E4 · cycle-002.e4-mys

Malaysia, scripted, 30 years from 1996

30 simulations of 30 rounds from the 1996 position. Malaysia decides by the scripted method; scripted events: Global funding seizure in year 1, Global funding seizure in year 12, Trade bloc rupture in year 22, Legitimacy rupture in year 24, Severe pandemic in year 24.

Model 0.1.0 · graph 0.2.0 · parameters 0.1.0 · seeds 1–30 · elasticity sampled within each bracket

The same configuration in other cycles: cycle 3.

Every simulation · Compare with another experiment · More experiments like this

Not a forecast

Simulations of a structural model with seed parameter values, to compare methods and test the model. Other nations stay at their 2026 ratings throughout.

Outcome across simulations

The median at 2026, with the 10th to the 90th percentile of the 30 simulations under it.

Capability index: 11.88 (−0.34); From 12.22; 11.88 to 11.89
Rank by index: 30; From 27; 30 to 30
Domains at Tier 1: 0; From 0; 0 to 0
Tier rank: 22; From 22; 21 to 22

Capability index by round

Rank by index, by round

Every objective at the position and at the horizon
Objective	1996	10th percentile	Median	90th percentile	Better
Capability indexprimary	12.22	11.88	11.88	11.89	higher
Rank by index	27	30	30	30	lower
Tier rank	22	21	22	22	lower
Domains at Tier 1	0	0	0	0	higher
Score sum	872.7	849.7	850.1	850.5	higher
Goal-weighted score	872.7	849.7	850.1	850.5	higher
Resilient score	872.7	850.1 (mean score sum over the simulations)			higher

The capability index is the mean of the three dimensions, each the mean of its domains, each the mean of its groups. Tiers are Pareto layers within each domain; the tier rank orders nations by domains held at Tier 1, then Tier 2, and so on. The goal-weighted score is the score sum unless the experiment sets a goal. The primary objective is the one the experiment is judged on; every other objective is reported beside it.

Credits

What each round’s credits went on in the median simulation. Across all simulations, build took 87 per cent, buy took 0 per cent, partner took 13 per cent of the 1,620 credits a simulation spent on actions.

Against history

Started from a reconstructed 1996 position and played along the pathway and shocks Malaysia actually had, then scored against the 2026 ratings. The position is an estimate, so part of any error is the reconstruction's. The calibration summary across all fifty nations is on the cycle's page under Loop.

Index 1996: 12.22; reconstructed
Observed 2026: 11.48; the Index's rating
Simulated 2026: 11.88; 11.85 to 11.91
Skill against persistence: −0.37; group error 1.35 against 0.98 if nothing changed

Domain	1996	Observed	Simulated	Bias	Direction
Critical Technology	7.34	7.51	6.23 6.2–6.3	−1.27	wrong
Strategic Infrastructure	14.06	12.90	14.48 14.4–14.5	+1.58	wrong
National Security	9.57	8.43	8.84 8.8–8.9	+0.41	right
Human Capital	12.35	12.14	12.99 13.0–13.0	+0.85	wrong
Information & Influence	12.42	10.89	10.88 10.8–10.9	0.00	right
Governance & Integrity	11.88	11.21	12.28 12.3–12.3	+1.07	wrong
Financial Strength	14.45	13.82	14.40 14.4–14.4	+0.58	wrong
Production & Innovation	13.18	12.49	13.05 13.0–13.1	+0.56	wrong
Trade & Investment	14.78	13.96	13.77 13.7–13.8	−0.19	right

The history it was played under: shocks and events

1997 · S02 × 0.9 The Asian financial crisis triggered capital flight, sharp ringgit depreciation and banking distress, forcing Malaysia's capital controls and bank restructuring.
2008 · S02 × 0.7 The global financial crisis hit Malaysia through collapsing exports and capital outflows, prompting fiscal stimulus and monetary easing.
2018 · S01 × 0.5 US-China trade tensions disrupted supply chains and created uncertainty, though Malaysia also attracted diverted electronics investment.
2020 · S05 × 0.6 The Sheraton Move collapsed the elected Pakatan Harapan government and produced a disputed change of government, but institutions continued to function.
2020 · S10 × 0.8 COVID-19 forced prolonged movement control orders, border closures and a sharp recession, straining public health and fiscal capacity.

