## Statistical research falls in the positivistic research paradigm

Statistical
Model

The
following regression model will be applied for estimation

ROE=
?0 + ?1 (FL)+ ?2 (OL)+ ?3 (Age)+?4 (AU)+?5 (NDTS)+ ?6(Pre/Post)+e

Where;

ROE=Return
of Equity

FL=
financial leverage

OL=Operating
leverage

?0
= Constant/Intercept

?1=Coefficient/Slop
of FL

?2=Coefficient/Slop
of OL

?3=Coefficient/Slop
of Age

?4=Coefficient/Slop
of AU/Asset Turnover

?5=Coefficient/Slop
of NDTS/Depreciation.

?6=Coefficient/Slop
of Pre/Post

3.3 Type of Research

The
explanatory/ co-relational research is carried out on the groundwork prepared
by the descriptiveresearch. During the descriptiveresearch, the researcher tries to
enrich the field prepared by exploratory research by finding some additional features
and facets. While in explanatory research,
the researcher attempts to connect the
ideas to understand the cause and effect. Here the researcher wants to know
“what is going on.”

Our research objectives are also
similar nature through which we will find the impact of leverages on
profitability in the presence of both adverse and favorable economic
conditions. Hence, we can conclude that our proposed research will be
explanatory.

3.4

In
the positivistic research paradigm, the research
must be value-free, i.e. the research is assumed to be free of subjective
bias. Since our proposed research is also based
on the objective reality and going to be free from subjective bias. Therefore,
our research paradigm will be positivistic. Other reasons due to which our
research falls in the positivistic research paradigm are; firstly, the study will generate the hypothesis/questions
and test them and, secondly, the research will use the quantitative methods to
analyze and evaluate the data.

3.5
Nature of Research Study

Quantitative
Research is used to quantify the problem by way of a collection of numerical data type or data that can be transformed into usable statistics. It is
used to quantify attitudes, opinions, behaviors, and other defined variables
and generalize results from a larger sample population. Quantitative Research
uses measurable/quantifiable data to formulate facts and uncover patterns in
research. Quantitative data collection methods are much more structured as
compared to the qualitative data collection methods, i.e. surveys focused interviews, questionnaires or observations, etc. Since our research will be making use of
data available in the numeric
form, therefore, the proposed study
will be quantitative.

3.6
Population and Sample Frame

The population of 29 firms listed under
the head of “Chemical Sector” on Pakistan Stock Exchange (PSX) will be included in the analysis. In addition to the
29 chemical sector firms, 14 more firms
having chemical related products will also be
included in the analysis. Hence
our population and sample will comprise 43 firms.

3.7
Sample Size and Sampling technique

The all twenty-nine (29) listed chemical
sector firms on PSX will be included in
the sample In addition to these 29 listed companies, 14 more firms having similar business nature like
pharmaceuticals & plastic, etc will
also be included by taking unbalanced
panel data to improve the study results. Hence the total firms under analysis will become 43. The study
will comprise the twelve years period data from 2004 to 2015

3.8
Data Collection Sources

All
the necessary data for this study will be collected/ gathered from the secondary
sources available on the internet. For
this study, the data will be used from
State bank of Pakistan (SBP)’s publication under the head “Financial Statement
Analysis of Joint Stock Companies listed on KSE/PSX.”

3.9
Data analysis techniques

Appropriate Panel Data Regression
model will be used to estimate the coefficients and relationships between the
dependent and independent variables. The statistical analysis software “Stata” will
be used for the regression and estimation.

3.10
Variables Description

3.10.1 Dependent Variable

Profitability =ROE= (EBT/Equity)

The most relevant independent variable that may be considered as the measure of the financial performance and
efficiency is Profitability.  Profitability
can be regarded as main independent variable that determines
capital structure because of the well-known
postulate that is represented by POT. According to the POT, the mostly the
firms try to fulfill its capital needs from the internal sources, and if the firms
need additional capital beyond the internal
arrangements, the firms go towards the
outside arrangements, i.e. debt and
equity issues. Therefore, according to POT, the financial leverage and
profitability are related negatively. Similarly,
according to the TOT, the firm identifies the target debt ratio by comparing
benefit from and cost of financial leverage.
Hence, according to the TOT, the profitability and financial leverage are
related positively.

