks Real estate banks Bank holding companies Private banking companies Loans to equity All types of banks Commercial banks Investment banks Real estate banks Bank holding companies Private banking companies Cash to equity All types of banks Commercial banks Investment banks Real estate banks Bank holding companies

Forecast horizon

Impulse Variable GDP growth

QE

Lending rate

Loans to equity

Cash to equity

Securities to equity

ROA

1 5 1 5 1 5 1 5 1 5 1 5

100.0% 95.1% 100.0% 94.1% 100.0% 89.6% 100.0% 90.8% 100.0% 90.6% 100.0% 94.1%

0.0% 0.5% 0.0% 0.8% 0.0% 0.3% 0.0% 1.5% 0.0% 0.7% 0.0% 0.8%

0.0% 3.8% 0.0% 3.2% 0.0% 8.8% 0.0% 1.8% 0.0% 2.8% 0.0% 2.0%

0.0% 0.1% 0.0% 0.1% 0.0% 0.1% 0.0% 0.1% 0.0% 0.3% 0.0% 2.9%

0.0% 0.1% 0.0% 0.0% 0.0% 0.2% 0.0% 0.2% 0.0% 2.9% 0.0% 0.3%

0.0% 0.3% 0.0% 1.6% 0.0% 0.5% 0.0% 0.4% 0.0% 2.5% 0.0% 0.7%

0.0% 0.0% 0.0% 0.2% 0.0% 0.5% 0.0% 5.1% 0.0% 0.2% 0.0% 1.7%

1 5 1 5 1 5 1 5 1 5 1 5

84.3% 92.3% 84.4% 91.8% 83.9% 85.7% 84.3% 83.4% 84.8% 85.5% 85.2% 89.9%

15.7% 4.4% 15.6% 3.5% 16.1% 5.5% 15.7% 2.9% 15.2% 3.8% 14.8% 3.5%

0.0% 2.3% 0.0% 1.3% 0.0% 7.7% 0.0% 0.2% 0.0% 1.7% 0.0% 0.6%

0.0% 0.2% 0.0% 0.1% 0.0% 0.2% 0.0% 0.3% 0.0% 0.1% 0.0% 0.8%

0.0% 0.2% 0.0% 0.2% 0.0% 0.1% 0.0% 0.5% 0.0% 4.5% 0.0% 0.4%

0.0% 0.7% 0.0% 2.8% 0.0% 0.6% 0.0% 0.7% 0.0% 4.2% 0.0% 0.3%

0.0% 0.0% 0.0% 0.4% 0.0% 0.2% 0.0% 12.0% 0.0% 0.2% 0.0% 4.6%

1 5 1 5 1 5 1 5 1 5 1 5

88.1% 92.2% 87.8% 91.1% 87.3% 86.0% 87.9% 82.3% 88.0% 84.9% 90.3% 89.9%

5.5% 2.8% 5.7% 2.4% 5.6% 3.5% 5.8% 2.1% 5.6% 2.4% 4.3% 1.9%

6.4% 3.9% 6.5% 3.1% 7.1% 9.0% 6.3% 2.0% 6.3% 3.5% 5.4% 2.3%

0.0% 0.2% 0.0% 0.1% 0.0% 0.1% 0.0% 0.2% 0.0% 0.1% 0.0% 0.8%

0.0% 0.2% 0.0% 0.1% 0.0% 0.1% 0.0% 0.3% 0.0% 4.3% 0.0% 0.3%

0.0% 0.7% 0.0% 2.9% 0.0% 0.8% 0.0% 1.1% 0.0% 4.6% 0.0% 0.5%

0.0% 0.0% 0.0% 0.2% 0.0% 0.4% 0.0% 12.0% 0.0% 0.2% 0.0% 4.2%

1 5 1 5 1 5 1 5 1 5 1 5

0.9% 3.0% 1.4% 3.8% 0.2% 2.5% 2.1% 1.6% 2.2% 8.2% 0.1% 0.5%

0.3% 0.2% 0.1% 0.2% 0.3% 0.3% 0.4% 0.4% 0.3% 0.9% 1.7% 1.6%

0.0% 0.4% 0.2% 0.5% 0.2% 1.2% 0.1% 1.1% 0.0% 0.1% 0.0% 1.1%

98.7% 96.1% 98.3% 92.4% 99.3% 84.9% 97.4% 87.1% 97.4% 88.0% 98.2% 91.7%

0.0% 0.1% 0.0% 0.0% 0.0% 8.4% 0.0% 6.6% 0.0% 0.7% 0.0% 1.2%

0.0% 0.1% 0.0% 0.0% 0.0% 2.4% 0.0% 3.0% 0.0% 2.1% 0.0% 0.6%

