Stationarity Check:
Regression of a non-stationary time-series variable may cause a spurious regression or non-sense regression (a symptom of spurious regression is that the R-square value is greater that Durbin-Watson test statistics) which is not desirable. After estimating the first regression, the R-square value is 0.99 while the Durbin-Watson statistic is 0.86. So if I continue with the data without correcting for non-stationarity, my estimated model will be a spurious regression model.
I further checked for the stationarity for all the variables using Augmented …show more content…
The result shows that the value of the estimated coefficient of unemployment rate is -0.20. This means that for 1% increase in unemployment rate, infant mortality will go down by 0.20%, other things being constant. Also the P-value of the estimated coefficient is 0.0395, which indicates that we are 95% sure that the estimated effect is significant. The estimated coefficient of poverty rate is 0.1810, which denotes that, ceteris paribus, for 1% increase in poverty rate, infant mortality will go up by 0.18%. Although this is consistent with our prior belief, the P-value of 0.2047 indicates that there is no significant relation between poverty rate and infant mortality in the U.S. previous literature found that infant mortality is negatively correlated with the income level of parents. This study also finds a similar relation. The estimated coefficient indicates that for 1% the per capita real income, infant mortality will decrease by 3.79%, other things being constant. However, the high probability (0.2497) indicates that estimated effect is not significantly different from zero. The estimated coefficient of inflation rate is 0.0257, indicating that, ceteris paribus, 1% increase in inflation rate is correlated with 2.57% increase in infant mortality rate. While this finding consistent with our belief, the high p-value indicates that even at 10% level of significance we fail to reject the null hypothesis that inflation rate has no significant influence on infant mortality rates. The estimated coefficient of health expenditure 0.1626, denoting that, ceteris paribus, for 1% increase in healthcare expenditure, infant mortality will do up by 0.1626%, although the high P-value (0.8806) denotes that this estimated value is not significantly different from zero. The estimated coefficient of food expenditure by individuals and families (-0.8736)