• 2018-07
  • 2018-10
  • 2018-11
  • 2019-04
  • 2019-05
  • The article proceeds as follows


    The article proceeds as follows. Section 2 presents some state of the art on the relation between R&D and employment. Section 3 discusses methodology and data. Section 4 the results. Section 5 concludes. Technical parts are included in Appendix A.
    State of the art First of all, the controversy is also theoretical: small differences in the framework of analysis give totally different predictions. If we rely on general equilibrium framework, the issue is almost a non-sense, because equilibrium implies full employment, by definition. This does not apply for classical economists, where the reference model assumes a given wage and perfectly elastic supply of labour (through the Malthusian rad51 inhibitor mechanism). As a result “classical” equilibrium does not necessarily imply absence of unemployment. Classical economists elaborate the theory of compensation mechanisms, putting forth that initial labour saving effects will be recovered by adjustment in demand (coming through prices and/or income effects, as explained in Vivarelli, 1995). However, even in neoclassical framework, adding small frictions in the wage setting mechanism will generate unemployment, thus technical change can generate alternative long run scenarios depending on the direction of change: e.g. in an efficiency wage case with monitoring, whether the technical change improves monitoring capacity or raise “potential” productivity generates different consequences. If we depart from equilibrium framework towards evolutionary (out of equilibrium) dynamics, then adjustment lags and continuous processes of variety and selection imply a number of controversial trade-offs. In this situation the time necessary to re-establish full employment can be considerable (a discussion is in Van Reenen, 1997). Empirically, the issue is even more controversial: the level of analysis is determinant in the sense that at micro level we should take into account the possibility that the positive employment effect of innovation is simply driven by business stealing; at industry level we can miss information depending on the possible bias towards services or manufacturing; finally at macro level there exist huge measurement problems due to aggregation, besides the obvious impossibility to comprehend the overall dynamics behind (Bogliacino and Vivarelli, 2012). If we focus on the micro level, the consensus can be summed up as follows: at firm level technological change creates employment; at industry level the direct employment effect is positive in the case of product innovation (and thus R&D), but can be negative for process innovation. If we consider also the indirect effect, i.e. compensation mechanisms, the full and instantaneous compensation cannot be assumed ex ante (Chennels and Van Reenen, 2002; Pianta, 2005; Vivarelli, 2007, for some recent contributions, Piva and Vivarelli, 2005; Harrison et al., 2008; Hall et al., 2008; Ping et al., 2008; Bogliacino and Pianta, 2010). Finally, since we mention the EU competitiveness policies (Europe 2020), genera exercise may contribute to the assessment exercise of the Agenda. There has been a very large interest on the productivity consequences of increasing R&D, but less focus has been put on the expected change in employment. As the former is concerned, there is now a large consensus that research driven innovation is a major force shaping growth (Ortega-Argilés et al., 2010). As the latter is concerned, there have been some articles trying to quantify the impact of reaching the targets both at industry level (Bogliacino and Vivarelli, 2012) and through general equilibrium computations (Chevallier et al., 2006; Gelauff and Lejour, 2006; Gardiner and Bayar, 2010). Although very sensible to the assumptions, all studies agree that the impact would be positive.
    Methodology and data In order to formulate our reduced form labour demand to estimate, we start from one of the workhorse of the literature on R&D and innovation, i.e. the “patent race” (Dasgupta and Stiglitz, 1980): firms compete in R&D to gain some market power from an innovation. This market power depends on the appropriability conditions, i.e. the extent of intellectual property rights protection or learning lags, the features of the innovation itself (basically if drastic or not) and the market structure, i.e. barriers to entry in research, competitive pressures from substitutable products and from cumulativeness.