Occupations under threat: Estimating the impact of automation on the Swedish labour market

Abstract: Building upon research which estimates the risk of automation for different occupational categories labour market outcomes of workers switching occupation due to automation is estimated using a data driven occupational mobility model. Models of this kind are constructed using network theory, agent based modelling and, a large data set, and have proven to be useful in analysing flows in the labour market to understand aggregate phenomena such as the Beveridge curve. In order to determine worker behaviour in the model, an occupational mobility network is constructed -which contains the probabilities of transition between ordered pairs of occupations -using all transitions that occurred in Sweden between 2016 and 2017. In addition, the model is calibrated using the Swedish Beveridge curve. A shock, based on estimates of occupation specific automation probability is implemented. In this way, the findings of del Rio-Chanona et al. [2019] are replicated using Swedish data instead of US data.Which means that it is shown that the structure of the occupational mobility network may affect unemployment levels in certain occupations more than the shock itself.Meaning that some occupations with a high probability of automation are unaffected, while others, with low probability of automation are adversely affected.