0.0% 0.1% 0.0% 2.9% 0.0% 0.2% 0.0% 0.1% 0.0% 0.0% 0.0% 3.4%

1 5 1 5 1 5 1 5 1 5

0.2% 1.1% 2.1% 7.0% 0.1% 2.0% 1.8% 2.7% 5.3% 17.5%

0.0% 0.0% 0.9% 0.8% 0.6% 0.7% 0.5% 1.6% 0.0% 0.1%

0.1% 0.2% 0.9% 1.4% 0.0% 4.4% 0.6% 1.3% 0.3% 1.0%

1.3% 4.9% 1.2% 2.4% 0.2% 0.9% 12.2% 23.9% 5.2% 7.9%

98.4% 89.5% 94.9% 86.7% 99.1% 70.3% 84.8% 63.6% 89.2% 69.6%

0.0% 4.1% 0.0% 1.6% 0.0% 21.3% 0.0% 5.4% 0.0% 3.8%

0.0% 0.1% 0.0% 0.2% 0.0% 0.4% 0.0% 1.5% 0.0% 0.0%

(continued on next page)

15

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Table 4 (continued) Response variable

Private banking companies Securities to equity All types of banks Commercial banks Investment banks Real estate banks Bank holding companies Private banking companies ROA All types of banks Commercial banks Investment banks Real estate banks Bank holding companies Private banking companies

Forecast horizon

Impulse Variable GDP growth

QE

Lending rate

Loans to equity

Cash to equity

Securities to equity

ROA

1 5

0.0% 2.4%

5.1% 3.9%

0.0% 1.7%

0.2% 4.5%

94.7% 84.7%

0.0% 2.7%

0.0% 0.2%

1 5 1 5 1 5 1 5 1 5 1 5

1.6% 4.5% 2.8% 4.3% 0.1% 5.4% 7.4% 7.7% 16.3% 21.2% 0.9% 6.2%

0.4% 0.4% 1.3% 1.1% 0.1% 0.4% 0.1% 1.7% 2.0% 1.9% 0.1% 1.0%

0.0% 0.1% 0.0% 0.2% 0.4% 1.1% 0.0% 0.4% 0.7% 0.7% 0.2% 1.2%

5.3% 5.1% 7.4% 7.0% 6.5% 6.3% 5.0% 3.1% 7.5% 7.1% 0.0% 0.6%

9.1% 9.1% 4.5% 4.4% 32.9% 35.8% 16.8% 9.8% 3.4% 3.9% 9.6% 7.3%

83.6% 80.8% 84.0% 82.8% 60.0% 50.6% 70.7% 76.2% 70.1% 65.2% 89.2% 81.9%

0.0% 0.0% 0.0% 0.2% 0.0% 0.4% 0.0% 1.2% 0.0% 0.0% 0.0% 1.8%

1 5 1 5 1 5 1 5 1 5 1 5

0.5% 1.3% 0.0% 0.5% 0.1% 1.2% 11.5% 10.5% 7.5% 12.9% 0.3% 1.8%

0.1% 0.1% 1.1% 0.8% 0.0% 0.2% 5.9% 5.4% 0.1% 0.2% 1.1% 1.5%

0.0% 0.0% 0.6% 3.5% 0.3% 1.3% 0.0% 2.6% 0.2% 0.4% 3.4% 3.0%

0.0% 0.2% 0.4% 0.5% 0.4% 1.2% 9.0% 8.7% 0.1% 0.4% 0.2% 4.4%

0.7% 1.1% 0.8% 1.0% 1.0% 1.9% 2.4% 2.1% 0.7% 1.7% 0.2% 2.6%

0.1% 0.8% 0.2% 0.1% 0.3% 1.1% 1.0% 0.8% 0.1% 1.3% 0.6% 15.0%

98.6% 96.5% 96.9% 93.5% 97.8% 93.1% 70.0% 70.0% 91.2% 83.2% 94.3% 71.7%

Note: The table presents the forecast error of variance decomposition (%) for different horizons (second column, horizons 1 and 5), which is derived from the panel VAR framework for all variables of interest (first row entitled “All types of banks”) and by type.

