Questions: Technological Change and Labor Markets

Question 1

The 'task-based' model of technological change differs from the skill-biased technological change (SBTC) hypothesis by...

• A. Arguing that technology has no effect on labor markets
• B. Focusing on which tasks can be automated (routine vs. non-routine) rather than which skill levels are affected (high vs. low)
• C. Claiming that only manufacturing jobs are affected by technology
• D. Assuming that all workers are equally affected by technological change

Answer: (B)

SBTC treats technology as uniformly complementing educated/skilled workers and substituting for uneducated/unskilled workers, predicting a monotonic relationship between skill and demand. The task model is more nuanced: technology automates routine tasks (which are concentrated in middle-skill jobs like bookkeeping, manufacturing assembly, and clerical work) while complementing non-routine cognitive tasks (analysis, management, creativity) and having little effect on non-routine manual tasks (janitor, home health aide). This explains job polarization — a U-shaped pattern of employment growth that SBTC cannot.

Question 2

Job polarization — the simultaneous growth of high-skill and low-skill employment with declining middle-skill employment — contradicts the simple skill-biased technological change hypothesis.

• T. True
• F. False

Answer: True

SBTC predicts that technology uniformly increases demand for high-skilled workers and decreases demand for low-skilled workers — a monotonic pattern. Job polarization shows a non-monotonic pattern: middle-skill jobs decline (routine tasks automated) while both high-skill (non-routine cognitive) and low-skill (non-routine manual) jobs grow. This pattern, documented by Autor and Dorn (2013) in the US and by Goos and Manning (2007) in Europe, is better explained by the task model's focus on routine vs. non-routine tasks than by SBTC's focus on high vs. low skill.

Question 3

What determines whether a particular task is susceptible to automation?

Model answer: A task is susceptible to automation when it can be described by explicit rules and procedures (codified), follows a predictable pattern, and operates in a structured environment. Routine cognitive tasks (data entry, bookkeeping, basic calculation) and routine manual tasks (assembly line work, sorting) meet these criteria. Non-routine tasks resist automation because they require flexibility, judgment, creativity (non-routine cognitive) or physical dexterity in unstructured environments (non-routine manual). However, advances in AI and machine learning are expanding the frontier of automatable tasks into previously non-routine domains.

Autor, Levy, and Murnane's (2003) 'routineness' criterion was the key theoretical contribution: the relevant distinction is not skill level but task structure. A PhD mathematician doing statistical analysis (non-routine cognitive) and a janitor navigating a cluttered room (non-routine manual) both resist automation — though for different reasons. A bookkeeper following accounting rules (routine cognitive) and a factory assembler performing repetitive motions (routine manual) are both automatable — though one is 'white collar' and the other is 'blue collar.' The task framework cuts across the traditional skill hierarchy.