77
Chapter 15 Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Embed Size (px)

Citation preview

Page 1: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Chapter 15

Panel Data Models

ECON 6002EconometricsMemorial University of Newfoundland

Adapted from Vera Tabakova’s notes

Page 2: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Chapter 15: Panel Data Models

15.1 Grunfeld’s Investment Data

15.2 Sets of Regression Equations

15.3 Seemingly Unrelated Regressions

15.4 The Fixed Effects Model

15.4 The Random Effects Model

Extensions RCM, dealing with endogeneity when we have

static variables

Slide 15-2Principles of Econometrics, 3rd Edition

Page 3: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Chapter 15: Panel Data Models

The different types of panel data sets can be described as:

“long and narrow,” with “long” time dimension and “narrow”, few

cross sectional units;

“short and wide,” many units observed over a short period of time;

“long and wide,” indicating that both N and T are relatively large.

Slide 15-3Principles of Econometrics, 3rd Edition

Page 4: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.1 Grunfeld’s Investment Data

The data consist of T = 20 years of data (1935-1954) for N = 10 large firms.

Let yit = INVit and x2it = Vit and x3it = Kit

Slide 15-4Principles of Econometrics, 3rd Edition

(15.1)

(15.2)

,it it itINV f V K

1 2 2 3 3it it it it it it ity x x e

Notice the subindices!

Value of stock, proxy for expected profitsCapital stock, proxy for desired permanentCapital stock

Page 5: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.2 Sets of Regression Equations

Slide 15-5Principles of Econometrics, 3rd Edition

(15.3a)

(15.3b)

, 1 2 , 3 , ,

, 1 2 , 3 , ,

1, ,20

1, ,20

GE t GE t GE t GE t

WE t WE t WE t WE t

INV V K e t

INV V K e t

1 2 2 3 3 1, 2; 1, ,20it it it ity x x e i t

For simplicity we focus on only two firmskeep if (i==3 | i==8) in STATA

Page 6: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.2 Sets of Regression Equations

Slide 15-6Principles of Econometrics, 3rd Edition

(15.4a)

(15.4b)

, 1, 2, , 3, , ,

, 1, 2, , 3, , ,

1, ,20

1, ,20

GE t GE GE GE t GE GE t GE t

WE t WE WE WE t WE WE t WE t

INV V K e t

INV V K e t

1 2 2 3 3 1, 2; 1, ,20it i i it i it ity x x e i t

Page 7: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.2 Sets of Regression Equations

Assumption (15.5) says that the errors in both investment functions (i) have zero mean, (ii) are homoskedastic with constant variance, and (iii) are not correlated over time; autocorrelation does not exist. The two equations do have different error variances

Slide 15-7Principles of Econometrics, 3rd Edition

(15.5)

2, , , ,

2, , , ,

0 var cov , 0

0 var cov , 0

GE t GE t GE GE t GE s

WE t WE t WE WE t WE s

E e e e e

E e e e e

2 2 and .GE WE

Page 8: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.2 Sets of Regression Equations

Slide 15-8Principles of Econometrics, 3rd Edition

reg inv v k if i==3scalar sse_ge = e(rss)

reg inv v k if i==8scalar sse_we = e(rss)

Page 9: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.2 Sets of Regression Equations

Let Di be a dummy variable equal to 1 for the Westinghouse

observations and 0 for the General Electric observations. If the

variances are the same for both firms then we can run:

Slide 15-9Principles of Econometrics, 3rd Edition

(15.6)1, 1 2, 2 3, 3it GE i GE it i it GE it i it itINV D V D V K D K e

* Create dummy variablegen d = (i == 8)gen dv = d*vgen dk = d*k

* Estimate dummy variable modelreg inv d v dv k dktest d dv dk

Page 10: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.2 Sets of Regression Equations

Slide 15-10Principles of Econometrics, 3rd Edition

Page 11: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.2 Sets of Regression Equations

Slide 15-11Principles of Econometrics, 3rd Edition

* Goldfeld-Quandt testscalar GQ = sse_ge/sse_wescalar fc95 = invFtail(17,17,.05)di "Goldfeld-Quandt Test statistic = " GQdi "F(17,17,.95) = " fc95

Goldfeld-Quandt Test statistic = 7.45338

F(17,17,.95) = 2.2718929

So we reject equality at the 5% level…=> we cannot really merge the two equations for now…

Page 12: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3 Seemingly Unrelated Regressions

This assumption says that the error terms in the two equations, at the same point in time, are correlated. This kind of correlation is called a contemporaneous correlation.

