ILPC 2027

Autor, Katz and Kearny (2006) and Acemoglu and Autor (2010) use a task-based model of computerization to explain the empirically observed "polarization" of the US labor market in the 1990s. This polarization was characterized by a rapid growth of high-wage jobs and a moderate growth of low-wage jobs at the expense of middle-wage jobs. By employing this task-based model, these authors "say goodbye" to the so-called "canonical" model of skill-biased technological change. This model only includes two skill groups – high and low skilled workers – which perform two imperfectly substitutable occupations. "Despite the canonical model's virtues and substantial empirical applicability, a satisfactory analysis of modern labor markets and recent empirical trends necessitates a richer framework." (Acemoglu and Autor 2010, pp. 1–2).

The task-based model building on the framework of Autor, Ley and Murnane (2003) differentiates between tasks, i.e. units of work activity that produce output, and skills, i.e. the worker's capabilities for performing various tasks. It distinguishes between three forms of tasks: non-routine cognitive tasks, routine (cognitive and manual) tasks and non-routine manual tasks. Computers complement cognitive tasks of high-wage jobs, and substitute routine tasks found in many traditional middle-wage jobs, whereas they have little direct impact on (non-routine) manual tasks in relatively low-paid jobs. Accordingly, falling computer prices are the driving force behind the development of employment and wages in different occupations.

This new framework has proven to be very successful empirically: First, it can explain why jobs with high skill requirements increased, while jobs with medium skill requirements decreased and jobs with low skill requirements stagnated (or slightly rose or fell). This accounts for the change in the occupational "landscape" in the United States, the European Union and Japan (Michaels, Natraj and Van Reenan 2010; Autor and Dorn, 2009; Goos, Manning and Salomons 2010). Second, the model gives a theoretical explanation for the polarization of wages that was not only observed for the United States (Autor, Katz and Kerarny 2006) but also for Britain (Goos and Manning 2003) and West Germany (Spitz-Oener 2006, Goebel, Gornig and Haeussermann 2010).

However, Acemoglu and Autor (2010) mention another autonomous factor that shapes the distribution of employment and earnings: organizational change. In this paper we argue that computerization has led to an enormous change of work organization, which in turn has led to a "polarization of skill requirements". The use of new information and communication technologies has reduced the costs of specialization, i.e. the costs of coordinating the activities of different employees and of different production sites. Furthermore, globalization and a higher need of flexibility have increased the returns of specialization. Both developments have led to an increase in specialization or "retaylorization" and have caused a polarization of skill requirements: On the one hand many "narrow" jobs with a small number of different and less complex tasks emerged. Hence the skill requirements for certain occupations, or rather, for many jobs in certain occupations, have decreased.[1] Workers do not need medium skills any longer, i.e. a de-qualification or "downgrading" has taken place. On the other hand specialization has led to some jobs (often in the same occupation) that include a multitude of different tasks and combine several complex tasks and thus do require the same or even a higher skill level ("upgrading").

In the paper we will proceed as follows: Section 2 presents some stylized facts on the polarization of employment and wages in the US and other countries, while Section 3 outlines the central ideas of the task-based model in a non-formal way and explains its advantages in comparison to the canonical model. In Section 4 and 5 we will demonstrate why an increasing specialization or "retaylorization" did not only take place within establishments, but also between different locations of the same corporations, and (even) between different countries. This specialization then led to the de-qualification (of many jobs) and thus to a polarization of labor markets. Finally, Section 6 summarizes and draws some conclusions.