The matching method for treatment evaluation does not balance selective unobserved differences between treated and non-treated. We derive a simple correction term if there is an instrument that shifts the treatment probability to zero in specific cases. Policies with eligibility restrictions,...

This paper investigates the demise of unionisation in British private sector workplaces over the last quarter century. We show that dramatic union decline has occurred across all types of workplace. Although the union wage premium persists it is quite small in 2004. Negative union effects on...