BANK-SPECIFIC DECISIONS AND BANKS’ PROVISIONING BEHAVIOUR IN NIGERIA: THE ROLE OF IFRS ADOPTION AND BANKS’ RISKINESS Текст научной статьи по специальности «Экономика и бизнес»

Bocci-ïiceiaïï xypHan MeHegxMenTa 22 (1): 51-85 (2024)

Russian Management Journal 22 (1): 51-85 (2024)

BANK-SPECIFIC DECISIONS AND BANKS' PROVISIONING BEHAVIOUR IN NIGERIA: THE ROLE OF IFRS ADOPTION AND BANKS' RISKINESS

A. A. SALAMI

Al-Hikmah University, Nigeria

Goal: this study examines the relationship between bank loan loss provisioning behaviour and various bank-specifie- decisions on loan loss reporting. The evidence is derived from several cases of non-performing exposures of Nigerian deposit money banks. Methodology: to investigate the relationship between bank loan loss provisioning behaviour and various bank-specific decisions on loan loss reporting there was created the loan loss bank-specific decisions index (BDI) according to the International Financial Reporting Standards (IFRSs). The hypotheses are tested by using the estimates of panel corrected standard errors regression model. Findings: the results showed that discretionary adjustments to loan loss provisions (LLPV) are typical of Nigerian banks although the earnings manipulations imbedded in discretionary LLPVs (DLPV) are not factored in. However, during IFRS a reversal is evident suggesting bank-specific decisions prompting increased DLPV except for seemingly financially-distressed banks. These findings also confirm the relationship between BDI components and provisioning behaviour in individual tests. The study also found out that the provisioning behaviour in Nigeria during global financial crisis was pro-cyclical (nonrational) while during the local economic recession of 2016-2017, it was counter-cyclical (rational). The latter, which happened during the IFRS period, was characterized by income smoothing while the former was addressed at the level of bank-specific decisions. Originality and contributions of the author: the study contributes to the literature by testing a list of bank-specific factors engendering adjustments to loan loss decisions. It is unique in terms of the empirical reference to numerous bank-specific decisions as a single factor in accounting for loan losses. The paper also establishes the parochial success of institutional changes in the corporate Nigeria and offers useful suggestions for the way forward.

Keywords: bank-specific decisions, loan loss provisions, institutional change, bank financial health, Nigeria.

JEL: M4, G21, C33, G32, G38

Postal address: Department of Accounting, Al-Hikmah University, Ilorin, Kwara State, P . M . B . 1601, Nigeria .

© A. A. Salami, 2024

https://doi . org/10 .21638/spbu18 .2024 . 103

INTRODUCTION

Loan loss provisions (LLPVs) are not only a key financial reporting tool in the banking industry but also use for numerous decisions which originate from two circumstances: bank financial condition and economic condition of a country [Salami, 2021] . The decisions in accounting for loan losses that require adjustments to bank financial condition which are referred to as loan loss bank-specific decisions include managerial discretionary use of LLPVs, funding and diversification strategies and corporate taxation [Amidu, Kuipo, 2015; Andries, Gallemore, Jacob, 2017; Soedarmono, Pramono, Tarazi, 2017].

In Nigeria, series of multi-dimensional issues necessitating an empirical study providing a nexus between bank-specific decisions and loan loss discretionary accounting are in the public domain The issues of financial reporting irregularities established against Stanbic IBTC Holdings Plc and capital smoothing of former Skye Bank Plc [FRC, 2015; Proshare, 2017] are testimonies . The sudden feeble financial condition and weak capital base that followed rosy financial health of former Diamond Bank Plc (with trillion naira assets) if not for its acquisition by Access Bank Plc would have caused heavy loss of funds and investments for depositors and investors respectively Without prejudice to the breakdown in the soundness of corporate governance practices in FBN Holdings Plc in 2021, the magnitude of non-performing loans identifiable with one of its subsidiaries, First Bank of Nigeria Limited, and level of regulatory forbearances allowed by the Central Bank of Nigeria (CBN) connect funding strategy with LLPVs decisions in a diversified firm

The risk of insolvency has been linked to the propensity of banks towards accounting manipulations embedded in LLPVs [Lev-entis, Dimitropoulos, Anandarajan, 2011; Yasuda, Okuda, Konishi, 2004] . This was revealed subsequent to the 2009 CBN special audit of Nigerian deposit money banks

(DMBs) . The majority of DMBs who had their management taken over and which were provided with bailouts post-CBN audit were found to have hidden their non-performing loans crisis via manipulative loan loss reporting [Otusanya, Uadiale, 2014; Sanusi, 2010; 2012]. The 2009 scenarios necessitated reforms of Prudential Guidelines [CBN, 2010], establishment of Assets Management Corporation of Nigeria (AM-CON) and early adoption of International Financial Reporting Standards (IFRSs) by Nigerian DMBs [Sanusi, 2012].

The exit to reporting in IFRSs according to [Sanusi, 2012] is to integrate Nigerian banks into global best practices in financial reporting, reduce uncertainties, improve market disciplines and enable greater transparency in financial disclosures . The CBN confidence in the efficacy of IFRS reporting is reinforced by similar experience reported in the literature [Gebhardt, Novotny-Farkas, 2011; Leventis, Dimitropoulos, Anandarajan, 2011]. However, the occurrence of traces of events established during 2009 CBN special audit in post-IFRS adoption era [FRC, 2015; Proshare, 2017] necessitates an empirical study investigating the role of IFRSs on the nexus between various bank-specific decisions and loan loss reporting in Nigeria

While it is acknowledged that similar attempts have been made previously in the literature, majority of these attempts focused on the bank-specific decisions embedded in the managerial discretionary use of LLPVs for income smoothing, signalling and regulatory capital management [Chen et al . , 2021; Leventis, Dimitropoulos, Anandarajan, 2011; 2012; Nikulin, Downing, 2021; Ozili, Outa, 2019] . Only few studies are available connecting funding and diversification strategies with loan loss reporting [Amidu, Kui-po, 2015; Salami, Uthman, 2022; Salami et al . , 2022]. Thus, this study advances literature via testing a long list of bank-specific factors causative to adjustments to LLPVs Another contribution of this study to knowledge is in the derivation of index

of bank-specific decisions (BDI) . The derivation facilitates the reference to loan loss bank-specific decisions as a single factor despite being represented by numerous measures in the literature .

There are four other sections of literature review, methodology, results presentation and concluding remarks in the study The literature review is structured into theory, concepts and previous empirical studies Methodology unveils the materials and methods applied in the conduct of the study while results presentation includes data analysis and discussion of the study's findings The final section of concluding remarks summarises, draws inferences and states practical implications and recommendations for policy making

LITERATURE REVIEW

This study is premised on three theories: prospect theory (PT); positive accounting theory (PAT); and institutional change theory (ICH) . While PT and PAT reinforce managerial opportunistic behaviour, the approaches to reducing the practice are encapsulated in institutional change theory (ICH) . The relevance of PT to the loan loss decisions is evident in its application to corporate inclination to manage earnings to exceed threshold [Wasiuzzaman, Sahafzadeh, Nejad, 2015] . This also reflects in banking institutions attempting to appear well-funded [Salami et al . , 2022], maintain holding company status in a conglomerate of financial services and avoid sanctions owing to failure to meet capital adequacy benchmark [Salami, Uthman, 2022; Salami, Uthman, Owoade, 2022]. Similarly, PAT explains managerial opportunistic behaviour which prompts choice of accounting methods as a result of managerial explicit contract [Ozili, 2017a; Watts, Zimmerman, 1986]. However, ICH encapsulates the improvement in the institutional arrangements and structures when there is institutional change [Samadi, Alipourian, 2021]

The adoption of IFRSs which is a global phenomenon is an example of institutional reform aimed at improving financial reporting quality and increasing corporate disclosures and access to financial information by capital providers [Eiler, Miranda-Lopez, Tama-sweet, 2021] . Though reforms following the 2009 banking crisis in Nigeria are many [Sanusi, 2012], IFRS reporting mandate is central as it extends to other sectors . Therefore, PT and PAT alongside ICH are used to explain the relationship between bank-specific decisions and bank provisioning behaviour in Nigeria upon adoption of IFRSs

Bank-specific decisions in the context of loan loss provisioning are better conceptualised when reference is made to managerial discretions, funding strategy and diversification strategy [Salami, 2021]. Managerial discretionary attributes of bank management involve three decisions of management/smoothing of capital and earnings and signalling of earnings in the literature [Curcio, Hasan, 2015]. The relationship between LLPVs and bank earnings before LLPVs and tax (BEBLT) will reveal smoothing of earnings if positive [Muriu, Josea, 2020] . For earnings signalling, banks will show their ability to absorb future losses if one-year-ahead percentage change in BEBLT (SINL) exerts positive influence on LLPVs [Leventis, Dimitropoulos, Anandarajan, 2012; Salami, Uthman, Abdulrauf, 2021].

Bank regulatory capital, which could be Tier 1 or core capital (CCR) and total regulatory capital incorporating core capital and Tier 2 capital (TCR), has to be inversely related to LLPVs for real case of capital management to be established [Salami, Uth-man, Owoade, 2022] . Like LLPVs, discretionary component of LLPVs (DLPV) can also be used to manage capital, smooth bank earnings and signal strength with the earlier established relationships subsisting [Salami, Uthman, Owoade, 2022] This is realistic subsequent to the segregation of LLPVs into discretionary and non-discretionary components [Zhang, Mclntyre, 2021] .

Using the approach of [Amidu, 2013; Amidu, Wolfe, 2013; Amidu, Kuipo, 2015] three funding strategies of wholesale/non-deposit funding (FNDP), internal funding (FIGR) and deposit funding (FDEP) are identified Another bank-specific decision prompting bank loan loss behaviour but considered another funding strategy [Salami, 2021] is the bank motive for external financing [Othman, Mersni, 2014] . As argued in the literature, banks' desire of external funds may prompt reduction of perceived risk (in this case LLPVs) to facilitate the increase in the reported earnings [Othman, Mersni, 2014; Seppanen, 2000]. As a favourable consequence, banks with stable and positive earnings overtime have higher chance of accessing market for funding [Kanagaretnam, Lobo, Mathieu, 2003] The proportion of bank total credit (loans) to total deposits from customers (FLD) is used to measure bank external funding motive in the relevant literature [Shawtari et al , 2015; Zoubi, Al-Khazali, 2007]. Based on the foregoing evidence from the literature, FNDP, FIGR, FDEP and FLD constitute funding strategies in loan loss reporting

Thus diversification strategy can be an instrument for unholy adjustments to bank earnings [Salami, Uthman, 2022] In bank financial reporting, diversification strategy can be devised using revenue, products, deposit (liability) and assets [Gambacorta, Scatigna, Yang, 2014; Olarewaju, Migiro, Sibanda, 2017] Diversification can be measured traditionally using simple ratio [Ozili, 2017b] but a better approach is the measurement via constructing Herfindahl-Hirschman Index (HHI) [Doumpos, Gaganis, Pasiouras, 2016] . Based on deductions from [Amidu, Kuipo, 2015; Doumpos, Gaganis, Pasiouras, 2016], diversification within bank revenue (RDIV), non-interest activities (NDIV), earnings assets (EDIV) and statement of financial position (BDIV) constructed using HHI are adopted and specified in the Appendix 1-3 .

To find index of bank-specific decisions, the study's main independent variable, con-

densation technique adopted by [Abdul-mumin et al . , 2019; Salami et al . , 2022; Salami, Uthman, 2022] is applied . This facilitates via application of principal component analysis (PCA) the reduction of a long list of decisions related to managerial discretions (BEBTL, SINL, CCR and TCR), diversification strategy (RDIV, NDIV, EDIV and BDIV) and funding strategy (FDEP, FNDP, FIGR and FLD) to BDI.

From the empirical literature of BDI components, only S . Ghosh reported the evidence of earnings smoothing, earnings signalling and capital management [Ghosh, 2007] in the managerial discretionary use of LLPVs. The majority of other evidence in the literature is mixed . The authors [Anandarajan, Hassan, McCarthy, 2007; Chang, Shen, Fang, 2008; Pinho, Martins, 2009; Misman, Ahmad, 2011; Abdullah, Bu-jang, Ahmad, 2013; Adzis, Anuar, Mohd Hishamuddin, 2015; Bryce et al . , 2015; Du-shku, 2016; Abdullah, Bujang, Sahudin, 2017; Curcio, De Simone, Gallo, 2017; Skala, 2018; Muriu, Josea, 2020; Salami, Uthman, Owoade, 2022] found the case of earnings and/or capital management except for authors [Ashour, 2011; Alessi, Di Colli, Lopez, 2014] with no clear-cut evidence . Only signaling of earnings was prioritized in the Middle-East and North Africa, Jordan and Italy as found by [Olson, Zoubi, 2014; Abu-Serdaneh, 2018; Caporale et al . , 2018] respectively . For D . Curcio and I . Hasan [Curcio, Hasan, 2015] findings revealed that Euro Area (EA) banks smooth earnings while non-EA banks manage capital and signal earnings using LLPVs

For discretionary use of DLPV, earnings smoothing and signaling and/or capital management are reported by [Kwak, Lee, El-dridge, 2009; Shawtari et al. , 2015; Tran, Hassan, Houston, 2022; Salami, Uthman, Abdulrauf; 2021; Salami, Uthman, Owoade, 2022] In the bank diversification literature, evidence of bank diversification strategy having inverse relationship with discretionary use of LLPVs is predominant [Amidu, Kuipo, 2015; Saona, Azad, 2020] except for

the authors [Salami, Uthman, 2022] reporting contrary evidence for activity mix within bank revenue The literature documentation related to funding strategy-earnings smoothing nexus of banks revealed positive signs for non-deposit and internal funding [Amidu, Kuipo, 2015], Syirkah funding (a source of funding to Islamic bank which entitles the holders to returns only if the bank makes profit) and debt funding [Mukh-ibad, Nurkhin, 2019] and deposit funding [Saona, Azad, 2020] for African, Indonesian Islamic and Asian banks respectively

Furthermore, evidence of reduction in LLPVs to bring down risk related to the banks' ability to attract external funding is reported by [Ashour, 2011; Zoubi, Al Khaz-ali, 2007]. Similarly but with the use of DLPV, increase in DLPV (income smoothing) was found by [Kanagaretnam, Lobo, Mathieu, 2003; 2004; Kwak, Lee, Eldridge, 2009; Othman, Mersni, 2014; Shawtari et al . , 2015; Bhattarai, 2018] as a result of increased motive for external funding As found by [Salami et al , 2022], smoothing of earnings is identifiable with deposit and non-deposit funding against internal funding and bank motive for external funding

Since the majority of evidence in the literature showed that one or more components of BDI prompt the adjustments to LLPVs for discretionary purpose, the study's first hypothesis is stated as:

Hypothesis H1 . Adjustment to LLPVs for various bank-specific decisions is typical of banks in Nigeria.

