L23 - Granger Causality and Deterministic Regressors

Question 1

What is a Granger Causality?

Answer 1

• used for forecasting rather than estimating

Question 2

How is strong exogeneity defined?

Answer 2

Engle, Hendry and Richard (1983) define strong exogeneity as follows. Consider a model of the form:

Y_t = βX_t + u_t

X is strongly exogenous if:

1. X is weakly exogenous for the purposes of estimating the parameter of interest β
2. Y does not Granger cause X.

Strong exogeneity means that we do not take into account feedback from Y to X when we use our estimated model for forecasting

Question 3

How do you test for Granger causality?

Answer 3

In general, we would regress Y_t on its previous lags

• use a t-test with a degree of freedom of n-1 ( as its a sample and all values are exogenous - no variables)

Question 4

What is super exogeneity?

Answer 4

super exogeneity entails that the. parameters of a conditional model are invariant to changes in the distribution of weakly exogenous conditioning variables.

When the equation is used to form policy

Question 5

What are deterministic regressors?

Answer 5

Question 6

What can we do to our equation if it contains an intercept value?

Answer 6

Question 7

Why does the presence of a trend create a problem for statistical analysis?

Answer 7

the classical statistical analysis assumes stationarity –> the moments of the distributions e.g. mean, variance and covariances are constant throughout time

to adjust for non-stationary, include a time trend in the data

Question 8

Why do you take logarithms of a linear time trend?

Answer 8

regressed the log of Q on the log of Y and a time trend