Endogenous Task Bundling, Skills and Automation, NBER Working Paper 35211 by Joshua S. Gans, examines how firms recombine activities after automation. The May 2026 paper argues that holding the pre-automation task bundle fixed can obscure which remaining human skills are rewarded. Its model shows that a wage regression based on a fixed bundle can misidentify the sign of an AI-exposure effect.
This is a theoretical result, not a measured forecast of wage changes across employers. The paper says the direction of omitted redesign bias is not unconditional. Its relevance to reward teams is a question about the job being evaluated: has its composition changed, and is that change visible in the analysis?
Record the role after the redesign This editor recommends that organisations document proposed changes to tasks and responsibilities before assessing their reward implications. A familiar job title may remain in place while the work changes. The review should identify which activities disappear, which remain and which responsibilities are added or moved elsewhere. That description gives workforce and reward teams a common basis for discussion.
The organisation can then examine whether its job-evaluation process is using the current role or an outdated description. A decision record should state what has been reviewed and who is responsible for confirming the relevant information. The aim is to understand the work being assessed rather than infer its value solely from a technology exposure score.
Separate an analytical model from a pay decision Research can identify a weakness in an assumption without determining an individual salary. This editor recommends making that distinction explicit when a model enters a workforce proposal. The team should identify which insight the model supplies, which evidence comes from the organisation and which decisions require its established evaluation process.
Employees and managers should also have an appropriate way to raise questions about a redesigned role. Communication can explain the responsibilities under review and the process used to assess them. It should avoid presenting a theoretical result as proof that a particular skill or employee has automatically gained or lost value.
The paper adds a useful issue to AI governance: job redesign is part of the analysis, not simply background context. A documented description of the new work can support a more transparent discussion of skills, training and reward as the organisation changes its operating arrangements.
Sources: Source primaire ou référence du sujet [1]
Share this:
- More [3]
