Scribd
Upload
8 views5 pages

Ôn Final KTL

Uploaded by

khonbaohs
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
8 views5 pages

Ôn Final KTL

Uploaded by

khonbaohs
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

Chapter 1

Methodology of Econometrics
- Statement of theory or hypothesis
- Specification of the mathematical model of the theory
- Specification of the statistical, or econometric model
- Collecting data
- Estimation of the parameters of the econometric model
- Hypothesis testing
- Forecasting or prediction
- Using the model for control or policy purposes

Terminology for simple regression


y x
Depedent variable Independent variable
Explained variable Explanatory variable
Response variable Control variable
Predicted variable Predictor variable
Regressand Regressor

Types of data
- Time series data: a variable - different times
- Cross-section data: one or more variables - same point in time
- Panel data/ Pooled data: time series + cross-section

Measurement scales of variables


- Ratio scale: GDP growth rate, interest rate
- Interval scale: distance between two time period
- Ordinal scale: income class (upper, middle, lower), grading systems (A,B,C)
- Nominal scale: gender, marital status

Chapter 2
Properties of OLS statistics
- The sum and the sample average of the OLS residuals is zero ↔ ∑ u^i=0
- The sample covariance between the regressors and the OLS residuals is zero ↔ ∑ X i u^i=0
- The point ( X , Y ) is always on the OLS regression line
Properties of least-squares estimators
- Gauss-Markov Theorem: the least-squares estimators have the minimum variance - BLUE (Best Linear
Unbiased Estimator)

The assumptions underlying OLS


- The regression model is linear in parameters
- X values are fixed in repeated samplings. X is assumed to be non-stochastic
- Zero mean value of disturbance ↔ E ( ui∨X i) =0 or Zero variance between ui and Xi ↔ Cov ( ui X i )=0
2
- Homoscedasticity or equal variance of disturbance ↔ Var ( u i∨X i ) =σ

- No autocorrelation between the disturbance ↔ Cov ( ui u j ) =0 , i≠ j


- The number of observations n must be greater than the number of parameters to be estimated
- Variability in X values. The X values in a given sample must not all be the same
- The regression model is correctly specified. There is no specification bias or error in the model used in
empirical analysis
- There is no perfect multicollinearity

A measure of “Goodness of fit”


- The Goodness of fit: how “well” the sample regression line fits the data measured by r2
- r2 measures the proportion or percentage of the total variation in Y explained by the regression model

Chapter 5
1. Multicollinearity
- High (but not perfect) correlation between two or more independent variables is called multicollinearity
- The estimator is indeterminate, the variances are infinite
- The t ratio of one or more coefficients tends to be statistically insignificant
- R2, the overall measure of goodness of fit, can be very high
- When collinearity is high, tests on individual regressors are not reliable
- Detecting multicollinearity: variance-inflating factor (VIF) and Tolerance and variance inflation factor (TOL)
- Curing the problem: do nothing, a priori information, additional or new data

2. Heteroscedasticity
- The variances of ui are not the same
- Heteroscedasticity can arise as a result of the presence of outliers
- Heteroscedasticity is likely to happen in cross-sectional analysis
- OLS estimators are still linear and unbiased
- Not BLUE → variance of coefficient of estimators is not minimum, σ^ 2 is biased, t and F statistics are
unreliable
- Detecting heteroscedasticity: graphical method, Park test, Breusch-Pagan test, White’s General
Heteroscedasticity test
- Correcting for heteroscedasticity:
+ When σ 2i is known: the method of weighted least squares
+ When σ 2i is unknown: use 1 of 4 assumptions → transformation

3. Autocorrelation
- Autocorrelation is most likely to occur in time series data
- Reason for autocorrelation: inertia, excluded variables case, incorrect functional form, Cobweb phenomenon,
lags, manipulation of data, data transformation
- The coefficient estimator is still linear and unbiased
- Not BLUE → variance of coefficient of estimators is not minimum, σ^ 2 is biased, t and F statistics are
unreliable
- Should use GLS instead of OLS
- Dectecting autocorrelation: graphical method, the runs test, the Durbin Waston test, Breusch-Godfrey (BG)
test
- Curing the problem: pure autocorrelation (add more important variables), use GLS, Newey-West, continue to
use OLS.

Chapter Dummy
- A dummy variable (binary variable) is a variable that takes on the value 1 or 0.
- Do not include male and female in a model with an intercept → dummy variable trap due to perfect
collinearity.
- If there are n categories, there should be n −1 dummy variables.

Giải đề
Đề 1 K59 27/12/2021
Section A: MCQ
1B 2C 3B 4A 5D 6B 7A 8D 9C 10C
Section B: Short answer
Problem 1:
(i) .
- rosneg; utility have negative relationships on lsalary.
- roe; finance; consprod; lsales have positive relationships on lsalary.
- R2 <50 % is not high
- Prob> F=0 % → reject H 0 → not at all the slope coefficients simultaneously zero
(ii) .
- ∆=0.161 −0.138=0.023=2.3 %
Let finance is a base group, we have equation:
lsalary =α 0+ α 1 lsales+ α 2 roe+α 3 rosneg +γ 2 consprod + γ 3 utility +v
Use t-test on consprod to test whether the difference is statistically significant

Problem 2:
(i) Hypothesis:
H 0 : β 1=0
H 1 : β1 ≠ 0
−0.163
t= =− 9 →this is fairly strong ¿ reject H 0 →There is a tradeoff between working∧sleeping
0.0181
(ii)
2
R /(k −1)
F= 2
=19.635 > F∗ → not at all the slope coefficients are simultaneously zero
(1 − R )/(n − k )
(iii) Hypothesis:
H 0 : β 1=−0.1
H 1 : β1 >− 0.1
−0.163+ 0.1
t= =− 3.48 →this is fairly strong ¿ reject H 0
0.0181
(iv)
Step 1: Exclude other factors, build a new equation including sleep (dependent variable), age and agesq
Step 2: Regression this equation
Step 3: State hypothesis that both slope coefficients are zero
Step 4: Use t-test to test

Problem 3:
β2 β3
Y i=β 1 X 2 i X 3 i ui →lnY =ln β1 + β 2 ln X 2 i+ β3 X 3 i +ln ui →lnY =β 0 + β 2 ln X 2 i+ β3 X 3 i
Y ∗
→ lnY − ln X 2i =β 0+(β 2 − 1) ln X 2 i + β 3 X 3 i → ln =β 0 + β 2 ln X 2i + β 3 X 3 i
X2i
Test hypothesis whether β ∗2 =0
∗
H 0 : β 2 =β 3=0
H1: ∃ β ≠ 0
0.995 /2 ∗
F= =1691 , 5> F 2, 17 ,0.05=3.2 → reject H 0
(1 −0.995)/17

Problem 4:
(i) 1 point increases in female*educ leads to the decrease in wage by 0.724%
(ii) 1 point increases in female leads to the

You might also like