Estimation of standard errors and treatment effects in empirical economics—methods and applications

Journal for Labour Market Research

Table 11 OLS estimates of Cobb-Douglas functions with artificial dummies (DV.) as regressor; dependent variable: logarithm of sales

	\(\hat{\beta}_{\ln L}\)	Std.err.	\(\hat{\beta}_{\ln K}\)	Std.err.	\(\hat{\beta}_{DV.}\)	Std.err.
\(\overline{DV1} =0.1692\)	0.9464	0.0043	0.2223	0.0032	0.0470	0.0128
\(\overline{DV2}=0.2952\)	0.9453	0.0043	0.2223	0.0032	0.0808	0.0105
\(\overline{DV3}=0.3672\)	0.9446	0.0043	0.2224	0.0032	0.0923	0.0099
\(\overline{DV4}=\boldsymbol{0.5388}\)	0.9434	0.0043	0.2225	0.0032	0.1334	0.0096
\(\overline{DV5}=0.6301\)	0.9432	0.0043	0.2226	0.0032	0.1285	0.0100
\(\overline{DV6}=0.7190\)	0.9438	0.0043	0.2226	0.0032	0.1124	0.0107
\(\overline{DV7}=0.8360\)	0.9449	0.0043	0.2226	0.0032	0.0979	0.0130
\(\overline{DV8}=0.9445\)	0.9448	0.0043	0.2226	0.0032	0.1599	0.0210
\(\overline{DV9}=1.0000\)	0.9472	0.0043	0.2225	0.0032	0.0000	–

Note: IAB Establishment Panel 2006–2010; n=34,308. DV is constructed in the following way: The interaction variable between the wave number (14,…,18) and the identification number of the establishments is split into nine classes and ordered from the smallest to the largest class (C1,…,C9). Then new cumulative variables are determined and transformed into dummy variables: DV1=1 if the establishment j belongs to C1, =0 otherwise; DV2=1 if j belongs to C1 or to C2, =0 otherwise; …; DV8=1 if j belongs to C1 or C2 or … or C8, =0 otherwise