# Neoliberal Index methodology

Version: run-seed-2026-06-04
Generated: 2026-06-07T14:43:49.209Z

The index measures whether official New Zealand parliamentary and legislative records move policy toward or away from economic market-liberalisation and market-oriented institutional settings.

It is not a complete ideology survey. It does not score campaign claims, speeches, public comments, social-media posts, coalition positioning, or social and cultural liberalism unless those questions appear as official market-regulation, trade, fiscal, monetary, labour, welfare, or public/private delivery records.

Scores use official House votes, bill records, legislation, and official government documents. News coverage, campaign rhetoric, social posts, speeches, and other public comments are excluded from scoring.

## Score areas

- Tax and spending: Tax, spending, debt, public balance sheet, and fiscal transparency settings.
- Reserve Bank: Central-bank mandate, inflation priority, and monetary-policy institutional design.
- Public/private delivery: Privatisation, state ownership, contracting, and market delivery of public functions.
- Rules and competition: Regulatory barriers, consenting, competition settings, and entry constraints.
- Work and welfare: Labour-market flexibility, collective bargaining, welfare conditions, and employment rules.
- Free trade and investment: Free trade agreements, foreign investment, market access, trade liberalisation, and capital controls.

## Attribution rules

- Party votes are attributed to MPs whose party membership was active on the vote date.
- Personal votes and named split-party votes are attributed individually.
- Minister, sponsor, amendment, and committee roles are displayed separately from vote totals.
- Unclassified and low-confidence items remain visible.
- Abstentions are treated as neutral on classified items; absences are excluded from the score calculation.

## Score calculation

- Classified votes are converted to alignment values: +1 for a vote in the classified market-liberal direction, -1 against it, and 0 for mixed or abstaining records.
- Each item is weighted by issue-area share, classification confidence, importance, and attribution specificity.
- Low-confidence classifications are down-weighted.
- Final scores are evidence-weighted by the current vote sample. They are not forced to give every score area equal weight.

## Confidence

- Confidence blends data coverage, classification confidence, and attribution specificity.
- Personal votes are more specific than party votes.
- Low-confidence classifications are down-weighted and labelled.

## Current sample balance

The table below shows event-level effective weight in the current public sample. It is a transparency diagnostic, not an extra scoring input.

| Score area | Effective share | Evidence weight |
| --- | ---: | ---: |
| Tax and spending | 15.4% | 6.85 |
| Reserve Bank | 4.7% | 2.08 |
| Public/private | 15.2% | 6.76 |
| Rules/competition | 27.8% | 12.38 |
| Work/welfare | 12.9% | 5.74 |
| Free trade/investment | 24.1% | 10.74 |

## Fairness notes

- Party and MP headline scores can be shaped by the current curated corpus, especially when many records are government-versus-opposition votes in one Parliament.
- Dimension scores and evidence tables should be read alongside the headline score. For example, Labour free-trade and investment-supporting votes are credited in the trade/investment area even when other categories lower the headline score.
- Low or neutral current-MP scores should not be read as clearing a party history. Major historical Labour market-liberal policies are outside the current parsed vote sample until matching official records are added.
- Environmental and social-regulation records are scored only through the stated market-regulation lens; ambiguous market-failure cases should remain mixed or low-confidence.

## Public files

- /data/scores.csv
- /data/dimensions.csv
- /data/policy-events.csv
- /data/sources.csv
- /api/summary.json
- /api/index.json