Table 4a

Estimates from the McFadden Conditional Logit State Choice Model

Normalized LMC MeasureBartikBartik5yrEd_bartikEd_bartik5
LMC main effect0.91
(−1.52)		0.97
(−1.84)		1.00
(− 0.06)		1.02
(1.32)
DO × LMC0.86**
(−3.75)		0.93**
(−3.52)		1.00
(0.24)		0.99
(− 1.00)
SC × LMC1.14**
(3.4)		1.08**
(8.69)		1.01
(0.36)		1.01
(0.45)
CG × LMC1.26**
(2.94)		1.14**
(6.87)		1.12*
(2.1)		1.08**
(3.93)
Black × LMC0.91
(− 0.63)		0.93
(−1.88)		0.89
(−1.03)		0.86**
(−9.74)
Hispanic × LMC2.07**
(4.74)		1.41**
(9.46)		1.35**
(5.74)		1.18**
(12.76)
Female × LMC0.98
(− 0.99)		0.99
(−1.04)		1.01
(1.14)		1.01
(1.82)
Distance from birth state0.96**
(− 46.98)		0.96**
(−48.89)		0.96**
(− 50.52)		0.96**
(−49.51)
Education-specific income2.97**
(10.11)		2.79**
(9.35)		2.91**
(9.8)		2.78**
(9.07)
In division of birth2.74**
(62.63)		2.73**
(61.84)		2.74**
(63.67)		2.73**
(64.3)
Education-specific flows253.16**
(169.17)		252.52**
(166.76)		253.62**
(168.83)		253.50**
(171.5)
Choice State FEsYesYesYesYes
Choice Region × birth year FEYesYesYesYes

  • ↵Notes: Top number in cell is implied odds ratio. Z-statistics computed using clustered standard errors on labor market entry year are in parentheses. * indicates significance at the 5 percent level, ** at the 1 percent level. Data are taken from IPUMS versions of U.S. Census 5 percent samples for 1980, 1990, and Sample is noninstitutionalized U.S. natives with 5–8 years of potential labor market experience and positive wages in the previous year. Observations used in estimation were a random 50 percent sample of the sample used in Table 3.