Under this assumption, the joint regression would be better than the separate simple OLS regressions

Slide 15-12Principles of Econometrics, 3rd Edition

(15.7) , , ,cov ,GE t WE t GE WEe e

Page 13: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3 Seemingly Unrelated Regressions

Econometric software includes commands for SUR (or SURE) that

carry out the following steps:

(i) Estimate the equations separately using least squares;

(ii) Use the least squares residuals from step (i) to estimate

;

(iii) Use the estimates from step (ii) to estimate the two equations jointly

within a generalized least squares framework.

Slide 15-13Principles of Econometrics, 3rd Edition

2 2,, and GE WE GE WE

Page 14: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3 Seemingly Unrelated Regressions

Slide 15-14Principles of Econometrics, 3rd Edition

Page 15: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3 Seemingly Unrelated Regressions

Slide 15-15Principles of Econometrics, 3rd Edition

* Open and summarize data (which is already in wide format!!!)use grunfeld2, clearsummarize

* SUR sureg ( inv_ge v_ge k_ge) ( inv_we v_we k_we), corrtest ([inv_ge]_cons = [inv_we]_cons) ([inv_ge]_b[v_ge] = [inv_we]_b[v_we]) ([inv_ge]_b[k_ge] = [inv_we]_b[k_we])

Page 16: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3.1 Separate or Joint Estimation?

There are two situations where separate least squares estimation is

just as good as the SUR technique :

(i) when the equation errors are not contemporaneously correlated;

(ii) when the same (the “very same”) explanatory variables appear in

each equation.

If the explanatory variables in each equation are different, then a test

to see if the correlation between the errors is significantly different

from zero is of interest.Slide 15-16Principles of Econometrics, 3rd Edition

Page 17: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3.1 Separate or Joint Estimation?

In this case we have 3 parameters in each equation so:

Slide 15-17Principles of Econometrics, 3rd Edition

22,2

, 2 2

ˆ 207.58710.53139

ˆ ˆ 777.4463 104.3079GE WE

GE WEGE WE

r

20 20

, , , , ,1 1

1 1ˆ ˆ ˆ ˆ ˆ

3GE WE GE t WE t GE t WE tt tGE WE

e e e eTT K T K

3.GE WEK K

Page 18: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3.1 Separate or Joint Estimation?

Testing for correlated errors for two equations:

LM = 10.628 > 3.84 (Breusch-Pagan test of independence: chi2(1))

Hence we reject the null hypothesis of no correlation between the

errors and conclude that there are potential efficiency gains from

estimating the two investment equations jointly using SUR.

Slide 15-18Principles of Econometrics, 3rd Edition

0 ,: 0GE WEH

2 2, (1) 0 under .GE WELM Tr H

Page 19: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3.1 Separate or Joint Estimation?

Testing for correlated errors for three equations:

Slide 15-19Principles of Econometrics, 3rd Edition

0 12 13 23: 0H

2 2 2 212 13 23 (3)LM T r r r

Page 20: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3.1 Separate or Joint Estimation?

Testing for correlated errors for M equations:

Under the null hypothesis that there are no contemporaneous

correlations, this LM statistic has a χ2-distribution with M(M–1)/2

degrees of freedom, in large samples.

Slide 15-20Principles of Econometrics, 3rd Edition

12

2 1

M i

iji j

LM T r

Page 21: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.3.2 Testing Cross-Equation Hypotheses

Most econometric software will perform an F-test and/or a Wald χ2–test; in the context of SUR equations both tests are large sample approximate tests.

