BusinessEconomic309

Project Part 3

Professor Ogunc

Economic 309

Alina Basnet

Date: 3/21/2021

Project Part 3

Nike Company

Above are the three different macroeconomic variables that I selected to run the revenue of my company i.e., Nike. I choose Population, Earnings and Unemployment. As we saw the diagram of correlation, we can rectify that the macroeconomic variables have linear relationships between the revenue of Nike as the lines of x and y variables are moving along the same direction. First, population and earnings have positive relation with the revenue, and they are significantly correlated with the y variable Revenue because the p-value is smaller than the correlation coefficient. But unemployment is not significantly correlated with revenue because it has greater p-value than the correlation coefficient. By this, we can verify that Earnings and Revenue are highly correlated to each other than other variables as the absolute value of the coefficient is greater than the p-value which means that we can reject the null hypothesis. This scenario tells us that unemployment is least related to the revenue as these two are negatively related.

The Scatterplot of Revenue with Population shows positive relation between each other as we can see the the variables is moving along the data of revenue and there is no trend as well. Thus, the pair is linear.

Likewise, we observe positive relationship as diagnosed by the scatterplot of Revenue and Earnings.

Now, the scatterplot of Revenue and Unemployment is a negative relation. This pair is nonlinear. As we see the data is moving opposite direction. As the unemployment is increasing, the revenue is decreasing.

From the above Regression analysis, we concluded that, holding earnings constant the earnings and unemployment, if the population increases by 100000, then revenue of Nike increases by 0.24. Whereas, holding population and unemployment constant, for every 10 increase in Earnings leads to increase in the revenue by 1924. Similarly, holding population and earnings constant, in the case of unemployment if its rate increases by 1 then there is decrease in revenue by 2567.

The P-value of all constant, population, earnings and unemployment are 0. This means that the variables are statistically significant. The regression coefficient is higher than the p-value.

This analysis shows that the variables of Earnings have high correlation with the revenue.

I do have a sign switch because the regression coefficient does not match with that of correlation coefficient. The coefficient correlation of constant and earnings are positive, but the regression correlation of population and unemployment are negative. Therefore, there is a negative sign switch.

After all the analysis that I have done, I see that there is outliers and I probably must run the analysis again which will result in change of the value and make it better than the present value.

By the above regression analysis, I would not settle for a good overall model because sign switch has been detected. This model has not satisfied the multicollinearity assumption as well. There is significance in F-test and R squared is percent variation which has been demonstrated by the above model. This test helps to figure out the linear regression of my model which is clearly depicted in the method but in overall the model is not satisfactory.