WEBINAR RECORDING · 2024-03
Staggered Difference-in-Differences in Gravity Settings
In this conversation: Arne Nagengast; discussant Rebecca Freeman; moderators Ben Shepherd and Hanna Norberg.
Arne Nagengast shows that allowing agreement effects to vary across cohorts and time substantially changes gravity estimates. Rebecca Freeman and the Q&A explore slow adjustment, agreement content and causal limits, cautioning against treating weak recent-cohort estimates as proof that new agreements are ineffective.
Session information draws on the recording and its original description. Affiliations and policy context refer to the time of the webinar.
FROM THE CONVERSATION
Three takeaways.
What they mean in practice.
Prepared from the session recording using automatic transcription and editorial review. Follow the timestamp links for the original wording and context.
Heterogeneity changes the estimated trade gain
Watch 17:14 ↗Nagengast reports roughly 46% higher bilateral trade under the extended estimator versus 18% under conventional estimation in the baseline sample. His OLS decomposition attributes much of the difference to excessive weight on small short-run and late-cohort effects; negative weights are not the main driver.
Policy interpretation: Report the estimator, aggregation weights and sample when communicating effects; the headline is not a forecast for every agreement.
Recent agreements need comparable evaluation horizons
Watch 31:52 ↗Freeman highlights effects building for over 15 years. Responding to her concern about weaker recent cohorts, Nagengast explains that these agreements have fewer observed years to mature. Freeman acknowledges this qualification in Q&A and proposes comparing cohorts over comparable horizons.
Policy interpretation: Compare agreements at similar ages before concluding that newer agreements deliver smaller gains.
Agreement depth needs more than a provision count
Watch 37:40 ↗Freeman questions dividing agreements by their number of provisions and proposes examining coverage, enforceability and commitment types. In Q&A she stresses that provisions often start together, making separate effects difficult to identify; some pursue standards rather than trade expansion.
Policy interpretation: Evaluate commitments against their objectives and distinguish hypotheses about provisions from identified effects.
From the discussion
Freeman asks how estimates translate into general-equilibrium analysis (34:11). Nagengast describes a possible extension, not completed welfare results (40:36). She questions uniform anticipation periods (35:45); he acknowledges that country-specific timing needs additional information (41:30). Asked about endogeneity, he says a shock coinciding with an agreement can still confound its effect (55:45).
