52nd Congress of the European Regional Science Association: "Regions in Motion - Breaking the Path", 21-25 August 2012, Bratislava, Slovakia
Large metropolitan regions with pollution, congestion and high rents remain the supernatural economic places. The mystery of what exactly raises productivity in large cities continues. Going back to the famous pin factory of Adam Smith, the answer might be that city workers are able to specialise more and that this extensive division of labour increases productivity. Duranton and Jayet (2011) indeed find that the division of occupations is limited by the extent of the market. But what happens within occupations? This research documents and interprets the division of job tasks of workers across German cities. We question to what extent city size limits the division of tasks in Germany. Furthermore, we analyse whether the division of labour explains (part) of the urban wage premium. To guide our empirical analyses, we first set up a simple model of the local division of labour. Each occupation consists of a bundle of job tasks. Workers who specialise in a few job tasks are more productive. Specialisation is however bounded by both coordination costs as in Becker and Murphy (1992) and by local demand as in Baumgardner (1988). Density lowers coordination costs. Thus, workers in large cities with low coordination costs and high local demand specialise more than workers in small cities. Second, we document the division of labour across German cities using the BIBB/IAB Surveys of the Working Population of 2006. The database includes information on the performance of 38 job tasks. Workers in large cities tend to achieve less job tasks than workers in small cities. Workers in the central area of the city are more specialised in certain job tasks than workers in the peripheral area of the same city. Controlling for education, occupation and industry does not affect these stylized facts. A high educated worker in the central area of one of the largest German cities (more than half a million inhabitants) performs 3 out of 38 job tasks less than a low educated worker in the same occupation in a rural German town (less than 5000 inhabitants). Third, urban wage premia in Germany are (partly) explained with the level of specialisation of workers. Wage regressions indicate that workers in the largest German cities earn, ceteris paribus, 16% of a standard deviation more wage than workers in German villages. The inclusion of the specialisation level of a worker and an interaction of specialisation and city size lowers the urban wage premium to 12% of a standard deviation.