Cyclicality of liquidity creation: Nonlinear evidence from US bank holding companies

• Published date: 01 December 2023
• Author: Ghulame Rubbaniy,Ali Awais Khalid,Shoaib Ali,Efstathios Polyzos
• Date: 01 December 2023
• DOI: http://doi.org/10.1111/jfir.12352

Received: 19 January 2022

|

Accepted: 5 September 2023

DOI: 10.1111/jfir.12352

ORIGINAL ARTICLE

Cyclicality of liquidity creation: Nonlinear

evidence from US bank holding companies

Ghulame Rubbaniy

1

|Ali Awais Khalid

2

|Shoaib Ali

3

|

Efstathios Polyzos

4

1

Bristol Business School, University of the

West of England, Bristol, UK

2

Lahore Business School, University of

Lahore, Lahore, Pakistan

3

Adnan Kassar School of Business, Lebanese

American University, Beirut, Lebanon

4

College of Business, Zayed University,

Abu Dhabi, UAE

Correspondence

Ghulame Rubbaniy, Bristol Business School,

University of the West ofEngland, BS16 1QY,

Bristol, UK.

Email: Ghulame.rubbaniy@uwe.ac.uk

Abstract

Using a panel smooth transition regression framework on a

new proxy of the business cycle (BC) index and quarterly data

of US bank holding companies from 1993Q1 to 2020Q1, our

results provide empirical support for the theory that the BC has

a nonlinear effect on liquidity creation. We find a positive and

highly significant nonlinear effect of the BC on liquidity

creation, which not only supports the pro‐cyclicality of liquidity

creation but also improves the liquidity creation estimation

compared to previous studies. The results are robust to

different proxies of the BC and model specifications. We also

document that US bank holding companies create liquidity

more during the expansion phase (normal times) than during

the recession phase (crisis times) of the BC, suggesting an

asymmetrical effect of BC changes on liquidity creation. Our

findings have important implications for financial market

participants by suggesting that banks should keep alternative

sources of funding on hand during the BC recession phase.

Insights from our study also provide policy implications for

central banks and prudent supervisors to consider when

incentivizing banks, for instance, by lowering regulatory

requirements, adjusting the policy rate, or implementing any

other quantitative easing policy during the BC recession phase

to keep the financial system efficient.

JEL CLASSIFICATION

G21, O40, O43

J Financ Res. 2023;46:1165–1185. wileyonlinelibrary.com/journal/JFIR

|

1165

© 2023 The Southern Finance Association and the Southwestern Finance Association.

1|INTRODUCTION

The primary function of banks is to create liquidity in the financial market (Allen et al., 2008; Berger &

Bouwman, 2015) through extending loans to market participants or investing in financial market securities. These

consumer loans are the illiquid assets on the balance sheets of the banks that require liquidity to finance these

illiquid assets. Traditionally, banks create liquidity through their liquid liabilities (demand deposits) (Bryant, 1980;

Gorton & Winton, 2017), asset sales, or equity issuances. However, in modern times banks also generate liquidity

through off‐balance‐sheet items (e.g., standby letter of credit, loan commitments, and guarantees) to help firms

operating in financial markets achieve their strategic goals (Berger & Bouwman, 2009; Holmström & Tirole, 1998).

Consequently, banks hold illiquid assets and loan commitments to foster the economy.

The global financial crisis (GFC) in 2008 provides substantial evidence on long‐lasting consequences of a quick

liquidity vaporization (Abbas et al., 2021;Diaz&Huang,2017). Since the onset of the GFC, liquidity management issues

became apparent even for adequately capitalized financial institutions. Theoretically, banks' liquidity creation (LC) is

expected to be cyclical because banking sector LC increases with higher bank deposits and lending (Davydov et al., 2018).

However, during the GFC, the behavior of financial institutions' LC tended to be pro‐cyclical, which amplified the financial

shocks and resulted in exacerbating the crisis in the banking system and the wider economy. Furthermore, the cyclicality of

LC not only adversely affects the economy by exacerbating recession but is also considered to be a primary driver of the

financial crisis (Tang, 2019). To combat the pro‐cyclical nature of financial shocks, the Basel Committee on Banking

Supervision proposed a regulatory framework of countercyclical capital buffers in Basel III in December 2010 (Bank for

International Settlements, 2010), which assists banks in becoming more resilient to pro‐cyclical dynamics. Since then,

cyclicality in the banking industry has received a lot of attention from academia and practitioners. For instance, some

recent studies (Berger & Bouwman, 2017; Davydov et al., 2018; Fidrmuc et al., 2015) investigate the cyclicality of LC of

banks using linear models and gross domestic product (GDP) as proxy of the business cycle (BC). However, the extant BC

proxies are criticized for putting more emphasis on the manufacturing sector and having low accuracy in forecasting BC

changes (Berge & Jordà, 2011; Golinelli & Parigi, 2007). Furthermore, linear models are not useful for observing the

asymmetric impacts of BC on bank‐specific variables in the presence of heterogeneity (Brei & Gambacorta, 2016;Davydov

et al., 2018; Rubbaniy et al., 2021). The presence of these criticisms calls for using a better proxy of the BC andapplying a

nonlinear approach to capture the nonlinear effect of the BC on bank LC. In addition, studies of the cyclicality of LC are

scant and restricted to non‐US banks. Our study fills this void in the literature and investigate s the cyclicality of LC in US

bank holding companies using the new measure of BC, developed by Brave et al. (2019),

1

and implements a nonlinear

panel smooth transition regression (PSTR) model, developed by Gonzalez et al. (2017), to deal with asymmetrical effects of

BC and heterogeneity issues.

Berger and Bouwman (2015) argue that investigators should use bank LC, rather than bank lending, as banks' output

measure because it is a broader measure of banks' output that captures the differences in loan categories and composition

on the liability side. Although the literature examines the cyclicality of banks' lending (Bertay et al., 2012)andLC(Davydov

et al., 2018),theliteratureonbankLCisscarceandlimitedtonon‐USbank markets (Davydovet al., 2018;Tang,2019). For

instance, Davydov et al. (2018) examine a sample of Russian banks from 2004 to 2015 and find that LC is pro‐cyclical.

They also argue that the pro‐cyclicality of bank LC can be explained by the fact that growth in deposits and lending

activities enhances bank LC, which in turn positively contributes to economic growth. Tan g (2019) studies the cyclicality of

LC using the data of Chinese banks from 2012 to 2018. He finds evidence in support of the pro‐cyclicalityof LC, which he

suggests can be mitigated by using regulatory capital pressure and interbank credit. Berger and Sedunov (2017) investigate

1

We use the US business activity index (cycle) component expressed in annualized real GDP growth equivalent units by Brave et al. (2019), which is the

sum of the leading and lagging subcomponents of the BC index. A zero value of the cycle component of the BC index indicates zero growth in economic

activity, a negative value indicates below‐average growth, and a positive value indicates above‐average growth. We also use quarterly real GDP growth to

observe the procyclicality of LC.

1166

|

JOURNAL OF FINANCIAL RESEARCH