The pathway: what it prioritised, period by period

1997–2001 · build · 5% each Malaysia imposed capital controls, fixed the ringgit and consolidated banks through Danaharta and Danamodal to stabilise the economy after the Asian financial crisis. (Financial Stability & Crisis Management, Monetary & Currency Capability, Fiscal Capacity & Public Finance, Regulatory Governance & Strategic Controls, Financial Infrastructure)
2002–2006 · partner · 2% each Recovery policy concentrated on electronics manufacturing, port and road expansion, and export promotion to restore growth and jobs. (Industrial & Sectoral Base, Advanced Manufacturing, Technology Adoption & Diffusion, Trade Facilitation & Logistics, Export Capability & Sophistication)
2007–2011 · build · 2% each The Government Transformation Programme and economic corridors improved public delivery, integrity and investment while deepening Islamic finance. (Government Effectiveness & Public Administration, Public Integrity & Anti-Corruption, Investment Attraction & Projection, Transport Infrastructure, Financial System Depth & Access)
2012–2016 · build · 2% each The Economic Transformation Programme, National Education Blueprint and MRT construction aimed to lift productivity, entrepreneurship and human capital. (Economic Output & Productivity, Business Dynamism & Entrepreneurship, Innovation & Commercialisation, Advanced Education, Transport Infrastructure)
2017–2021 · build · 2% each Industry4WRD, 5G and digital economy initiatives were pursued alongside COVID-19 health and economic responses. (Digital & AI Competence, Communications & Digital Infrastructure, Semiconductors & Advanced Computing, Health System Capability, National Resilience & Crisis Management)
2022–2026 · build · 2% each Malaysia Madani and the National Energy Transition Roadmap prioritised clean energy, institutional integrity and post-pandemic resilience. (Energy Technologies, Energy Infrastructure, Public Integrity & Anti-Corruption, Population Health, Digital & AI Competence)

Reconstruction notes (deepseek-v4-pro): The 1996 baseline is reconstructed from national accounts, trade and sectoral data, but service-sector and governance scores are more uncertain because of data opacity. Event magnitudes are approximate and mapped to library categories; Malaysia's exposure to shocks varied with global and domestic conditions. The policy periods simplify overlapping national plans and actual spending shares were not consistent across all listed priorities.

Report

Written from the simulation record by the loop's narrator. Every figure names the query it came from, in square brackets.

Batch cycle-002. Model 0.1.0, graph 0.2.0, parameters 0.1.0. 30 simulations, seeds 1 to 30; probes on 10. Every figure is a distribution over simulations of a structural model with seed parameter values; it shows what the model implies under these settings, not what will happen. Other nations are frozen at their 2026 ratings. No reconstructed historical data is used in this experiment.

Headline

Malaysia played 30 rounds from 1996 under the scripted method, with scripted events, 30 times. Its capability index moved from 12.22 to a median of 11.88 (−0.34), with 90 per cent of simulations between 11.87 and 11.89 [experiment_objectives].

Starting position

In 1996 the nation ranks 27 on the index against the frozen field, holds 0 of nine domains at Tier 1 and has a tier rank of 22. Its base income is 429 credits a year, calibrated so that its 2026 position is affordable with headroom for investment.

What was decided

A simulation spent a median of 1,620 credits on moves. The five groups that took most were Regulatory Governance & Strategic Controls (build, 105 credits, 6.5 per cent); Financial Infrastructure (build, 105 credits, 6.5 per cent); Financial Stability & Crisis Management (build, 105 credits, 6.5 per cent); Fiscal Capacity & Public Finance (build, 105 credits, 6.5 per cent); Monetary & Currency Capability (build, 105 credits, 6.5 per cent) [experiment_decisions_by_group]. By mode: build 87 per cent, partner 13 per cent [experiment_decisions_by_group]. Concentration of spending (HHI) had a median of 0.044, and the most repeated action appeared in 33 per cent of rounds.

The leading domain by period: rounds 1 to 5, Financial Strength; rounds 6 to 10, Trade & Investment; rounds 11 to 15, Governance & Integrity; rounds 16 to 20, Production & Innovation; rounds 21 to 25, Human Capital; rounds 26 to 30, Human Capital [experiment_spend_by_period].

What the world did

A median simulation saw 5 shocks and 0 upside events. The most damaging per firing were S10 (30 firings in 30 simulations, −30.4 points each); S01 (30 firings in 30 simulations, −19.3 points each); S02 (60 firings in 30 simulations, −16.7 points each) [experiment_shocks].

The groups that lost most to shocks, per simulation: Fiscal Capacity & Public Finance −2.69; Labour & Employment −2.62; Investment Attraction & Projection −2.55; Monetary & Currency Capability −2.50 [experiment_shock_damage_by_group].

The edges that carried most change through the graph, per simulation: Economic Output & Productivity to Fiscal Capacity & Public Finance (+0.04 points); Fiscal Capacity & Public Finance to Social Infrastructure (+0.08 points); Fiscal Capacity & Public Finance to Force Generation & Mobilisation (+0.06 points) [experiment_propagation_by_edge].

Where it ended

At 2026 the median rank is 30 (5th to 95th percentile 30 to 30), the median count of Tier 1 domains 0 (0 to 0) and the median tier rank 22; the resilient score, the mean score sum over simulations, is 850 [experiment_objectives].

Why

Without shocks the index ends +1.615 higher on average, so the shocks cost about 1.615 index points [experiment_probe_attribution].

Holding instead of following this method changes the index by −1.359: that is the method's worth against doing nothing [experiment_probe_attribution].

Removing any single one of the five largest actions changes the index by between −0.043 and −0.006, for example 1997 build Financial Stability & Crisis Management (21 credits) [experiment_probe_attribution]. The largest actions by credits come late, when income is highest, and are rarely the most valuable.

Removing the largest shock of a simulation raises the index by +0.479 on average [experiment_probe_attribution].

