Table 4.

IV estimation with lagged job losses

	Dependent variable: involuntary job separations
	Persol	Line	DN	DN-JONET
	(1)	(2)	(3)	(4)
Mobility${}_{i,j}$	− 6.797${}^{\ast }$ (3.732)	− 6.096${}^{\ast }$ (3.655)	− 9.855${}^{\ast \ast }$ (4.373)	− 7.908 (5.565)
Mobility${}_{i,j}\times$ Tel${}_i$	6.393${}^{\ast }$ (3.512)	5.133 (3.115)	27.696${}^{\ast \ast }$ (11.821)	14.342 (9.635)
Tel${}_i$	−0.173 (0.240)	−0.259 (0.202)	− 0.168 (1.150)	− 0.198 (0.886)
Job Loss${}_{i,j-1}$	0.113 (0.085)	0.103 (0.091)	0.119 (0.079)	0.164${}^{\ast \ast }$ (0.080)
Observations	282	282	282	282
R${}^2$	0.467	0.480	0.502	0.479
F-stat: mobility	31.706	32.656	53.661	115.051
F-stat: mobility $\times$ Tel	189.160	169.398	72.135	314.598

All regressions include prefecture-specific controls and month dummies and are weighted by prefecture population. $\text{Job}{\text{Loss}}_{i,j-1}$ represents the lagged dependent variable. The rows “F-stat: Mobility” and “F-stat: Mobility $\times$ Tel” report the first-stage F-statistics on the relevance of instruments for Mobility${}_{i,j}$ and Mobility${}_{i,j}\times$ Tel${}_i$, respectively. Standard errors clustered at the prefecture level are shown in parentheses

${}^{\ast }p<0.1$; ${}^{\ast \ast }p<0.05$; ${}^{\ast \ast \ast }p<0.01$