Essays on innovation, productivity and labor economics

Résumé

In the following thesis I study how innovation affects labor markets: how it originates and what are its consequences for the labor force. In the first two chapters the source of an increase in productivity and innovative activity are co-worker networks. The third chapter focuses on automation, which itself is a product of an increase in innovative activity. In that chapter I study how automation affects employment structure, with emphasis on displaced workers: their occupational choices and human capital. In Chapter 1 of this thesis, To work or to network? - a study of firm hiring decisions, I investigate the effect of social network on firm performance. What renders informal contacts attractive to employers? Does firm's social network simply speed up the hiring process or it additionally facilitates selection of high-skilled individuals? Using matched employer-employee data from Veneto, an industrial region in northern Italy, this chapter studies the role of co-worker links in firms' hiring decisions. Novel empirical findings show that the hires from firm's own co-worker network increase significantly its productivity. I find that 10% surge in connected hires increases productivity by approximately 1%. The event study analysis reveals that the effect lasts up to three years following the hire. The evidence points that the co-worker links increase firm productivity mainly through industry-specific skills, which suggests that employers may use informal contacts to poach high-skilled workers. Hence, social networks might facilitate the transmission of job-specific skills and knowledge diffusion. In Chapter 2, Inventors' Coworker Networks and Innovation (joint with Sabrina Di Addario and Michel Serafinelli), we build on the previous chapter by studying the role of coworker network in plants' innovative activity and knowledge diffusion. This chapter presents direct evidence showing the extent to which plants' innovation is affected by access to knowledgeable labor connected through the co-worker network. We use a unique dataset that matches administrative employer-employee records from north-central Italy, a region with many successful industry clusters, to patent data for the period 1987-2008. Displacements of inventors due to plant closures generate labor supply shocks to plants that employ their previous co-workers. We estimate (a) event-study models where the treatment is the displacement of a connected inventor and (b) IV specifications where we use the displacement of a connected inventor as instrument for the hire of a connected inventor. Estimates indicate that the improved capacity to employ inventors within their employees’ network increases plants’ patenting activity. In Chapter 3, I study the effects of job automation on labor markets and displaced workers. How does job automation affect reallocation decisions of displaced workers? I show that displaced workers at risk of automation have on average 10 percentage points higher probability of changing their broad occupational category. The mobility rates within high exposure occupations are monotone, pointing that low earners switch their occupations more frequently. Furthermore, the direction of mobility is downward: individuals at risk of automation switch into occupations with lower average wages. To evaluate the role of job automation in the evolution of occupational mobility, this chapter proposes a search and matching model with technological acceleration and human capital accumulation. The reallocation decision of unemployed individuals depends on their human capital level and skill transferability between two occupations. The results show that the response of the economy to automation shock follows closely patterns observed in the data between 1996 and 2012. Job automation accounts for 79 percent of the increase in mobility gap. This in turn leads to output losses due to skill transferability mechanism and the fact