Part 1

Question 1

Censored and Truncated log likelyhoods

Answer 1

p1

Question 2

Tobit likelyhood

Answer 2

p1

Question 3

Censored regression assumptions

Answer 3

zero cond, homo, normal. MLE inconsistent without

Question 4

Truncated and Censored mean

Answer 4

p2

Question 5

Partial effect censored regression + mean

Answer 5

p2

Question 6

OLS consistent with sample selection

Answer 6

U V indep or X Z uncorrel

Question 7

Sample selection model 2 equations

Answer 7

p3

Question 8

Conditional density V|U and V|Y

Answer 8

p3

Question 9

rho = 0 sample selection pro

Answer 9

easier maximize separate likelyhood

Question 10

Log likelyhood sample selection

Answer 10

p3

Question 11

Sample selection E(Y|I=1,X,Z)

Answer 11

xb + rho sigma pdf/cdf of z gamma inv mills p3

Question 12

Heckman estimation sample selection

Answer 12

Probit on I get gamma, get mills, OLS on selected. less eff than MLE. formula p3

Question 13

Identification sample selection

Answer 13

exclusion restr

Question 14

Bias IV theory and approx

Answer 14

corr ZU / corr XZ var U/Var X
#instrumrhoUV(1-R)/#obs * R

Question 15

Bias IV to OLS

Answer 15

1/F, #instr/#obs*R

Question 16

Quantile population def 2

Answer 16

p 4.5

Question 17

Estimate pop quantile

Answer 17

qN, round, order stat, or minimize

Question 18

Conditional quantile def 2

Answer 18

p 4.5

Question 19

Equivalence OLS median regression

Answer 19

symm, X indep U

Question 20

Indentifying ass. Median regression

Answer 20

med U|X = 0

Question 21

Quantile regression estimate

Answer 21

p4.5 forula, check not cross

Question 22

Pros quantile reg

Answer 22

p 4.5

Question 23

Cons quantile reg

Answer 23

p 4.5

Question 24

When use quantile reg

Answer 24

picture whole distribution, skewed data like earnings

Question 25

Pro panel data

Answer 25

#obs, dynamic can, omitted var easy deal also attrition

Question 26

Strict exog, what rule out

Answer 26

E U all X, eta = 0 (weak only Xit). No lagged endog & feedback

Question 27

Static FE estimation

Answer 27

p5 (normalization, corrrect SE!)

Question 28

Static FE consistency and identific

Answer 28

need suff variation in Xit over time; T both beta and eta cons, N only beta

Question 29

First diff estimator estimation and assumption

Answer 29

p6

Question 30

Feedback break of strict exog test (T 2 or more)

Answer 30

p6

Question 31

Serial correlation test, solution, why 0.5

Answer 31

p6

Question 32

Cons FE

Answer 32

time invarant, predict out of sample, ignore info of time invariant, if most varition between i and not t can imprecise. Very robust but low efficiency

Question 33

Additional assumptions RE

Answer 33

uncorr eta all Xit, homo, uncorr etas, eta U indep

Question 34

EStimating RE

Answer 34

p7

Question 35

Mudlak test

Answer 35

p7

Question 36

Hausman test

Answer 36

p7

Question 37

T=2 attrition + test

Answer 37

p8

Question 38

Verbeek Nijman test

Answer 38

p8