Revisiting German labour market reform effects—a panel data analysis for occupational labour markets

IZA Journal of European Labor Studies

Table 3 Fixed effects estimation results based on the data set disaggregated by occupations and NUTS3 regions

	Dependent variable: logM
	FE 1	FE 2	FE 3	FE 4
β _Us	0.580***	0.592***	0.624***	0.626***
	(0.013)	(0.012)	(0.009)	(0.010)
β _Vs	0.065***	0.058***	0.039***	0.044***
	(0.006)	(0.005)	(0.004)	(0.004)
Year dummies, effect coded (reference year: 2000):
d ₂₀₀₁				−0.116***
				(0.031)
d ₂₀₀₂				−0.149***
				(0.026)
d ₂₀₀₃				−0.122***
				(0.017)
d ₂₀₀₄				−0.111***
				(0.019)
d ₂₀₀₅				−0.082***
				(0.022)
d ₂₀₀₆				−0.030**
				(0.014)
d ₂₀₀₇				0.066***
				(0.020)
d ₂₀₀₈				0.143***
				(0.015)
d ₂₀₀₉				0.146***
				(0.017)
d ₂₀₁₀				0.178***
				(0.021)
d ₂₀₁₁				0.152***
				(0.020)
γ		1.681***	1.363***	1.377***
		(0.557)	(0.272)	(0.271)
a	−4.533***	−4.785***	−0.896**	−0.826**
	(0.633)	(0.614)	(0.397)	(0.410)
Control variables	Yes	Yes	Yes	Yes
Monthly time dummies	No	No	Yes	No
Quarter dummies	No	No	No	Yes
Observations	2,393,683	2,393,683	2,393,683	2,393,683
Number of groups	55,316	55,316	55,316	55,316
Within R-squared	0.236	0.240	0.304	0.275

Driscoll-Kraay standard errors in parentheses. Column FE 3 includes monthly time fixed effects with effect coding (reference period is January 2000); compare with Fig. 2, left panel. Control variables are the share of female workers, the shares of workers at different skill levels (‘vocational training’ and ‘academic degree,’ reference group is ‘without vocational training’) in each NUTS3 region and occupation, and the average age by the population by NUTS3 regions. The estimated coefficients are provided on request. The estimated coefficients are provided on request
*** p <0.01, ** p <0.05, * p <0.1