Short-time work (STW) has been widely used, both during the Great Recession and the COVID crisis, to preserve jobs. In most European countries, the implementation of STW schemes is often the result of bargaining between trade unions and employers, yet very little is known about the role of unions. In this paper, we investigate the effects of STW schemes on a number of firms' economic outcomes, considering the role of unions and collective bargaining. We use firm-level panel data, for the metal--engineering industry (from 2009 to 2015), with information on firms characteristics, STW use, industrial relations attributes, merged with accounting data. We estimate the elasticity of employment, working hours, wages and labour productivity to STW hours using an IV-FE estimator. We find that STW is an effective policy to preserve jobs in all firms. The positive effect on employment is supported by quite different mechanisms, which depend on unions presence and power in the firm. In low unionized firms wage cuts are the prevailing adjustment mechanisms, while in highly unionized firms, per-capita wages are insensitive to STW and adjustment mainly occurs through a reduction in working hours. These results are coherent with the use of STW as a work sharing device to protect incumbent workers who are mainly union members.