0% found this document useful (0 votes)
156 views19 pages

Econometrics Model With Panel Data: Dinh Thi Thanh Binh Faculty of International Economics, FTU

This document discusses econometrics models for panel data. It defines panel data as data that observes the same individuals or groups over multiple time periods. There are three main econometrics models for panel data: pooled OLS, fixed effects model, and random effects model. The fixed effects model controls for time-invariant characteristics of individuals and is preferred when these characteristics are correlated with independent variables. The random effects model treats individual characteristics randomly and is preferred when they are uncorrelated with independent variables. The document outlines tests in Stata to select the appropriate model.

Uploaded by

Lâm Bulls
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
156 views19 pages

Econometrics Model With Panel Data: Dinh Thi Thanh Binh Faculty of International Economics, FTU

This document discusses econometrics models for panel data. It defines panel data as data that observes the same individuals or groups over multiple time periods. There are three main econometrics models for panel data: pooled OLS, fixed effects model, and random effects model. The fixed effects model controls for time-invariant characteristics of individuals and is preferred when these characteristics are correlated with independent variables. The random effects model treats individual characteristics randomly and is preferred when they are uncorrelated with independent variables. The document outlines tests in Stata to select the appropriate model.

Uploaded by

Lâm Bulls
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Econometrics Model with Panel Data

Dinh Thi Thanh Binh


Faculty of International Economics, FTU

1
1. Definition
• Consumption and Income of 5 households in 3 years

Obs Id year Consumption Income


1 1 2007 1.200 2.000
2 1 2008 1.400 2.100
3 1 2009 1.500 2.200
-- .. .. . .
13 5 2007 2.000 3.000
14 5 2008 2.100 3.500
15 5 2009 2.300 3.400
2
1. Definition
• Panel data: the same groups of observation (N)
(household, enterprise, individuals, countries…) are
observed over time (T)
• The panel data can have:
– Variable has different values for each obs, but
does not change overtime (location, gender..)
– Variable has different values for each obs, but
change overtime (exchange rate, FDI,
consumption, income..)

3
2. Advantages of Panel Data
• For example: ( Cross-section dataset) Analyze the relationship
between quantity of fertilize and productivity
• The sample is of 20 fields (thửa ruộng)
Productivity^ = 4 +0.5Fertilize
• Q: Are there any problems with the estimated result?
• A: Yes
– The productivity also depends on the rich (độ phì nhiêu) of
the land that cannot be observed  the model cannot
evaluate the effects of the unobserved variables (u).
– If the variable “the rich of the land” has correlation with
the independent variable “Fertilize”  We have the
problem of the correlation between X and u

4
2. Advantages of Panel Data

– If 20 fields distributed in 20 different provinces,


we want to observe the effect of province 
generate 19 dummy variables  too complicated

 Some conclusions:
– There are missing unobserved variables (u) and u
correlate with X
– These variables are different for each unit
=> Biased OLS estimation

5
2. Advantages of Panel Data

• Overcome the problem of missing unobserved


variables
• The data (units) are observed overtime, so the
sample is larger and we can track all the changes
of units overtime.

6
3. Econometric Model for Panel Data

The unobserved variables can be:


– Change/ unchange by unit (i)
– Change/ unchange by time (t)
– Change by both unit (i) and time (t)

7
3. Econometric Model for Panel Data

• We have model:
Yit = b0 + δ0d2t+ β1Xit+ ai+ uit t=1,2

uit satisfy all the assumptions of the OLS


ai: unobserved and unchanged overtime (ie.
geography location of provinces)

 Fixed effects model (Mô hình này được gọi là


mô hình ảnh hưởng cố định)
8
3. Econometric Model for Panel Data

• Let: vit = ai+uit


• Depends on the characteristics of ai, we have
3 models:
– Pooled OLS (Mô hình hồi qui gộp - POLS)
– Fixed effect (Mô hình tác động cố định - FE)
– Random effect (Mô hình tác động ngẫu nhiên-RE)

9
Pooled OLS - POLS
1. If there is no existence of ai:
• Apply the OLS method. The panel data is considered
as pooled cross section data  POLS
• OLS is the best method because error term v satisfy
all the assumption of the OLS)

10
Random Effect Model

2. If there is the existence of ai:


2.1 If ai does not correlate with X => vit = ai+uit
(composite error term) has no correlation with X

Cov(ai , X )  0
RE - random effects model

11
Random Effect Model

• We have model:
Yit = b0 + β1X1it+..+ βkXkit + ai+ uit,
• vij = ai+uij: composite error
Yit = b0 + β1X1it+..+ βkXkit + vij
RE analyze the effect of ai on the change of Y We
have:
Var(vij)=sa² + su²  homoskedasticity

12
Random Effect Model

cov(vij; vis) = cov(ai + uij; ai + uis)= var(ai) +cov(ai, uis) +


cov(ai;uij) when j ≠ s
cov(vij; vis) = var(ai) +var(uij)+cov(ai, uis) + cov(ai;uij) khi j
=s
• These factors are different from 0, even when a and
u do not correlate
 There is auto-correlation in u
 Estimation Method: GLS for pooled cross-section
data

13
Fixed Effect Model - FE

2.2 When ai correlate with X, both RE and POLS are


biased
Cov(ai , X )  0

 Cannot group ai into vit


Use FE – fixed effects model with vertical
estimation method
** Note: if X is unchanged over time, we can not use
FE model.
14
FE- vertical estimation

Y it
b b X it
 a i  u it
0 1

Y it
b b X it
 a i  u it
0 1

Y Y
it it
b
1
 X  X   u  u 
it it it it

Can delete the effect of unobserved variable


ai

15
FE vs. RE
FE: RE
• Cannot evaluate the • Assumption about the
effect of unchanged- un-correlation between
overtime variable: a and X are too strong.
gender, capability, initial
characteristics

16
How to choose the model?
• Nếu biến bị bỏ sót là không đáng kể => MH hồi qui
gộp là tốt nhất
• Nếu biến bị bỏ sót không tương quan với X => RE là
hiệu quả hơn FE (nhưng phải giả thiết về sự không
tương quan giữa a và u)
• Nếu biến bị bỏ sót là tương quan với X thì RE là
chệch và không vững=> chọn FE
• Lựa chọn giữa POLS và RE: sử dụng xttest0
• Nếu RE được lựa chọn => sẽ chọn giữa FE hay RE:
Hausman
17
Câu lệnh trong STATA
Bước 1: cần test để xem nên chọn RE/FE hoặc POLS.
 Breusch-Pagan test for the significant difference
across units.
H0: no significant difference across units (no panel
effect) [không tồn tại ai )
H1: significant difference across units (panel effect) [có
tồn tại ai )
xtset [panel variable] [time variable]
Xtreg Y X, re
Xttest0
p-value < 0.05  bác bỏ H0, chấp nhận H1 
RE/FE
p-value >0.05  chấp nhận H0  POLS
18
Câu lệnh trong STATA
Nếu trong Bước 1, H0 bị bác bỏ  HAUSMAN TEST để
chọn FE or RE
H0: Cov(ai,xit) = 0
H1: Cov(ai,xit) ≠ 0
xtreg Y X, fe
est store fe
xtreg Y X, re
hausman fe
• Nếu p-value < 0.05  bác bỏ H0  chọn FE
• Nếu p-value > 0.05  chấp nhận H0  choose RE
19

You might also like