The F-statistic has J numerator degrees of freedom and (MTK) denominator degrees of freedom, where J is the number of hypotheses, M is the number of equations, and K is the total number of coefficients in the whole system, and T is the number of time series observations per equation. The χ2-statistic has J degrees of freedom.

Slide 15-21Principles of Econometrics, 3rd Edition

(15.8)0 1, 1, 2, 2, 3, 3,: , ,GE WE GE WE GE WEH

Page 22: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4 The Fixed Effects Model

SUR is OK when the panel is long and narrow, not when it is short and wide. Consider instead…

We cannot consistently estimate the 3×N×T parameters in (15.9) with only NT total observations. But we can impose some more structure…

Slide 15-22Principles of Econometrics, 3rd Edition

(15.9)

(15.10)

1 2 2 3 3it it it it it it ity x x e

1 1 2 2 3 3, ,it i it it

We consider only one-way effects and assume common slopeparameters across cross-sectional units

Page 23: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4 The Fixed Effects Model

All behavioral differences between individual firms and over time are

captured by the intercept. Individual intercepts are included to

“control” for these firm specific differences.

Slide 15-23Principles of Econometrics, 3rd Edition

(15.11)1 2 2 3 3it i it it ity x x e

Page 24: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.1 A Dummy Variable Model

This specification is sometimes called the least squares dummy

variable model, or the fixed effects model.

Slide 15-24Principles of Econometrics, 3rd Edition

(15.12)

1 2 3

1 1 1 2 1 3, , , etc.

0 otherwise 0 otherwise 0 otherwisei i i

i i iD D D

11 1 12 2 1,10 10 2 2 3 3it i i i it it itINV D D D V K e

Page 25: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.1 A Dummy Variable Model

Slide 15-25Principles of Econometrics, 3rd Edition

Page 26: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.1 A Dummy Variable Model

These N–1= 9 joint null hypotheses are tested using the usual F-test

statistic. In the restricted model all the intercept parameters are equal.

If we call their common value β1, then the restricted model is:

Slide 15-26Principles of Econometrics, 3rd Edition

(15.13)0 11 12 1

1 1

:

: the are not all equal

N

i

H

H

1 2 3it it it itINV V K e

So this is just OLS, the pooled model

Page 27: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.1 A Dummy Variable Model

Slide 15-27Principles of Econometrics, 3rd Edition

reg inv v k

Page 28: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.1 A Dummy Variable Model

We reject the null hypothesis that the intercept parameters for all

firms are equal. We conclude that there are differences in firm

intercepts, and that the data should not be pooled into a single model

with a common intercept parameter.

Slide 15-28Principles of Econometrics, 3rd Edition

1749128 522855 948.99

522855 200 12

R U

U

SSE SSE JF

SSE NT K

Page 29: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.2 The Fixed Effects Estimator

Slide 15-29Principles of Econometrics, 3rd Edition

(15.14)1 2 2 3 3 1, ,it i it it ity x x e t T

(15.15)

1 2 2 3 31

1 T

it i it it itt

y x x eT

1 2 2 3 31 1 1 1

1 2 2 3 3

1 1 1 1T T T T

i it i it it itt t t t

i i i i

y y x x eT T T T

x x e

Page 30: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.2 The Fixed Effects Estimator

Slide 15-30Principles of Econometrics, 3rd Edition

(15.16)

1 2 2 3 3

1 2 2 3 3

2 2 2 3 3 3

( )

( ) ( ) ( )

it i it it it

i i i i i

it i it i it i it i

y x x e

y x x e

y y x x x x e e

(15.17)2 3it it it ity x x e

Page 31: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.2 The Fixed Effects Estimator

Slide 15-31Principles of Econometrics, 3rd Edition

Page 32: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.2 The Fixed Effects Estimator