If banks' reporting behaviour during economic crisis and while they are threatened by risk of default is comparable [Salami, Uthman, Owoade, 2022], the approaches to the adjustments to LLPVs for discretionary purposes by banks can be relatively similar Given the argument, the study's review of previous studies incorporates discretionary loan loss behaviour of banks when threatened by solvency risk and during financial crisis The evidence of discretionary use of LLPVs during economic crisis is documented by the authors [El-Sood, 2012; Skala,

2014; Curcio, De Simone, Gallo, 2017; Car-bo-Valverde, Rodriguez-Fernandez, 2018]. There are also empirical findings of riskier banks indulging in managerial discretionary use of LLPVs [Leventis, Dimitropoulos, Anandarajan, 2011; 2012; Bryce et al . , 2015; Salami, Uthman, Abdulrauf, 2021; Salami et al , 2022; Salami, Uthman, Owoade, 2022] Similar evidence is also obtainable from the bank diversification and funding strategy literature [Salami et al , 2022; Salami, Uth-man, 2022]

Having noted that there are higher cases of discretionary use of LLPVs for components of BDI by seemingly financially distressed banks as documented in the literature, the study hypothesis as follows:

Hypothesis H2. Nigerian banks threatened by risk of insolvency are prone to adjusting LLPVs for various bank-specific decisions.

From previous empirical studies of [Van Oosterbosch, 2009; Gebhardt, Novotny-Far-kas, 2011; Leventis, Dimitropoulos, Anandarajan, 2011; 2012] to those of [Duru, Tsitinid-is, 2013; Ozili, 2015; Adzis, Tripe, Dunmore, 2016; Arbak, 2017; Ozili, Outa, 2019], evidence of improvement in discretionary use of LLPVs during IFRSs is reported No evidence of capital management but earnings management via LLPVs is reported during IFRS by [Atoyebi, Simon, 2018] . As found by [Amidu, Issahaku, 2019; Saona, Azad, 2020] adoption of IFRSs improves diversification strategy-discretionary LLPVs nexus of African and Asian banks respectively Similar evidence is also obtained by [Salami, Uthman, 2022] for Nigerian banks including riskier ones except for revenue and non-interest activity mix .

There is no evidence of any measure of funding strategy being incidental to discretions embedded in LLPVs during IFRSs except internal funding which is insignificant [Salami et al . , 2022]. In contrast, management of earnings via LLPVs is identifiable with banks reporting in IFRS within Organisation of Islamic Cooperation [Ashraf, Hassan, Basher, 2015] and in South Africa [Ozili, Outa, 2018] .' Similar scenario of discretions in

LLPVs except to signal earnings is also reported in the global study of [Ashraf et al . , 2019] for banks reporting in principles-based accounting standards like IFRSs. Nigerian banks (including riskier ones) are found to use LLPVs to smooth earnings and capital but not to signal earnings [Salami, Uthman, Abdulrauf, 2021; Salami, Uthman, Owoade, 2022] .

Since IFRS reporting appears to bring more blessing than curse as previously documented in the literature regarding discretionary use of LLPVs, the third and fourth hypotheses are stated as:

Hypothesis H3. There is considerable reduction in discretionary adjustments to LLPVs for various bank-specific decisions by Nigeria banks upon switch to IFRS reporting;

Hypothesis H4. Discretionary adjustments to LLPVs for various bank-specific decisions are not characteristic of financially unhealthy Nigerian banks subsequent to adoption of IFRSs.

METHODOLOGY

Data and methods of estimation

Since all variables required to attain the objective of the study are bank-specific in nature, a hand-extraction of relevant data from audited financial statements of Nigerian banks is carried out over the period 2007-2017 . The duration for data collection covers both pre-IFRS and IFRS reporting era as the adoption of IFRSs by public interest entities including banks took effect on 1 January, 2012 in Nigeria . A halt in data collection in 2017 is premised on avoiding the distortion of the study's findings as switch to IFRS 9: Financial Instruments from International Accounting Standards No . 39 (IAS 39): Financial Instruments: Recognition and Measurement was based on partial implementation for four years from 1 January 2018 as instructed by CBN .

The halt is also attributed to the fact that the partial reporting in IFRS 9 for loan losses in Nigeria was in embryo as

at the time of extracting data for the study The basic criterion for inclusion of a bank among the sampled DMBs is the access to annual report and accounts of the bank for not less than 60 % of the study's sampled period covering both pre-IFRS and IFRS regimes . This results in the use of 169 bank-year observations instead probable 176 bank-year events (based on 16 sampled DMBs for 11 years) given unavailability of annual reports of a couple of banks for seven years The unavailability of the said annual audited financial statements was due to the period useful annual financial records of one of the sampled banks began to be in public domain which is 2011 . Another bank was absorbed by a bigger bank within the sampled period resulting in the acquiree having useful financial records up to 2014 . The unavailability of the financial records of the former for four years (2007-2010) and the latter for three years (2015-2017) account for the use of 169 bank-year events .

The study's data are estimated using both univariate and multivariate techniques . The univariate techniques adopted are summary statistics of mean, median, standard deviation, minimum and maximum values used to facilitate detection of basic attributes of the study's variables The basic multivariate techniques adopted are PCA used for deriving index of loan loss bank-specific decisions and panel regression model given the nature of data collection which is longitudinal The choice of PCA, despite existence of plethora of other condensation techniques including principal axis factor analysis (PAFA), maximum likelihood estimate and cluster analysis [Abeyasekera, 2005; Mabel, Olayemi, 2020], is not attributed to its popularity [De Winter, Dodou, 2016] but its superiority [Mabel, Olayemi, 2020] The superiority of PCA can be inferred from its ability to produce factors loading heavily on the study's variables which the factors actually represent [Mabel, Olayemi, 2020]. The probable increase in interpretability makes

PCA a viable extraction technique using economic datasets [Tsoulfidis, Athanasiadis, 2022] as used in this study . As found by [De Winter, Dodou, 2016], PAFA, a close substitute of PCA, is somewhat subject to a higher improper outcomes . This superiority informs its popularity in accounting for loan losses' studies [Olszak et al . , 2017; Salami, Uthman, Owoade, 2022].

The study followed basic procedures in panel data modelling These include the choice between fixed-effects model (PFE) and random-effects model (PRE) using results of Hausman statistics (HST) and the choice between PRE (in case of insignificance of HST at p-value < 0 . 05) and pooled ordinary least squared regression (POLS) using the results of Breusch-Pagan Langrange Multiplier statistics (BPLM) . However, using concurrent significance at p-value < 0 05 of the tests of heteroscedastic-ity, cross-sectional dependence and serial correlation, panel corrected standard errors (PCSE) regression analysis is majorly opted for as presented in Tables 8, 9, 10, 11 and 12

In PCSE estimates, problems of disturbances being autocorrelated within the panel at first order, heteroscedastic and contemporaneously correlated across panels and temporally dependent are rectified [Beck, Katz, 1995; Salami, 2021]. For het-eroscedasticity test, Wooldridge test for heteroscedasticity (PFE-HET) for PFE and Breusch-Pagan/Cook-Weisberg test for heteroscedasticity with fitted values of dependent variable (POLS-H1) and independent variables (POLS-H2) for POLS are adopted . Wooldridge panel data first-order autocorrelation test (PAR-1) and Breusch-Pagan LM test of independence (CR-DPD) are performed for serial correlation and cross-sectional dependence respectively . As an aid to proper specification of study's econometric models, pairwise correlation analysis and variance inflation factor (VIF) are used to detect cases of multi-colline-arity problems among the explanatory variables

Econometric models

The BDI which is the study's main independent variable is derived from the identified loan loss bank-specific factors of managerial discretions, funding strategy and diversification strategy using PCA as follows:

BDIit = 51CCRit + 52TCR + + 53 BEBLTit + 54 SINLit + 5S FDEPit + + Z6FNDPit +57 FIGRit + d&FLDn + + S9RDIVu + SWNDIVU + dnEDIVit +

+ dl2 BDIVU. (1)

The BDIit in equation (1) represents an index of bank-specific factors of DMB i at time period t; where i ranges from 1-16 and t ranges from 2007-2017; 5 is the is the eigenvector or factor loading used as the weight in the computation BDI using PCA . The measurements of other variables are presented in the Appendix 1

Following the approach of [Salami, 2021], banks' loan loss provisions are assumed to be actual (LLPVs) and discretionary (DLPV) . In the literature, DLPV is derived using various non-discretionary loan loss models [Amidu, Kuipo, 2015; Beaver, Engel, 1996; Kanagaretnam, Lobo, Mathieu, 2003; 2004]. However, this study adopts K . Kanagaretnam, G . Lobo, R . Mathieu's [Kanagaretnam, Lobo, Mathieu, 2003; 2004] model as it contains all variables sufficiently accessible in the financial records of Nigerian DMBs . The model which makes provision for loan losses a function of three non-discretionary components is as specified in equation (2):

PLLSit =p0 + VlNPFLu - + + V2CHNPFLit +P3 + LOANu +eu. (2)

In equation (2), the dependent variable and three explanatory variables (non-discretionary components) are measured as:

PLLSit — loan loss provisions of current period scaled by beginning loans and advances;

NPFLit— — non-performing loans of year t-1 scaled by loans and advances of year t-1;

CHNPFLit — difference between nonper-forming loans of year t and year t-1 divided by gross loans and advances at year t-1;

DLOANit — change in value of gross loans and advances between year t and year t-1 scaled by loans and advances at year t-1.

The residuals (sit) of equation (2) represent DLPV but the absolute value of DLPV (ADLPV) is used as second measure of bank provisioning behaviour since distinction is not made between income-increasing and income-decreasing earnings smoothing in this study.

Subsequent to derivation of BDI and DLPV, bank provisioning behaviour is made a function of BDI following the approach of [Salami, 2021] as follows:

LLPVit = «0 + «1 BDIit + «2 + +NPLit + «3 LVRit + «4 LTAit +

+ «5 LTNU , (3)

ADLPVU =«, +<*! BDIlt + + «2 LASit + «3 LVRit + «4 + + LTAit + «5 LTNU + Vit, (4)

LLPVt =x0 + BDIit + IFRSit + + «3 (IFRS • BDI).t + «4 RISit +

+ X,

( RIS • BDI )

+ «6 (IFRS • RIS • BDI).t + x7 + + NPLit +«8 LVRit +«9 LTAit +

+ «io LTN^it, (5)

ADLPVt =«0 +«i BDIit + + «2 IFRSit + «3 (IFRS • BDI) .t + + «4 RISit + «5 (RIS • BDI) ,t + + «6 (IFRS • RIS • BDI) ,t +

+ x7 LASit + X8 LVRit + + X9 LTAit +xw LTNit it.

(6)

The equations (3) and (4) are specified to test the first hypothesis while equations

(5) and (6) are specified to test the second, third and fourth hypotheses . The specification of dependent variables (LLPV and ADLPV), independent variable (BDI), moderating variables (IFRS, RIS and their interactions with independent variables) and control variables is based on deductions from [Leventis, Dimitropoulos, Anandarajan, 2011; 2012; Salami, 2021] .

The control variables included in the study's econometric models are relevant to loan loss provisioning decisions based on previous literature [Salami, 2021]. Both change in non-performing loans (ANPF) and gross loans to total assets (LAS) are included to control for riskiness in banks' financial statements and loan defaults cy-clicality [Gebhardt, Novotny-Farkas, 2011; Ozili, 2017a], A positive sign is expected for the coefficient of ANPF because higher growth in loan defaults when loan supply increases is causative to increased LLPVs [Ozili, 2017a] . The behaviour of LAS in relation to provisioning practices is comparable to that of growth in bank lending where issue of loan loss cyclicality is considered [Soedarmono, Pramono, Tarazi, 2017]. Banks' size often measured as natural logarithm of banks' total assets (LTA) is often included in the LLPV models [Amidu, Kui-po, 2015; Leventis, Dimitropoulos, Anandarajan, 2011].

The larger proportion of "loans and advances" in the assets structure of depository financial institutions reinforces the argument that banks having larger assets are prone to having higher level of business and by implication having higher LLPVs to compensate for increased levels of activity and risk [Anandarajan, Hassan, McCarthy, 2007] The inclusion of banks' cross listing status (LTN) following the approach of [Leventis, Dimitropoulos, Anandarajan, 2011] is subject to the fact that manipulation of earnings prompted by the urge to increase stocks price or improve financial condition is identifiable with banks listed in other jurisdictions other than their places of incorporation . According to [Amidu, Kuipo,

+

2015], level of leverage (LVR) is also an important component of loan loss model The subsisting arguments in this regard are two First, high leverage is causative to earnings smoothing [Jaggi, Lee, 2002; Othman, Zhegal, 2006] since the reduction in creditors' perceived risks and increase in firms' bargaining power are often prioritised Second, financially-weak firms or those being threatened by risk of insolvency are identifiable with high leverage [Mohd Saleh, Ahmed, 2005] Nonetheless, considering the place of "deposits", the largest part of depository financial institutions' liabilities structure, and their complementary role in banks' lending ability from which higher interest income accrues, inverse relationship between level of leverage and DLPV is probable [Salami, 2021].

