Hypotheses Testing in Econometrics

In this course, you will learn why it is rational to use the parameters recovered under the Classical Linear Regression Model for hypothesis testing in uncertain contexts. You will: – Develop your knowledge of the statistical properties of the OLS estimator as you see whether key assumptions work. – Learn that the OLS estimator has some desirable statistical properties, which are the basis of an approach for hypothesis testing to aid rational decision making. – Examine the concept of null hypothesis and alternative hypothesis, before exploring a statistic and a distribution under the null hypothesis, as well as a rule for deciding which hypothesis is more likely to hold true. – Discover what happens to the decision-making framework if some assumptions of the CLRM are violated, as you explore diagnostic testing. – Learn the steps involved to detect violations, the consequences upon the OLS estimator, and the techniques that must be adopted to address these problems. Before starting this course, it is expected that you have an understanding of some basic statistics, including mean, variance, skewness and kurtosis. It is also recommended that you have completed and understood the previous course in this Specialisation: The Classical Linear Regression model. By the end of this course, you will be able to: – Explain what hypothesis testing is – Explain why the OLS is a rational approach to hypothesis testing – Perform hypothesis testing for single and multiple hypothesis – Explain the idea of diagnostic testing – Perform hypothesis testing for single and multiple hypothesis with R – Identify and resolve problems raised by identification of parameters.

• Properties of the OLS Approach
• This week we are going to look at the properties of the OLS approach as a basis for the hypothesis testing, focussing on linearity, unbiasedness, efficiency and consistency.
• Hypothesis Testing
• This week we shall be exploring hypothesis testing, looking at the t-test and the F-test, and considering the problems raised by hypothesis testing.
• Diagnostic Testing I
• This week we shall be discussing diagnostic testing as we look at non-linearity, violation of full rank and errors correlated with regressors.
• Diagnostic Testing II
• This week we will continue to look at diagnostic testing as we consider spherical errors, heteroscedasticity, autocorrelation, Stochastic Regressors, and the non-normality of errors.