Slide 15-32Principles of Econometrics, 3rd Edition

(15.18) .1098 .3106

(se*) (.0116) (.0169)

itit itINV V K

2*ˆ 2e SSE NT

2 2 198 188 1.02625NT NT N

Page 33: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.2 The Fixed Effects Estimator

Slide 15-33Principles of Econometrics, 3rd Edition

Page 34: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.2 The Fixed Effects Estimator

Slide 15-34Principles of Econometrics, 3rd Edition

(15.19)

1 2 2 3 3i i i iy b b x b x

1 2 2 3 3 1, ,i i i ib y b x b x i N

Page 35: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.4.2 The Fixed Effects Estimator

Slide 15-35Principles of Econometrics, 3rd Edition

ONE PROBLEM: Even with the trick of using the within estimator, we still implicitly (even if no longer explicitly) include N-1 dummy variables in our model (not N, since we remove the intercept), so we use up N-1 degrees of freedom.

It might not be then the most efficient way to estimate the common slope

ANOTHER ONE. By using deviations from the means, the procedure wipes out all the static variables, whose effects might be of interest

In order to overcome this problem, we can consider the random effects/or error components model

Page 36: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5 The Random Effects Model

Slide 15-36Principles of Econometrics, 3rd Edition

(15.20)

(15.22)

1 1i iu

(15.21) 20, cov , 0, vari i j i uE u u u u

1 2 2 3 3

1 2 2 3 3

it i it it it

i it it it

y x x e

u x x e

Randomness of the intercept

Usual error

Average intercept

Page 37: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5 The Random Effects Model

Because the random effects regression error has two components, one

for the individual and one for the regression, the random effects

model is often called an error components model.

Slide 15-37Principles of Econometrics, 3rd Edition

(15.23)

(15.24)

1 2 2 3 3

1 2 2 3 3

it it it it i

it it it

y x x e u

x x v

it i itv u e

a composite error

Page 38: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.1 Error Term Assumptions

Slide 15-38Principles of Econometrics, 3rd Edition

(15.25)

0 0 0it i it i itE v E u e E u E e

2

2 2

var var

var var 2cov ,

v it i it

i it i it

u e

v u e

u e u e

v has zero mean

v has constant varianceIf there is no correlation betweenthe individual effects and the error term

Page 39: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.1 Error Term Assumptions

Slide 15-39Principles of Econometrics, 3rd Edition

But now there are several correlations that can be considered.

The correlation between two individuals, i and j, at the same

point in time, t. The covariance for this case is given by

cov , ( )

0 0 0 0 0

it jt it jt i it j jt

i j i jt it j it jt

v v E v v E u e u e

E u u E u e E e u E e e

Page 40: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.1 Error Term Assumptions

Slide 15-40Principles of Econometrics, 3rd Edition

The correlation between errors on the same individual (i) at

different points in time, t and s. The covariance for this case is

given by

(15.26)

2

2 2

cov , ( )

0 0 0

it is it is i it i is

i i is it i it is

u u

v v E v v E u e u e

E u E u e E e u E e e

Page 41: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.1 Error Term Assumptions

Slide 15-41Principles of Econometrics, 3rd Edition

The correlation between errors for different individuals in

different time periods. The covariance for this case is

cov , ( )

0 0 0 0 0

it js it js i it j js

i j i js it j it js

v v E v v E u e u e

E u u E u e E e u E e e

Page 42: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.1 Error Term Assumptions

Slide 15-42Principles of Econometrics, 3rd Edition

(15.27)

2

2 2

cov( , )corr( , )

var( ) var( )it is u

it isu eit is

v vv v

v v

The errors are correlated over time for a given individual, but are otherwiseuncorrelatedThis correlation does not dampen over time as in the AR1 model

Page 43: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.2 Testing for Random Effects

Slide 15-43Principles of Econometrics, 3rd Edition

(15.28)

1 2 2 3 3it it it ity x x e

1 2 2 3 3it it it ite y b b x b x

2

1 1

2

1 1

ˆ1

2 1 ˆ

N T

iti t

N T

iti t

eNT

LMT e

This is xttest0 in Stata if H0 is not rejected you can use OLS

Page 44: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.3 Estimation of the Random Effects Model

