# Panel Data Analysis in EViews

Crime and deterrence, US NC counties 1981-87
Example 13.9 and problem 14.7 (using crime4.wf1)
structure the databes (identify i and t)
(but do not let EViews balance anything unless you explicitly want that)

Pooled OLS, clustered errors
log(crimeit) = β’xit+ uit
Cov(xit, uit) = 0

Dependent Variable: LOG(CRMRTE)
Method: Panel Least Squares
Sample: 1981 1987
Cross-sections included: 90
White cross-section standard errors & covariance (d.f. corrected)
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
LOG(PRBARR)
LOG(PRBCONV)
LOG(PRBPRIS)
LOG(AVGSEN)
LOG(POLPC)

-2.206729
-0.721511
-0.549277
0.237972
-0.065201
0.362523

0.196612
0.023373
0.025941
0.055093
0.024699
0.024715

-11.22375
-30.86915
-21.17440
4.319484
-2.639795
14.66827

0.0000
0.0000
0.0000
0.0000
0.0085
0.0000

County fixed-effects, iid errors
log(crimeit) = β’xit+ ai+ vit
Cov(xit, ai) ≠ 0
Cov(xit, vis) = 0

Dependent Variable: LOG(CRMRTE)
Method: Panel Least Squares
Sample: 1981 1987
Cross-sections included: 90
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
LOG(PRBARR)
LOG(PRBCONV)
LOG(PRBPRIS)
LOG(AVGSEN)
LOG(POLPC)

-1.872858
-0.383537
-0.305976
-0.195451
0.035664
0.413771

0.172931
0.033467
0.021858
0.033364
0.026125
0.027469

-10.83007
-11.46009
-13.99847
-5.858201
1.365157
15.06334

0.0000
0.0000
0.0000
0.0000
0.1728
0.0000

Effects Specification
Cross-section fixed (dummy variables)

County random effects, iid errors
log(crimeit) = β’xit+ ai+ vit
Cov(xit, ai) = 0
Cov(xit, vis) = 0

Dependent Variable: LOG(CRMRTE)
Method: Panel EGLS (Cross-section random effects)
Sample: 1981 1987
Cross-sections included: 90
Swamy and Arora estimator of component variances
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
LOG(PRBARR)
LOG(PRBCONV)
LOG(PRBPRIS)
LOG(AVGSEN)
LOG(POLPC)

-1.929440
-0.448597
-0.346917
-0.187692
0.027629
0.418481

0.167097
0.030760
0.020209
0.032802
0.025909
0.025433

-11.54681
-14.58367
-17.16632
-5.722010
1.066421
16.45441

0.0000
0.0000
0.0000
0.0000
0.2866
0.0000

0.299778
0.146818

0.8065
0.1935

Effects Specification
Cross-section random S.D. / Rho
Idiosyncratic random S.D. / Rho

County & year FE, iid errors
log(crimeit) = β’xit+ ai+ dt + wit
Cov(xit, ai) ≠ 0
Cov(xit, dt) ≠ 0
Cov(xit, wis) = 0
Dependent Variable: LOG(CRMRTE)
Method: Panel Least Squares
Sample: 1981 1987
Cross-sections included: 90
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
LOG(PRBARR)
LOG(PRBCONV)
LOG(PRBPRIS)
LOG(AVGSEN)
LOG(POLPC)

-1.663947
-0.359794
-0.285873
-0.182781
-0.004488
0.424114

0.165974
0.032419
0.021217
0.032461
0.026447
0.026366

-10.02535
-11.09819
-13.47359
-5.630777
-0.169694
16.08559

0.0000
0.0000
0.0000
0.0000
0.8653
0.0000

Effects Specification
Cross-section fixed (dummy variables)
Period fixed (dummy variables)

County & year FE, cluster errors
log(crimeit) = β’xit+ ai+ dt + wit
Cov(xit, ai) ≠ 0
Cov(xit, dt) ≠ 0
Cov(xit, wis) = 0
Dependent Variable: LOG(CRMRTE)
Method: Panel Least Squares
Sample: 1981 1987
Cross-sections included: 90
White cross-section standard errors & covariance (d.f. corrected)
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
LOG(PRBARR)
LOG(PRBCONV)
LOG(PRBPRIS)
LOG(AVGSEN)
LOG(POLPC)

-1.663947
-0.359794
-0.285873
-0.182781
-0.004488
0.424114

0.328918
0.037349
0.042916
0.032330
0.042343
0.065212

-5.058854
-9.633410
-6.661205
-5.653590
-0.105989
6.503644

0.0000
0.0000
0.0000
0.0000
0.9156
0.0000

Effects Specification
Cross-section fixed (dummy variables)
Period fixed (dummy variables)

County FD & year FE, cluster errors
Δlog(crimeit) = β’Δxit+ Δdt + Δwit
Cov(xit, dt) ≠ 0
Cov(xit, wis) = 0

Dependent Variable: D(LOG(CRMRTE))
White cross-section standard errors & covariance (d.f. corrected)
Variable

Coefficient

Std. Error

t-Statistic

Prob.

C
D(LOG(PRBARR))
D(LOG(PRBCONV))
D(LOG(PRBPRIS))
D(LOG(AVGSEN))
D(LOG(POLPC))

-0.006299
-0.327494
-0.238107
-0.165046
-0.021761
0.398426

0.000701
0.030836
0.038689
0.018626
0.023078
0.048313

-8.988490
-10.62060
-6.154312
-8.861234
-0.942913
8.246702

0.0000
0.0000
0.0000
0.0000
0.3462
0.0000

Effects Specification
Period fixed (dummy variables)

Summary of results

log(prbarr)
log(prbconv)
log(prbpris)
log(avgsen)
log(polpc)
Observations

OLS
+ cluster
-0.722

Cty FE

Cty RE

Cty+Y FE
-0.360

Cty+Y FE
+ cluster
-0.360

Cty FD +Y FE
+ cluster
-0.327

-0.384

-0.449

(6.59)**

(11.46)**

(13.74)**

(11.10)**

(6.05)**

(5.80)**

-0.549

-0.306

-0.347

-0.286

-0.286

-0.238

(7.80)**

(14.00)**

(16.18)**

(13.47)**

(5.55)**

(6.02)**

0.238

-0.195

-0.188

-0.183

-0.183

-0.165

(2.23)*

(5.86)**

(5.39)**

(5.63)**

(4.04)**

(3.60)**

-0.065

0.036

0.028

-0.004

-0.004

-0.022

(0.63)

(1.37)

(1.00)

(0.17)

(0.13)

(0.84)

0.363

0.414

0.418

0.424

0.424

0.398

(3.03)**

(15.06)**

(15.51)**

(16.09)**

(5.00)**

(3.87)**

630

630

630

630

630

630

t statistics in parentheses
* significant at 5%; ** significant at 1%