RESULTS PRESENTATION Derivation of BDI

While measurements of all variables specified in equations (3), (4), (5) and (6) are presented in Table 1, it is important to note that RIS (risk of insolvency) is derived from banks' Z-index computed from the addition of return on assets and equity-assets ratio scaled by standard deviation of return on assets following Salami's approach [Salami, 2018] .

The PCA eigenvalues and proportions are presented in Table 2 while eigenvectors are presented in Table 3 based on estimation of equation (1) . There are four components having eigenvalues more than 1 but the first principal component has the highest value of 3 . 41 which explains 28 . 41 % of total variance as presented in Table 2 This allows the factor loadings or eigenvectors of first principal component to be used for computation of BDI adopted as the independent variable For calculations we used the Stata 15 database1 .

1 Stata 15 . URL: www . stata . com/stata15/ (accessed: 15 .12 .2023) .

Derivation of discretionary LLPVs

The regression estimates of K Kanagaret-nam, G . Lobo and R . Mathieu's loan loss model which facilitate the derivation of DLPV are presented in Table 4 [Kanaga-retnam, Lobo, Mathieu, 2003; 2004] POLS which has the capacity to correct problem of autocorrelated residuals, Cochrane-Orcutt regression (Corc-OLS), is opted for following the approach of [Chang, Shen, Fang, 2008; Salami, Uthman, Owoade, 2022] As revealed in Table 4, upward change in non-performing loans, positive loan growth and higher level of previous period non-performing loans are all causative to increase in provision for loan losses This is evident in the significant and positive coefficients of CHNPFL, ALOAN and NPFLt-1 presented in Table 4 . These findings are relatively similar to those of [Kanagaretnam, Lobo, Mathieu, 2003; 2004; Salami, Uthman, Owoade, 2022] .

Summary statistics

The descriptive statistics of the study are based on sampled banks' financial health (Table 5) and accounting regime (Table 6) following the approach of [Salami, 2021] . As revealed in Table 5, statistics appear to favour less risky DMBs than riskier ones Lower LLPV, negative DLPV and lower ADLPV, in terms of mean values, are characteristics of less risky DMBs This may indicate increase profitability, income-increasing earnings smoothing and lower earnings management for the DMBs' category The negative value of BDI (-0 36) for riskier DMBs may suggest use of loan loss bank-specific decisions to increase the level of profitability while the positive value (0 36) for less risky DMBs may show their attempt to reduce their profitability

While riskier DMBs prioritise signalling financial strength, less risky DMBs' focus might be reducing their liability to tax There is higher growth in non-performing loans (ANPL) of less risky DMBs relative to riskier banks if mean (maximum) values of their

Table 1

Definition and measurement of variables

Variable Variable name Description

LLPVit Actual provision for loan losses LLPVs divided by total loans

DLPVit Abnormal LLPVs Disturbances of equation (2)

ADLPVit Absolute values of DLPV Absolute values of equation (2) disturbances

BDIit Index of loan loss bank-specific decisions Index of bank-specific decisions derived from equation 1 using PCA

IFRSit Reporting in IFRS "1" is assigned for IFRS period, otherwise "0"

RISit Risk of insolvency "1" is assigned for bank with Z-index that is less than median Z-index of all sampled banks and "0" otherwise

IFRS • BDIit IFRS and BDI Interaction between bank-specific decisions index and IFRS reporting

RIS • BDIit Default risk and BDI Interaction between bank-specific decisions index and financial distress status

IFRS • RIS • BDIit IFRS, insolvency risk and BDI IFRS, bank default risk level and BDI interactions

ANPLit Change in non-performing loans (credit risk) Difference between non-performing loans in the year t and year t — 1 divided by the year t — 1 loans defaults

LASit Credit risk Gross loans to total assets ratio

LVRit Leverage Ratio total debts to total equity

LTAit Bank size Natural Logarithm of bank total assets

LTNit Bank cross listing status An indicator variable represented by "1" when a bank is listed in other outside Nigeria but "0" otherwise

ANPL are compared . The higher ANPL of less risky DMBs does not prevent them from appearing more stable in terms of higher mean, median, minimum and maximum Z-index. Nonetheless, in terms of bank-year events, there is no palpable difference between less risky and riskier DMBs given 85 and 84 respectively . In Table 6, lower LLPV and leverage are noticeable pre-IFRS while lower ADLPV, Z-index and ANPL are evident during IFRS in terms of their mean values The negative mean and median values of DLPV during IFRS suggest a relatively pure income-increasing earnings smoothing in the period while a mix of negative and positive values of DLPV pre-IFRS gives an indication of a relative balance

between income-increasing and income-decreasing earnings smoothing. Other variables' descriptive statistics are as presented in both tables

Correlation matrix

and Variance Inflation Factor

The results for diagnostics for multi-collinearity are presented in Table 7 using pairwise correlation analysis and VIF for the study's main explanatory variables while correlation matrix among the components of BDI are presented in the Appendix 2 . Using correlation coefficient more than 0 . 80 as a benchmark for multi-collinearity problem to set in, there is no any two variables with

Table 2

Principal ^mponents eigenvalue and proportion for Bank-Specific Decisions Index

Component Eigenvalue Difference Proportion Cumulative

Comp 1 3 .40868 0.909788 0 .2841 0 .2841

Comp 2 2 .49889 0.823937 0 .2082 0 .4923

Comp 3 1 .67495 0 . 510172 0 . 1396 0 . 6319

Comp 4 1 .16478 0.298956 0 . 0971 0 . 7289

Comp 5 0.865827 0. 188016 0 . 0722 0 . 8011

Comp 6 0.677811 0 .0842931 0 . 0565 0 .8576

Comp 7 0. 593517 0. 123882 0 . 0495 0 .9070

Comp 8 0.469636 0 .0464366 0 . 0391 0 .9462

Comp 9 0.423199 0 . 314412 0 . 0353 0 . 9814

Comp 10 0. 108787 0 .0185438 0 . 0091 0 .9905

Comp 11 0 .0902432 0 .0665704 0 . 0075 0 .9980

Comp 12 0 .0236728 — 0 . 0020 1 . 0000

Note: the last cell of the "difference" column of the principal components/correlation is left empty based on Stata 15 outputs .

correlation coefficient less than 0 . 80 in Table 7 . Also, no VIF and tolerance (HVIF) have coefficient more than 10 and less than 0 . 10 respectively . This represents an indication of no multi-collinearity problem . However, in the Appendix 2, correlation coefficient between TCR and CCR is 0 88 which is more than 0 80 The case of multi-collinearity between TCR and CCR necessitates their separation in the regression estimates presented in Tables 10 and 11 .

Empirical regression results

The results for the tests of the study's four hypotheses are presented in Table 8 . As evident in the table, estimates are presented with and without interaction of the moderating variables of IFRS and RIS with respect to the study's dependent variables: LLPV and ADLPV. While negative coefficient of BDI in the LLPV s models indicates that Nigerian banks' loan loss bank-specific decisions are targeted at reducing LLPVs to ensure increased net income, similar negative coefficient in the ADLPVs models

shows that increased profitability as a result of reduced LLPVs is not imbued with earnings smoothing motive Thus, despite attempts to increase profits Nigerian banks do not prioritise earnings manipulations over the study's sampled period

However, during IFRS, loan loss bank-specific decisions bring about increased LLPV and ADLPV given the significantly positive coefficient of IFRS • BDI in both models The increased LLPV which may be targeted at reducing these DMBs' income tax is reinforced by increased earnings smoothing during IFRS. The increased earnings smoothing practices are also identifiable with financially unhealthy Nigerian DMBs based on significantly positive coefficient of RIS but somewhat in their various provisioning-inclined bank-specific decisions as coefficient of RIS • BDI which is positive is not significant in ADLPV model The financially unhealthy DMBs' motive to increase net income based on negative coefficient RIS • BDI in the LLPV model cannot be completely said to be causative to earnings smoothing as the same variable in the ADLPV model is not signifi-

Principal component eigenvectors for Bank-Specific Decisions Index

Variable Comp 1 Comp 2 Comp 3 Comp 4 Comp 5 Comp 6 Comp 7 Comp 8 Comp 9 Comp 10 Comp 11 Comp 12 Unexplained

CCB 0.486 0.172 -0.043 -0.054 0.096 -0.058 -0.251 0.160 -0.146 0.261 -0.616 0.400 0

TCB 0.497 0.108 0.041 -0.093 -0.034 -0.190 -0.075 -0.007 -0.250 0.441 0.648 -0.103 0

BEBLT 0.026 0.534 0.229 0.142 0.056 0.132 -0.225 0.399 0.257 -0.374 0.321 0.321 0

SINL 0.105 -0.411 -0.273 0.000 0.343 -0.395 0.032 0.484 0.479 0.033 0.074 -0.027 0

BDIV 0.181 0.426 0.106 -0.150 -0.158 -0.247 0.567 -0.235 0.506 0.104 -0.141 -0.058 0

NDIV -0.073 0.063 0.028 -0.737 0.498 0.355 0.218 0.100 -0.101 0.007 0.056 0.038 0

EDIV 0.307 -0.132 -0.166 0.317 -0.181 0.588 0.504 0.349 -0.054 0.050 0.009 -0.025 0

BDIV 0.030 0.250 -0.199 0.484 0.689 -0.100 0.216 -0.274 -0.218 -0.091 0.019 -0.027 0

ELD 0.119 -0.262 0.499 0.193 0.277 0.373 -0.210 -0.300 0.437 0.300 0.009 0.001 0

FDEB -0.335 0.102 -0.545 0.018 -0.085 0.175 -0.062 -0.189 0.218 0.419 0.192 0.493 0

FNDB -0.296 -0.194 0.487 0.116 0.014 -0.272 0.394 0.175 -0.255 0.206 0.056 0.504 0

FIGB -0.397 0.351 0.087 0.130 0.064 0.031 -0.086 0.404 -0.021 0.517 -0.169 -0.472 0

Table 4

Regression model of K. Kanagaretnam, G. Lobo, R. Mathieu

Variable Dependent variable: PLLS

Coefficient Std. error t P > t

CHNPFL 0 .0123769* 0 .0025912 4 . 78 0 . 000

ALOAN 0 . O24O9410 0 . 0140127 1 . 72 0 . 088

npfl,_1 0 . O5981540 0. 0313902 1 . 99 0 . 059

Const 0 .0192356* 0. 0065295 2 . 97 0 . 003

R2 0 .1526

Adj R2 0 .1356

F-test 8.94 (0 .0000)*

RMSE 0 . 04758

PAR 1 30 .87 (0 .0001)*

Observation 169

Model type Corc-OLS

Note: 0 and * indicate significance at 90 % and 99 % confidence levels respectively . Based on: [Kanagaretnam, Lobo, Mathieu, 2003; 2004],

cant though positive . However, the improvement in the financial reporting quality of riskier Nigerian banks is noticeable during IFRS given the significantly negative coefficient of IFRS ■ RIS • BDI in the ADLPV model This is regardless of reduction in LLPVs which may be targeted at increased profitability as indicated by significantly negative estimate of IFRS • RIS • BDI in the LLPV model

The higher credit risk prompts increased LLPVs and smoothing of earnings by Nigerian DMBs as indicated by significantly positive coefficients of ANPL and LAS respectively in the models with interaction variables . However, DMBs with higher leverage are known for increased profitability achievable via lower LLPVs but not for earnings smoothing/management with reference to negative coefficient of LVR in all models . For LTA and LTN, the remaining control variables, bigger Nigerian DMBs and those cross-listed respectively are not indulging in earnings smoothing despite being identifiable with the lower LLPVs based on negative coefficients of both variables

except in the LLPV model with interaction variables .