Slide 15-44Principles of Econometrics, 3rd Edition

(15.29)

(15.30)

* * * * *1 1 2 2 3 3it it it it ity x x x v

* * * *1 2 2 2 3 3 3, 1 , ,it it i it it it i it it iy y y x x x x x x x

(15.31)2 21 e

u eT

Is the transformation parameter

Page 45: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.4 An Example Using the NLS Data

Slide 15-45Principles of Econometrics, 3rd Edition

2 2

ˆ .1951ˆ 1 1 .7437

5 .1083 .0381ˆ ˆe

u eT

Is the transformation parameter

Page 46: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Summary for now

Pooled OLS vs different intercepts: test (use a Chow type, after FE or run RE and test if the variance of the intercept component of the error is zero (xttest0))

You cannot pool onto OLS? Then…

FE vs RE: test (Hausman type)

Different slopes too perhaps? => use SURE or RCM and test for equality of slopes across units

Page 47: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Summary for now

Note that there is within variation versus between variation

The OLS is an unweighted average of the between estimator and the within estimator

The RE is a weighted average of the between estimator and the within estimator

The FE is also a weighted average of the between estimator and the within estimator with zero as the weight for the between part

Page 48: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Summary for now

The RE is a weighted average of the between estimator and the within estimator

The FE is also a weighted average of the between estimator and the within estimator with zero as the weight for the between part

So now you see where the extra efficiency of RE comes from!...

Page 49: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Summary for now

The RE uses information from both the cross-sectional variation in the panel and the time series variation, so it mixes LR and SR effects

The FE uses only information from the time series variation, so it estimates SR* effects

Page 50: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Summary for now

With a panel, we can learn about dynamic effects from a short panel, while we need a long time series on a single cross-sectional unit, to learn about dynamics from a time series data set

Page 51: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.5a Endogeneity in the Random Effects Model

If the random error is correlated with any of the right-hand side

explanatory variables in a random effects model then the least squares and

GLS estimators of the parameters are biased and inconsistent.

This bias creeps in through the between variation, of course, so the FE model

will avoid it

Slide 15-51Principles of Econometrics, 3rd Edition

it i itv u e

Page 52: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.5b The Fixed Effects Estimator in a Random Effects Model

Slide 15-52Principles of Econometrics, 3rd Edition

(15.32)

(15.33)1 2 2 3 3

1 1 1 1 1

1 2 2 3 3

1 1 1 1 1T T T T T

i it it it i itt t t t t

i i i i

y y x x u eT T T T T

x x u e

1 2 2 3 3 ( )it it it i ity x x u e

Page 53: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.5b The Fixed Effects Estimator in a Random Effects Model

Slide 15-53Principles of Econometrics, 3rd Edition

(15.34)

1 2 2 3 3

1 2 2 3 3

2 2 2 3 3 3

( )

( ) ( ) ( )

it it it i it

i i i i i

it i it i it i it i

y x x u e

y x x u e

y y x x x x e e

Page 54: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.5c A Hausman Test

We expect to find

because Hausman proved that

Slide 15-54Principles of Econometrics, 3rd Edition

(15.35) , , , ,

1 2 1 22 2

, ,, ,se sevar var

FE k RE k FE k RE k

FE k RE kFE k RE k

b b b bt

b bb b

, ,var var 0.FE k RE kb b

, , , , , ,

, ,

var var var 2cov ,

var var

FE k RE k FE k RE k FE k RE k

FE k RE k

b b b b b b

b b

, , ,cov , var .FE k RE k RE kb b b

Page 55: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.5c A Hausman Test

The test statistic to the coefficient of SOUTH is:

Using the standard 5% large sample critical value of 1.96, we reject the hypothesis that the estimators yield identical results. Our conclusion is that the random effects estimator is inconsistent, and we should use the fixed effects estimator, or we should attempt to improve the model specification.

