FIML estimation of a bivariate probit selection rule: an application on firm growth and subsidisation

This study applies a full information maximum likelihood (FIML) estimator of the sample selection model with bivariate selection rule for the investigation of the impact of subsidised firm foundation from unemployment on employment growth of the firm. The empirical analysis is based on the ZEW Firm Start-up Panel using a cohort of firms founded in 15 labour market districts during 1993 and 1995. Estimation results show that the use of the FIML estimator is clearly warranted, compared to a two-step estimator. The FIML model yields a significant negative impact of bridging allowance on employment growth, whereas the two-step estimator underestimates the impact.