CHAPTER THREE
3.1 Research Design
The study is an ex-post facto research design in a linear specification via the
partial adjustment approach making use of generalized theories and
empirical studies.
3.2 Population and Sample
This is purely time series study across banking sector in this regard, the
population of the study covers the entire banking sector- their assets,
profitability, credit default and other related variables. From this population
of commercial banks in Nigeria, fourteen (14) banks were selected for the
analyses, these sample were drawn randomly making use of bank variables
of loan to deposit ratio, non-performing loan and interest margin span from
2002 to 2017. The Deposit Money banks selected are; First bank of Nigeria
Plc, Sterling Bank Plc, Diamond Bank Plc, Access Bank Plc, Zenith Bank Plc,
Unity Bank Plc, United Bank for Africa Plc, Union Bank of Nigeria, Guaranty
Trust Bank Plc, Skye Bank Plc, WEMA Bank Plc, First City Monument Bank Plc,
Fidelity Bank Plc and Ecobank Plc.
3.3 Data Collection Method
The data employed in this study are typically secondary or already published
data. These data are collected in times series and across banking sector
which then turn it to panel data. Therefore data on loan to deposit ratio,
nonperforming loan, return on equity was collected from annual publication
of Nigeria Deposit Insurance Corporation (NDIC) and statistical bulletin of the
Central Bank of Nigeria (CBN) all via their websites.
3.4 Model Specification
The model relevant to the study is adopted from the studies of Donia (2011);
Yu & Gan (2010); Benyah (2010); Seetanah et al., (2009); Ali and Iva (2013).
This study introduced static OLS equation follows a standard ARDL model is
quoted in four models:
Following the research objectives, the functional form of the variable are
hereby developed in form of
Y=β0 +β1X (3.1)
roe = f(npc, inmg, dtl) (3.2)
roe =α0 + α1npc + α2inmg + α3dtl+μ (3.3)
Then put in panel format;
roeit =α0 + α1npcit + α2inmgit + α3dtlit+μit (3.4)
Equation (3.4) is the general model specification for objective. This static
model assumes that all the variables are well
behaved. That is, each of the variables is stationary at order zero.
3.4.1 Data Analysis Techniques
Dependent Variables
Return on Equity (roe):
This represents the rate of return received from equity invested in banks. It is
the amount of net income returned as a percentage of shareholders equity.
Return on equity measures profitability by revealing how much profit a bank
can generate with the money shareholders have invested.
Independent Variables
Non performing loan (npcs):
These are loans that are outstanding both in its principal and interest for a
long period of time contrary to the terms and conditions under the loan
contract. Thus, the amount of non-performing loan represents the quality of
bank assets (Tseganesh, 2012).
Deposit to Loan Ratio (dtl):
This ratio examines bank liquidity by measuring the funds that a bank has
utilized into loans from the collected deposits. It demonstrates the
association between loans and deposits. Besides, it provides a measure of
income source and also measures the liquidity of bank asset tied to loan
(Makri et al.2014)).
Net Interest Margin (int):
Lending rates are one of the primary economic determinants of NPCs and
PCs. Interest rate spread is a measure of profitability between the cost of
short term borrowing and the return on long term lending. Interest rate
spread affect performing assets in banks as it increases the cost of loans
charged on the borrowers (Joseph, 2011). While Net interest margin is a
measure of the difference between the interest income generated by
commercial banks and the amount of interest paid out to depositors relative
to the amount of their (interest-earning) assets.
Table 3.1 Priori Expectation
3.5 Limitations of the Methodology
The model is a single equation approach to cointegration and it is only
applicable to variables with mixed integrations but none must be I (2)
variable. Also, the model must be weakly stationary or ergodic and the
residuals must follow a Gaussian process. Thus, the model is developed to
take care of the weakness of the multivariate equation approach to co-
integration which stipulates that all the variable series must be I (1).