Why are firms willing to locate in dense areas although production costs are much higher in metropolitan than in rural areas? As first hypothesized by Marshall (1890), this may be explained by cost or productivity advantages enjoyed by firms when they locate near other firms. The goal of this project is to provide novel evidence on the existence and sources of such agglomeration effects.In the first part of the project, I use job creations and job destructions at a large scale (like the opening or closing of a large plant) to identify agglomeration effects in the local labour market. Local governments are often willing to substantially subsidise plants that may locate within their jurisdiction or loss-making plants that are at risk of failing. The main rational behind these incentives is that the opening of a new plant may lead to additional job creations, while the closing of an existing plant may trigger additional job destructions in the area. The questions I address in this project include: May a plant closing trigger a region’s decline? Can the opening of a large plant create regional prosperity?In the second part of the project, I focus on knowledge spillovers in the labour market as one source of agglomeration effects. While most of the existing literature investigates knowledge spillovers at the city or state level, I investigate knowledge spillovers at the level of the firm where, through face-to-face interactions, they are likely to originate. The questions I address include: do workers learn from their co-workers? How do workers learn from their peers? Is it the quality of human capital, the similarity of skills or the skill diversity among peers that matter?My data are uniquely suited to address these questions, and cover the universe of all workers and establishments in Germany for more than three decades.