1. The canonical model

The canonical model was the standard framework throughout the 1990s and 2000s to explain the rise in returns to skill and wage inequality.

Acemoglu and Autor (2011) - Part 1

This section discusses the first part in Acemoglu and Autor (2011), “Skills, Tasks and Technologies: Implications for Employment and Earnings”, Handbook of Labor Economics, Sections 1-3, p. 1044-1117.

The discussion summarizes the success of the canonical model in explaining the skill premium, but also its challenges that have lead to the emergence of task models.

2. Task models with skilled and unskilled labor

The first task-based models aimed to provide a more nuanced explanation of the relationship between technological progress and the skill premium.

Acemoglu and Autor (2011) - Part 2

This is a discussion of the second part in Acemoglu and Autor (2011), “Skills, Tasks and Technologies: Implications for Employment and Earnings”, Handbook of Labor Economics, Sections 4-6, p. 1117-1166.

In contrast to the canonical model, it shows that a task-based model of the skill premia can be consistent with declining wage levels for the lowest skilled workers and with job polarization.

Autor, Levy and Murnane (2003)

The paper discussed is Autor, Levy and Murnane (2003), “The Skill Content of Recent Technological Change: An Empirical Exploration”, Quarterly Journal of Economics.

The discussion explains how occupations intensive in routine labor tasks have been automated by computers. Embedding their task model into the canonical framework, the authors show that these underlying changes in task demand can explain a substantial part of the increase in the skill premium.

3. Task-based models with capital and labor

Today’s task models focus on the allocative efficiency of labor and capital.

Acemoglu and Restrepo (2019a)

The paper discussed is Acemoglu and Restrepo (2019a), “Automation and New Tasks: How Technology Displaces and Reinstates Labor”, Journal of Economic Perspectives.

The discussion provides an introduction to task modeling.

Acemoglu and Restrepo (2019b)

The paper discussed is Acemoglu and Restrepo (2019b), “Artificial Intelligence, Automation, and Work”, NBER Conference Volume The Economics of Artificial Intelligence, Chapter 8.

The discussion focusses on the importance of labor market imperfections such as skill mismatch or wage rents in the context of automation.

4. The declining labor share

Task models provide an intuitive explanation for the recent decline in the labor share due to automation.

Grossman and Oberfield (2022)

The paper discussed is Grossman and Oberfield (2022), “The Elusive Explanation for the Declining Labor Share”, Annual Review of Economics.

The discussion gives an overview of recent papers with different explanations for the declining labor share.

Autor, Dorn, Katz, Patterson and Van Reenen (2020)

The paper discussed is Autor et al. (2020), “The Fall of the Labor Share and the Rise of Superstar Firms”, Quarterly Journal of Economics.

The paper argues that the labor share has fallen because sales in product markets have shifting towards the most productive firms, which have higher markups and lower labor shares.

5. Empirical evidence

Following the paradigm shift from the canonical to task models, interest in finding empirical evidence supporting the task framework has surged. This literature builds on ongoing developments in causal inference designs such as SSIV, RD, and DiD.

Bessen, Goos, Salomons and van den Berge (2025)

The paper discussed is Bessen, Goos, Salomons and van den Berge (2025), “What Happens to Workers at Firms that Automate?”, Review of Economics and Statistics.

Coding lab DiD with staggered treatment timing

Here is a link to a coding lab about DiD with staggered treatment timing based on a simulated dataset.