Skip to content

Add conference poster for the VAT microsimulation paper#14

Merged
vahid-ahmadi merged 12 commits into
mainfrom
poster
Jun 29, 2026
Merged

Add conference poster for the VAT microsimulation paper#14
vahid-ahmadi merged 12 commits into
mainfrom
poster

Conversation

@vahid-ahmadi

Copy link
Copy Markdown
Contributor

Summary

Adds an A0 portrait conference poster for A Firm-Level Microsimulation for VAT Policy Analysis, under paper/poster/.

Built on the Gemini beamerposter theme (the same template family as the author's other posters) with a custom PolicyEngine colour theme (brand blue + teal).

Layout (3 columns, top-to-bottom narrative)

  • Col 1 — the VAT notch, the contribution (open microsim + synthetic-data generator), the method, and validation against HMRC (£85k→£90k costed to within £10m).
  • Col 2 — the notch and its £21,250 dominated region; the mechanical-bunching placebo (8,712 → ~99 firms); the reform menu (level / shape / rate on a common static base).
  • Col 3 — the headline result (only the graduated taper removes the distortion, and for less), behavioural sensitivity over elasticity, policy conclusions, scope/next steps, and contact + QR.

Five figures from paper/figures/ are reused; headline numbers follow the committed paper text (common static base £183.6bn).

Self-contained build

  • xelatex poster.tex (run twice). Requires XeLaTeX for fontspec.
  • Raleway/Lato faces are vendored under fonts/ so no system font install is needed.
  • Compiled poster.pdf is included for convenience.

Note on numbers

The fix-vat-liability-scaling branch regenerates results with a different base (~£184.65bn) and larger reform costs. The poster uses the published paper numbers; if the regenerated figures become final, the reform-menu table and behavioural figures here should be updated to match (flagged in paper/poster/README.md).

🤖 Generated with Claude Code

vahid-ahmadi and others added 4 commits June 28, 2026 18:28
A0 portrait poster built on the Gemini beamerposter theme with a custom
PolicyEngine colour theme. Three-column narrative: the VAT notch and
dominated region, validation against HMRC, the reform menu (level/shape/
rate on a common static base), the mechanical-bunching placebo, and the
behavioural sensitivity range.

Self-contained build: Raleway/Lato fonts vendored under fonts/, figures
copied from paper/figures/. Compiles with `xelatex poster.tex`.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…al logo

- Landscape A0, 4 balanced columns covering every paper section
  (notch/motivation, contribution, synthetic data, static costing + HMRC
  validation, threshold sweep, dominated region + secondary-notch, value-added
  robustness, mechanical bunching + placebo + recovery, reform-menu table,
  conditional behavioural layer, fiscal drag, conclusions, future work).
- Repalette blue -> official PolicyEngine teal (#2C7A7B); teal wordmark + square
  mark in header (from policyengine-app-v2 brand assets); QR to repo.
- Pull main into the branch so figures/numbers match the referee-revised paper.
- 9 figures; builds cleanly with xelatex (single A0 landscape page).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Remove the three blocks sourced from the unmerged PR #16 (not in the published
paper): value-added robustness, the bunching recovery test, and the fiscal-drag
projection. The bunching block now ends with the paper's own framing (bunching
non-identified; Liu et al. 2021 remains the behavioural fact). Drop the
taper_reform_distribution figure (not used in main.pdf); use firms_impact_2025_26
in the reform-menu block and enlarge column-4 figures to rebalance.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@vahid-ahmadi vahid-ahmadi self-assigned this Jun 28, 2026
@vahid-ahmadi vahid-ahmadi requested a review from MaxGhenis June 28, 2026 20:14
vahid-ahmadi and others added 8 commits June 28, 2026 21:23
The teal-tinted alert/example block bodies produced faint horizontal lines
through the text when the A0 page is rasterised at low zoom (dark text
antialiased against the tint). Switch all alert/example block bodies to white
(keeping teal title bars for hierarchy) so previews are clean at any zoom; the
underlying PDF was always crisp at full resolution.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The alert/example block titles ('Contribution', 'The problem', 'Conclusions',
'Static costing', 'Method...') used white text on a filled teal bar, which
showed the same low-zoom rasterisation lines through the title letters. Make
those titles teal/dark-teal bold text on white with a teal underline rule
(matching the standard-block titles) so every section title is artifact-free
and the poster reads uniformly.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Header: two-line subtitle — line 1 name + envelope icon + email; line 2
  PolicyEngine + globe icon + www.policyengine.org + 'free, open-source policy
  analysis' tagline (fontawesome5 icons).
- Smaller PolicyEngine wordmark (4.2cm -> 2.9cm); remove the redundant square PE
  mark from the top-right.
- Footer: drop 'branch poster'; use www.policyengine.org.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make the four columns read as one Problem -> Method -> Distortion -> Reforms
arc: add short italic teal lead-in cues and column-start arrows linking each
block to the next; move Contribution up under the problem with a one-line
Headline ('only the graduated taper removes the distortion; the level move is
the dearest and least efficient'); reorder within columns for logic. No content
or numbers changed -- still exactly the main.pdf results; titles stay
teal-on-white, header/footer/logo unchanged. Figure widths trimmed slightly to
keep columns balanced.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Several poster figures (notably dynamic_notch_fit_e017.png, which showed a
step-UP density that is not the published version) were stale/regenerated copies
that did not match main.pdf. Wipe paper/poster/figures and re-copy every figure
verbatim from origin/main:paper/figures so the poster uses ONLY the figures that
appear in the published paper. Also drop the two unused extras
(obr_vat_bunching, taper_reform_distribution). All 9 figures now md5-match main;
all numeric claims already verified present in main's text.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Read column-by-column the blocks now follow main's section order: notch/intro ->
reform families -> contribution -> method -> synthetic data -> static costing
(+HMRC validation) -> threshold sweep -> reform menu -> bunching (placebo) ->
dominated region -> how reforms change it -> behavioural -> conclusions ->
future work. Key fix: bunching (sec 5) now precedes the dominated region (sec 6),
and the reform menu (sec 4) precedes bunching. Rebalanced columns and resized
figures so every column clears the footer (no clipping). No content, numbers, or
figures changed; figures still match main exactly; teal-on-white titles and
header/footer/logo unchanged.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…ayout polish

- Fix currency: every monetary value now renders with the pound sign; \pounds
  was inside math mode (rendered as $) -> moved to text mode / \text{\pounds}.
- Header: uniform-size two-line subtitle; 'PolicyEngine | (globe) www.policyengine.org
  | Free, open-source tax and benefit analysis'; big QR in the top-right corner
  encoding https://firm-microsim-paper.vercel.app/.
- Footer: 'Beamm.conf26 | Conference on Public Finance & Public Policy Evaluation'.
- Show one figure in the intro (firm-size distribution) and one in the threshold
  sweep; reform menu is the table only; remove the 'Future work & contact' block.
- Enlarge key figures and distribute blocks so all four columns fill the page with
  no cut-off into the footer. Content unchanged and still matches main.pdf.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
… titles

- All block titles render in peteal (example blocks no longer darker)
- LARGE block titles + scale 1.05 to use whitespace
- Trim behavioural-layer text by ~3 lines (keep all numbers + caveat)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@vahid-ahmadi vahid-ahmadi merged commit 2626f46 into main Jun 29, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant