Dividend policy and bank opacity

• DOI: http://doi.org/10.1002/ijfe.1611
• Author: Dung Viet Tran,Badar Nadeem Ashraf
• Date: 01 April 2018
• Published date: 01 April 2018

RESEARCH ARTICLE

Dividend policy and bank opacity

Dung Viet Tran

1,2,3

| Badar Nadeem Ashraf

4

1

Banking Faculty, Banking Academy of

Vietnam, Hanoi, Vietnam

2

Institute of Research and Development,

Duy Tan University, Da Nang, Vietnam

3

University Grenoble Alpes, CNRS,

Grenoble INP, CERAG, 38000 Grenoble,

France

4

School of Management, Huazhong

University of Science and Technology,

1037 Luoyu Road, Hongshan District,

Wuhan, China

Correspondence

Dung Viet Tran, University Grenoble

Alpes, CNRS, Grenoble INP, CERAG,

38000 Grenoble, France.

Email: viet‐dung.tran@univ‐grenoble‐

alpes.fr

JEL Classification: G21; G28; G34; G38

Abstract

Does dividend policy increase or decrease bank opaqueness? Under the Divi-

dend–Transparency Channel, paying dividends involves banks having greater

discipline from the markets due to external financing and reduces private bene-

fits of control, leading to lower earnings management concerns. Under the Div-

idend–Opacity Channel, due to a hesitancy to change dividend policy, banks

have high vocations to engage earnings management to circumvent payout pol-

icy restrictions in debt covenants or to keep their dividend target unchanged.

Employing a large sample of 2,483 U.S. bank holding companies from 2001:

Q1–2013:Q4, the study documents the double‐edged sword of dividends on the

discretionary behaviours of banks. Paying dividends makes banks less opaque

to compare with nonpayers, which is consistent with the Dividend–Transparency

Channel, however among dividend paying banks, excessive dividends involve

banks to manage more their numbers, which is consistent with the Dividend–

Opacity Channel. The findings are robust under different specifications. The

results are of important interest to bank regulators.

KEYWORDS

banks, discretionary loan loss provisions,dividends, earnings management,opacity, payout policy

1|INTRODUCTION

Banks are black boxes in the rationale of regulators and

are inherently more opaque than other firms (Morgan,

2002). When banks manipulate their reported numbers,

this induces greater opacity and interferes with the private

governance and official regulation of banks (Jiang,

Levine, & Lin, 2016), leading to higher opportunities for

expropriation. Previous literature documents the main

motives that drive banks'discretionary behaviours

through manipulating loan loss provisions (LLPs), such

as earnings smoothing, capital management, tax reduc-

tions, and private information conveyance (Ahmed,

Takeda, & Thomas, 1999). Recent studies point out

accounting discretion is associated with greater vulnera-

bility, markets valuation, and lending (Beatty & Liao,

2011). Dividends are considered as a way to mitigate the

asymmetric information and agency problems, because

paying dividends is a costly signal to convey information

and is used to discipline managers. This study aims to pro-

vide an empirical investigation into whether dividend pol-

icy is associated with earnings management within the

banking context.

There are two strands of literature that are related to

the question of how dividends affect bank opacity. First,

the agency theory suggests that dividends help to reduce

the agency costs derived from the separation of ownership

and control (Easterbrook, 1984). Dividend‐paying banks

have to raise funds from capital markets more frequently

than nonpayers, which leads in turn to closer scrutiny of

managers by outside professionals, and consequently, less

leeway to manipulate reported numbers. Additionally, the

higher the private benefits of control, the higher the incen-

tives to manipulate the account numbers in order to con-

ceal diversion activities (Leuz, Nanda, & Wysocki, 2003).

Paying dividends reduces the private benefits of control,

Received: 26 September 2017 Accepted: 12 December 2017

DOI: 10.1002/ijfe.1611

186 Copyright © 2018 John Wiley & Sons, Ltd. Int J Fin Econ. 2018;23:186–204.wileyonlinelibrary.com/journal/ijfe

thus, attenuating the desires to engage earnings manage-

ment for personal benefits of controlling insiders. Hence,

one could expect that dividend‐paying banks are less

opaque by having fewer incentives to engage earnings

management than dividend nonpaying banks. We call this

channel the Dividends–Transparency Channel. Skinner

and Soltes (2011) document that dividend‐paying firms

report more persistent earnings than nonpaying firms.