Based on the findings presented in Table 8, both first and second hypotheses are retained . However, the evidence of adjustments to LLPVs for various bank-specific decisions established as proposed in the first hypothesis is not laden with earnings manipulations via DLPV. Since the improvement envisaged upon adoption of IFRSs is only identifiable with Nigerian DMBs having questionable financial health, the fourth hypothesis is retained while the third one is rejected

These results, though with different motives, are comparable with findings of [Salami, Uthman, Owoade, 2022]. There are also similarities between the findings of this study and those of [Leventis, Dimitropoulos, Anandarajan, 2011; 2012]. More so, sharing the same findings with the present study are those of [Ashraf, Hassan, Basher, 2015; Ozili, Outa, 2018; Ashraf et al. , 2019] on the positive relationship between discretionary provisioning and bank-specific decisions upon reporting in IFRS. However, contrary

Descriptive statistics based on Banks' Financial Health

DMBs' Category

Variable All sampled banks (169) Riskier Nigerian DMBs (84) Less risky Nigerian DMBs (85)

Mean Median Std. dev. Min Max Mean Median Std. dev. Min Max Mean Median Std. dev. Min Max

LLPV 0.05 0.02 0.23 -0.28 2.93 0.07 0.03 0.32 -0.28 2.93 0.02 0.01 0.02 -0.01 0.08

DLPV 0.00 -0.01 0.05 -0.28 0.30 0.01 0.00 0.07 -0.28 0.30 -0.01 -0.02 0.02 -0.07 0.04

ADLPV 0.03 0.02 0.04 0.00 0.30 0.04 0.02 0.06 0.00 0.30 0.02 0.02 0.01 0.00 0.07

BDI 0.00 0.07 1.00 -10.05 1.55 -0.36 -0.07 1.27 -10.05 1.19 0.36 0.28 0.36 -0.11 1.55

ANBL 0.54 0.13 1.54 -0.99 8.11 0.38 0.11 1.20 -0.99 5.79 0.70 0.15 1.81 -0.76 8.11

LAS 0.45 0.45 0.13 0.06 1.01 0.44 0.43 0.14 0.06 1.01 0.45 0.46 0.11 0.17 0.65

LVR. 7.47 6.05 14.94 -9.64 191.21 9.60 7.28 20.98 -9.64 191.21 5.36 5.66 1.52 2.50 9.75

LTA 20.67 20.76 0.85 18.68 22.45 20.46 20.68 0.87 18.68 22.28 20.87 20.80 0.79 19.20 22.45

LTN 0.33 0.00 0.47 0.00 1.00 0.36 0.00 0.48 0.00 1.00 0.29 0.00 0.46 0.00 1.00

Z-index 14.84 15.29 10.64 -38.34 43.08 6.59 7.19 7.47 -38.34 14.94 22.99 22.11 5.99 15.29 43.08

Max 0 . 31 0 30 0 30 1 . 55 8 11 1 . 01 35 03 21 77 1 00 43 08

o ^ Min -0 .28 -0 . 28 0 . 00 -3 . 80 -0 . 77 0 .18 -9 . 64 18 .68 0 . 00 -4 . 93

0 • pH h « cc ! £ Std. dev. 0 . 06 0 . 07 0 . 06 0 . 76 2 . 01 0 .14 6 .49 0 . 76 0 . 44 11 . 83

Median 0 . 02 -0 . 01 0 . 02 0 .20 0 .10 0 .42 5 .10 20 . 28 0 . 00 17 .13

Mean 0 . 04 0 . 00 0 . 04 0 . 16 0 . 84 0 . 45 6 . 09 20 . 29 0 .26 16 . 53

Max 2 . 93 0 . 07 0 . 07 0 . 74 6 .91 0 . 77 191 . 21 22 .45 1 . 00 29 . 52

no « Min -0 . 02 -0 . 04 0 . 00 -10 . 05 -0 . 99 0 . 06 -1 . 65 18 . 87 0 . 00 -38 . 34

TS 0 • pH h « a Std. dev. 0 . 30 0 . 03 0 . 01 1 .15 0 .95 0 .11 19 .25 0 . 80 0 . 49 9 . 39

cc 6 Median 0 . 02 -0 . 01 0 . 02 0 . 00 0 .15 0 .46 1 io co 20 . 95 0 . 00 14 . 38

Mean 0 . 06 0 . 00 0 . 02 -0 . 13 0 . 30 0 . 45 8 . 60 20 .98 0 . 38 13 .46

Max 2 .93 0 . 30 0 . 30 1 . 55 8 . 11 1 . 01 191 . 21 22 .45 1 . 00 43 . 08

« (S 1—1 TS Min -0 . 28 -0 .28 0 . 00 -10 . 05 -0 . 99 0 . 06 -9 . 64 18 .68 0 . 00 -38 . 34

0 • pH h « a « p-H t ä M 1—1 i-H s Std. dev. 0 .23 0 . 05 0 . 04 1 . 00 4 io ,—i 0 . 13 14 .94 0 . 85 0 . 47 10 .64

Median 0 . 02 -0 . 01 0 . 02 0 . 07 0 .13 0 .45 6 . 05 20 . 76 0 . 00 15 .29

h Mean 0 . 05 0 . 00 0 . 03 0 . 00 0 . 54 0 . 45 7 . 47 20 .67 0 . 33 14 . 84

Variable LLPV DLPV ADLPV BDI ANPL LAS LVR LTA LTN Z-index

Table 7

Correlation matrix and Variance Inflation Factor

BDI IFRS RIS ANPL LAS LVR LTA LTN

BDI 1 . 00

IFRS -0 . 14 1 . 00

RIS -0 . 36* 0 . 07 1 . 00

1NPL 0 . 06 -0 .17* -0 . 10 1 . 00

LAS -0 . 07 0 . 01 -0 . 02 0 . 18* 1 . 00

LVR -0 . 09 0 . 08 0 .14 -0 . 05 -0 . 12 1 . 00

LTA 0 .23* 0 . 41* -0 .24* -0 . 05 0 . 03 -0 . 12 1 . 00

LTN 0 . 06 0 .12 0 . 07 0 . 03 0 .10 -0 . 03 0 .47* 1 . 00

VIF 1 . 26 1 . 37 1 . 28 1 . 08 1 . 07 1 . 06 1 . 87 1 . 38

1/VIF 0 . 79 0 . 73 0 . 78 0 .93 0 . 94 0 .94 0 . 53 0 . 73

Notes: * — indicates significance at 95 % confidence level; mean VIF is 1. 30 .

scenarios are reported by [Gebhardt, Novo-tny-Farkas, 2011; Leventis, Dimitropoulos, Anandarajan, 2011; 2012; Adzis, Tripe, Dun-more, 2016; Arbak, 2017; Amidu, Kuipo, 2019; Ozili, Outa, 2019].

Economic crisis effect of loan loss bank-specific decisions

Since the study's sampled period (20072017) falls within time of global financial crisis (GFC) of 2007-2009 and local economic recession (LFC) of 2016-2017 in Nigeria, it is deemed necessary to examine the economic crisis effect of Nigerian DMBs' provisioning behaviour via various bank-level decisions . The results of the regression estimates (models specified in the Appendix 3) are presented in Table 9 with "crisis effect" and "crisis and insolvency risk effect" in separate columns for both LLPV and ADLPV models . As obtainable in Table 9, it is evident that Nigerian DMBs charge higher LLPVs during global financial crisis (GFC) indicating loan loss pro-cyclicality which eventually results in higher income smoothing as the coefficients of GFC are significantly positive in both LLPV and ADPLV models . During local economic recession (LFC), provisioning behaviour appears to be counter-cyclical as

the coefficient of LFC in the LLPV models is negative except in the model without interaction variables .

However, counter-cyclical provisioning during LFC as evident under LLPV models could not lead to clear-cut income-smoothing practices as the coefficients of LFC in all ADLPV models which are positive are not statistically significant . The evidence of pro-cyclical provisioning identifiable with Nigerian DMBs during GFC is similar to the findings of [Adzis Anuar, Mohd Hishamuddin, 2015; Arbak, 2017] for Malaysian and Belgian banking respectively while loan loss counter-cyclicality found during LFC is comparable to the conclusion of [Caporale et al . , 2018] regarding 2011-2015 Italian recession . Both findings are, however, in contrast with those of [Salami, Uthman, Ajape, 2022] .

The pro-cyclical provisioning identifiable with Nigerian banking environment during GFC can also be confirmed in the Nigerian DMBs' loan loss bank-specific decisions during GFC as the coefficients of GFC • BDI and RIS • GFC • BDI are positive and statistically significant in LLPV models This shows that all categories of Nigerian banks, including financially-weak ones, are involved in imprudent provisioning practices in their bank-specific decisions during GFC. Not-

Table 8

Bank-Specific Decisions Index (BDI) and provisioning behaviour

Dependent variable: LLPV Dependent variable: ADLPV

Variable Without moderation With moderation Without moderation With moderation

BDI -0 .2174 (0 .0595)* -0 . 0178 (0 . 0128) -0 . 0052 (0 .0021)X -0 .0105 (0. 0051)X

IFRS -0 . 0513 (0 . 0146)* -0 . 0221 (0 .0052)*

IFRS • BDI 0 .0638 (0 .0219)* 0 .0402 (0 .0148)*

RIS -0 . 0069 (0 . 0090) 0 .0167 (0 .0038)*

RIS • BDI -0 .0455 (0 .0182)X 0. 00013 (0 .0097)

IFRS • RIS • BDI -0 .2786 (0 . 0349)* -0 .0349 (0 .0189)0

ANPL -0 .0006 (0 .0072) 0. 0032 (0 . 0019)0

LAS 0 .0320 (0 .0131)X 0 . 0263 (0 . 0127)X

LVR -0. 00115 (0 . 00042)X -0 . 0013 (0 . 0004)* -0. 00013 (0 .00006)X -0 .00008 (0. 0002)

LTA -0. 0664 (0 . 0373)0 0 . 0270 (0 . 0111)X -0 .0149 (0 .0026)* -0 . 0024 (0 .0025)

LTN -0 .0242 (0 .025)) 0. 0083 (0 . 01281) -0 .0001 (0 .0025) -0 .0066 (0. 0026)X

Const 1 .4361 (0 .7714)0 -0 .5074 (0 .2219)X 0 . 3249 (0 . 0550)* 0 .0729 (0 .0544)

HST 39 .56 (0 .0000)* 19 .98 (0 .0294)X 9 .37 (0 .0953)0 13 .28 (0 .2084)

PFE-HET 6802 .62 (0 .0000)* 24252 . 34 (0 .0000)*

BPLM 0. 00 (1 . 0000) 0. 00 (1 . 0000)

POLS-H1 8 . 54 (0 . 0035)* 19 .97 (0 . 0000)*

POLS-H2 9 .39 (0 .0946)0 23 .39 (0 . 0094)*

PAR-1 0 .133 (0 .7204) 19 .191 (0 .0005)* 36 . 509 (0 . 0000)* 24 .551 (0 .0002)*

CR-DPD 273. 345 (0. 0000)* 332. 761 (0. 0000)* 435 .155 (0,0000)* 474. 575 (0. 0000)*

R2 0 . 7145 0. 8795 0 .1849 0 .1282

F-Stat/Wald 10.11 (0. 0002)* 549. 30 (0. 0000)* 100. 80 (0. 0000)* 120. 53 (0. 0000)*

Observation 169 169 169 169

Model type PFE with RSE PCSE PCSE PCSE

Notes: regression coefficients are reported with standard errors in parentheses; diagnostic tests other than R2 report statistics with p-value in parentheses; 0, X, and * indicate significance at 90, 95 and 99 % confidence levels respectively; where a cell is empty, the item in the extreme left of the row is not applicable .

withstanding imprudent provisioning and evidence of income smoothing during GFC, bank-specific decisions of Nigerian DMBs are not geared towards income smoothing during GFC as the coefficients of GFC • BDI and RIS • GFC • BDI are negative in the ADPLV models .

Similarly, counter-cyclical/prudent provisioning and no clear-cut earnings-smoothing practices during LFC appears not to be sustainable as bank-specific decisions are

imbued with income smoothing for all banks (given positive coefficient of LFC • BDI) other than those threatened by risk of insolvency (based on positive but insignificant coefficient of RIS • LFC • BDI) . Since LFC fell within IFRS period, it is a fact based on the positive coefficients of LFC • BDI and RIS • LFC. BDI that banks' maneuvering of IFRSs in their loan loss decisions does not have economic boundary . This shows that IFRS reporting of loan losses during eco-

BDI, provisioning behaviour and economic crisis effect

Dependent variable: LLPV Dependent variable: ADLPV

Variable Without moderation With moderation Without moderation With moderation

Crisis effect Crisis effect Crisis & risk effect Crisis effect Crisis effect Crisis & risk effect

BDI -0.212 (0.06)* -0.048 (0.005)* -0.046 (0.006)* -0.006 (0.002)* -0.008 (0.003)* -0.004 (0.003)

GFC -0.004(0.038) 0.024 (0.011)A 0.030 (0.009)* 0.014 (0.004)* 0.016 (0.004)* 0.011 (0.005)A

LFC 0.049 (0.019)A -0.005 (0.011) -0.036 (0.013)* 0.002 (0.004) 0.004 (0.004) 0.002 (0.004)

GFC-BDI 0.025 (0.006)* -0.014 (0.004)*

LFC-BDI -0.217 (0.005)* 0.005 (0.003)"

BIS -0.011 (0.009) 0.012 (0.004)*

BIS- GFC-BDI 0.020 (0.009)A -0.018 (0.005)

BIS-LFC-BDI -0.218 (0.006)* 0.002 (0.004)

ANPL -0.001 (0.009) 0.005 (0.001)* 0.006 (0.001)*

LAS 0.024 (0.012) -0.006 0.010) -0.002 (0.012)

L\B -0.001 (0.0004)A -0.0004 (0.0002)" -0.0002 (0.0002) -0.0001 (0.0001) -0.0002 (0.0001)A -0.0002 (0.0001)°

LTA -0.089 (0.037)A 0.005 (0.002)A 0.008 (0.004)A -0.013 (0.002)* -0.010 (0.002)* -0.008 (0.002)*

LTN -0.019 (0.025) 0.0002 (0.004) 0.004 (0.007) 0.0001 (0.002) -0.001 (0.002) -0.002 (0.002)

Const 1.893 (0.771)A -0.074 (0.049) -0.135 (0.081)" 0.285 (0.046)* 0.235 (0.038)* 0.186 (0.051)*

HST 42.12 (0.0000)" 643.22(0.0000)* 11.89 (0.2196) 8.52 (0.5786) 10.74 ( 0.1504) 10.76 (0.2926) 8.88 ( 0.5434)

PFE-HET 9194.6 (0.000)*

BPLM 0.00 (1.0000) 0.00 (1.0000) 0.00 (1.0000) 0.00 (1.0000) 0.00 (1.0000)

POLS-HI 24.63 (0.0000)* 21.82 (0.0000)* 10.51 (0.0012)* 18.25 (0.0000)* 19.49 (0.0000)*

POLS-H2 29.58 (0.0005)* 26.53 (0.0031)* 21.06 (0.0119) À 21.03 (0.0125)À 22.24 ( 0.0139)À

PAB-1 0.107 (0.7479) 5.089 (0.0394)À 5.971 (0.0274) 27.556 (0.0001)* 9.484 (0.0076)* 8.996 (0.0090)*

CB-DPD 278.3 (0.0000)" 379.5 (0.0000)* 407.5 (0.0000)* 435.98 (0.0000)* 436.2 (0.0000)* 445.6 (0.0000) *

B2 0.7196 0.8909 0.9102 0.1931 0.1287 0.1381

F-Stat/Wald 21.38 (0.0000)" 33180.96 (0.0000)* 57180.95 (0.0000)* 142.08(0.0000)* 231.47 (0.0000)* 152.62 (0.0000)*

Observation 169 169 169 169 169 169

Model type PFE with BSE PCSE PCSE PCSE PCSE PCSE

Notes: regression coefficients are reported with standard errors in parentheses; diagnostic tests other than E" report statistics with p-value in parentheses; o, À, and * indicate significance at 90, 95 and 99% confidence levels respectively; GFC is measured "1" for GFC period and "0" otherwise; LFC is also dichotomous like GFC; where a cell is empty the item in the extreme left of the row is not applicable.

nomic crisis is a function of income smoothing practices . The behaviour of other variables, the control variables, is similar to what is obtainable in Table 8 .

