The Impact of Financial Metrics on Earnings Quality in Nigerian Banks

This report presents an analysis of earnings quality in Nigerian banks from 2009 to 2018. It begins with descriptive statistics of key variables like earnings quality (EQ), board size (BDS), firm leverage (FML), and ownership structure (OWS). The study employs Augmented Dickey-Fuller (ADF) tests to assess stationarity, revealing that EQ and FML are stationary at the zero difference. The study then proceeds to utilize ARDL models to test the hypotheses. The findings indicate a significant relationship between board size, ownership structure, firms leverage, and earnings quality. The results indicate that board size, ownership structure, and firm leverage significantly influence earnings quality. The report concludes by discussing these findings, attributing the negative relationship between the variables to a lack of regulatory controls, and suggesting further analysis. The analysis suggests that Nigerian banks use board structure, firm leverage, and ownership structure to manipulate earnings quality.

4.1: Data Presentation
The employed predictor and criterion variables of banks in Nigeria (2009 to 2018) are
presented as follows
Descriptive
This section further proceeds to evaluate the underlying trend of employed data in this subsection
based on various attributes of the employed data.
EQ BDS FML OWS
Mean 0.383863 4.292308 5.696532 6.000000
Median 0.259326 4.000000 5.437776 6.000000
Maximum 8.735455 7.000000 38.99508 8.000000
Minimum -21.18190 1.000000 -7.219780 4.000000
Std. Dev. 2.338341 1.123713 6.237580 0.373544
Skewness -5.735753 0.131125 2.901713 -0.895806
Kurtosis 58.71467 3.476544 15.27821 21.66667
Jarque-Bera 17526.81 1.602624 999.0180 1904.794
Probability 0.000000 0.448740 0.000000 0.000000
Sum 49.90219 558.0000 740.5492 780.0000
Sum Sq. Dev. 705.3513 162.8923 5019.056 18.00000
Observations 130 130 130 130

Stationarity Unit Root Test
Table 4.2 below shows the ADF stationarity unit root tests output of variables in this study via
E-View version 9.
Table 4.2: ADF Stationarity Unit Root Test Extract
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(EQ)
Method: Least Squares
Date: 09/16/21 Time: 15:21
Sample (adjusted): 4 130
Included observations: 127 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
EQ(-1) -0.642151 0.169940 -3.778693 0.0002
D(EQ(-1)) -0.558856 0.146255 -3.821102 0.0002
D(EQ(-2)) -0.223216 0.093420 -2.389374 0.0184
R-squared 0.648082 Mean dependent var 0.041496
Adjusted R-squared 0.642406 S.D. dependent var 3.712740
S.E. of regression 2.220187 Akaike info criterion 4.456398
Sum squared resid 611.2244 Schwarz criterion 4.523584
Log likelihood -279.9813 Hannan-Quinn criter. 4.483695
Durbin-Watson stat 1.996212

The table above indicates the result of Stationarity using Augmented Dickey Fuller (ADF) unit
root test. The results revealed that earnings quality became stationary at the zero difference (0)
with (ADF t-statistic value of --3.778693 with the critical value of 0.0002 at 5% level),
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(BDS)
Method: Least Squares
Date: 09/16/21 Time: 15:49
Sample (adjusted): 5 130
Included observations: 126 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
BDS(-1) -0.021778 0.022735 -0.957896 0.3400
D(BDS(-1)) -0.593758 0.088848 -6.682815 0.0000
D(BDS(-2)) -0.327113 0.098565 -3.318767 0.0012
D(BDS(-3)) -0.220759 0.087095 -2.534679 0.0125
R-squared 0.293893 Mean dependent var -0.015873
Adjusted R-squared 0.276530 S.D. dependent var 1.308337
S.E. of regression 1.112833 Akaike info criterion 3.082926
Sum squared resid 151.0844 Schwarz criterion 3.172967
Log likelihood -190.2243 Hannan-Quinn criter. 3.119507
Durbin-Watson stat 2.013564

The table above indicates the result of Stationarity using Augmented Dickey Fuller (ADF) unit
root test. The results revealed that board size became non-stationary at the zero difference (0)
with (ADF t-statistic value of -0.957896 with the critical value of 0.3400 at 5% level),
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(OWS)
Method: Least Squares
Date: 09/16/21 Time: 15:51
Sample (adjusted): 5 130
Included observations: 126 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
OWS(-1) 0.001775 0.005452 0.325488 0.7454
D(OWS(-1)) -0.841871 0.079249 -10.62315 0.0000
D(OWS(-2)) -0.536099 0.095071 -5.638953 0.0000
D(OWS(-3)) -0.223284 0.079149 -2.821048 0.0056
R-squared 0.484842 Mean dependent var 0.015873
Adjusted R-squared 0.472174 S.D. dependent var 0.505713
S.E. of regression 0.367409 Akaike info criterion 0.866550
Sum squared resid 16.46871 Schwarz criterion 0.956590
Log likelihood -50.59263 Hannan-Quinn criter. 0.903130
Durbin-Watson stat 1.987937