monetary policies. Consequently, this regulatory and supervisory transition determined a trade-off between deleveraging and revenue generation, as well as a trade-off between financial stability and the cost of financial intermediation. 5. Conclusion Considerable efforts have been made by the central banks in recent years to provide effectively a sufficient monetary stimulus to their economy during the recent global and domestic downturns and ensure the sound functioning of financial sectors. In the UK, the financial institutions are the main collectors of funds and suppliers to the non-financial and households’ sectors; therefore, a strong understanding of the UK financial institutions’ role during the implementation of the BoE QE strategy is vital, because it raises a series of concerns regarding the economic spin-off that could be triggered by these monetary policy decisions. The paper gauges how the different types of UK financial institutions’ leverage responded to the incentives determined by the QE decisions realized in BoE asset purchases, using a panel VAR framework. We find that QE decisions are driven mainly by real economic activity, lending rates and to a diverging degree the disaggregated leverage with different effects on the five main types of UK financial institutions. The findings highlight the crucial role played by commercial banks in explaining these interrelationships. When the BoE proceeds to instigate a positive shock to asset purchases, the financial institutions’ profitability is significantly reduced. Turning to the relationship between an unconventional monetary policy and the financial institutions’ leverage, we find that QE rounds seem to have a positive effect on the leverage components, particular to security holdings to equity for commercial banks, implying riskier behaviour during QE rounds for boosting the real economy. The quantitative easing policies aim to increase the money supply by inundating financial institutions with capital in a struggle to encourage lending and implicitly liquidity. Our paper shows that, during the implementation of the QE strategy, the overall leverage of the banking sector is increased. This implies a signal of credit easing conditions that disappeared during the involvement of the financial crisis. The decrease in banks’ profitability implied negative signals from the financial sector to the monetary authorities to reduce unconventional easing strategies and assess financial stability, which is the main goal derived from these policies. Moreover, given the high uncertainty and low interest rates, the heightened risk-taking behaviour of financial institutions as a response to a possible restraint on their policy choices can be observed. This pro-risk attitude has high potential to influence the market price of risk in the economic system. Likewise, a higher level of risk affects the financial sector’s stability and soundness, particularly if the additional risk is condensed in systemically important financial institutions. As a result, these issues accentuate the policy makers’ concerns related to the limitation of financial institutions’ risk-taking behaviour. 16

