IESET.
Policies·eu_ai_act_2024

EU AI Act (Regulation 2024/1689)

DEU, FRA, ITA, ESP, NLD, BEL, POL, SWE, IRL, AUT·2021 – present·enacted 2024-07-12candidate
movessectoral licensing↑product market competition↓trade openness↓

What the policy did

First horizontal statutory framework for AI systems in a major jurisdiction. Regulation (EU) 2024/1689 published in the Official Journal on 12 July 2024, entering force 1 August 2024 with a phased application schedule (prohibited practices at 6 months, GPAI obligations at 12 months, high-risk system obligations at 24-36 months). Core architecture: risk-tier classification (unacceptable / high / limited / minimal), conformity assessment for high-risk systems, general-purpose AI (GPAI) obligations including systemic-risk tier for models trained above a 10^25 FLOP compute threshold, transparency obligations for generative systems, and an AI Office under DG CONNECT coordinating enforcement. Penalties up to 7% of global turnover for prohibited-use violations. Framework codes this as a strong sectoral-licensing movement: the environmental-stringency axis does not apply, but product-market competition and compliance- cost implications for foundation-model training are central.

Policy-content fingerprint — what this policy moved, on which axes

Per invariant 3, reforms are scored by what they did on each channel-separated axis, not by the party that enacted them. This fingerprint is how the policy-match engine finds historical analogues.

intended
↑
sectoral licensing
regulatory.sectoral_licensing
Sector-specific licensing regimes, concentration / quota allocation, state-controlled entry (energy, telecoms, healthcare, banking).
increased · strong
tighter sectoral licensing / more state gating
Conformity assessment, CE-marking, post-market monitoring, and GPAI systemic-risk duties constitute a full sectoral licensing architecture.
↓
product market competition
regulatory.product_market_competition
Product-market regulation, entry barriers, licensing burdens, network-industry regulation, price controls.
decreased · moderate
more restrictive regulation, higher entry barriers
Compliance-cost floor raises entry barriers in foundation-model training; disputed whether GPAI systemic-risk threshold is level-playing-field or incumbent-friendly.
↓
trade openness
regulatory.trade_openness
Trade policy openness — tariffs, non-tariff barriers, FTAs, industrial protection.
decreased · weak
more protectionist
Extraterritorial reach to any provider placing a system on the EU market; adequacy-like frictions foreseeable.

Enacted by

Empirical evidence — linked hypotheses

Explicit links are curated by the author. Inferred links are hypotheses in the library that test the same axes this policy moved — the framework's answer to "what does the data say about a policy like this?".

