This

chapter of the study aims to explain the results of the time series data. In

this study E-views 9 software is used for the analysis of data to obtain the

objectives of the study. The chapter begins with explaining the results of

stationarity tests on data which is necessary to check the long run and short

run co-integration of selected variables with demand for life insurance. 5.2. Stationarity TestBefore

applying the ARDL approach to cointegration, data of all variables are required

to test for stationarity. To know whether the data is stationary or

nonstationary, the Augmented Dickey Fuller test or unit root test is used. A

series is said to be stationary, when its mean and variance are constant over

time. 5.2.1. Results of ADF Test

Thus,

according to the results of ADF test which is applied on selected variables: price

of life insurance, crude death rate and education are stationary at level at 5%

significance level. The variables: gross saving, inflation and sum assured are

non-stationary at level. After checking the stationarity at level, data are

checked at first difference. The results of the first difference showed that

these variables became stationary after taking the log of education and sum

assured, therefore we could apply ARDL approach to cointegration. The spurious

regression problem occurs when the error terms (the residuals) of the

regression model have a unit root. Therefore, results show that error term have

no unit root which means that there is no spurious regression problem. The

results are mentioned in table 5.1 below.5.3. ARDL Bounds tests for

cointegration:for

the empirical analysis of the long-run relationships and short run dynamic relations

among the selected variables, the autoregressive distributed lag (ARDL)

cointegration technique is used. The ARDL cointegration approach was first time

used by Pesaran and Shin (1999) and Pesaran et al. (2001). It is applied on

data as all the variables are not integrated at the same order, some variables

are integrated of order one, some of order zero and as the sample size of data

is less than 30 years therefore, the ARDL test is relatively more efficient. The estimates of the long-run model are

unbiased which are obtained with ARDL technique (Harris and Sollis, 2003). 5.3.1. ARDL Bound Test for

Co-Integration

Lag-length

selection is very important for correct results of long-run relationship in the

model (Bahmani-Oskooee and Bohal, 2000). Table 5.2 presents the computed

F-statistic to select optimal lag-length in the model. with lag of order 4 the

lower and upper bound values at 90 percent significance level are 2.75 and 3.79

respectively. Table 5.2 shows that the computed value of F-statistic (3.09) is

in between lower and upper bound value of F-statistic at 10 % which indicates

inconclusive. Therefore, we conclude that there is may or not long-run

relationship among the variables.The

joint F-statistic is mostly used in bound test. first, six equations (1, 0, 2,

0, 1, 1) are estimated with the help of ordinary least squares (OLS) by

conducting an F-test for the joint significance of the coefficients of the

lagged levels of the variables.

AIC

is used to select the orders of the ARDL (1, 0, 2, 0, 1, 1) model in the six

variables. Thus, equations are found with orders of ARDL (1, 0, 2, 0, 1, 1) model.

The results obtained by normalizing on lnlidd in the long run are given in

Table 5.2.Table

5.3 indicates the long run results of ARDL model. The coefficient shows that

the relationship is either positive or negative between dependent and

independent, whereas the probability value shows that whether the relationship

between independent and dependent variables is significance or insignificance.

According to rule of thumb, if p < 5% it indicates the significance there is
positive relationship between dependent and independent variables.
Findings
confirm that there is statistically positive and significance relationship
between gross saving (gst) and demand for life insurance (lnliddt) at 5% level
of significance. It indicates that if we increase 1unit in gst variables in
this response there will increase of 0.031484 in lnliddt.This results supports
the findings of Steven Haberman
and Chee Chee Lim's study
conducted in 2011
The
result shows that the price of life insurance is significantly negative
relationship with the demand for life insurance as the p values is less than 5%
at 1. it indicates that if we increase 1unit in plit variables in this response
there will decrease of 0.00227 in lnliddt. This result supports the literature
present on the impact of price of life on the demand of life insurance and is
matched with finding of Steven
Haberman and Chee
Chee Lim conducted in 2011.
According to results there is positive and
significance relationship education (lnedt) and demand for life insurance because
of p values is less than 5%.it indicates that if we increase 1unit in education
(edt) variable in this response there will increase of 0.612579 in life
insurance demand lnliddt. This
finding is lined with Kjosevski study conducted in 2010, Amrot Yilma's
study conducted in 2014, study
of Steven Haberman and Chee Chee Lim conducted in 2011 and is opposite to the study of Celik and Kayali, higher education
influences positively life insurance demand.
According
to the results the inflation influences the demand for life insurance significantly
negative. The result on inflation rate was consistent with the theoretical
propositions and is not lined with Neumann analysis
conducted in 1946- 1964 and is lined with Kjosevski study conducted in 2010 and
Amrot Yilma (2014), Nesterova (2008), study of Steven Haberman and Chee Chee Lim's study conducted in 2011 which showed that
inflation had significant negative influence and a damping impact on the
purchase of life insurance.
The
result also shows that there is positive and significance relationship between
crude death rate life insurance demand at 5 % level of significance as the
value of p is less than 5% at lag 2 which indicates that when there is 1 unit increase
in crude death rate (cdrt), the demand for life insurance will increase by 4.899575
which is lined with study of Steven Haberman and Chee Chee Lim conducted in 2011 and all literature of determinants
of life insurance demand in different countries.
The
R2 is 0.99; implying that approximately 99% of variations in life
insurance demand are explained by all the independent variables while the
remaining 1% is captured by the error term. There is a significant linear
association between the life insurance market demand and the economic
variables.
The overall significance test of
model:
The P value for the F-test of overall significance test is
0.000 which is less than significance level 0.05, therefore, reject the
null-hypothesis and concluded that the model of study gives a good fit than the
intercept-only model.