Supplementary empirical regression estimates

Though this study advances relevant literature via conceptualisation of BDI, continued specification of components of BDI by recent studies [Muriu, Josea, 2020; Salami et al . , 2022; Salami, Uthman, 2022; Tran, Hassan, Houston, 2020] necessitates further analyses as presented in Tables 10, 11 and 12 . While Tables 10 and 11 depict the results of managerial discretionary components of BDI following the approach of [Curcio, Hassan, 2015], Table 12 presents the results of estimates related to funding and diversification components as adopted by [Amidu, Kuipo, 2015] . This is in addition to the interaction variable approach of [Salami, 2021]. Similar approach adopted in the estimates presented in Table 8 is followed in Tables 10, 11 and 12 but relevant econometric models are specified in the Appendix 3 .

From Tables 10 and 11, use of actual provisions (LLPVs) to smooth earnings indicated by positive coefficient of BEBLT and manage bank capital indicated by negative coefficients of CCR and TCR might be responsible for the negative coefficient of BDI in Table 8 The management of capital and earnings has the potential to prompt increase in net income as engendered by negative coefficient of BDI. However, the case of non-use of DLPV to smooth earnings and signal might explain the negative coefficient of BDI in ADLPV models . The non-use of DLPV to manage capital and earnings alongside the negative coefficient of BDI in ADLPV model are consequent upon improved financial reporting quality of Nigeria banks . The rejection of third hypothesis owing to positive coefficient of IFRS • BDI in ADLPV model may have something to do with two cases of management and signalling of

earnings indicated by positive coefficients of IFRS • BEBLT and IFRS • SINL respectively in Tables 10 and 11 .

There is also a connection between no clear-cut increased discretions in provisioning practices of riskier Nigerian DMBs prompted by provisioning-inclined bank-specific decisions in Table 8 and virtually no discretionary use of LLPVs found for financially unhealthy banks in Tables 10 and 11 . However, similar circumstance cannot also be adduced for nexus between IFRS • RIS • BDI in Table 8 and findings for managerial discretionary use of LLPVs by financially comatose DMBs in Tables 10 and 11 .In Table 8, for instance, the significantly negative coefficient of IFRS • RIS • BDI provides that BDI of riskier Nigerian DMBs is an ingredient of improved financial reporting quality during IFRS. In contrasts, signs of coefficients of each component of BDI especially in ADLPV models are not in any way causative to loan loss reporting quality owing to increased DLPV as evident in Tables 10 and 11 .

In Table 12, the higher number of significant coefficients of funding and diversification components of BDI in the ADLPV model (without interaction variables) confirms the evidence of adjustments to LLPVs as a result of bank-specific decisions reported in Table 8 Similarly, the case of adjustments to LLPVs not imbued with earnings manipulations given the higher number of insignificant and negative components of BDI in the ADLPV model with moderating variables is another confirmation Second, the fact that there is no improvement in the financial reporting quality of banks in Nigeria during IFRS can be inferred from having six out of eight components in that category with positive coefficients Third, the significantly negative coefficients of three out of four diversification components and insignificant two and one coefficients of funding and diversification components respectively for riskier DMBs in Table 12 confirm that RIS • BDI is not favourable to earnings smoothing/management as contained in

Table 10

Discretionary attributes and provisioning behaviour of Nigerian DMBs (CCR)

Dependent variable: LLPV Dependent variable: ADLPV

Variable Without moderation With moderation Without moderation With moderation

CCR -0.8925 (0 .3241)x 0 . 1268 (0 .0759)0 -0 .0223 (0. 0098)x -0 .0034 (0 . 0248)

BEBLT 1 . 5634 (0 .2896)* 2 .2743 (0 .8222)* -0 . 7010 (0 .0784)* -0 .4761 (0 ,1991)x

SINL -0 .1629 (0 .6317) 0 .6323 (0 . 5229) -0 . 3094 (0 .0844)* 0 . 1197 (0 .1304)

IFRS 0. 1641 (0 . 0201)* -0. 0353 (0 .0130)*

IFRS•CCR -0 .3185 (0 .1074)* 0 . 0221 (0 .0356)

IFRS • BEBLT -2 .7829 (0 .7216)* 0 .7582 (0 .2491)*

IFRS• SINL -0 .9194 (0 .6003) 0 . 0414 (0 .1921)

RIS 0 .1552 (0 . 0344)* 0 .0179 (0 . 0100)0

RIS• CCR -0 .00481 (0 . 0855) 0 .0985 (0 .0369)*

RIS • BEBLT -3 .2619 (0 .8701)* -0.9020 (0 .1722)*

RIS•SINL -1 .0506 (0 . 5346)x -0.8298 (0 .1667)*

IFRS• RIS• CCR -1 .0803 (0 .1408)* -0. 1368 (0 .0337)*

IFRS • RIS • BEBLT 1 .4453 (0 .8128)0 0 .8000 (0 .2552)*

IFRS • RIS • SINL -0 . 5506 (0 .8208) 0 .6764 (0 .2321)*

ANPL 0 .0050 (0 . 0071) 0 .0061 (0 . 0019)*

LAS -0. 0307 (0 .0200) 0 . 0007 (0 .0229)

LVR -0 .0001 (0 .0003) -0 .0019 (0 .0002)* -0. 0003 (0 .00012)x -6 . 95e-06 (0 .0001)

LTA -0 .0712 (0 .0448) 0 .0203 (0 . 0067)* -0 .0060 (0. 0026)x -0 .0049 (0 .0024)x

LTN -0 .0215 (0 .0243) 0 .0149 (0 . 0155) 0 .0036 (0 .0040) -0 .0017 (0 . 0038)

Const 1 .6154 (0 .9638) -0 . 5071 (0 .1567)* 0 .1896 (0. 0537)* 0 .1407 (0 . 0554)x

HST 24 .47 (0 .0009)* 33 .72 (0 .0091)* 6 .65 (0 .4663) 5 .83 (0 .9943)

PFE-HET 27266 .95 (0 . 0000)* 23631 . 70 (0 .0000)*

BPLM 0 .00 (1 .0000) 0 .00 (1 .0000)

POLS-H1 37 .94 (0 .0000)* 22 .28 (0 . 0000)*

POLS-H2 38 . 05 (0 .0000)* 38 .95 (0 . 0029)*

PAR-1 3 .550 (0 . 0791)0 6 .156 (0 . 0254)x 13 .504 (0. 0023)* 13 .579 (0 .0022)*

CR-DPD 313 .744 (0. 0000)* 346 .923 (0 .0000)* 375 . 125 (0 .0000)* 452 .356(0 .0000)*

R2 0 .6390 0 .8912 0 .4832 0 . 3728

F-Stat/Wald 136 .95 (0 .0000)* 1380 .36 (0 .0000)* 588 .63 (0. 0000)* 230 .06 (0 .0000)*

Observation 169 169 169 169

Model type PFE with RSE PCSE PCSE PCSE

Notes: regression coefficients are reported with standard errors in parentheses; diagnostic tests other than R2 report statistics with p-value in parentheses; 0, À, and * indicate significance at 90, 95 and 99 % confidence levels respectively; where a cell is empty the item in the extreme left of the row is not applicable .

Table 11

Discretionary components and provisioning behaviour of Nigerian DMBs (TCR)

Dependent variable: LLPV Dependent variable: ADLPV

Variable Without moderation With moderation Without moderation With moderation

TCR -1 .0234 (0 .1956)* 0 . 1103 (0 . 0714) -0 .0200 (0 .0105)0 0 .0055 (0. 0251)

BEBLT 1 .4828 (0 .6010)x 2 .4229 (0 .6753)* -0 . 7790 (0 . 0703)* -0. 3948 (0 . 1980)x

SINL -0 .4355 (0 .7642) 0 . 5111 (0 .4769) -0 . 3675 (0 .0749)* 0 .1352 (0. 1309)

IFRS 0 . 1745 (0 .0227)* -0 .0357 (0 .0113)*

IFRS • TCR -0 .4176 (0. 1062)* 0 .0443 (0. 0379)

IFRS • BEBLT -2 .590 (0 .6587)* 0 .5920 (0.2061)*

IFRS • SINL -0 .7170 (0 .5733) 0 .0735 (0 .1955)

RIS 0 . 1991 (0 .0310)* 0 .0202 (0 .0097)x

RIS•TCR -0 .4525 (0. 0791)* 0 . 0639 (0 .0354)0

RIS • BEBLT -2 .3261 (0. 7666)* -0 .7190 (0 .2464)*

RIS•SINL -0.6217 (0 .5801) -0 .7864 (0 .1786)*

IFRS • RIS • TCR -0 .4945 (0. 1205)* -0 .1323 (0 .0375)*

IFRS • RIS • BEBLT -0 .6473 (0 .8398) 0 . 7760 (0 .3334)x

IFRS • RIS • SINL -1 .8033 (0 .8778)x 0 .6026 (0 .2424)x

ANPL -0 .00001 (0. 0061) 0 .0041 (0 .0019)x

LAS -0 . 0232 (0 . 0195) -0 .0127 (0 .0218)

LVR -0 .0014 (0 .0002)* -0 .0016 (0. 0003)* -0. 00024 (0 . 00006)* -0 .00002 (0 . 0001)

LTA -0 .0809 (0. 0334)x 0 . 0295 (0 .0074)* -0 .0070 (0 . 0024)* -0 .0016 (0 .0023)

LTN -0 . 0139 (0 .0266) 0 . 0129 (0 . 0181) 0 .0038 (0 .0048) 0 .0003 (0. 0044)

Const 1 .8832 (0 .7156)x -0 .6973 (0. 1675)* 0 .2099 (0 .0503)* 0 .0723 (0. 0496)

HST 46 .97 (0 .0000)* 25 . 30 (0 . 0882)0 4 .84 (0 .6800) 4 .92 (0 .9980)

PFE-HET 5462 .68 (0 .0000)*

BPLM 0 . 00 (1 . 0000) 0 .00 (1 .0000) 0. 00 (1 .0000)*

POLS-H1 35. 33 (0 .0000)* 34 . 37 (0 . 0000)* 32 .99 (0 .0000)*

POLS-H2 55.27 (0 .0000)* 35 .23 (0 . 0000)* 41 .00 (0 .0015)*

PAR-1 0 .618 (0 .4440) 7 . 553 (0 . 0149)x 9 . 051 (0 . 0088)* 5 .995 (0 .0254)x

CR-DPD 300 .320 (0 .0000)* 268.682 (0. 0000)* 369 . 535 (0 . 0000)* 404 .069 (0 .0000)*

R2 0 . 7612 0.8898 0 .3888 0 . 5181

End of Table 11

Dependent variable: LLPV Dependent variable: .ADLPV

Variable Without moderation With moderation Without moderation With moderation

F-Stat/Wald 216 .99 (0 .0000)* 2086 .72 (0 .0000)* 468 .19 (0 . 0000)* 607 .18 (0 .0000)*

Observation 169 169 169 169

Model type PFE with RSE PCSE PCSE PCSE

Notes: regression coefficients are reported with standard errors in parentheses; diagnostic tests other than R2 report statistics with p-value in parentheses; 0, X, and * indicate significance at 90, 95 and 99 % confidence levels respectively; where a cell is empty the item in the extreme left of the row is not applicable .

Table 8 . While three positive and significant coefficients of BDI diversification components of riskier DMBs in the ADLPV model during IFRS presented in Table 12 reverse similar results presented in Table 8, two negative coefficients on one hand and two insignificantly positive coefficients on the other hand of the funding components reinforce, in comparisons with estimates in Table 8, that there is considerable improvement in financial reporting quality of Nigerian riskier banks during IFRS.

CONCLUDING REMARKS

This study addresses the problems revolving around non-performing loans crisis which has become a major feature of Nigerian banking system in the recent time The problems which have overall influence on accounting for loan losses are linked to quite a number of issues in the literature . These issues, though having a long list, can be categorised into those affecting bank-specific decisions and macroeconomic matters in loan loss reporting The bank-specific decisions which also have a long list of pressing decisions principally classified as managerial discretionary, funding strategy and diversification strategy matters are empirically addressed in this study These bank-specific decisions are condensed into an index (BDI), as advancement in the literature, and its relationship with loan loss provisioning

behaviour, measured as loan loss provisions (LLPVs) as reported in the banks' financial records and discretionary loan loss provisions (DLPV), having reference to risk of insolvency and reforms imbedded in IFRSs is examined. The data are extracted from the audited annual financial reports of a sample of 16 Nigerian DMBs over a period 20072017 divisible into pre-IFRS and IFRS periods .

