Explanation

The Breusch-Pagan test is used to detect heteroskedasticity in regression models:

• Null Hypothesis: Homoskedasticity (constant variance of errors)
• Alternative Hypothesis: Heteroskedasticity (non-constant variance of errors)

At 5% significance level:

Model 1:

• BP statistic = 5.458, p-value = 0.0195
• Since p-value (0.0195) < 0.05, we reject the null hypothesis
• Conclusion: Model 1 exhibits heteroskedasticity

Model 2:

• BP statistic = 0.046, p-value = 0.8302
• Since p-value (0.8302) > 0.05, we fail to reject the null hypothesis
• Conclusion: Model 2 does not exhibit significant heteroskedasticity

Analysis of options:

• A: Incorrect - Model 1 shows heteroskedasticity, not homoskedasticity
• B: Correct - When heteroskedasticity is detected (Model 1), White's correction for heteroskedasticity-consistent standard errors should be applied
• C: Incorrect - Model 2 shows no significant heteroskedasticity, so t-statistics can be relied upon

Correct Answer: B - White's correction should be applied to Model 1 to address the detected heteroskedasticity.