North American Journal of Economics and Finance xxx (xxxx) xxx–xxx

D. Philippas et al.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.najef.2018.04. 014. References Adrian, T., & Hyun, S. S. (2010). Liquidity and leverage. Journal of Financial Intermediation, 19, 418–437. Andrews, D. W. K., & Lu, B. (2001). Consistent model and moment selection procedures for GMM estimation with application to dynamic panel data models. Journal of Econometrics, 101, 123–164. Anselin, L. (2010). Thirty years of spatial econometrics. Papers in Regional Science, 89, 3–25. Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies, 58, 277–297. Athanasoglou, P. P., Brissimis, S. N., & Delis, M. D. (2008). Bank-specific, industry-specific and macroeconomic determinants of bank profitability. Journal of International Financial Markets, Institutions and Money, 18, 121–136. Beetsma, R., & Giuliodori, M. (2011). The effects of government purchase shocks: Review and estimates for the EU. Economic Journal, 121, F4–F32. Bhattacharya, K. (2003). How good is the Bankscope database? A cross-validation exercise with correction factors for market concentration measures. BIS working paper, no. 133. Bordeleau, E., Crawford, A., & Graham, C. (2009). Regulatory constraints on bank leverage: Issues and lessons from the Canadian experience. Bank of Canada Discussion Paper No 2009-15. Borio, C., & Haibin, Z. (2012). Capital regulation, risk-taking and monetary policy: A missing link in the transmission mechanism? Journal of Financial Stability, 8, 236–251. Bowman, D., Cai, F., Davies, S., & Kamin, S. (2015). Quantitative easing and bank lending: Evidence from Japan. Journal of International Money and Finance, 57, 15–30. Canova, F., & Ciccarelli, M. (2013). Panel vector autoregressive models: a surveyWorking paper series 1507. European Central Bank. Canova, F., Ciccarelli, M., & Ortega, E. (2012). Do institutional changes affect business cycles? Journal of Economic Dynamics and Control, 36, 1520–1533. Chen, H., Cúrdia, V., & Ferrero, A. (2012). The macroeconomic effects of large-scale asset purchase programmes. Economic Journal, 122, F289–F315. Christensen, J. H. E., & Rudebusch, G. D. (2012). The response of interest rates to US and UK quantitative easing. Economic Journal, 122, F385–F414. Chung, H., Laforte, J. P., Reifschneider, D., & Williams, J. C. (2012). Have we underestimated the likelihood and severity of zero lower bound events? Journal of Money, Credit and Banking, 44, 47–82. Ciccarelli, M., Maddaloni, A., & Peydró, J. L. (2013). Heterogeneous transmission mechanism: Monetary policy and financial fragility in the Eurozone. Economic Policy, 28, 459–512. D’Amico, S., English, W., López-Salido, D., & Nelson, E. (2012). The Federal Reserve’s large-scale asset purchase programmes: Rationale and effects. Economic Journal, 122, F415–F446. Delis, M. D., Iftekhar, H., & Mylonidis, N. (2017). The risk-taking channel of monetary policy in the US: Evidence from corporate loan data. Journal of Money. Credit and Banking, 49, 187–213. Delis, M. D., & Kouretas, G. (2011). Interest rates and bank risk-taking. Journal of Banking & Finance, 35, 840–855. Demirer, M., Diebold, F., Liu, L., & Yilmaz, K. (2018). Estimating global bank network connectedness. Journal of Applied Econometrics, 33, 1–15. Diamond, D. W., & Rajan, Raghuram G. (2012). Illiquid banks, financial stability, and interest rate policy. Journal of Political Economy, 120, 552–591. Dragomirescu-Gaina, C., & Philippas, D. (2015). Strategic interactions of fiscal policies in Europe: A global VAR perspective. Journal of International Money and Finance, 59, 49–76. Eser, F., & Schwaab, B. (2016). Evaluating the impact of unconventional monetary policy measures: Empirical evidence from the ECB’s securities markets programme. Journal of Financial Economics, 119, 147–167. Farhi, E., & Tirole, J. (2012). Collective moral hazard, maturity mismatch, and systemic bailouts. American Economic Review, 102, 60–93 https://www.aeaweb.org/ articles?id=10.1257/aer.102.1.60. Frame, W. S., Hancock, D., & Passmore, W. (2012). Federal home loan bank advances and commercial bank portfolio composition. Journal of Money, Credit and Banking, 44, 661–684. Gambacorta, L. (2005). Inside the bank lending channel. European Economic Review, 49, 1737–1759. Gilchrist, S., & Zakrajšek, E. (2013). The impact of the Federal Reserve’s large-scale asset purchase programs on corporate credit risk. Journal of Money, Credit and Banking, 45, 29–57. Hamilton, J. D., & Wu, J. C. (2012). The effectiveness of alternative monetary policy tools in a zero lower bound environment. Journal of Money, Credit and Banking, 44, 3–46. Hancock, D., & Passmore, W. (2011). Did the Federal Reserve’s MBS purchase program lower mortgage rates? Journal of Monetary Economics, 58, 498–514. Holtz-Eakin, D., Newey, W., & Rosen, H. S. (1988). Estimating vector autoregressions with panel data. Econometrica, 56, 1371–1395. Istiak, K., & Serletis, A. (2016). A note on leverage and the macroeconomy. Macroeconomic Dynamics, 20, 429–445. Joyce, M., & Spaltro, M. (2014). Quantitative easing and bank lending: A panel data approach. Bank of England working papers 504, Bank of England. Joyce, M., Miles, D., Scott, A., & Vayanos, D. (2012). Quantitative easing and unconventional monetary policy – An introduction. Economic Journal, 122, F271–F288. Krishnamurthy, A., & Vissing-Jorgensen, A. (2011). The effects of quantitative easing on interest rates: Channels and implications for policy. Brookings Papers of Economic Activity, 43, 215–287. Lambert, F., & Ueda, K. (2014). The effects of unconventional monetary policies on bank soundness. IMF working paper no. 14/152. Love, I., & Zicchino, L. (2006). Financial development and dynamic investment behaviour: Evidence from a Panel VAR. The Quarterly Review of Economic and Finance, 46, 190–210. Mamatzakis, E., & Bermpei, T. (2016). What is the effect of unconventional monetary policy on bank performance? Journal of International Money and Finance, 67, 239–263. Mamatzakis, E., Matousek, R., & Vu, A. N. (2015). What is the impact of bankrupt and restructured loans on Japanese bank efficiency? Journal of Banking and Finance, 72, S187–S202. Neely, C. J. (2015). Unconventional monetary policy had large international effects. Journal of Banking and Finance, 52, 101–111. Papadamou, S., & Siriopoulos, C. (2012). Banks' lending behavior and monetary policy: Evidence from Sweden. Review of Quantitative Finance and Accounting, 38, 131–148. Pesaran, M. H., & Smith, R. P. (2016). Counterfactual analysis in macroeconometrics: An empirical investigation into the effects of quantitative easing. Research in Economics, 70, 262–280. Putnam, B. H. (2013). Essential concepts necessary to consider when evaluating the efficiency of quantitative easing. Review of Financial Economics, 22, 1–7. Steeley, J. M. (2015). The side effects of quantitative easing: Evidence from the UK bond market. Journal of International Money and Finance, 51, 303–336. Weale, M., & Wieladek, T. (2016). What are the macroeconomic effects of asset purchases? Journal of Monetary Economics, 79, 81–93. Yang, Z., & Zhou, Y. (2013). Credit risk spillovers among financial institutions around the global credit crisis: Firm-level evidence. Management Science, 59, 2343–2359. Yang, Z., & Zhou, Y. (2017). Quantitative easing and volatility spillovers across countries and asset classes. Management Science, 63, 333–354.

17