Serial correlation Simplified

Get easy to understand CFA Level 2 material regularly delivered to your inbox
📪📪📪📪https://sendfox.com/lp/1gjgo6

CFA level 2: Quantitative Methods 2020

This channel is targeted at candidates for the CFA™ (Chartered Financial Analyst) Level 2 2020 exam.

In this video, we are going to discuss Serial Correlation. The issues we face in regression models, and solutions to rectify it.

The framework we are discussing about Serial correlation relates to Assumptions of Regression Model which is
1. Dependent and independent variables are assumed to have a linear relationship
2. Error Terms are assumed to be not serially correlated.

Serially correlation of error terms means the errors terms have a pattern ,that is, positive error terms are followed by positive error terms and negative error terms are followed by negative error terms. This makes variance of standard errors very high and when we do the test of significance t-stat = (coefficient of independent variable-0)/ standard error. With a high error term, we get a t-stat which is very low, hence meaning the coefficient of error terms have very less significance and we fail to reject the null hypothesis. ( Null hypothesis is fail to reject null, hence coefficient is insignificant, Alternative hypothesis is reject the null, hence coefficient is significant)

Test Serial Correlation:
To Test Serial Correlation we use Durbin watson (DW) = 2(1-r) , where r is correlation coefficient.

Therefore, if
r= 1 , then DW= 0. Hence, serial correlation exists
r=0, then DW= 2, no serial correlation
r=-1, then DW= 4 .Hence, serial correlation exists.

Other Quantitative Methods Topics of interest:

Assumptions of Linear Regression :    • CFA Level 2: Quantitative Methods: As...  
Multicollinearity :    • CFA Level 2: Multicollinearity  
Heteroskedasticity :    • CFA Level 2: Heteroskedasticity Simpl...  
Quantitative Method Playlist:    • CFA Level 2 2020 :Quantitative Methods  




Please like and subscribe to my channel :)

Follow me on twitter:   / watchliststock  
Email me for mentoring : [email protected]