CHAPTER THREE

RESEARCH METHODOLOGY

3.1 Introduction

This chapter describes the methodology will use to address the research problem. The chapter
describes the research design and approach, population and sample, Data source and collection,
Method of data analysis, Model variables (input and output) and Model specification.

3.2 research design and approach

This study will conduct using descriptive research method. appproch A descriptive study
defines a subject by constructing a profile of people, groups or events through tabulation and the
collection of data on the frequencies on study variables (Cooper & Schindler, 2007). And
quantitative approach is preferred since “is associated with exploring connections between
variables” (Bryman & Bell, 2007)

3.3 Population and sample

According to national bank of Ethiopia at the end of 2017 there are 16 private commercial banks
and the study will considers all the private commercial banks which are operating in the country
as decision making units. There are two important aspects of DEA, following which it is
preferable to study the population of banks. Firstly, it is sample specific, thus implying that
results obtained for the sample cannot be generalized for the entire population. Secondly, it gives
the relative efficiency scores and not the absolute efficiency scores." This means that the best
performing DMU out of the group will be shown as 100 per cent efficient. The rest of the DMUs
will be benchmarked against this one. According to Dyson et al., 1998 cited in Gunjan M.,
(2006) In case of DEA, the sample size should be generally larger than the product of the number
of inputs and outputs.
3.4 Data source and collection

Due to the nature of the study only secondary data will be used. The data will be found from the
banks financial statements for the period 2005-2009. The data will be collect from the published
and audited annual report of the all private commercial banks and from the records held by
National bank of Ethiopia.

The study used panel data to measure the TE of commercial banks in Ethiopia for the operating
years from 2011 to 2014. For this purpose secondary data was used. Secondary data was
obtained from the balance sheet and the income statement of commercial banks covered in the
study.
3.5 Method of data analysis

According to Raphael, (2013) as cited in Tadesse, (2016) There are two common approaches to
measure the efficiency of DMUs, which are the parametric (econometric) and non-parametric
(mathematical) techniques. The popular parametric technique is the Stochastic Frontier
Approach (SFA), while that of non-parametric is the Data Envelopment Analysis (DEA)
According to Rajput and Handa (2011), the wide acceptance of DEA as a measurement tool for
measuring efficiency of the financial institution can be attributed to certain strengths of this
approach. According to Debasish (2006) DEA has been widely used to measure efficiency
performance of different financial institutions like banks, insurance and mutual funds.
Particularly in the banking sector, it has been applied to benchmark the efficiency performance
of different banks or to study the efficiency estimates of different branches of a particular bank.
Data Envelopment Analysis (DEA) will use in carrying out this study,
Charnes, Cooper and Rhodes (1978) generalized Farrell’ singe input single output efficiency
measure to multiple - input multiple - output situations and operationalized it using mathematical
programming (Emrouznejad & Emilyn, 2015). Data Envelopment Analysis (DEA) developed by
Charnes et al. (1978) is a linear programming based technique. DEA occasionally called frontier
analysis is a performance measurement technique which can be used for analyzing the relative
efficiency of productive units, having the same multiple inputs and multiple outputs (Majid,
2012)
There are two assumptions under DEA model, the constant return to scale (CRS) and the variable
return to scale (VRS).
Constant return to scale (CRS) model: The original DEA approach by Charnes et al. (1978)
assumed constant returns to scale of activities by DMUs. The CCR model is the most widely
used DEA model. It is used in frontier analysis when a constant return to scale relationship is
assumed between inputs and outputs. Being the first DEA model to be developed, this model
calculates the overall efficiency for each unit, where both pure technical efficiency and scale
efficiency are aggregated into one value (Takeda, 2000). This assumption is only appropriate
when all DMUs operate at an optimal scale. Factors such as imperfect competition, or limited
financial resources, may prevent decision-making units from operating at an optimal scale
(Coelli et al., 1998). Consequently, the use of the constant returns to scale specification might
result in measures of technical efficiency, which are confounded by scale efficiencies.

Variable return to scale (VRS) Model: The BCC model is the DEA model used in frontier
analysis when a variable returns to scale relationship is assumed between inputs and outputs.
This model focuses primarily on the technological aspects of production correspondences, and
can be used to estimate technical and scale efficiency without requiring estimates of input and
output prices. Thus, this approach has been used extensively in the regulated sector

Since the study assessing how efficient DMUs use inputs to produce outputs both CRS and VRS
assumptions are necessary. Why using both ccr and bcc mobel together

3.6 Model variables (input and output)

The DEA does not require in advance assumptions about the production function’s analytical
form and is not affected by the unit of measure. However, it is sensitive to extreme observations
and choice of variables as inputs and outputs (Tesfaye, 2014).Regarding the appropriate inputs
and outputs variables to be employed by DEA model for banks, as mentioned in several studies,
there are two main approaches that can be used to determine the bank inputs and outputs. They
are production or intermediation approaches (G. A. Mousa1, 2015).

Intermediate approach: by this approach the selection is made based on the bank's assets and
liabilities, bank assets including labor represent the inputs and liabilities represent the outputs.
In this approach, banks perform an intermediary role between borrowers and depositors and
hence accept deposits and other funds in order to provide loans and alternative investments.
Output is measured by interest income, total loans, total deposits and non-interest income, while
inputs are usually represented by operating and interest costs. (G. A. Mousa1, 2015)

Production approach: which considers the bank as normal company or producer, and hence the
inputs are the physical elements such as labor and capital and all other assets and liabilities are
outputs, this approach argued that all deposits (which are assets) should be treated as output since
they are produced by capital and labor. According to Johnes et al. (2009, p.14) as cited in (G. A.
Mousa1, 2015) in production approach the bank is treated as a firm that provides services, such
as loans, through the use of capital and labor inputs. Output is generally represented by the
number of deposit accounts or transactions and inputs are defined as number of employees
(labor) and capital expenditures on fixed assets (capital).
This study will follow the intermediation approach. According to Berger and Humphrey (1997)
as cited by Raj Yadav,(2015) intermediation approach is well suited to analyzing firm level
efficiency, whereas the production approach is suited to measuring branch level efficiency. The
study use two inputs and two outputs listed below

Input

I. Non-interest expense
II. Non-interest income

Output

I. Net-interest income
II. Non-interest income

3.7 Model specification