The results from panel regression analyses subsequent to the tests of four research hypotheses reveal overall that adjustments to LLPVs while making various bank-specific decisions are characteristic of Nigerian DMBs but not laden with earnings smoothing/management in the form DLPV. For financially unhealthy DMBs, the adjustments, as found, are geared towards increased profitability but do not have conclusive evidence of being imbued with earnings manipulations . For the reforms imbedded in IFRSs, the impact is favourably parochial as the improvements are only identifiable with financially unhealthy DMBs rather than all sampled DMBs These overall findings are also evident, to a large extent, in the relationship between provisioning behaviour and individual components of BDI. In relation to economic crisis, the study establishes that the loan loss counter-cyclicality, the hallmark of Nigerian DMBs' provisioning behaviour during LFC (which fell within IFRS period) is manipulated to smooth earnings

Table 12

Diversification and funding components and provisioning behaviour of Nigerian DMBs

Dependent variable: LLPV Dependent variable: ADLPV

Variable Without moderation With moderation Without moderation With moderation

RDIV 0 .0519 (0 .0306)0 0 .0068 (0 .1722) 0 . 0446 (0 . 0092)* 0 .0887 (0 .0813)

NDIV 0 . 1142 (0 .0404)* -0 . 0058 (0 .0238) -0 .0002 (0 .0144) 0. 0414 (0 . 0172)x

EDIV -0 . 3278 (0 .1754)0 0 .0330 (0 .1293) -0 .0512 (0 .0400) -0. 0521 (0 . 0490)

BDIV 0 . 0821 (0 .0636) 0 .0035 (0 .0543) -0 .0710 (0 .0170)* 0 .0328 (0 .0226)

FLD 0 . 0058 (0 .0419) 0 . 0446 (0 . 0239)0 0 .0576 (0 .0327)0 -0 . 1027 (0 . 0287)*

FDEP 1 . 0074 (0 .1110)* 0 .2350 (0 .0704)* 0 .2355 (0 . 0356)* -0 .1446 (0 .0607)x

FNDP 0 .8894 (0 .1221)* 0 .1534 (0 .0959) 0 . 1550 (0 . 0216)* 0 .0021 (0 .0464)

FIGR 0 . 7062 (0 .1230)* 0 .2687 ( 0. 1201)x -0 .2131 (0 .0284)* 0 .1274 (0. 0752)0

IFRS -0 .2881 (0 .2335) -0. 1476 (0 . 0936)

IFRS • RDIV 0 . 0145 (0 .1616) -0 .1701 (0 .0968)0

IFRS • NDIV 0 .0416 (0 .0306) -0. 0151 (0 . 0221)

IFRS • EDIV 0 .2550 (0 . 1536)0 0 .1049 (0 .0651)

IFRS • BDIV 0 .0061 (0 .0502) 0 .0058 (0 .0287)

IFRS • FLD 0 .0492 (0 .0385) 0 .0074 (0 .0363)

IFRS • FDEP 0 .1079 (0 .2145) 0 .1987 (0 .0761)*

IFRS • FNDP 0 .1572 (0 .2051) 0. 1681 (0 . 0828)x

IFRS • FIGR 0 .0472 (0 .1676) 0 .1856 (0 .1234)

RIS -0 .3726 (0. 1631)x -0. 0909 (0 . 0880)

RIS • RDIV 0 .0573 (0 .1741) 0 .0013 (0 .0703)

RIS • NDIV 0 .2990 (0 .0682)* -0 . 1621 (0 . 0321)*

RIS • EDIV 0 .1831 (0 .1879) -0 .3052 (0 ,1238)x

RIS • BDIV 0 .0507 (0 .0884) -0 .2906 (0 . 0289)*

RIS • FLD 0 .0129 (0 .0637) 0. 0917 (0 . 0372)x

RIS • FDEP 0 .2315 (0 . 1269)0 0 .4791 (0 .0939)*

RIS • FNDP -0 . 0399 (0 .1274) 0 .0530 (0 .0645)

RIS• FIGR -0 .4835 (0 .1372)* -0 .0783(0 .0808)

IFRS • RIS • RDIV -0 .2091 (0 .1767) 0 .1339 (0 .1337)

IFRS • RIS • NDIV -0 .2809 (0 .0810)* 0 .1255 ( 0 .0403)*

IFRS • RIS • EDIV -0 . 1988 (0 .2276) 0 .5078 (0 .1362)*

IFRS • RIS • BDIV -0 .1190 (0 . 1018) 0 .1842 (0 .0567)*

IFRS • RIS • FLD -0 . 0811 (0 .0790) -0. 0597 (0 . 0506)

End of Table 12

Dependent variable: LLPV Dependent variable: ADLPV

Variable Without moderation With moderation Without moderation With moderation

IFRS • RIS • FDEP 0 .2699 (0 ,1248)x -0 . 5601 (0 . 0969)*

IFRS • RIS • FNDP 0 .7615 (0. 1942)* 0 .0398 (0 .1307)

IFRS • RIS • FIGR 1 .2355 (0.2808)* 0 .0135 (0 .1793)

ANPL 0 . 0080 (0 .0024)* 0 .0082 (0. 0009)*

LAS -0 .0742 (0. 0563) 0. 1335 (0 . 0563)x

LVR -0 .0002 (0 .0004) -0 .00039 (0. 00019)x -0. 0005 (0 . 0001)* -0 .0005 (0 .00018)*

LTA -0 .0102 (0 .0085) -0 . 0056 (0 .0041) -0. 0109 (0 . 0036)* -0 . 0077 (0 . 0025)*

LTN -0 .0275 (0 .0107) -0 .0117 (0. 0057)x 0 . 0024 (0 .0047) 0 .0025 (0 .0036)

Cons -0 . 5758 (0 ,2712)x -0 .1079 ( 0 . 1596) 0 . 0915 (0 .0742) 0.25244 (0 . 0873)*

HST 39.81 (0 .0001)* 52 .90 (0 .0547)0 12 . 75 (0 .3873) 10 .20 (1 . 0000)

PFE-HET 3853 . 58 (0. 0000)*

BPLM 0 .00 (1 .0000) 0 .00 (1 . 0000) 0 .00 (1 .0000)

POLS-H1 0 .04 (0 .8420) 41 .69 (0 .0000)* 96 . 05 (0 . 0000)*

POLS-H2 66 .65 (0 .0028)* 43 .81 (0. 0000)* 124 .78 (0 .0000)*

PAR-1 11 .036 (0 .0046)* 28. 029 (0. 0001)* 5 .945 (0 .0279)x 5 .839 (0 .0289)x

CR-DPD 333.850 (0. 0000)* 420 .893 (0 .0000)* 407 . 081 (0 . 0000)* 420 . 794 (0 . 0000)*

R2 0 . 9111 0 .9811 0.4393 0 .6457

Wald 5692 . 50 (0. 0000)* 487602 (0 .0000)* 401 .70 (0 .0000)* 14424. 74 (0 .0000)*

Observation 169 169 169 169

Model type PCSE PCSE PCSE PCSE

Notes: regression coefficients are reported with standard errors in parentheses; diagnostic tests other than R2 report statistics with p-value in parentheses; 0, X, and * indicate significance at 90, 95 and 99 % confidence levels respectively; where a cell is empty the item in the extreme left of the row is not applicable

when pressing loan loss bank-specific decisions are made In contrast, the unfavourable provisioning behaviour of loan loss pro-cyclicality evident during GFC is not found to warrant practices of income smoothing when loan loss bank-specific decisions are considered

Though IFRS period has a general feature of reduction in DLPV as found, the reduction is however not reflected in the Nige-

rian DMBs' loan loss reporting decisions . This is an indication of failure of reforms in corporate Nigeria The inability of reforms to materialize fully may affect the leadership role of Nigerian banks who have built a niche in the West African Sub-region . This may also inhibit their access to global capital markets and funds as many Nigerian banks are known for accessing global depository receipts to boost their capital

base . For the case of Nigerian DMBs threatened by risk of insolvency that appears to be favourable, the situation is attributed to the level of attention of regulators on them The role being played by CBN, AMCON and Nigeria Deposit Insurance Corporation in this perspective is evident in the corporate Nigeria . The Nigerian banking scenario, as a matter of urgency, beckons similar attention being paid on all banks to save investors' and depositors' funds . These regulators can employ the services of independent consultants if they are short of competent hands that can carry out the assignments on regular basis

The need for requisite competencies may also be prompted by the extent of discretions imbedded in the application of IFRS 9. Having established the situation in IAS 39 regime, managerial discretionary tendencies may be pervasive leading to increased earnings smoothing practices by Nigerian DMBs in the IFRS 9 regime . To avoid a replica of IAS 39 scenario or worsening one, CBN should develop a more rule-based guidelines of application of IFRS 9 with little or no opportunity for "comply or explain" nature of IFRS standards . This is

without prejudice to the previous ones developed by the body . This has the capacity to enhance the ability of regulators to detect areas of lapses and/or infringements and impose sanctions where appropriate

The palpable evidence of increased earnings manipulations as a result of provision-ing-inclined bank-specific decisions beckons the return of disclosure of off-balance sheet items below statement of financial position as used to be prior to the IFRS adoption . The present approach of disclosure only in the notes appears an instrument of earnings smoothing since it is too sophisticated for investors/users without requisite knowledge to discern The access to the financial records of primary mortgage banks and related financial institutions majority of which is not in public domain will better the generalisations of findings of future research It will also enable comparisons with those of DMBs established in this study The findings of this study also call for a comparative analysis, in a future study, between IAS 39 and IFRS 9 regimes in accounting for loan losses but become meaningful post-partial IFRS 9 adoption era .

REFERENCES

Abdullah H. , Bujang I . , Ahmad I . 2013 . Loan loss provisions and earnings management in Malaysian banking industry. Global Journal of Business and Social Science Review 1 (1): 93-104.

Abdullah H. , Bujang I . , Sahudin Z . 2017 . Empirical investigation on loan loss provisions: Evidence from Hong Kong and Japan . Advanced Science Letters 23 (9): 8766-8770. https://doi. org/10 ,1166/asl. 2017. 9967

Abdulmumin B .A. , Etudaiye-Muhtar O .F . , Jimoh A. T . , Sakariyahu O . R . 2019 . An investigation into the level of financial inclusion in Sub-Saharan Africa. Scientific Annals of Economics and Business 66 (1): 41-63 . https://doi. org/10. 2478/saeb-2019-0004

Abeyasekera S . 2005. Chapter XVIII: Multivariate methods for index construction . Reading (UK): University of Reading. [Electronic resource]. https://unstats. un. org/ unsd/hhsurveys/pdf/Chapter_18 . pdf (accessed: 12. 04. 2022) .

Abu-Serdaneh J . A . R . 2018 . Bank loan-loss accounts, income smoothing, capital management, signaling, and procyclicality: Evidence from Jordan . Journal of Financial Reporting and Accounting 16 (4): 677693. https://doi. org/10 . 1108/JFRA-06-2016-0041

Adzis A. A. , Anuar H . S . , Mohd Hishamuddin N . S . 2015 . Malaysian commercial banks: Do income smoothing, capital management, signaling, and pro-cyclicality exist

through loan loss provisions? International Journal of Economics, Finance and Management 4 (2): 68-76 . https://www . ejournalofbusiness . org/archive/vol4no2/vol-4no2_4.pdf (accessed: 15. 09. 2022) . Adzis A. A. , Tripe D . W . L . , Dunmore P . 2016 . IAS 39, income smoothing, and pro-cycli-cality: Evidence from Hong Kong banks . Journal of Financial Economic Policy 8 (1): 80-94. https://doi. org/10 . 1108/JFEP-05-2015-0026

Alessi M . , Di Colli S . , Lopez J. S . 2014. Loan loss provisioning and relationship banking in Italy: Practices and empirical evidence Journal of Entrepreneurial and Organizational Diversity 3 (1): 111-129 . http://dx . doi. org/10. 5947/jeod. 2014. 006 Amidu M . 2013 . The effects of the structure of banking market and funding strategy on risk and return . The International Review of Financial Analysis 28 (1): 143-155 . http://dx . doi. org/10 ,1016/j. irfa . 2013. 03 . 001 Amidu M . , Issahaku H . 2019 . Do globalization and adoption of IFRS by banks in Africa lead to less earnings management? Journal of Financial Reporting and Accounting 17 (2): 222-248. https://doi. org/10 .1108/ JFRA-05-2017-0035 Amidu M . , Kuipo R . 2015 . Earnings management, funding and diversification strategies of banks in Africa Accounting Research Journal 28 (2): 172-194. http:// dx. doi. org/10 . 1108/ARJ-07-2013-0045 Amidu M. , Wolfe S . 2013 . The impact of market power and funding strategy on bank-interest margins The European Journal of Finance 19 (9): 888-908. http://dx . doi . org/10. 1080/1351847X . 2011 . 636833 Anandarajan A. , Hasan I . , McCarthy C . 2007. Use of loan loss provisions for capital, earnings management and signalling by Australian banks . Accounting and Finance 47 (3): 357-379. https://doi.org/10 . 1111/j.1467-629X . 2007. 00220. x Andries K , Gallemore J. , Jacob M. 2017 . The effect of corporate taxation on bank transparency: Evidence from loan loss provisions Journal of Accounting and Economics 63 (2): 307-328. https://doi. org/10 ,1016/j. jacceco. 2017. 03. 004

Arbak E . 2017 . Identifying provisioning policies of Belgian banks. National Bank of Belgium Working Paper, Research No. 326. https://www . nbb . be/doc/ts/publications/wp/ wp326en . pdf Ashour M . O . 2011 . Banks Loan Loss Provision Role in Earnings and Capital Management — Evidence From Palestine. Doctoral dissertation, dissertation for the degree of master in accounting finance . Islamic University Gaza. http://library. iugaza. edu. ps/ thesis/98935. pdf Ashraf A . , Hassan M . K , Basher A . S . 2015 . Loan loss provisioning in organization of Islamic cooperation countries: Evidence from conventional vs . Islamic banks . Munich Personal RePEc Archive (MPRA) Working Paper No. 61687. https://mpra . ub. uni-muenchen. de/61687/1/MPRA_paper_61687. pdf Ashraf A , Hassan M . K , Putnam K , Turunen-Red A. 2019 . Prudential regulatory regimes, accounting standards, and earnings management in the banking industry Bulletin of Monetary Economics and Banking 21 (3): 367-394. htps://doi. org/10. 21098/ bemp . v21i3. 975. Atoyebi T . A. , Simon A. J. 2018 . The impact of International Financial Reporting Standards (IFRSs) adoption on financial reporting practice in the Nigerian banking sector . Journal of Accounting, Finance and Auditing Studies 4 (3): 170-197. Beaver W . H . , Engel E . E . 1996. Discretionary behaviour with respect to allowances for loan losses and the behaviour of security prices . Journal of Accounting and Economics 22 (1-3): 177-206. https://doi. org/10 . 1016/s0165-4101(96)00428-4 Beck N . , Katz J . N . 1995. What to do (and not to do) with time-series cross-section data American Political Science Review 89 (3): 634-647. Bhattarai B P 2018 Determinants of loan loss provisions of commercial banks in Nepal European Journal of Accounting, Auditing and Finance Research 6 (9): 23-37.

