Draft Paper

Generated Output

• output/tables/*.tex — Regression tables (modelsummary → LaTeX with tabularray)
• output/figures/ — Figures (maps, event studies, scatter plots)
• book/ — Quarto website with exploratory analysis (.qmd files)

Econometric Approaches

The paper uses multiple identification strategies:

1. Hedonic regressions — Cross-sectional: spill count/hours (continuous + bins) × prior/full period, with LSOA FE
2. Repeat sales — Within-property variation (Palmquist 1982 approach)
3. Long difference — 250m grid-level changes, weighted/unweighted × all/exposed grids
4. News/media DiD — Google Trends and LexisNexis media coverage as treatment
5. Upstream/downstream — Directional spillover via river network topology
6. Dry spills — Spills occurring without rainfall as identification variation

Key Variables and Notation

• Outcome: log(price) for sales, log(rent) for rentals
• Treatment: spill_count, spill_hrs, spill_count_daily_avg, spill_hrs_daily_avg
• Geography: n_spill_sites within radius, min_dist / mean_dist to nearest site
• Radii: 250m, 500m, 1000m, 2000m, 5000m, 10000m
• Fixed effects: LSOA (lsoa), MSOA (msoa), year-quarter
• Standard errors: heteroskedasticity-robust (vcov = "hetero") via fixest::feols()

LaTeX Conventions

• Citations: \textcite{} for textual, \parencite{} for parenthetical (APA biblatex)
• Tables: tabularray format with booktabs, [H] float placement
• Equations: numbered with \label{eq:...}

Workflow

Step 1: Context Gathering

Before drafting:

1. Read the existing section file in docs/overleaf/ (if it exists)
2. Read _main.tex to understand document structure and preamble
3. Scan output/tables/ for available regression output
4. Scan output/figures/ for available figures
5. Check refs.bib for available citations
6. Read relevant .qmd files in book/ for analysis context and results interpretation
7. Read relevant analysis scripts in scripts/R/09_analysis/ for methodology details

Step 2: Section Routing

Based on $ARGUMENTS:

• intro: Draft 01_introduction.tex
• background: Draft 02_background_context.tex
• motivating-evidence: Draft 03_motivating_evidence.tex
• instrument: Draft 04_hydraulics_instrument.tex
• research-question: Draft 05_research_question.tex
• results: Draft results section(s) from regression output
• conclusion: Draft conclusion
• abstract: Draft abstract (requires other sections to exist)
• appendix: Draft or extend appendix sections
• full: Draft all sections in sequence, pausing between for user feedback
• No argument: Ask which section to draft

Step 3: Drafting Standards

Introduction (~1,000-1,500 words)

• Hook with UK sewage crisis context → research question → methodology overview → key findings → contribution → road map
• Contribution paragraph names specific papers being advanced (environmental disamenity capitalisation, hedonic pricing literature)
• Effect sizes with magnitudes and units (percentage impact on house prices per additional spill)

Background & Context (~800-1,200 words)

• UK privatised water industry structure
• EDM monitoring requirements and data availability (2021+)
• Scale of the sewage spill problem
• Policy and regulatory responses

Empirical Strategy (~800-1,200 words)

• Identification assumption stated formally
• Estimating equation displayed and numbered
• Each approach (hedonic, repeat sales, long diff) with clear specification
• Threats to identification addressed (sorting, omitted variables, measurement error)

Results (~800-1,500 words per approach)

• Main specification with economic interpretation
• Effect sizes in meaningful terms (£ impact, percentage change)
• Robustness across radii, time periods, and specifications
• Heterogeneity results

Conclusion (~500-700 words)

• Restate headline effect sizes
• Policy implications for water regulation
• Limitations and future work (rental market, long-run effects)

Step 4: Humanizer Pass

Apply automatically as final pass:

• Strip AI writing patterns (see /humanizer skill for full 24-pattern checklist)
• Preserve formal academic structure where genuinely needed
• Maintain citation density and technical vocabulary

Step 5: Quality Self-Check

Before presenting the draft:

• Every displayed equation is numbered (\label{eq:...})
• All \textcite{} / \parencite{} keys exist in refs.bib
• Effect sizes stated with units and magnitudes
• No banned hedging phrases ("interestingly", "it is worth noting")
• Notation consistent with project conventions (LSOA, MSOA, radius distances)
• All referenced tables/figures actually exist in output/
• LaTeX compiles cleanly with the KOMA-Script setup

Step 6: Present to User

Present each section for feedback. Flag:

• TBD: Where empirical results are needed but not yet available
• VERIFY: Citations that need user confirmation
• PLACEHOLDER: Effect sizes awaiting final estimates

Principles

• This is the authors' paper (Balboni & Dhingra), not Claude's. Match their voice and style from existing sections.
• Never fabricate results. Use TBD placeholders. Cross-check numbers against output/tables/.
• Citations must be verifiable. Only cite papers confirmed in refs.bib or the literature review in docs/lit_review/.
• Humanizer is automatic. Every draft gets de-AI-ified before presentation.
• Match existing conventions. Read existing .tex files first to match style, formatting, and level of formality.