Hence
the profitability is taken to be the Return on Equity (ROE). The ROE is measured as Earnings before Taxes (EBT) divided by Total Shareholders’ Equity of
the firm

3.10.2
Independent Variables

a).
FL = (Debt/Equity)

We
will use the ratio of debt to equity as financial leverage

b).
OL =C M/EBIT=Gross Profit/EBIT

OL
is the degree to which a company’s operating costs are fixed. The OL is measured as the ratio of contribution margin
(CM) to earnings before interest & Taxes (EBIT). Since the CM calculation requires the data from cost accounts and those
are not readily available on the secondary sources. Therefore, we will
use the “Gross Profit” Figures from financial data as the proxy of CM.

3.10.3
Control Variables

We
will introduce three control variables in our model to investigate some
overlapping relationship effects (if any). These three control variables are;
assets utilization (assets turnover), age, and non-debt tax shields
(depreciation & investment tax credits)

3.10.4
Dummy Variable (Pre and Post Crisis Measurement)

We
will introduce one dummy variable named “Pre/Post” in our model corresponding
to Pre-Crisis and Post Crisis Period. The variables “Pre /Post” will take the
value “0” for the years falling in the Pre-Crisis Period. Similarly, the value taken by this
variable will be “1” for the years falling in the Post-Crisis Period.

RESULTS

CHAPTER No. 04

4.1 Background

The
study has utilized the data collected from the Pakistan Stock Exchange (PSX)
formerly known as Karachi Stock Exchange (KSE). The total number of companies
included in the analysis was 43. The 29 companies
relate directly to the chemical sector while other 14 companies whose products
are either of chemical oriented or utilized
by the chemical companies as input. The study has included these additional
firms in the analysis to improve the
analysis results. A comprehensive list of these companies/firms appends in the
annexure at the end.

The
data used for this analysis is taken from
the SBP’ s publication titled “Financial Statements Analysis of Joint Stock
Companies list on PSX for the period from 2004 to 2015. Therefore our analysis
confined the 12 years period.

Our
data consists of both types, i.e. Cross Section Data (CSD) and Time-Series
Data (TSD). This combination of both types is
referred as Panel Data (PD). The CSD consists of the data gathered from
multiple individuals at the same time. Whereas,
the TSD is a data type collected from the same individual at different times.
The PD data type exhibits the qualities of both
the types, i.e. CSD & TSD.  Therefore, in the PD data type, the data is
gathered from multiple individuals at different times.

The
OLS (Ordinary Least Square) Model could
not be applied to
Panel Data due to correlated errors occurring in the presences of time-series
and cross-sectional
components.

As
we know that the model form for OLS is “Yi= a+bXi+e” whereas for Panel Data is “Yit=a+bXit+et” where “i” stands for
cross section and “t” for time series

Therefore,
we have used the Panel Linear Model (PLM) Instead of OLS model.

1).
Our data consists of an unbalanced panel. It is due to the facts that
SBP does not include substantially all
the firms throughout the analysis period, i.e.
from 2004 to 2015. Since this is a prolonged
period and companies are added/ frequently removed from to/from the listing of
stock exchange during a short period. Similarly, some firms are listed
again after some period either under the same name or changed one. Therefore,
the number of firms mentioned in the
first few years’ analysis is not the same as having
been included in the last few years’ of analysis. While in the balance panel
data the number of the firms and data variables necessarily
have to remain same throughout the analysis period.

2).
Similarly our analysis comprises random
effect model because, in the fixed effect
model, it is assumed that every firm included in the analysis is substantially
same. While in the real since the every firm has
different set of characteristics and merits/demerits. Therefore, the firms included in the analysis can’t be 100% of
the same nature. This required applying a
random effect model rather than fixed effect model.

4.2
Regression Method

To
analyze the data, the “Stata” Statistical Software has been used. The “Enter” method has been applied rather than the
“Step” method. The dependent variable is
ROE%  and other six variables, i.e., FL, OL, AU, Age, NDTS and Pre/Post are
taken as independent variables. Regression has produced the following results.

4.3
Results

4.3.1
Tabular Form

The
actual results produced by running Stata
Multiple Regression are given in the seven tables given below. The
detailed discussion about these results is
provided in the next chapter