Credit Crunches from Occasionally Binding Bank Borrowing Constraints

• Published date: 01 March 2020
• Author: TOM D. HOLDEN,PAUL LEVINE,JONATHAN M. SWARBRICK
• Date: 01 March 2020
• DOI: http://doi.org/10.1111/jmcb.12601

DOI: 10.1111/jmcb.12601

TOM D. HOLDEN

PAUL LEVINE

JONATHAN M. SWARBRICK

Credit Crunches from Occasionally Binding Bank

Borrowing Constraints

We present a model in which banks and other financial intermediaries face

both occasionally binding borrowing constraints, and costs of equity is-

suance. Near the steady state, these intermediaries can raise equity finance

at no cost through retained earnings. However,even moderately large shocks

cause their borrowing constraints to bind, leading to contractions in credit

offered to firms, and requiring the intermediaries to raise further funds by

paying the cost to issue equity.This leads to the occasional sharp increases in

interest spreads and the countercyclical, positively skewed equity issuance

that are characteristics of the credit crunches observed in the data.

JEL codes: E22, E32, E51, G2

Keywords: occasionally binding constraints, credit crunches, financial

crises, interest spreads, bank equity, banking.

ECONOMIC DOWNTURNS ARE USUALLY ACCOMPANIED by sharp

increases in interest spreads as the effects of financial frictions worsen. This is

We are grateful to Matt Baron for sharing his bank balance sheet data set, and would like to thank

an anonymous referee, Luke Buchanan-Hodgman, Federico di Pace, Martin Ellison, Andrea Ferrero, Raf

Wouters,and participants at a number of workshops and conferences for helpful comments and suggestions.

Thispaper was partly written while Swarbrick was a doctoral candidate. Swarbrick gratefully acknowledges

the financial support from the Economic and Social Research Council (grant number ES/J500148/1)

during this period. Holden and Levine acknowledge support from the EU framework 7 collaborative

project “Integrated Macro-Financial Modelling for Robust Policy Design (MACFINROBODS),”grant no.

612796. The views expressed in this paper are those of the authors and do not represent the views of the

Bank of Canada, the Deutsche Bundesbank, the Eurosystem, or its staff. No responsibility for the views

expressed here should be attributed to any of these institutions.

The copyright line for this article was changed on 15 October 2019 after original online publication.

TOM D. HOLDEN is a Researcher in the Research Department, Deutsche Bundesbank (E-mail:

thomas.holden@gmail.com). PAUL LEVINE is a Professor of Economics in the School of Economics,

University of Surrey (E-mail: p.levine@surrey.ac.uk). JONATHAN M. SWARBRICK is a Senior Economist in

the Canadian Economic Analysis Department, Bank of Canada (E-mail: jswarbrick@bankofcanada.ca).

Received March 1, 2018; and accepted in revised form November 27, 2018.

Journal of Money, Credit and Banking, Vol. 52, Nos. 2–3 (March–April 2020)

C

2019 The Ohio State University

550 :MONEY,CREDIT AND BANKING

particularly true during banking crises when the financing costs faced by inter-

mediaries rise dramatically.1In this paper, we present a model in which financial

intermediaries face occasionally binding borrowing constraints that cause spreads to

rise when the value of assets declines sufficiently,due to the costs these intermediaries

face in issuing new equity. The increased spread between the savings rate and the

return on capital implies a drop in the marginal efficiency of investment, generating

declines in aggregate investment relative to the efficient benchmark, and introducing

asymmetries in macroeconomic time series. However, in our model, in the vicinity

of the steady state, financial constraints are slack and financial intermediation is

efficient. This allows for the characterization of normal times and credit crunches.

Our model differs importantly from the existing literature in this dimension. While

the presence of occasionally binding constraints usually depends on calibration,2

in our model, borrowing constraints are always occasionally binding, essentially

irrespective of parameter values.This holds since financial intermediaries, henceforth

simply known as “banks,” choose to borrow to the edge of the constrained region.3

