Answer-first summary for fast verification
Answer: Not use linear regression.
## Explanation A unit root in a time series indicates the presence of a stochastic or random trend, meaning the series is **non-stationary**. Linear regression assumes that the data is **stationary** - that the mean, variance, and autocorrelation structure do not change over time. ### Key Issues with Unit Roots: - **Non-stationarity**: When a time series has a unit root, its mean and variance change over time - **Spurious regression**: Using linear regression on non-stationary data can lead to misleading results that appear statistically significant but are actually meaningless - **Violated assumptions**: Linear regression's statistical properties (like t-statistics and R-squared) become unreliable with non-stationary data ### Why Other Options Are Incorrect: - **Option A**: While co-integration can sometimes allow for valid regression, this doesn't address the fundamental problem of the unit root in one series - **Option C**: Changing significance levels doesn't solve the underlying non-stationarity issue - **Option D**: Similar to A, this doesn't properly handle the unit root problem ### Correct Approach: The analyst should first difference the series to remove the trend and make it stationary before proceeding with any modeling. This transforms the non-stationary series into a stationary one suitable for regression analysis.
Author: Tanishq Prabhu
Ultimate access to all questions.
No comments yet.
An analyst intends to use linear regression to model the relationship between two-time series. After some testing, she finds out that one of the time series has a unit root. She should:
A
Not use linear regression if the two time series are not co-integrated.
B
Not use linear regression.
C
Perform another test on a higher level of significance before proceeding to use linear regression.
D
Only use linear regression if the time series are co-integrated.