Table 4.

Results of using the Bartik shock defined in Eq. 9 as an instrument for the industry-specific knowledge brought to a pioneer firm by its first hires ($\mathrm{\Phi }$)

	Dependent variable
	Industry knowledge	3-y growth rate
Independent variables	First stage: 1	Reduced form: 2	Instrumental variable: 3	OLS: 4
Industry knowledge			0.${\mathrm{502}}^{**}$	0.${\mathrm{177}}^{***}$
(${\mathrm{\Phi }}^{\left(t\right)}$)			(0.256)	(0.032)
Bartik shock	−6.${\mathrm{899}}^{***}$	−3.${\mathrm{465}}^{**}$
($B_{ri}^{\left(ind\right)}$)	(1.568)	(1.686)
Growth of industry	0.${\mathrm{282}}^{**}$	−0.003	−0.144	0.056
($g_{i;r}$)	(0.134)	(0.144)	(0.162)	(0.161)
Constant	−0.${\mathrm{634}}^{***}$	0.${\mathrm{496}}^{***}$	0.${\mathrm{814}}^{***}$	1.${\mathrm{898}}^{***}$
	(0.075)	(0.081)	(0.243)	(0.549)
Observations	1,380	1,380	1,380	1,380
$R^2$	0.016	0.003		0.234
Adjusted $R^2$	0.015	0.002		0.089
$F$ statistic	11.${\mathrm{236}}^{***}$	2.129		1.${\mathrm{609}}^{***}$
	(df = 2)	(df = 2)		(df = 220)

Our two-stage least-squares estimates confirm the direction of the effect on growth found using OLS. The F test for the strength of the instrument yields a statistic of 18.339***. (28). Industry knowledge is expressed in SD units. *P<0.1; **P<0.05; ***P0.01. df, degrees of freedom.