The model is in the spirit of the banking model proposed in Gertler and Kiyotaki

(2010) (henceforth GK), but while these authors prevent equity issuance to ensure

that banks are always financially constrained,4endogenous-dividend payments

and equity issuance costs in our model imply that the financial constraint is only

occasionally binding. To raise funds when further debt finance is unavailable, banks

can reduce dividend payments and use retained profits for free. However, if banks

are unable to raise sufficient funds via retained earnings, they are restricted to costly

equity issuance. This introduces a spread between the risk-free saving rate, and the

risky return to capital, often described as an investment or capital wedge.5Because

the investment wedge appears only during downturns, the effects of the financial

friction are inherently asymmetric. This enables our model to better explain a

number of key facts as compared with other models such as GK. In particular, we are

able to match the large positive skewness in spreads and to provide an explanation

for the observation that crises are occasional phenomena during which the adverse

effects of financial frictions worsen significantly.Furthermore, because it is desirable

to issue equity only when all other sources of finance are exhausted, bank equity

1. See Babihuga and Spaltro (2014) for a discussion of bank funding costs during the 2007–08

financial crisis.

2. Compare, for example, Gertler and Kiyotaki (2010) with Bocola (2016). The constraint is always

binding in the former but only binds occasionally in the latter due to different calibrations of banking

sector parameters.

3. As discussed below, we consider this to be appealing feature given that financial crises occur

across many countries and under a range of banking regulations.

4. There is an extension discussed in GK, pursued further in Gertler, Kiyotaki, and Queralto (2012)

(henceforth GKQ), which introduces bank equity issuance by extending the same agency problem in debt

finance to equity finance by differentiating between inside and outside shareholders. But in doing so, the

setup generates counterfactual dynamics with respect to equity issuance. Specifically, equity issuance is

procyclical, whereas the data indicate that this is countercyclical.

5. SeeChari, Kehoe, and McGrattan (2007) as an example of the former, and Hall (2010) of the latter.

TOM D. HOLDEN, PAULLEVINE, AND JONATHAN M. SWARBRICK :551

issuance is strongly countercyclical, consistent with the data,6but missed in other

models of bank equity issuance, such as GKQ.7Additionally, modeling occasionally

binding financial constraints eliminates the financial accelerator mechanism during

normal times, in line with the evidence that models without a financial accelerator

perform better in normal times Del Negro, Hasegawa, and Schorfheide (2016); in

our model, only during sufficiently deep downturns do the financial constraints bind,

further amplifying the recession. However, once allowing banks to choose the level

of dividend payments and issue new equity, we find that the financial accelerator

mechanism is significantly dampened, even during crisis episodes.

As well as modeling the financial structure of banks more realistically, we improve

upon the GK agency problem. The GK borrowing constraint emerges due to limited

contract enforceability; banks have an outside option to divert assets and declare

bankruptcy. By parameterizing the proportion of assets that can be reclaimed by

creditors, the authors set the outside option to a fixed amount of the current value

of bank assets. In our model, we carefully specify off-equilibrium play and use U.S.

bankruptcy law to implement the amount recoverable by creditors. In particular,

while GK place timing restrictions when banks can choose to default in order to

prevent banks making a large, unrecoverable dividendat the end of one period before

defaulting in the next, the restriction is not required in our approach as, according to

U.S. bankruptcy law, the amount paid out wouldalso be liable to be reclaimed by the

courts. This mechanism also gives an additional motive for dividendpayments; since

recent dividend payments are reclaimable during bankruptcy, dividend payments act

to relax the present and future borrowing constraint, and consequently can sometimes

be paid even if the bank is issuing equity, helping to explain a long-discussed

puzzle (see, e.g., Myers 1984, Loderer and Mauer 1992, Fama and French 2005). In

particular, while the borrowing constraint in GK is given by:

Vj

t≥θAj

t,

for bank jwhere Vj

tis the value of bank equity and Aj

tis the value of bank assets,

in our model, this becomes:

Vj

t≥θj

1,tAj

t−θ2,t¯

Dj

t−1,

where ¯

Dj

t−1is a weighted average of previous, reclaimable divided payments. This

captures both a time-varying, bank-specific, weight on bank assets, and the role of

past dividends in relaxing the constraint.8

We compare our model both with the standard real business cycle (RBC) model,

which provides an efficient benchmark, and with the always-binding borrowing

6. Itis widely accepted that equity issuance by most nonfinancial firms is procyclical; however, recent

studies have shown that bank equity issuance is countercyclical (see, e.g., Baron 2017).

7. As mentioned in footnote 4. The introduction of differing costs of equity and debt is similar to

that proposed in Jermann and Quadrini (2012) who include tax benefits of debt finance; however, in our

model, the tightness of the borrowing constraint is endogenous, and only occasionally binding.

8. See Appendix A for further details.