Đinh Nguyễn Thu Thảo_BAFNIU21576

Trần Uyên Thi_BAFNIU21579

Trần Phan Như Quỳnh_BAFNIU21557
Nguyễn Trâm Huyền_BAFNIU21302
HOMEWORK

1. Explain the difference between: Pooled Regression, Fixed Effect Method (FEM),
Random Effect Method (REM) for Panel Regression.
Pooled Regression Fixed Effect Random Effect
Method (FEM) Method (REM)

Description Assumes no individual Controls for Assumes

(cross-sectional) or unobserved individual-specific
temporal (time-series) heterogeneity by effects are random
differences in the data. allowing each entity and uncorrelated
All observations are (or time) to have its with the
treated as a single intercept. independent
pooled dataset. variables.

Key Assumption Homogeneity across Differences across Variation across

entities and time; no entities (or time entities is random
unobserved periods) are and not correlated
heterogeneity. captured in the with the explanatory
intercept and are variables.
fixed (constant) over
time.

How It Works Includes dummy Models the error

variables or term as having two
de-meaned components: one
variables to account due to
for individual-specific
individual-specific effects and one due
effects. to idiosyncratic
error.

Usage When the data does not When you suspect When individual
exhibit variation across that unobserved effects are
entities or time. factors are uncorrelated with
correlated with the regressors and are
independent randomly
variables. distributed.

Drawback Ignores the panel Cannot estimate If the assumption of

structure, potentially time-invariant no correlation is
 leading to biased results variables because violated, estimates
if heterogeneity exists. they are absorbed may be biased and
into the fixed effects. inconsistent.

2. Briefly explain the 3 tests to choose the right panel regression model. (Hausman
Test, F-Test, Bresch Pagan Lagrange Multiplier Test)

Hausman Test F-Test Breusch-Pagan

Lagrange
Multiplier Test (LM
Test)

Purpose Determines whether Determines whether Determines whether

Fixed Effect (FE) or Pooled Regression Pooled Regression
Random Effect (RE) or Fixed Effect or Random Effect
is appropriate. Model is Model is
appropriate. appropriate.

How It Works Tests if individual Tests if the fixed Tests for the
effects are effects presence of random
correlated with the (entity-specific effects
independent intercepts) are
variables. statistically
significant.

Null Hypothesis RE is appropriate Pooled regression is Pooled regression is

(no correlation appropriate (no appropriate (no
between individual significant fixed random effects).
effects and effects).
independent
variables).

Decision If p-value > 0.05, If p-value > 0.05, If p-value > 0.05,
use REM. use Pooled use Pooled
If p-value ≤ 0.05, Regression. Regression.
use FEM. If p-value ≤ 0.05, If p-value ≤ 0.05,
use FEM. use REM.

3. Briefly explain the 3 main diagnostic tests for panel regression (multicollinearity,
heteroskedasticity, autocorrelation)

● Multicollinearity:
- This problem occurs when the explanatory variables are very highly correlated with
each other.
- Perfect multicollinearity: exact relationship between 2 or more variables. Then we
cannot estimate all the coefficients.
- Near Multicollinearity: Problems if Near Multicollinearity is present but ignored
+ R2 will be high but individual coefficients will have high standard errors.
+ The regression becomes very sensitive to small changes in the specification.
+ Thus confidence intervals for the parameters will be very wide, and
significance tests might therefore give inappropriate conclusions.
- Consequences: Multicollinearity makes it difficult to isolate the individual impact of
each predictor on the dependent variable. It can lead to unstable and unreliable
coefficient estimates, making it hard to interpret the results.
- Tests:
+ Variance Inflation Factor (VIF): A high VIF (generally above 5 or 10) for a
predictor variable indicates high multicollinearity.
+ Correlation Matrix: Examining the correlation matrix between independent
variables can reveal high correlations.
● Heteroskedasticity:
- This assumption is violated when the variance of the error term in the regression
model is not constant across all observations. In simpler terms, the spread of the
residuals (the difference between the predicted and actual values) is not equal for all
observations.
- Consequences: Heteroskedasticity can lead to inefficient and biased coefficient
estimates, making it difficult to draw accurate inferences.
- Test:
+ Breusch-Pagan Test: Detects heteroskedasticity by regressing squared
residuals on independent variables.
+ White Test: A general test that identifies heteroskedasticity without assuming
its specific form.
+ Modified Wald Test (for panel data): Tests for groupwise heteroskedasticity
in panel regression.
● Autocorrelation:
- Autocorrelation occurs when the error terms in the regression model are correlated
with each other over time. This is common in time series data.
- Consequences: Autocorrelation can lead to inefficient and biased coefficient
estimates, and it can invalidate statistical tests.
- Tests:
+ Durbin-Watson test: A common test for autocorrelation in time series data.
+ Breusch-Godfrey test: A more general test for autocorrelation that can be
used in various regression models.