Slide 15-55Principles of Econometrics, 3rd Edition

, ,

1 2 1 22 2 2 2

, ,

.0163 (.0818) 2.3137

.0361 .0224se se

FE k RE k

FE k RE k

b bt

b b

Page 56: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.5c A Hausman Test

The Hausman test assumes that the RE estimator used in the comparison is fully efficient, which requires that the unobserved effect and the idiosyncratic error are both i.i.d. (Cameron & Trivedi MMA page 719)

often not the case => the hausman command yields incorrect statistic

Example: If the error terms are cluster, (e.g. due to autocorrelation across time for an individual, then the RE estimator is not efficient)

Solutions: do a panel bootstrap of the Hausman test or use the Wooldridge (2002) robust version of Hausman test.

Slide 15-56Principles of Econometrics, 3rd Edition

Page 57: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Test for gamma =0 in:

To run in Stata, generate the RE differences and the mean differences

Principles of Econometrics, 3rd Edition

Page 58: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

To run in Stata, generate the RE differences and the mean differences manually

See an example here: pages 267-268 of Cameron&Trivedi’s MUS book

Principles of Econometrics, 3rd Edition

Page 59: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

15.5.5a Endogeneity in the Random Effects Model

If the random error is correlated with any of the right-

hand side explanatory variables in a random effects model then the

least squares and GLS estimators of the parameters are biased and

inconsistent.

Then we would have to use the FE model

But with FE we lose the static variables?

Solutions? HT, AM, BMS, instrumental variables models could help

Slide 15-59Principles of Econometrics, 3rd Edition

it i itv u e

Page 60: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

We can generalise the random effects idea and allow for different

slopes too: Random Coefficients Model

Again, the now it is the slope parameters that differ, but as in RE

model, they are drawn from a common distribution

The RCM in a way is to the RE model what the SURE model is to the

FE model

Slide 15-60Principles of Econometrics, 3rd Edition

Further issues

Page 61: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Unit root tests and Cointegration in panels

Dynamics in panels

Slide 15-61Principles of Econometrics, 3rd Edition

Further issues

Page 62: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Of course it is not necessary that one of the dimensions of the panel

is time as such Example: i are students and t is for each quiz they take

Of course we could have a one-way effect model on the time

dimension instead

Or a two-way model

Or a three way model! But things get a bit more complicated there…

Slide 15-62Principles of Econometrics, 3rd Edition

Further issues

Page 63: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Another way to have more fun with panel data is to consider

dependent variables that are not continuous

Logit, probit, count data can be considered

STATA has commands for these

Based on maximum likelihood and other estimation techniques we

have not yet considered

Slide 15-63Principles of Econometrics, 3rd Edition

Further issues

Page 64: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Another extension is to consider mixed linear models (Cameron&Trivedi MUS

page 305)

Stata’s xtmixed fits linear mixed models. From Stata;s help:

Mixed models contain both fixed effects and random effects.

The fixed effects are analogous to standard regression coefficients and are

estimated directly

Slide 15-64Principles of Econometrics, 3rd Edition

Further issues

Page 65: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

The random effects are not directly estimated but are summarized according to their estimated variances and covariances

Although random effects are not directly estimated, you can form best linear unbiased predictions (BLUPs) of them (and standard errors) by using predict after xtmixed

Random effects may take the form of either random intercepts or random coefficients, and the grouping structure of the data may consist of multiple levels of nested groups.

Mixed models are also known as multilevel models and hierarchical linear models

Quite rare in the econometric literature

Mixed Linear Models

Undergraduate Econometrics, 3rd Edition

Principles of Econometrics, 3rd Edition

Page 66: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Some particular specifications of the mixed linear models result in more standard models

OLS, RE are special cases of mixed linear models

Another one is known as the Random Coefficients Model

RCM also allows groupwise heteroskedasticity rather than imposing homoskedasticity like its mixed linear model equivalent

Random Coefficients Model

Undergraduate Econometrics, 3rd Edition

Principles of Econometrics, 3rd Edition

Page 67: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Example in Cameron & Trivedi MUS page 310

