7/7 👉Valuation & Conclusion:

🔹Use labor tax shocks to estimate effects on wages

🔹Bringing estimations together: One-standard-deviation increase in worker rights equals ~5.7% wage increase

🔹Conclude: Worker rights have real economic value, not merely symbolic

6/7 👉Empirical Results II: Employment

🔹Use leave-one-out employment rates as instruments for bargaining power shocks in DiD estimation

🔹Higher employment rate leads to more worker rights clauses

🔹Improved employment opportunities strengthen worker bargaining power

5/7 👉Empirical Results I: Labor Tax

🔹Use provincial labor income tax changes as shocks in DiD estimation

🔹Higher provincial labor taxes lead to more worker rights clauses

🔹Demonstrates substitution from taxed income toward untaxed amenities

4/7 👉Validation:

🔹Human and LLM-scaled pairwise clause comparisons confirm our worker rights measure benefits workers

🔹Correlation of rights measure with pro-worker HR practices (World Management Survey by
@I_Am_NickBloom
et al.)

🔹Case study: 2005 Auto Workers Crisis

3/7 👉NLP Pipeline & Data:

🔹30,000 Canadian CBAs (1986-2015)

🔹Dependency parsing identifies linguistic structure of text which we map to legal categories ("employees shall receive...")

🔹LLM embeddings & clustering to classify rights into distinct topics

2/7 👉 Motivation:

🔹CBAs contain extensive worker rights beyond wages

🔹We investigate if these rights have real economic value

🔹Key question: Are these legal rights mere "cheap talk" or valued amenities?

🚨New Working Paper:

The Value of Worker Rights in Collective Bargaining

w/ @elliottash.bsky.social , W. Bentley MacLeod & Suresh Naidu

We propose novel NLP methods applied to 30,000 collective bargaining agreements to quantify the value of worker rights.

👉 nber.org/papers/w33605

🧵1/7