Tong and Miao (2011) report that dividend‐paying status

is related to higher quality of accruals, whereas Caskey

and Hanlon (2013) examine the association between divi-

dends and fraudulent reporting and find that dividend‐

paying status is negatively associated with financial fraud-

ulent reporting. He, Ng, Zaiats, and Zhang (2017) support

these findings, pointing out that dividend payers manage

earnings less than dividend nonpayers. They report that

this effect is more pronounced in countries with weak

investor protection and high opacity. Lawson and Wang

(2016) examine the information content of dividends

through audit pricing and find that dividend‐paying firms

pay lower audit fees than nonpayers. The authors explain

that dividends provide information to auditors, reducing

the concerns over the earnings quality of clients.

Second, research documents dividends could be used

as a way to transfer wealth from creditors to shareholders.

To prevent these opportunistic incentives, creditors

impose payout policy restrictions in debt covenants.

Because banks are reluctant to change their dividend pol-

icy (Lintner, 1965), such covenants render reported earn-

ings a critical threshold for paying dividends, leading to

higher incentives for dividend‐paying banks to manage

earnings upwards to avoid dividend cuts. In addition,

because dividend‐paying banks target their dividend level

and their payout ratio, earnings variation would vary both

dividend level and payout ratio. Hence, dividend‐paying

banks are more likely to smooth their earnings to keep

their dividend target smooth. We call this channel the

Dividends–Opacity Channel. Daniel, Denis, and Naveen

(2008) show that dividend‐paying firms tend to manage

earnings upward when their earnings are lower than

expected dividends. The authors report that this effect is

evident only in firms with positive debt and is more pro-

nounced prior to the Sarbanes–Oxley Act. Liu and

Espahbodi (2014) report that dividend‐paying firms

engage in more earnings smoothing than nonpayers.

Louis and Sun (2015) state that firms with conservative

reporting practice pay less dividends than firms without

conservatism reporting practice.

In this paper, we conduct a two‐stage approach to

shed light on the straightforward question of how divi-

dend‐paying status impacts on the earnings management

of U.S. bank holding companies (BHCs). We first measure

banks'earnings management by focusing on the

provisions on loans losses (LLPs). These items are by far

the most critical accruals in bank accounts. They are typ-

ically large relative to net income and equity capital,

which are served as a sign of health for banks'stake-

holders such as creditors or regulators (Bushman & Wil-

liams, 2012). Due to the high dependence on the

judgment of managers, LLPs reflect information asymme-

try, which lies at the heart of the banking literature. Com-

pared with previous studies that use pooled samples of

industrial firms, the measure of earnings management

in a single industry such as banking is more accurate

due to the homogeneous sample composition (such as

accounting, disclosure regulation, …), thus mitigating

the bias arising from measurement errors. There is a large

body of literature on earnings management by shaping an

underlying model to capture the LLPs characteristics and

using the residual LLPs

1

as proxies of earnings manage-

ment; however, there is no consensus on how to best

model discretionary provisions (Beatty & Liao, 2014).

Therefore, we use the preferred model of Beatty and Liao

(2014) in our main analysis. This model allows us to

achieve a better separation of the normal LLPs, which

are supposed to capture all adjustments reflecting banks'

fundamental performance, from the abnormal LLPs,

which are, at least in part, due to managerial discretion.

This unexplained portion of LLPs reflects the degree of

earnings management, that is, a greater unexplained com-

ponent implies a higher level of earnings management.

In the second stage, we examine the effects of divi-

dend‐paying status on earnings management for a large

sample of 2,483 BHC during 2001:Q1 and 2013:Q4. Our

empirical analysis provides consistent evidence of a lower

accounting management through discretionary LLPs

(DLLPs) for dividend‐paying banks versus dividend non-

paying banks. For example, our baseline model indicates

that DLLPs of dividend‐paying banks are 0.1% lower than

nonpayers, and economically, DLLPs are about 22.7%

lower for dividend‐paying banks, compared to dividend

nonpaying banks. This evidence is consistent with the

Dividend–Transparency Channel. We further investigated

how dividend‐paying banks should behave if were they a

nonpayer bank, and the evidence points to dividend‐pay-

ing banks being more likely to manage their numbers if

they were the same bank but a nonpayer.

To ensure the robustness of our findings, we provide a

battery of sensitivity tests. We first re‐estimate our baseline

model using alternative samples: excluding public banks,

excluding Top 10th dividend‐paying banks in each quar-

ter, different bank sizes, average analysis, annual data,

and balance data. Our results hold in all specifications.

Second, we address the potential endogeneity of the

decision to pay dividends, which is a deliberate decision

made by bank managers. We begin by adding the bank

TRAN AND ASHRAF 187