The table above indicates the result of Stationarity using Augmented Dickey Fuller (ADF) unit
root test. The results revealed that ownership structure became non-stationary at the zero
difference (0) with (ADF t-statistic value of 0.325488 with the critical value of 0.7454 at 5% level),
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(FML)
Method: Least Squares
Date: 09/16/21 Time: 15:53
Sample (adjusted): 2 130
Included observations: 129 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
FML(-1) -0.311938 0.064358 -4.846878 0.0000
R-squared 0.155057 Mean dependent var 0.028603
Adjusted R-squared 0.155057 S.D. dependent var 6.714671
S.E. of regression 6.172179 Akaike info criterion 6.485703
Sum squared resid 4876.262 Schwarz criterion 6.507872
Log likelihood -417.3278 Hannan-Quinn criter. 6.494711
Durbin-Watson stat 2.135205

The table above indicates the result of Stationarity using Augmented Dickey Fuller (ADF) unit
root test. The results revealed that firms leverage became stationary at the zero difference (0)
with (ADF t-statistic value of -4.846878 with the critical value of 0.0000 at 5% level),
Going by the ADF test statistics and the p-value, it can be seen that two employed variables are
stationary while the other two are not stationary at the first and second difference. The study will
proceed to ARDL models. Since the prerequisite of ARDL models is applicable for both non-
stationary time series as well as for times series with mixed order of integration.
Hypotheses Testing
The formulated hypotheses were statistically tested as shown in their null form.
Hypothesis one
H01: Ownership structure does not have significant effect on earnings quality of firms in Nigeria.
Dependent Variable: EQ
Method: ARDL
Date: 09/16/21 Time: 16:09
Sample (adjusted): 5 130
Included observations: 126 after adjustments
Dependent lags: 4 (Fixed)
Dynamic regressors (4 lags, fixed): BDS
Fixed regressors: C
Variable Coefficient Std. Error t-Statistic Prob.*
EQ(-1) -0.244589 0.094336 -2.592742 0.0107
EQ(-2) 0.297569 0.097854 3.040963 0.0029
EQ(-3) 0.234007 0.101446 2.306709 0.0228
EQ(-4) 0.067152 0.097940 0.685641 0.4943
BDS 0.093112 0.193505 0.481190 0.6313
BDS(-1) 0.251306 0.197676 1.271300 0.2062
BDS(-2) -0.132080 0.200164 -0.659860 0.5107
BDS(-3) -0.260617 0.197724 -1.318087 0.1901
BDS(-4) -0.026856 0.193116 -0.139064 0.8896
C 0.636942 1.205140 0.528521 0.5981
R-squared 0.180352 Mean dependent var 0.396014
Adjusted R-squared 0.116759 S.D. dependent var 2.374442
S.E. of regression 2.231522 Akaike info criterion 4.519283
Sum squared resid 577.6440 Schwarz criterion 4.744385

Log likelihood -274.7148 Hannan-Quinn criter. 4.610735
F-statistic 2.836026 Durbin-Watson stat 1.969434
Prob(F-statistic) 0.004716
*Note: p-values and any subsequent tests do not account for model
selection.
The result indicates that board size significantly influence earnings quality with β = 0.0107. From
the analysis above and in appendix iv-vi, board size and earnings quality produced (t = --2.592742)
relationship. Board size significantly relates to earnings quality. The critical value level was less
than 0.05 significant level. Therefore, the null hypothesis was therefore rejected and the study
concluded that board size significantly influence earnings quality in the selected banks in the
period of this study.
Hypotheses two
H02: Ownership structure does not have significant effect on earnings quality of firms in Nigeria.

