Contribution Framing (jhr-contribution-framing)

When to trigger

• The paper has credible estimates but the JHR contribution is not obvious
• The policy lesson is buried
• The manuscript does not explain how it reconciles with prior work

Contribution formula

We study [policy-relevant empirical-micro question] using [data/design]. Relative
to [closest prior work], we [new variation/data/population/reconciliation] and
show [result with magnitude]. The findings matter for [policy or human-resource
outcome in the economics sense] because [mechanism].

Must include

• Field: labor, education, health, development, discrimination, retirement, or adjacent empirical micro.
• Credible design or disciplined descriptive contribution.
• Magnitude and policy relevance.
• Relationship to prior estimates.
• Boundary conditions: population, time, institution, and external validity.

Reconciliation paragraph

JHR framing is strongest when the contribution paragraph does not merely say "we add evidence." Add a reconciliation sentence:

Our estimate differs from [prior estimate] because [sample/design/institution/period], and when we
re-estimate the prior specification on [shared sample or comparable sample], the gap narrows/widens in
the predicted direction.

If the paper cannot explain why its magnitude differs from prior work, the contribution is vulnerable even when the design is credible. Route to jhr-data-analysis for comparative estimation.

JHR claim stress test

Before finalizing the introduction, test the claim against four reviewer questions:

• What is learned about human resources? Name the labor, education, health, family, retirement, inequality, discrimination, or development margin in economics terms.
• Why is the estimate credible? State the design feature in one sentence without jargon.
• Why is the magnitude meaningful? Translate the coefficient into dollars, months, percentile points, employment probability, schooling units, or another natural policy unit.
• Why does it differ from prior work? Identify whether the difference is design, sample, institution, period, outcome construction, or policy context.

If the answer to any question requires an appendix table to be intelligible, the main contribution paragraph is not yet ready for JHR.

Magnitude translation menu

JHR contribution claims land when the coefficient is restated in the natural unit of the field, paired with a benchmark such as the control mean, cost per treated unit, or the closest prior estimate:

• Labor: percentage-point employment changes, log wages converted to annual dollars, weeks of unemployment duration.
• Education: test-score standard deviations, completion or enrollment percentage points, years of attainment.
• Health: per-1,000 birth outcomes, coverage points, utilization counts.
• Retirement: months of claiming delay, replacement-rate points.
• Development: schooling years, height-for-age, household consumption.

A claim like "enrollment rose 4.2 percentage points off a 31 percent base, comparable to halving posted tuition" is doing the venue's work for the reader.

Worked vignette: tuition-waiver framing

Illustrative paper: staggered adoption of community-college tuition waivers across 23 states, 2014-2019, linked administrative enrollment and earnings records, Callaway-Sant'Anna ATT. All numbers are placeholders for the pattern.

• Draft claim: "We study the effect of tuition waivers on enrollment." Too thin for JHR — no magnitude, no policy unit, no reconciliation.
• Reframed claim: "Waivers raise first-time enrollment by 4.2 percentage points (13 percent of base), concentrated in the bottom family-income quartile; prior cross-state OLS estimates near 1 point understate the effect because non-adopting states were already trending upward."
• Boundary sentence: identified for adopting states' public two-year systems; spillovers to four-year colleges are bounded, not estimated.

Desk-screen contribution risks

Output format

[One-sentence contribution] ...
[Policy relevance] ...
[Prior-work reconciliation] ...
[Magnitude] ...
[Boundary] ...
[Next step] jhr-tables-figures