Bryce C . , Dadoukis A . , Hall M . , Nguyen L . , Simper R 2015 An analysis of loan loss provisioning behaviour in Vietnamese

banking . Finance Research Letters 14: 69-75 . https://doi. org/10 . 1016/j. frl . 2015 . 05 . 014

Caporale G . M. , Alessi M. , Di Colli S . , Lopez J. S . 2018 . Loan loss provisions and mac-roeconomic shocks: Some empirical evidence for Italian banks during the crisis . Finance Research Letters 25: 239-243 . https://doi . org/10 .1016/j. frl. 2017 .10. 031 Carbo-Valverde S . , Rodriguez-Fernandez F . 2018 . Do banks game on dynamic provisioning? Journal of Financial Management, Markets and Institutions 6 (1): 1-22 . htt-ps://doi . org/10 . 1142/S2591768418500022 CBN . 2010 . Prudential guidelines for deposit money banks in Nigeria. [Electronic resource]. https://www . cbn . gov . ng/out/2010/ publications/bsd/prudential%20guide-lines%2030%20june%202010%20final%20 %20_3_. pdf (accessed: 12 . 04. 2022) . Chang R . D . , Shen W . H . , Fang C . J . 2008. Discretionary loan loss provisions and earnings management for the banking industry. International Business and Economics Research Journal 7 (3): 9-20 . https://doi . org/10,19030/iber . v7i3. 3230 Chen L . , Emanuel D . M . , Li L . Z . , Yang M . 2021 . Regulatory changes and loan loss provisions management by Chinese banks . Journal of Accounting in Emerging Economies 11 (4): 651-675. doi: 10 .1108/ JAEE-05-2020-0116 Curcio D . , De Simone A. , Gallo A. 2017 . Financial crisis and international supervision: New evidence on the discretionary use of loan loss provisions at Euro Area commercial banks . The British Accounting Review 49 (2): 181-193. https://doi. org/10 .1016/j. bar . 2016. 09. 001 Curcio D . , Hasan I . 2015 . Earnings and capital management and signaling: The use of loan-loss provisions by European banks The European Journal of Finance 21 (1): 26-50. https://doi. org/10. 1080/1351847X . 2012 762408 De Winter J. C . , Dodou D . 2016 . Common factor analysis versus principal component analysis: A comparison of loadings by means of simulations . Communications in Statistics: Simulation and Computation

45 (1): 299-321 . https://doi. org/10.1080/03 610918. 2013. 862274 Doumpos M. , Gaganis C . , Pasiouras F . 2016 . Bank diversification and overall financial strength: International evidence . Financial Markets, Institutions & Instruments 25 (3): 169-213 . https://doi. org/10.1111/ fmii .12069

Duru K. , Tsitinidis A. 2013 . Managerial incentives and earnings management: An empirical examination of the income smoothing in the Nordic bankinglndustry. Master's Thesis. Uppsala University . [Electronic resource]. https://www . divaportal . org/smash/get/diva2:630941/FULL-TEXT01 .pdf (accessed: 12. 04. 2022) . Dushku E . 2016 . Some empirical evidence of loan loss provisions for Albanian banks . Journal of Central Banking Theory and Practice 5 (2): 157-173. https://doi. org/10 . 1515/jcbtp-2016-0016 Eiler L . A. , Miranda-Lopez J . , Tama-Sweet I . 2021. The impact of IFRS on earnings management: Evidence from Mexico . Journal of Accounting in Emerging Economies 12 (1): 77-96. https://doi . org/10 ,1108/JAEE-11-2020-0316 El Sood H . A. 2012 . Loan loss provisioning and income smoothing in US banks pre and post the financial crisis . International Review of Financial Analysis 25: 64-72. https://doi. org/10. 1016/j. irfa. 2012. 06 007

FRC 2015 . Regulatory decision in the matter of financial statements of Stanbic IBTC Holdings PLC for years ended 31st December 2013 and 2014. [Electronic resource] . https://drive. google.com/file/ d/0BxB1-bqcIt35aHh20XBFNFBneWM/ view?resourcekey=0-5fG0Zpp0KBwKNNB-WQZ7n9A (accessed: 12. 04. 2022) . Gambacorta L . , Scatigna M . , Yang J . 2014 . Diversification and bank profitability: A nonlinear approach Applied Economics Letters 21 (6): 438-441 . https://doi. org/10. 1080/13504851 2013 866196 Gebhardt G . , Novotny-Farkas Z . 2011 . Mandatory IFRS adoption and accounting quality of European banks . Journal of Business Finance & Accounting 38 (3-4): 289-333.

https://doi . org/10. 1111/j. 1468-5957 . 2011. 02242. x

Ghosh S 2007 Loan loss provisions, earnings, capital management and signalling: Evidence from Indian banks . Global Economic Review 36 (2): 121-136. https://doi. org/10. 1080/12265080701374040 Jaggi B . , Lee P . 2002. Earnings management response to debt covenant violations and debt restructuring . Journal of Accounting, Auditing and Finance 17 (4): 295-324. https://doi. org/10 . 1177 %2F0148558X0201700402 Jin J . Y . , Kanagaretnam K . , Liu Y . 2018 . Banks' funding structure and earnings quality . International Review of Financial Analysis 59: 163-178. https://doi. org/10 . 1016/j. irfa . 2018. 08. 009 Kanagaretnam K , Lobo G J , Mathieu R 2004 Earnings management to reduce earnings variability: Evidence from bank loan loss provisions Review of Accounting and Finance 3 (1): 128-148. https://doi. org/10 . 1108/eb043399 Kanagaretnam K , LoboG J , Mathieu R 2003 Managerial incentives for income smoothing through bank loan loss provisions . Review of Quantitative Finance and Accounting 20 (1): 63-80. https://doi. org/10. 1023/a:1022187622780 Kwak W , Lee Y H , Eldridge S W 2009 Earnings management by Japanese bank managers using discretionary loan loss provisions . Review of Pacific Basin Financial Markets and Policies 12 (1): 1-26 . https:// doi . org/10 . 1142/S0219091509001526 Leventis S . , Dimitropoulos P . E . , Anandarajan A .

2011 . Loan Loss Provisions, earnings management and capital management under IFRS: The case of EU commercial banks . Journal of Financial Services Research 40 (1): 103-122. https://doi. org/10,1007/s10693-010-0096-1

Leventis S . , Dimitropoulos P . E . , Anandarajan A .

2012 . Signalling by banks using loan loss provisions: The case of the European Union Journal of Economic Studies 39 (5): 604-618. https://doi. org/10.1108/01443581211259509

Mabel O . A. , Olayemi O . S . 2020. A comparison of principal component analysis, maximum likelihood and the principal axis in

factor analysis American Journal of Mathematics and Statistics 10 (2): 44-54. https://doi . org/10. 5923/j. ajms . 20201002. 03 Misman F . N . , Ahmad W . 2011. Loan loss provisions: Evidence from Malaysian Islamic and conventional banks International Review of Business Research Papers

7 (4): 94-103 .

Mohd Saleh N . , Ahmed K . 2005. Earnings management of distressed firms during debt renegotiation Accounting and Business Research 35 (1): 69-86. https://doi. or g/10. 1080/00014788. 2005. 9729663 Mukhibad H , Nurkhin A 2019 Islamic business ethics disclosure and earnings management: Evidence from Islamic banks in Indonesia Journal of Islamic Finance

8 (2): 31-42 .

Muriu P . , Josea K . 2020. What explains provisioning behaviour in the banking industry? Evidence from an emerging economy. Working Paper submitted to African Economic Research Consortium, May 2020 [Electronic resource]. https://aercafrica . org/ wp-content/uploads/2020/06/C2-Peter-Mu-riu-and-Josea-Kiplangat-FR- . pdf (accessed: 12. 04. 2022) . Nikulin E . , Downing J . 2021 . Loan-loss provisions, earnings management, and capital management by Russian banks: The impact of changes in banking regulation and oversight Eurasian Business Review 11 (4): 659-677 . https://doi. org/10.1007/ s40821-020-00163-9 Olarewaju O M , Migiro S O , Sibanda M 2017 . Operational diversification and financial performance of Sub-Saharan Africa commercial banks: Static and dynamic approach . Acta Universitatis Danubius Oeco-nomica 13 (5): 84-106. Olson D . , Zoubi T. A. 2014. The determinants of loan loss and allowances for MENA banks: Simultaneous equation and three-stage approaches . Journal of Islamic Accounting and Business Research 5 (1): 98-120. https://doi. org/10. 1108/ji-abr-07-2013-0027 Olszak M. , Pipien M. , Kowalska I . , Roszkows-ka S 2017 What drives heterogeneity of cyclicality of loan-loss provisions in the

EU? Journal of Financial Services Research 51 (1): 55-96. Othman B . H . , Mersni H . 2014. The use of discretionary loan loss provisions by Islamic banks and conventional banks in the Middle East region: A comparative study. Studies in Economics and Finance 31 (1): 106-128 . https://doi. org/10 . 1108/sef-02-2013-0017

Othman B . H . , Zhegal D . 2006. A study of earnings-management motives in the Anglo-American and Euro-Continental accounting models: The Canadian and French cases . The International Journal of Accounting 41 (4): 406-435 . https://doi. org/10 .1016/j. intacc . 2006. 09. 004 Otusanya O . J . , Uadiale O . M . 2014 . Creative accounting and financial scandals in Nigeria: Structures and actors analysis . International Journal of Economics and Accounting 5 (3): 191-215. https://doi. org/10. 1504/ijea . 2014. 067216 Ozili P . K 2015 . Loan loss provisioning, income smoothing, signaling, capital management and procyclicality: Does IFRS matter? Empirical evidence from Nigeria Mediterranean Journal of Social Sciences 6 (2): 224-232 . http://dx. doi. org/10. 5901/ mjss . 2015 . v6n2p224 Ozili P . K 2017a. Bank Income Smoothing and Loan Loss Provisioning Practices in Africa. Doctoral dissertation . University of Essex. https://repository . essex . ac. uk/20804/1/ Final%20Corrected%20Thesis%20for%20li-brary%20submission . pdf Ozili P . K 2017b . Bank earnings management and income smoothing using commission and fee income: A European context. International Journal of Managerial Finance 13 (4): 419-439. https://doi . org/10 . 1108/ IJMF-11-2016-0213 Ozili P . K , Outa E. R . 2018 . Bank income smoothing in South Africa: Role of ownership, IFRS and economic fluctuation . International Journal of Emerging Markets 13 (5): 1372-1394. https://doi. org/10 .1108/ IJoEM-09-2017-0342 Ozili P .K. , Outa E .R . 2019 . Bank earnings smoothing during mandatory IFRS adoption in Nigeria . African Journal of Economic and

Management Studies 10 (1): 32-47. https:// doi . org/10 . 1108/AJEMS-10-2017-0266 PinhoP . S . , Martins N . C . 2009. Determinants of Portuguese bank's provisioning policies: Discretionary behaviour of generic and specific allowances . Journal of Money, Investment and Banking 10: 43-56. Proshare 2017 . Skye bank Plc: One year after CBN takeover 1 (152). August . [Electronic resource]. https://www. proshareng. com/ admin/upload/reports/SkyeBankPlcOne-YearAfterCBNTakeover . pdf (accessed: 12 . 04. 2022) . Salami A . A . 2021 . Bank Corporate Decisions and Loan Loss Provisioning Practices Among Deposit Money Banks in Nigeria. Doctoral dissertation, Kwara State University Nigeria . Publication No . 28491780. ProQuest Dissertations & Theses Global.

