SW 10 Econometrics

Essentials of EconometricsTutorial Exercises SW 10
10.1 This exercise refers to the drunk driving panel data regression summarized in Table 1 (see
Table 1 at the last page).
(a) New Jersey has a population of 8.1 million people. Suppose that New Jersey increased
the tax on a case of beer by $1 (in 1988 dollars). Use the results in column (4) to predict
the number of lives that would be saved over the next year. Construct a 95% confidence
interval for your answer.
(b) The drinking age in New Jersey is 21. Suppose that New Jersey lowered its drinking
age to 18. Use the results in column (4) to predict the change in the number of traffic
fatalities in the next year. Construct a 95% confidence interval for your answer.
(c) Suppose that real income per capita in New Jersey increases by 1% in the next year. Use
the results in column (4) to predict the change in the number of traffic fatalities in the
next year. Construct a 90% confidence interval for your answer.
(d) Should time effects be included in the regression? Why or why not?
(e) A researcher conjectures that the unemployment rate has a different effect on traffic fatalities in the western states than in the other states. How would you test this hypothesis?
(Be specific about the specification of the regression and the statistical test you would
use.)
10.2 Consider the following binary variable version of the fixed effects model:
 =  0 +  1  +  1 1 +  2 2 +  +    + 
where  and  are observed regressors. In particular, let 1 be a binary variable that
equals 1 when  = 1 and equals 0 otherwise, let 2 equal 1 when  = 2 and equal 0 otherwise,
and so on.
(a) Suppose that  = 3. Show that the binary regressors and the “constant” regressor are
perfectly multicollinear; that is, express one of the variables 1 , 2 , 3 , and 0 as
a perfect linear function of others, where 0 = 1 for all , .
(b) Show the result in (a) for general .
(c) What will happen if you try to estimate the coefficients of the regression by OLS?
10.3 Recall the list of five potential threats to the internal validity of a regression study. Apply
this list to the empirical analysis of the Effect of Drunk Driving Laws on Traffic Deaths in
textbook Section 10.6 and thereby draw conclusions about its internal validity.
10.4 Using the regression in the following equation
 =  0 +  1  +  2 2 +  3 3 +  +    + 
where  and  are observed regressors. In particular, let 2 be a binary variable that
equals 1 when  = 2 and equals 0 otherwise, let 3 equal 1 when  = 3 and equal 0 otherwise,
and so on. What is the slope and intercept for
1
(a) Entity 1 in time period 1?
(b) Entity 1 in time period 3?
(c) Entity 3 in time period 1?
(d) Entity 3 in time period 3?
10.5 Consider the model with a single regressor  =  1 1 +  +  +  . This model also can
be written as
 =  0 +  1 1 +  2 2 +  +    +  2 2 +  +    +  
where 2 = 1 if  = 2 and 0 otherwise, 2 = 1 if  = 2 and 0 otherwise, and so forth. How
are the coefficients ( 0 ,  2 ,…,  ,  2 ,…,  ) related to the coefficients (1 ,…, , 1 ,…, )?
10.6 Do the fixed effects regression assumptions in Key Concept 10.3 in textbook imply that
(˜  ˜ ) = 0 for  6=  in Equation (10.28) (page 421 in the textbook)? Explain.
10.7 A reseracher believes that traffic fatalities increase when roads are icy and so states with more
snow will have more facilities than other states. Comment on the following methods designed
to estimate the effect of snow on fatalities:
(a) The researcher collects data on the average snowfall for each state and adds this regressor
( ) to the regressions given in Table 10.1.
(b) The researcher collects data on the snowfall in each state for each year in the sample
( ) and adds this regressor to the regressions.
10.8 Consider observations (   ) from the linear panel data model
 =   1 +  +  +  
where  = 1   ;  = 1   ; and  +   is an unobserved individual-specific time trend.
How would you estimate  1 ?
10.10 In a study of the effect on earnings of education using panel data on annual earnings for a large
number of workers, a researcher regresses earnings in a given year on age, education, union
status, and the worker’s earnings in the previous year using fixed effects regression. Will this
regression give reliable estimates of the effects of the regressors (age, education, union status,
and previous year’s earnings) on earnings? Explain. (Hint: Check the fixed effects regression
assumptions in Section 10.5.)
2
Table 1 Regression Analysis of the Effect of Drunk Driving Laws on Traffic Deaths
D ep en d ent va ria b le: tra ffi c fata lity rate (d ea th s p er 1 0,0 0 0 )
R eg resso r
(1 )
036∗∗
B e er ta x
(005)
(2 )
(3 )
−066∗
−064+
(029)
(036)
D rin k in g a g e 1 8
(4 )
−045
(030)
0028
(0070)
−0018
D rin k in g a g e 1 9
(0050)
0032
D rin k in g a g e 2 0
(0051)
(5 )
−069∗
(035)
−0010
(6 )
−046
(031)
(7 )
−093∗∗
(034)
0037
(0083)
(0102)
−0076
−0065
(0056)
−0113
(0068)
−0100+
D rin k in g a g e
(0099)
(0125)
−0002
(0021)
0038
0085
A vera g e veh ic le m ile s p er d river
0008
0017
U n e m p loy m ent ra te
−0063∗∗
M a n d a to ry ja il o r co m m u n ity serv ic e?
(0103)
(0007)
(0112)
(0011)
(0013)
0089
0009
0124
−0063∗∗
−0091∗∗
(0103)
(0007)
(0013)
182∗∗
R ea l in co m e p e r ca p ita (lo g a rith m )
0039
179∗∗
(0164)
(0049)
(0021)
100
(064)
(068)
Ye ars
1 98 2 -8 8
1 9 8 2 -8 8
1 9 8 2 -8 8
1 9 8 2 -8 8
1 9 8 2 -8 8
1 9 8 2 -8 8
S ta te eff e cts?
T im e eff ects?
C lu ste red sta n d a rd erro rs?
no
no
no
ye s
no
ye s
yes
yes
yes
ye s
ye s
ye s
ye s
ye s
ye s
yes
yes
yes
1982
& 8 8 o n ly
yes
yes
yes
(064)
−
−
S ta tistic a n d
Va lu e s
Testin g E x clu sio n o f G ro u p s o f Va ria b le s
422
T im e eff ects= 0
(0002)
1012
(0001)
035
D rin k in g a g e co e ffi cie nts= 0
(0786)
348
(0006)
1028
(0001)
141
3749
(0001)
042
(0253)
(0738)
U n e m p loy m ent ra te,
2962
in co m e p e r c ap ita = 0
3196
(0001)
̄2
0 .09 1
0 .8 8 9
0 .8 9 1
0 .9 2 6
0 .8 9 3
2520
(0001)
(0001)
0 .9 2 6
0 .8 9 9
T h e se re g ressio n s w ere e stim a ted u sin g p a n e l d a ta fo r 4 8 U .S . sta tes. R e g ressio n s (1) th ro u g h (6 )u se d a ta fo r a ll ye ars 1 9 8 2 to 1 98 8 , a n d reg ressio n (7 ) u ses
d a ta fro m 1 9 82 a n d 1 9 8 8 o n ly. T h e d a ta set is d e sc rib e d in A p p en d ix 1 0 .1 . S ta n d a rd erro rs a re g iven in p a renth esis u n d er th e co e ffi cie nts, a n d  − 
a re g ive n in p a renth e se s u n d er th e  −sta tistic s. T h e in d iv id u a l co e ffi cie nt is sta tistic a lly sig n ifi c a nt a t th e + 10%, ∗ 5%, o r ∗ 1% sig n ifi c a n ce le vel.
3