Estonia adopted among the most radical market-liberalisation packages of any post-Soviet state — flat tax (26% universal rate, 1994), currency board (EEK pegged to DM/EUR, 1992), rapid privatisation, unilateral free trade, and minimal capital controls — and by 2007 had recovered to Soviet-era GDP per capita levels and substantially exceeded them, while Belarusian and Ukrainian peers had not recovered comparably.
estonia_market_reform_post_soviet_growth_1991_2007inferred
viaregulatory.trade_opennessregulatory.product_market_competition
PARTIAL — recovery threshold pass=True (year_recovered=1998, 2007 vs 1991 = 70.53282727739165); Baltic−CIS gap pass=False (gap=5.1509956229348575)
partial
Across a broad panel of economies 1980-2020, market reforms (privatisation, trade liberalisation, and price decontrol) produce durable gains in real GDP per capita growth only when rule-of-law scores exceed a minimum threshold (WGI Rule of Law > -0.5, approximately the 40th percentile of the global distribution).
rule_of_law_market_reform_complementarityinferred
viaregulatory.product_market_competitionregulatory.trade_openness
REFUTED — coef=-0.1483 (sign opposite claim +), p=0.00481
refuted
Among high-income economies 1990-2020, services-sector competition — measured by low barriers to entry, low incumbent-protection scores, and high churn in retail, transport, communications, and professional services — predicts long-run prosperity (real GDP per capita growth and labour-productivity growth) better than manufacturing-specific industrial policy spending.
sectoral_competition_services_productivityinferred
viaregulatory.product_market_competitionregulatory.sectoral_licensingregulatory.trade_openness
PARTIAL — coef=+0.000842, p=0.361 (above α=0.05); direction inconclusive
partial
Canada’s long-run prosperity after the Canada–US Free Trade Agreement (1988) and NAFTA (1994) is more associated with market openness than with national industrial-policy initiatives.
canada_market_liberalisation_vs_state_industry_1988_2024inferred
viaregulatory.trade_opennessregulatory.product_market_competition
INCONCLUSIVE_DATA_PENDING — treatment 'canada_post_1988' has no within-country variation under country fixed effects
run pending
Australia’s long expansion after the Hawke-Keating reforms (1983–1996) — including tariff cuts, financial deregulation, competition-policy introduction, and fiscal consolidation — is better predicted by market liberalisation than by sector-specific state direction.
australia_hawke_keating_reform_long_runinferred
viaregulatory.trade_opennessregulatory.product_market_competition
PARTIAL — coef=-0.03935, p=0.076 (above α=0.05); direction inconclusive
partial
The Soviet central-planning system, having already exhibited TFP stagnation 1970-1989, underwent a canonical institutional and economic collapse 1989-1998 as plan-enforcement was withdrawn without functioning market institutions in place.
soviet_union_central_planning_gdp_collapse_1989_1991inferred
viaregulatory.product_market_competition
INCONCLUSIVE_DATA_PENDING — no outcome variable loaded; missing: ['derived: count of canonical_metrics with threshold met']
run pending
NZ Rogernomics 1984–1993 liberalisation (tariff removal, SOE corporatisation, financial-market liberalisation) produced productivity acceleration and real-income gains over 1990s–2000s relative to pre-reform trend.
nz_rogernomics_productivity_effectinferred
viaregulatory.trade_opennessregulatory.product_market_competition
refuted — NZ synthetic-control log-TFP gap mean over 1995-2005 = -3.80% (<= 0); informative log GDP-pc gap = -22.70%. Productivity acceleration claim not suppor…
refuted
Singapore's long-run prosperity and frontier convergence are better predicted by extreme trade openness, strong rule of law, competitive product and services markets, and high economic freedom than by state ownership or industrial targeting alone.
singapore_state_capacity_market_openness_comboinferred
viaregulatory.trade_opennessregulatory.product_market_competition
PARTIAL — coef=-0.0001143, p=0.713 (above α=0.1); direction inconclusive
partial

Similar historical policies

Ranked by axis-fingerprint overlap with this policy. Direction match bolded — those are the closest historical analogues. Shape of the match is what drives policy-outcome comparison, not the country or party label.

State Foreign Trade Monopoly and Autarkic Controls
MMR·1964
match
2.3
trade opennessproduct market competitionsectoral licensing
France — nationalisations of banks and industry (February 1982)
FRA·1982
match
1.9
product market competitionsectoral licensing
Enterprise Nationalisation Law and State Takeovers
MMR·1963
match
1.9
product market competitionsectoral licensing
US oil + gasoline price controls 1973-1981
USA·1973
match
1.9
sectoral licensingproduct market competition
Guinea local-content law 2022
GIN·2022
match
1.8
sectoral licensingtrade opennessproduct market competition
Equatorial Guinea hydrocarbons local-content regulation 2020
GNQ·2020
match
1.8
sectoral licensingtrade opennessproduct market competition
Italy INA life-insurance monopoly 1912
ITA·1912
match
1.7
sectoral licensingproduct market competition
Italy IRI creation 1933
ITA·1933
match
1.7
sectoral licensingproduct market competition

References

Notes

Migrated from movements/eu_ai_act_2024.yaml (action=CONVERT). This entity is a single policy/legislation, not a coalition era; reclassified to policies/. Original movement file deleted.