Random Coefficients Model

Undergraduate Econometrics, 3rd Edition

Principles of Econometrics, 3rd Edition

Page 68: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Data (available through Cameron & Trivedi’s MUS textbook ancillary files) :

mus08psidextract.dta (PSID wage data 1976-82 from Baltagi and

Khanti-Akom (1990))

I cut for you the first 994 observations mus08psidextract994

Random Coefficients Model

Undergraduate Econometrics, 3rd Edition

Principles of Econometrics, 3rd Edition

Page 69: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Random Coefficients Model

Principles of Econometrics, 3rd Edition

Test of parameter constancy: chi2(423) = 68970.73 Prob > chi2 = 0.0000

_cons 4.525957 .2825222 16.02 0.000 3.972224 5.07969

wks .0032176 .0050025 0.64 0.520 -.0065872 .0130223

exp .0973225 .0041396 23.51 0.000 .0892091 .1054359

lwage Coef. Std. Err. z P>|z| [95% Conf. Interval]

Prob > chi2 = 0.0000

Wald chi2(2) = 553.05

max = 7

avg = 7.0

Obs per group: min = 7

Group variable: id Number of groups = 142

Random-coefficients regression Number of obs = 994

. xtrc lwage exp wks, i(id)

DO we have a name for this test? Econometrics, 3rd Edition

Page 70: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

You can understand the use of the FE model as a solution to omitted variable bias

If the unmeasured variables left in the error model are not correlated

with the ones in the model, we would not have a bias in OLS, so we

can safely use RE

If the unmeasured variables left in the error model are correlated with

the ones in the model, we would have a bias in OLS, so we cannot

use RE, we should not leave them out and we should use FE, which

bundles them together in each cross-sectional dummy

Slide 15-70Principles of Econometrics, 3rd Edition

Further issues

Page 71: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Another criterion to choose between FE and RE

If the panel includes all the relevant cross-sectional units, use FE, if only a random sample from a population, RE is more appropriate (as long as it is valid)

Slide 15-71Principles of Econometrics, 3rd Edition

Further issues

Page 72: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Wooldridge’s book on panel data

Baltagi’s book on panel data

Greene’s coverage is also good

Slide 15-72Principles of Econometrics, 3rd Edition

Readings

Page 73: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Keywords

Slide 15-73Principles of Econometrics, 3rd Edition

Balanced panel Breusch-Pagan test Cluster corrected standard errors Contemporaneous correlation Endogeneity Error components model Fixed effects estimator Fixed effects model Hausman test Heterogeneity Least squares dummy variable

model LM test Panel corrected standard errors Pooled panel data regression

Pooled regression Random effects estimator Random effects model Seemingly unrelated regressions Unbalanced panel

Page 74: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Chapter 15 Appendix

Slide 15-74Principles of Econometrics, 3rd Edition

Appendix 15A Estimation of Error Components

Page 75: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Appendix 15A Estimation of Error Components

Principles of Econometrics, 3rd Edition Slide 15-75

(15A.1)

(15A.2)

(15A.3)

1 2 2 3 3 ( )it it it i ity x x u e

2 2 2 3 3 3( ) ( ) ( )it i it i it i it iy y x x x x e e

2ˆ DVe

slopes

SSE

NT N K

Page 76: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Appendix 15A Estimation of Error Components

Principles of Econometrics, 3rd Edition Slide 15-76

(15A.4)

(15A.5)

1 2 2 3 3 1, ,i i i i iy x x u e i N

1

22 2

2 21

22

var var var var var

1var

T

i i i i i itt

Te

u it ut

eu

u e u e u e T

Te

T T

T

Page 77: Panel Data Models ECON 6002 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s notes

Appendix 15A Estimation of Error Components

Principles of Econometrics, 3rd Edition Slide 15-77

(15A.6)

(15A.7)

22 e BEu

BE

SSE

T N K

2 2

2 2 ˆˆ e e BE DV

u uBE slopes

SSE SSE

T T N K T NT N K