Life Insurance Demand

Gross Saving Rate

Saving is that part of income which is not spent, which means that when income rises, the saving will increase. The gross saving (as % of GDP) was 23.29% during December 2017 in Pakistan. In this study, gross saving is taken as an independent variable to find the effect of saving and income on the demand for life insurance. Saving is closely related to investment; therefore, the income left after the consumption of goods and services is invested in life insurance. Savings, therefore, have a positive effect on the demand for life insurance and contribute to economic growth. Gross saving in this study is represented by “GS”.

Life Insurance Demand

Level of Education

According to previous studies, the level of education has a significant and positive effect on the demand for life insurance (Truett & Truett, 1990; Browne & Kim, 1993; Li et al., 2007; Kakar & Shukla, 2010; Mahdzan & Victorian, 2013). It was found that when the education level is higher, people are more aware of the types of life insurance and they attempt to secure themselves and their dependent relatives by consuming it. It is denoted by “ED” in this study.

Crude Death Rate

The crude death rate stands for the average annual number of deaths during a year per 1,000 persons in the population at mid-year, also referred to as the crude death rate. The death rate is 7.5 deaths per 100 people. In general, the crude death rate has a positive relationship with the demand for life insurance. In this study, the crude death rate is denoted by “CDR”.

Price of Insurance

The price of insurance is one of the important determinants of life insurance demand. It is the premium rate of life insurance which is charged annually, quarterly, or monthly. More formally, the price is the cost per 1,000 of ordinary life insurance coverage, defined as the ratio of the total annual premium in force to the total sums insured in force in a year. The price of insurance has a significant and inverse relationship with the demand for life insurance because high life insurance costs tend to decrease demand. The price of life insurance, which is taken as a measure to determine demand in this study, is based on the model used by Browne and Kim (1993). It is represented by “PLI” in this study.

Econometric Modelling for Life Insurance Demand

Keeping in view the above arguments, a multiple econometric model is predicted to find the determinants of life insurance demand in Pakistan. The studies of Babbel (1981), Truett & Truett (1990), Browne & Kim (1993), Hwang & Greenford (2005), Li et al. (2007), Nesterova (2008), Çelik & Kayali (2009), Ibiwoye et al. (2010), Kakar & Shukla (2010), and Mahdzan & Victorian (2013) form the basis for the econometric model design. They designed life insurance demand as a function of explanatory variables:

LIDD = f (ß0 + ß1 GSt + ß2 INFt + ß3 PLIt + ß4 EDt + ß5 CDRt)

LIDD = ß0 + ß1 GSt + ß2 INFt + ß3 PLIt + ß4 EDt + ß5 CDRt + et

Where ß1 > 0, ß2 > 0, ß5 > 0

Here in this study, LIDD = Life Insurance Demand. Sums insured (total business in force of life insurance) is used to measure life insurance demand. GS = Gross saving, INF = Inflation, i.e., consumer price index, PLI = Price of life insurance, i.e., annual gross premium of life insurance, ED = Education, i.e., Government education spending as a percent of GDP, CDR = Crude death rate, And e is the error term of the model, ß0 is the constant value of the regression surface. ß1, ß2, ß3, ß4, ß5 are parameters to be estimated and t = time-period.

The transformation of some variables is carried out in the empirical analysis of data collected (Koop, 2000; Gujarati, 2003) for obtaining an appropriate model in this study. The sum insured (demand for life insurance) and education level are subject to transformation by taking the natural logarithm of their level values. The transformed variables are named as LNLIDD and LNEDT. Hence, these transformed variables are used in the model of analysis.

LNLIDD = ß0 + ß1 GSt + ß2 INFt + ß3 PLIt + ß4 LNEDt + ß5 CDRt + et

Analysis Technique

Before analyzing the time-series data collected through different sources for estimation of the Time-series Model in Statistical Package EVIEWS 9, the stationarity of data is checked by using the ADF test or unit root test.

Unit-Root Test

It is one of the assumptions of standard regression analysis that all the variables being tested should be stationary at level or at first difference. In statistics, a unit root test is a test to check the stationarity of time series variables for using an autoregressive model because this problem is very common in time series data. A well-known test that is valid in large samples is the augmented Dickey–Fuller test. These tests use the existence of a unit root as the null hypothesis. That is, the series with a unit root present in it is said to be non-stationary, and a series when no unit root is present is said to be a stationary series. There are many methods to test unit roots:

• Dickey Fuller
• Augmented Dickey Fuller
• Philips-Perron (PP) Test

In this study, ADF is used as it is the most commonly used unit root test by econometricians.

Co-Integration Test

The two-stage approaches are used to test the co-integration of the variables and confirm whether there exists a long-term balanced association or not (Engle & Granger, 1987). The concept of co-integration implies that even if many economic variables are non-stationary, their linear combination may be stationary through time (Greene, 2006). Spurious results will be obtained in case of having no stationary variable and having no co-integration between the variables (Chan & Lee, 1997). For checking the co-integration, this study used the ARDL technique because some variables are stationary at level and some at first difference.

The Error Correction Model (ECM)

After finding that variables have long-run co-integration, the short-term relationship among variables is found by applying the ECM. This approach is useful in finding both short-term and long-term responses of time-series on other time-series. It finds the speed with which the data of the dependent variable restores to equilibrium by change in other time series. The ECM is made by combining the error term with the first difference of the variables (short-run indicators). This shows that the variables have long-run relationships.

Coefficient of Multiple Determination (R²)

The R squared (R²) statistic shows how well the model fits and confirms the strength of the joint explanatory variables in forecasting the values of the dependent parameter. It also displays how well the regression line fits the sample data in the model. The definition of R-squared is the percentage of the response variable variation that is explained by a linear model. In general, the higher the R-squared, the better the model fits your data. R-squared is always between 0 and 100%: 0% indicates that the model explains none of the variability of the response data around its mean, while 100% indicates that the model explains all the variability of the response data around its mean.

F-Statistic and Probability Value

The F statistic test shows the statistical significance of the overall model, while the probability value (p-value) indicates whether the parameters are statistically significant. The thumb rule for statistical significance for a parameter at 5% or 10% is that its p-value must be less than 0.05 or 0.1, respectively.