Salami A. A. , Uthman A. B . 2022. Bank diversification strategy and discretionary provisioning in Nigeria: Do risk and IFRS adoption matter? Babcock Journal of Accounting and Finance 2 (1): 1-35 . Salami A . A. , Uthman A. B . , Abdulrauf L . A. O . 2021 . Signaling behaviour and bank provisioning policies in Nigeria: The conditional effect of the IFRS adoption and solvency risk . Trends Economics and Management 15 (38): 35-59. http://dx. doi. org/10 . 13164/trends . 2021 . 38. 35 Salami A. A. , Uthman A. B . , Ajape K . M . 2022. Bank loan loss cyclicality in Nigeria: Global versus local economic crises effect Renaissance University Journal of Management and Social Sciences 8 (1): 1-16 . Salami A A , Uthman A B , Ibrahim R O , Nageri K I 2022 Bank funding strategy and income smoothing practices in Nigeria: IFRS and solvency risk analysis . Global Journal of Accounting 8 (1): 13-30 . http:// gja . unilag . edu . ng/article/view/1548 Salami A A , Uthman A B , Owoade R O 2022 Managerial discretions and loan loss provisions in Nigerian banks: Empirical IFRS and risk evidence . The Central European Review of Economics and Management 6 (2): 71-115 . http://dx. doi . org/10. 29015/ cerem 943

Samadi A. H . , Alipourian M. 2021 . Measuring institutional quality: a review . In N . Faghih, & A. H . Samadi (Eds . ) . Dynamics of Institutional Change in Emerging Market Economies: Theories, Concepts and Mechanisms 143-171. Springer . https://doi. org/10.1007/ 978-3-030-61342-6_6 Sanusi L . S . 2010 . The Nigerian banking industry in the aftermath of the recent global financial crisis. Being a text of the lecture delivered at the Chartered Institute of Bankers of Nigeria (CIBN) 44th Annual Bankers' Dinner . [Electronic resource], www . cibng . org/admin/publica-tions/CBN %20G0VN0R . pdf (accessed: 12. 04. 2022) . Sanusi L. S . 2012 . Banking reform and its impact on the Nigerian economy . CBN Journal of Applied Statistics 2 (2): 115-122 . Saona P . , Azad M . A. K 2020. Examining managerial misbehaviour in Asian banks through loan loss provisions . Asia-Pacific Journal of Financial Studies 49 (4): 581624. https://doi. org/10 . 1111/ajfs . 12307 Seppanen H . J . 2000. Discretionary disclosure and external financing in a relationship financing environment. [Electronic resource]. https://papers . ssrn . com/sol3/Deliv-ery. cfm/000206202. pdf?abstractid= 204668&mirid=1 (accessed: 12. 04. 2022) . Shawtari F . A. , Saiti B . , Abdul Razak S . H . , Ariff M . 2015 . The impact of efficiency on discretionary loans/finance loss provision: A comparative study of Islamic and conventional banks . Borsa Istanbul Review 15 (4): 272-282. https://doi. org/10 ,1016/j. bir. 2015 . 06. 002 Skala D . 2014 . Credit policy in small Polish banks-is there room for income smoothing? Collegium of Economic Analysis Annals 34: 183-196. http://rocznikikae . sgh . waw . pl/p/ roczniki_kae_z34_12 . pdf Skala D . 2018 . Loan growth strategies and income smoothing in Central European Banks . Annales Universitatis Mariae Curie-Sklodowska Lublin — Polonia 51 (1): 161169 . https://doi.org/10,17951/h. 2018. 52.1.161. Soedarmono W . , Pramono S . E . , Tarazi A. 2017 . The procyclicality of loan loss provisions in

Islamic banks Research in International Business and Finance 39: 911-919 . https:// doi. org/10.1016/j. ribaf. 2016. 05. 003 Tran D V , Hassan M K , Houston R 2020 Discretionary loan loss provision behaviour in the US banking industry Review of Quantitative Finance and Accounting 55 (2): 605-645 . https://doi. org/10.1007/ s11156-019-00854-z Tsoulfidis L , Athanasiadis I 2022 A new method of identifying key industries: A principal component analysis Journal of Economic Structures 11 (1): 1-23 . htt-ps://doi . org/10 . 1186/s40008-022-00261-z Van Oosterbosch R . J . J . 2009. Earnings management in the banking industry: The consequences of IFRS implementation on discretionary use of loan loss provisions. Master Thesis Rotterdam, Netherlands: Erasmus School of Economics . https://the-sis. eur . nl/pub/5611/M596-Oosterbosch-van_288138. doc Wasiuzzaman S . , Sahafzadeh I . , Nejad N . R . 2015 Prospect theory, industry characteristics and earnings management: A study of Malaysian industries Review of Accounting and Finance 14 (3): 324-347. https:// doi. org/10 . 1108/RAF-07-2014-0075 Watts R L , Zimmerman J L 1986 Positive Accounting Theory New Jersey: Prentice Hall Englewood Cliffs Yasuda Y . , Okuda S . , Konishi M . 2004. The relationship between bank risk and earnings management: Evidence from Japan Review of Quantitative Finance and Accounting 22: 233-248. https://doi. org/10. 1023/B:REQU . 0000025762 . 89848. 41 Zhang Y. , McIntyre M . L . 2021 . Discretionary loan loss provisioning and stock trading liquidity Journal of Banking Regulation 22 (2): 97-11. https://doi . org/10 . 1057/ s41261-020-00130-4 Zoubi T . A . , Al-Khazali O . 2007. Empirical testing of the loss provisions of banks in the GCC region . Managerial Finance 33 (7): 500-511 . https://doi . org/10.1108/03074350710753771

Initial Submission: October 11, 2023 Final Version Accepted: March 5, 2024

Банковские решения по формированию резервов в Нигерии: внедрение МФСО и банковские риски

А. А. Салами

Университет «Аль-Хикма>>, Нигерия

Цель исследования: проверка наличия взаимосвязи между поведением банков в части формирования резервов на покрытие возможных потерь по ссудам и разнообразными банковскими решениями по учету кредитных потерь на примере нескольких случаев незакрытых кредитов, которые затрагивают нигерийские депозитные банки . Методология исследования: в рамках международных стандартов финансовой отчетности (МСФО) исследуемая взаимосвязь сопряжена с риском неплатежеспособности . Для ее анализа построен индекс банковских решений по кредитным потерям. Проверка гипотез осуществляется на основе оценок параметров регрессионной модели с учетом корректировки стандартной ошибки, применяемой в случае панельных данных . Результаты исследования: показано, что дискреционные корректировки резервов на возможные потери по ссудам характерны для банков Нигерии, однако при этом не учитывается манипулирование прибылью, свойственное дискреционным корректировкам. В случае использования МСФО наблюдается увеличение дискреционных резервов на покрытие потерь по ссудам при принятии решений банками Исключением являются финансово неустойчивые банки. Эти результаты в значительной степени подтверждаются при индивидуальном тестировании связи компонентов индекса банковских решений с поведением по формированию резервов Согласно исследованию, поведение по формированию резервов в Нигерии в период глобального финансового кризиса является проциклическим (нерациональным), а во время местной экономической рецессии 2016—2017 гг . — контрциклическим (рациональным) . Для последнего случая, когда уже был принят набор МСФО, стало характерно сглаживание прибыли на уровне банковских решений . Оригинальность и вклад автора: работа развивает исследования посредством проверки более длинного списка специфичных для банка факторов, вызывающих корректировку решений, касающихся кредитных потерь Она уникальна тем, что опирается на эмпирическую базу, включающую в себя многочисленные банковские решения, которые рассматриваются как единый фактор при учете кредитных потерь . Также констатируется локальный успех институциональных изменений в организациях Нигерии и предлагаются полезные предложения по дальнейшей деятельности

Ключевые слова: банковские решения, резервы на возможные потери по ссудам, институциональные изменения, финансовое благополучие банков, Нигерия

For citation: Salami A. A. 2024. Bank-specific decisions and banks' provisioning behaviour in Nigeria: The role of IFRS adoption and banks' riskiness . Russian Management Journal 22 (1): 51-85 . https://doi .org/10. 21638/spbu18. 2024.103

Для цитирования: Salami A. A . 2024. Bank-specific decisions and banks' provisioning behaviour in Nigeria: The role of IFRS adoption and banks' riskiness . Российский журнал менеджмента 22 (1): 51-85. https://doi. org/10. 21638/spbu18. 2024.103

Статья поступила в редакцию 11 октября 2023 г.

Принята к публикации 5 марта 2024 г.

APPENDIX Appendix 1

Variable Variable name Description

CCR Core capital component Tier 1 capital-risk-weighted assets ratio

TCR Total capital component Addition of Tier 1 and Tier 2 capital scaled by risk-weighted assets

BEBLT Earnings management component Bank earnings before LLPVs and taxes scaled by total assets

SINL Earnings signalling component (BEBLT of year t+1 — BEBTL of year t) scaled by total assets

RDIV Revenue diversification component Bank activity mix within revenue as defined in equation (A1)

NDIV Non-interest income diversification component Bank activity mix within non-interest income as defined in equation (A2)

EDIV Earnings assets diversification component Bank activity mix within earnings assets as defined in equation (A3)

BDIV Statement of financial position diversification component Bank activity mix within bank statement of financial position as defined in equation (A4)

FLD Demand for external financing component Ratio of gross loans to customers' deposits

FDEP Deposit funding component Bank customers' deposits — total assets ratio

FNDP Non-deposit funding component Debts other than customers' deposits scaled by total assets

FIGR Internal funding component Net earnings less extraordinary items and LLPVs divided by gross loans

Appendix 2

Correlation matrix of the components of Bank-Specific Decisions Index

CCR TCR BEBLT SINL RDIV NDIV EDIV BDIV FLD FDEP FNDP FIGR

CCR 1 . 00

TCR 0 . 88* 1 . 00

BEBLT 0 . 27* 0 .15* 1 . 00

SINL 0 . 06 0 . 04 -0 . 52* 1 . 00

RDIV 0 . 36* 0 .40* 0 . 50* -0 . 34* 1 . 00

NDIV 0 . 04 -0 . 08 0 . 00 -0 . 04 0 . 07 1 . 00

EDIV 0 . 36* 0 . 35* -0 .13 0 .19* 0 . 01 -0 . 23* 1 . 00

BDIV 0 .16* 0 . 06 0 . 26* -0 . 03 0 .14 -0 .10 0 . 06 1 . 00

FLD 0 . 03 0 . 07 -0 . 06 0 . 08 -0 .19* -0 . 06 0 .13 -0 .10 1 . 00

FDEP 0 . 50* -0 . 59* -0 .10 -0 . 04 -0 .17* 0 . 04 -0 .20* 0 .15* -0 . 54* 1 . 00

FNDP 0 .63* -0 .48* -0 . 15* -0 . 06 -0 .22* -0 . 02 -0 . 30* -0 . 17* 0 .26* -0 .22* 1 . 00

FIGR 0 .45* -0 . 57* 0 . 55* -0 .44* 0 . 04 0 . 09 -0 .46* 0 .19* -0 . 30* 0 . 45* 0 . 33* 1 . 00

Note: * — indicates significance at 95 % confidence level .

Appendix 3

The model specification for the derivation of the diversification components of BDI

Diversification within revenue activities for each bank is presented as:

RDIV = \NCI2 + ( NIC 2

NOPI

NOPI

(A1)

where NOPI=NIIC + NIC; NIIC — non-interest income (net); while NIC is the net interest income and NOPI represents net operating income

Diversification within non-interest activities for each bank is presented as:

NDIV =

FCR

NIIC

TRDI + \ NIIC ' +

OPI NIIC

(A2)

where NIIC which is non-interest income stands for the aggregate of FCR, TRDI and OPI; FCR represents fee and commission revenue; TRDI captures trading income while OPI is the other operating income

Diversification within earnings assets for each bank is presented as:

(

EDIV =

(

+

NLAD

Y

( NLAD + OEAT ) OEAT

( NLAD + OEAT )

(A3)

where NLAD stands for net loans and advances; OEAT is the other earning assets other than net loans and advances

Diversification index within the bank statement of financial position or balance sheet (BDIV):

BDIV =

(

+

TAS

(TAS + OFBS ) OFBS2

(TAS + OFBS )

(A4)

where TAS represents total bank assets excluding off-balance sheet, OFBS stands for off-balance sheet items

When 1 is deducted from each index, the higher index of each component represents higher diversification

Model specification for BDI components

Model specification for managerial discretionary components of BDI:

LLPV / ADLPVit = »0 +®j CCRit + +»2 TCRit + »3 BEBLTit + »4 SINLit + + »5 IFRSit + »6 (IFRS ■ CCRit) + +»7 (IFRS ■ TCRit) + + »8 (IFRS ■ BEBLTit) + + »9 (IFRS ■ SINLit)+»10 RISit + + »11 (RIS ■ CCRit)+»12 (RIS ■ TCRit) + +»13 (RIS ■ SINLit)+»14 (RIS ■ SINLit) + +»15 (IFRS ■ RIS CCRit) + +»16 (IFRS ■ RIS TCRit) + + »17 (IFRS ■ RIS ■ BEBLTit) + + »18 (IFRS ■ RIS ■ SINLit) + +»19 ANPL / LASit + »20 LVRit +

+ »21 LTAit +»22 LTNit +Vit. (B1)

Model specification for funding and diversification components of BDI:

LLPV / ADLPVit =»0 + »1RDIVit +

+ »2 NDIVu +»3 EDIVit +»4 BDIVt +

+»5 FLDt +»6 FDEPt +»7 FNDPit +

+ »8 FIGRU +»9IFRSU +

,( IFRS RDIV )

+

+»i

+ <»

J10V,---------->it

+ »n (IFRS NDIV)it + + »12 (IFRS EDIV )t + + »13 (IFRS BDIV )t + +»14 (IFRS FLD).t + + »15 (IFRS ■ FDEP )tt + +»16 (IFRS FNDP)tt + + »17 (IFRS ■ FIGR ).t + »18 RISit + + »19 (RIS RDIV).t + +»20 (RIS NDIV).t + 1 (RIS ■ EDIV ).t + »22 (RIS ■ BDIV ).t 23 (RIS FLD). +»24 (RIS FDEP).

+ »

i (RIS ■ FNDP ).t + »26 (RIS ■ FIGR)

25 V. " " ' Ht ' ~26 V " ' it

+ »27 (IFRS ■ RIS ■ RDIV ).t+ + »28 (IFRS ■ RIS ■ NDIV )t + +»29 (IFRS RIS EDIV)tt + +»30 (IFRS RIS ■ BDIV).t + +»31 (IFRS RIS ■ FLD)it + + »32 (IFRS RIS FDEP)it + +»33 (IFRS RIS FNDP)it + + »34 (IFRS ■ RIS ■ FIGR)it + + »35ANPL / LASit +»36LVRit +

+ »37 LTAit +»38 LTNit + Vit. (B2)

Model specification for economic crisis effect of loan loss bank-specific decisions

Model specification for GFC and LFC effect:

LLPV / ADLPVit =»0 +»xBDIit + + ^2GFCit +»3 LFCit + +»4 (GFC ■ BDI )tt + »5 (LFC ■ BDI )it + + »6ANPL / LASit +»7LVRit +

+»^ LTAit +»9 LTNit +^it. (C1)

Model specification for crisis and insolvency risk effect:

LLPV / ADLPVit =»0 +»lBDIit +

+»2GFCit + »3 LFCit+»ARISit +

( RIS GFC BDI )

+ »6 (RIS LFC BDI).t +

+ »7 ANPLit + »8 LVRit + + »9 L TAit +»w LTNÜ + **. (C2)

+

+