Lecture 8 - Time Series Analysis B

Question 1

What are beta 0 and beta 1 in the linear trend model?

Answer 1

The intercept and slope parameters respectively.

Question 2

What are the 5 consequences of autocorrelation?

Answer 2

1) Variance of error terms will be higher than measured by the MSE
2) Standard error regression coefficients will be high
3) The least squares method will not produce efficient estimates
4) Hypothesis tests will not be valid
5) The model fit will be poor and forecasts will not be valid

Question 3

When should the autoregressive model be used?

Answer 3

When significant autocorrelation is present

Question 4

What is the autoregressive model?

Answer 4

A multiple regression model in which independent variables are time lagged versions of the dependent variables.

Question 5

Does the autoregressive model have an intercept and an error term like the linear regression model?

Answer 5

Yes

Question 6

What three measures can be used to evaluate the effectiveness of the AR, quadratic and linear regression models?

Answer 6

1) Forecast errors
2) Regression Statistics (R2, F and t stats)
3) Residual analysis (Residual plots, Durbin Watson)

Question 7

Can the DW test be used for the AR model?

Answer 7

No because of the lags in the independent variables.