Sunteți pe pagina 1din 11

# Duy-econometrics

11:17 AM

## 3. Life expectancy at birth public health expenditure per

capita
a. Before conducting any empirical analysis, briefly discuss your
expectations about the relationship between the two variables.
We expect that "Life expectancy at birth" and "Public health
expenditure per capita" have correlated relationship which can be
modelled regressively where the first is dependent variable and the
latter is independent one.
b. what do summary descriptive statistics and histograms of public
expenditure on health and life expectancy indicate about their
distributions.
Statistical results:

## Some notes can be drawn:

The variable of "Life expectancy at birth" might follow the rule of
normal distribution where the skewness and kurtosis are
approximately 0 and 3 respectively.

## The variable of "Public health expenditure per capita" is not likely

to normally distributed because the values of skewness and
kurtosis are significantly larger than the value of 0 and 3,
respectively.

## We use Shapiro - Wilk W test to examine whether or not those

variables are normally distributed.

With P-value < 0.05, all results has statistical significance. As the
above table can be seen, value of W of "Life expectancy at birth" is
approximately equal to one whereas value of W of "Public health
expenditure per capita" is just 0.59 .

## In conclusion, by observing results from statistical viewpoint, "Life

expectancy at birth" has normally distributed while the other "Public
health expenditure per capita" has non-normally distributed.

## a Before running a regression analysis draw some broad inferences

about the plausibility of the model by creating a scatterplot of life
expectancy and public expenditure on health. How would you
describe the relationship between the two variables? Assess the
relationship by computing a correlation ratio and interpret your
results.

## As we can see, there are a highly correlated relationship between Life

expectancy at birth, denoted by dependent variable Y, and Public
health expenditure per capita which is denoted by independent
variable X. Some inferences are drawn:
The graph shows that a non-linear association between X and Y.
We can use the model created based on the set of data of "Life
expectancy at birth" and "Public health expenditure per capita" for
estimation purpose due to the fitted line.

X and Y.

## Thus, correlation coefficient between X and Y is 0.5426 and indicates a

positively correlated relationship among 2 variables.

## Then, we need to test this result by using Spearman test because

variable X is non-normally distributed.

## The result shows that there is a close bond between Y, "Life

expectancy at birth", and X, "Public health expenditure per capita".

Although the fitted line is fit to the plots in the graph except a few of
outliners, it seems that the coefficient of correlation is not really high,
which is 0.5426 . There are two reasons to explain this:
Relationship among those variables is non-linear whereas the
correlation coefficient reflects linear one.

## And according to result of Spearman test, we obtain that the "real"

correlation coefficient is 0.8440 .

## a Generate a new variable ln(Health expenditure per capita). Create

a scatter plot of life expectancy and ln(Health expenditure per
capita) and compute the correlation ratio between them. Do these
results influence the potential regression specification to be
chosen?

Graph:

## We can see the relationship among two variables changed from

non-linear to almost linear.

## The correlation coefficient value is 0.7731 which is far larger than

0.5426 which is from the previous results.

## To conclude, choosing the potential regression specification is

strongly impact to value of coefficient of correlation and might
make researchers reach some wrong decisions.

## a Assess the impact of public health expenditure on life expectancy

by undertaking the following bivariate regression analyses and

## Because P-value < 0.05, the estimation of and is significant at

level of 5%. Thus, we obtain the equation:

## The value of coefficient of determination is 0.2944 means that the

estimated regression equation explains 29,44% variations of
values of Y. And this is also significant at 5% by doing F-test which
results in the fact that p-value < 0.05.

## Because P-value < 0.05, the estimation of and is significant at

level of 5%. Thus, we obtain the equation:

## This regression equation explains 59.76% differences of values of Y

with level significance of 5%.

## To summarize, both regression model illustrates there is a

positively correlated effect of Health expenditures to variations of
Life expectancy.

## a Conduct a literature survey and select two economic studies that

have analyzed the influence of health expenditure on life
expectancy. Briefly summarize their main findings and review
whether your results are different from them, and if so, why this
might be the case.

## First of all, it is important to mention an economic study "The

relationship between life expectancy at birth and health
expenditures estimated by a cross-country and time-series
analysis" was researched by Elisabeta Jaba, Christiana Brigitte
Balan and Ioan-Bogdan Robu from "Alexandru Ioan Cuza"
University of Iasi, 22 Carol I Avenue, Iasi 700505, Romania and

108 to page 114. Some keys is noted:

## The set of data which consisted of two variables, Health care

expenditures per capita (USD) and Life expectancy at birth (years),
were collected for 175 countries, grouped according to geographic
position and income level, over 16 years from 1995 to 2010.
They applied a panel data analysis to estimate Life expectancy by
a function of Health expenditures.
They used regression model without function of nature logarithm
of explanation variable Health care expenditure.
There is a significant relationship between Health expenditures
and Life expectancy.
Country effects are significant and show the existence of
important differences among the countries.

## Secondly, "Public and private pharmaceutical spending as

determinants of health outcomes in Canada" of Pierre
that:

## The paper measures the statistical relationship between drug

spending in Canadian provinces and overall health outcomes.
The analysis relies on more homogenous data and includes a
more complete set of controls for confounding factors than
previous studies.
Results show a strong statistical relationship between drug
spending and health outcomes, especially for infant mortality and
life expectancy at 65.
The analysis further indicates that substantially better health
outcomes are observed in provinces where higher drug spending
occurs.
Simulations show that if all provinces increased per capita dug
spending to the levels observed in the two provinces with the
highest spending level, an average of 584 fewer infant deaths per
year and over 6 months of increased life expectancy at birth would
result.

Overall, our results are the same as both studies above except the
fact that we highly recommend to use function of logarithm in the
model in order to reflect more accurately the relationship between
Life expectancy and Health expenditures.