Digital Transformation of the Italian and US Automotive Supply Chains: Evidence from Survey Data

Keywords:

Abstract:

Purpose – Recent advances in technologies for data collection (e.g. sensors, parts tracking) and analysis (e.g. algorithms and artificial intelligence) applied to automation technologies like robotics and machine vision will likely lead to many innovations, including autonomous vehicles and smart manufacturing (CEA, 2016). In manufacturing, the combination of these technologies might involve using sensors on robotics and other equipment to engage in continuous collection of data in real time, and use of sophisticated software tools to analyze them and predict performance (Agrawal, Gans and Goldfarb, 2018). However, the adoption and use of these new technologies may vary across firms. Following literature summarized in Helper, Martins, Seamans (2019), we distinguish two broad “organizational architectures”. For “Taylorist” firms, workers and technology are substitutes; such firms maintain a strict division of labor both between the “hand work” of transforming materials and the “brain work” of planning and problem-solving. “Pragmatist” firms in contrast value experimentation and believe that front-line workers have expertise that no one else has. Technology can complement these workers’ skills, by improving their access to data and time to engage in problem-solving. These different organizational architectures have significant implications for a variety of aspects of the firm, including operations strategies, technology, worker skills, and compensation (Bresnahan, Brynjolfsson, and Hitt 2002; Brynjolfsson and Milgrom 2013). Moreover, the experience in one country may differ dramatically from that in another country, as a result of different national institutions. In the past, for example, Japanese, German, and American firms have adopted automation in different ways. For instance, in adopting computerized machine tools (CNC) German firms were more likely to combine the functions of programmer and machine operator, while American firms typically separated them (Kelley 1994). Zahra and Covin (1993) find that “technology policy decisions should be evaluated in terms of their collective fit with business strategy rather than as independent decisions”.
In order to investigate these issues, we focus on the automotive industry as it dwarfs other industries in the number of robots shipped annually and the adoption of other automation technologies (CEA, 2016). We analyze plants in the automotive supply chain of the United States and Italy, the second largest robotic market in Europe after Germany and the seventh globally. By 2017, there were 200 industrial robots installed per 10,000 employees in the US, and 190 in Italy (IFR, 2018). We address the following research questions:
1. Under what conditions do organizations adopt and use robots and other digital technologies?
2. How do organizational architectures affect adoption and use of robots and digital technologies?
3. How do these findings differ across two major industrialized nations: US and Italy?

Research Methodology – We conducted a detailed survey of US auto supply firms in 2018-2019, which built on an earlier (2011) survey wave. We provided separate surveys for plant, sales and HR managers. The survey was carried with major industrial automotive associations both in Italy and US. The survey response rates were 1-2% for 2011 survey resample, and 15-30% for the sample of firms that were part of the automakers’ parts suppliers’ associations. A fully comparable survey of Italian firms active in the automotive sector started in October 2018 and will end in April 2020. The survey response rates were 4-5% for the survey resample, and 15-20% for the sample of firms affiliated to suppliers’ associations. As dependent variables, we constructed several measures of robot use: robot reduces labor cost, robot reduces total cost, robot increases quality, robot increases safety. As independent variables, we used responses to the survey to construct a measure of “pragmatism” in the plant’s operations strategy. Our pragmatism measure indexes involvement of production workers in problem-solving and data interpretation. We also constructed variables for data-driven decision making and for use of a system integrator, which literature and theory suggest are important. Our extensive fieldwork (involving dozens of plant visits) further leads us to believe these are important variables.

Findings – Analysis of the collected data showed that robots play a central role in linking process automation technologies to digital technologies. We find evidence that use of the technology does vary by organizational architecture and by country. For example, pragmatist firms are more likely to pursue quality enhancements rather than labor cost reductions in the US, while in Italy such firms are more likely to reduce total costs. In the U.S., production worker compensation rises when pragmatist firms adopt robots, but falls when Taylorist firms adopt robots. We find similar adoption rates for new technologies (families of sensors, machine vision equipment, tools for the simulation of manufacturing processes) across the two countries. We also find that there is heterogeneity in organizational architecture across firms, and this holds across both countries, though the percentage of Pragmatist firms is higher in Italy. This might be explained by their size, as well as the lower number of robots and relatively lower investments in digital technologies. On average, however, firms in each country consider digital technologies to complement worker skills. Preliminary results show interesting differences across the two countries for what concerns the impact of robots on performance, product costs, quality, and safety; we will investigate the sources of these differences, for example whether they are explained by differences in organizational architecture.