Dependent Variable: EQ
Method: ARDL
Date: 09/16/21 Time: 16:19
Sample (adjusted): 5 130
Included observations: 126 after adjustments
Dependent lags: 4 (Fixed)
Dynamic regressors (4 lags, fixed): OWS
Fixed regressors: C
Variable Coefficient Std. Error t-Statistic Prob.*
EQ(-1) -0.233436 0.094345 -2.474263 0.0148
EQ(-2) 0.294858 0.098182 3.003179 0.0033
EQ(-3) 0.222033 0.101734 2.182495 0.0311
EQ(-4) 0.050975 0.099049 0.514643 0.6078
OWS 0.229339 0.620052 0.369871 0.7122
OWS(-1) 0.138807 0.542173 0.256020 0.7984
OWS(-2) -0.178534 0.542242 -0.329252 0.7426
OWS(-3) -0.118405 0.543021 -0.218048 0.8278
OWS(-4) 0.001616 0.542561 0.002978 0.9976
C -0.130963 6.582716 -0.019895 0.9842
R-squared 0.154132 Mean dependent var 0.396014
Adjusted R-squared 0.088504 S.D. dependent var 2.374442
S.E. of regression 2.266934 Akaike info criterion 4.550772
Sum squared resid 596.1226 Schwarz criterion 4.775873
Log likelihood -276.6986 Hannan-Quinn criter. 4.642224
F-statistic 2.348584 Durbin-Watson stat 1.967878
Prob(F-statistic) 0.017970
*Note: p-values and any subsequent tests do not account for model
selection.

The result indicates that ownership structure significantly influence earnings quality with β =
0.0148. From the analysis above and in appendix iv-vi, ownership structure and earnings quality
produced (t = -2.474263) relationship. Ownership structure significantly relates to earnings
quality. The critical value level was less than 0.05 significant level. Therefore, the null
hypothesis was therefore rejected and the study concluded that ownership structure significantly
influence earnings quality in the selected banks in the period of this study.
Hypotheses Three
H03: Firms leverage has no significant effect on earnings quality of firms in Nigeria.
Dependent Variable: EQ
Method: ARDL
Date: 09/16/21 Time: 16:23
Sample (adjusted): 5 130
Included observations: 126 after adjustments
Dependent lags: 4 (Fixed)
Dynamic regressors (4 lags, fixed): FML
Fixed regressors: C

Variable Coefficient Std. Error t-Statistic Prob.*
EQ(-1) -0.238515 0.095215 -2.505000 0.0136
EQ(-2) 0.322980 0.099038 3.261183 0.0015
EQ(-3) 0.224724 0.102731 2.187487 0.0307
EQ(-4) 0.039115 0.099451 0.393305 0.6948
FML 0.045837 0.035505 1.291007 0.1993
FML(-1) -0.066074 0.039014 -1.693583 0.0930
FML(-2) 0.008867 0.039290 0.225691 0.8218
FML(-3) 0.014618 0.039369 0.371298 0.7111
FML(-4) -0.011208 0.035378 -0.316800 0.7520
C 0.350234 0.366105 0.956649 0.3407
R-squared 0.176545 Mean dependent var 0.396014
Adjusted R-squared 0.112656 S.D. dependent var 2.374442
S.E. of regression 2.236699 Akaike info criterion 4.523918
Sum squared resid 580.3275 Schwarz criterion 4.749020
Log likelihood -275.0068 Hannan-Quinn criter. 4.615370
F-statistic 2.763311 Durbin-Watson stat 1.969318
Prob(F-statistic) 0.005770
*Note: p-values and any subsequent tests do not account for model
selection.
The result indicates that firms leverage significantly influence earnings quality with β = 0.0148.
From the analysis above and in appendix iv-vi, firms leverage and earnings quality produced (t =

-2.474263) relationship. Firms leverage significantly relates to earnings quality. The critical value
level was less than 0.05 significant level. Therefore, the null hypothesis was therefore rejected
and the study concluded that firms leverage significantly influence earnings quality in the
selected banks in the period of this study.
Discussion of findings
The following observations were made:
1) There is a significant relationship between board size, ownership structure, firms leverage and
earnings quality.
2) This Study establishes direction of relationship between BDS, OWS, FML and EQ. A negative co-
efficient exist between BDS, OWS, FML and EQ. This means that there is a strong and negative
significant relationship between BDS, OWS, FML and EQ; rise in BDS, OWS, FML decreases EQ.
Similarly, increase in EQ decreases BDS, OWS, FML. The negative significant relationship between
BDS, OWS, FML and EQ could be attributed to lack of regulatory and controls measures put in place
by banks.
Conclusion
The study examined BDS, OWS, FML and tried to establish if it exerts significant effect on earnings
quality. This was to enable us ascertain if banks used BDS, OWS, FML to manipulate earnings quality.
The research showed strong and significant relationship between BDS, OWS, FML and EQ and showed.
We attribute this finding on EQ to the ownership size, board structure, FML and suggest further analysis
which is not considered in this study. The implication is that bank in Nigeria tends to use BDS, OWS, FML
to bloat EQ, which will have an adverse effect on the long run. However, test result established a
direction of influence between the variables of study. A negative co-efficient exist between BDS, OWS,

FML and EQ and EPS. This Imply that increase in any of the variables has potential of decreasing the
banks earning quality although banks in Nigeria are presently not using the variable to manipulate
earnings.