The replay probe reproduces the simulations exactly (index change 0.000000), so the attributions are clean [experiment_probe_attribution].

Surprises

tier_jump (warn): largest gain in one domain within five years: 5 tiers (30 simulations above 2).

retired_words (warn): e4-mys/report.md: 'simulations' 1 time(s), first near line 38; use 'simulations'.

The simulation furthest from the median (seed 10) ends +0.01 from it, with 5 shocks and 0 upside events [experiment_outlier_simulations].

Comparison

This is the baseline (hold) experiment, or there is none for this nation in the batch.

Against history: from a reconstructed index of 12.22 the model ends at 11.88, where the 2026 ratings put it at 11.48. Group scores miss by 1.35 points on average, against 0.98 for assuming nothing changed (skill −0.37); the change in each domain has the right sign in 33 per cent of domains. Furthest below history: critical technology (−1.27); furthest above: strategic infrastructure (+1.58) [experiment_against_history]. The starting position is reconstructed, so part of the error is the reconstruction's.

What each decision and shock was worth

Counterfactual probes on the first 10 seeds: the same simulations with one thing changed, paired seed by seed. A removed decision or shock is taken out and the other decisions replayed unchanged, so the difference is that item's worth.

Probe	Index	Score sum	Tier 1	For example
Hold instead of the method	−1.359	−94.7	0.00
No shocks	+1.615	+111.9	+3.00
Replay of the recorded actions (must change nothing)	0.000	0.0	0.00
Without the 1st largest decision	−0.017	−1.0	0.00	1997 build Financial Stability & Crisis Management (21 credits)
Without the 2nd largest decision	−0.017	−0.9	0.00	1997 build Monetary & Currency Capability (21 credits)
Without the 3rd largest decision	−0.043	−2.9	0.00	1997 build Fiscal Capacity & Public Finance (21 credits)
Without the 4th largest decision	−0.018	−1.3	0.00	1997 build Regulatory Governance & Strategic Controls (21 credits)
Without the 5th largest decision	−0.006	−0.5	0.00	1997 build Financial Infrastructure (21 credits)
Without the 1st largest shock	+0.479	+33.8	+1.00	2020 S10 (-30.4 points)
Without the 2nd largest shock	+0.418	+28.2	+0.10	1997 S02 (-19.8 points)
Without the 3rd largest shock	+0.322	+22.6	+1.00	2018 S01 (-19.3 points)

Simulations

Three simulations told round by round, each on its own page: the one nearest the median index at the horizon, and those nearest the 10th and the 90th percentile. Every one of the 30 is listed with its outcome on the simulations page.

All 30 simulations

Events

How often each shock and upside event began across the simulations. Shocks come from the ten downside scenarios and upside events from the ten upside ones, at their yearly likelihood.

Event	Kind	Simulations	Mean round	Points
Trade bloc rupture S01	Shock	100 per cent	22.0	−19.3
Global funding seizure S02	Shock	100 per cent	6.5	−16.7
Legitimacy rupture S05	Shock	100 per cent	24.0	−12.4
Severe pandemic S10	Shock	100 per cent	24.0	−30.4

Groups across simulations

Every capability group: its position, and its median and spread at the horizon across the simulations, with what the method spent on it.

Find a group

Domain

73 of 73 groups

Capability group	Domain	1996	2026	Change	Levels invested	Credits
Digital & AI Competence	Human Capital	9.85	16.23 16.19 to 16.26	+6.38	6.94	88
Regulatory Governance & Strategic Controls	Governance & Integrity	12.91	16.63 16.62 to 16.63	+3.72	4.54	105
Financial Infrastructure	Strategic Infrastructure	15.64	19.16 19.09 to 19.21	+3.52	4.76	105
Public Integrity & Anti-Corruption	Governance & Integrity	10.40	12.89 12.87 to 12.92	+2.49	3.42	87
Transport Infrastructure	Strategic Infrastructure	14.72	17.09 17.08 to 17.10	+2.37	3.01	88
Advanced Education	Human Capital	12.90	14.99 14.98 to 15.00	+2.09	2.73	44
Financial Stability & Crisis Management	Financial Strength	14.55	16.31 16.25 to 16.37	+1.76	3.96	105
Communications & Digital Infrastructure	Strategic Infrastructure	15.12	16.61 16.54 to 16.68	+1.49	2.06	44
Monetary & Currency Capability	Financial Strength	14.82	16.30 16.28 to 16.34	+1.48	3.93	105
Technology Adoption & Diffusion	Production & Innovation	13.40	14.76 14.72 to 14.79	+1.36	2.72	44
Business Dynamism & Entrepreneurship	Production & Innovation	12.26	13.45 13.39 to 13.48	+1.19	2.42	44
National Resilience & Crisis Management	Governance & Integrity	13.13	14.14 14.12 to 14.16	+1.01	2.03	44
Trade Facilitation & Logistics	Trade & Investment	15.67	16.42 16.34 to 16.52	+0.75	2.65	44
Export Capability & Sophistication	Trade & Investment	14.04	14.77 14.74 to 14.79	+0.73	2.97	44
Energy Technologies	Critical Technology	7.42	8.13 8.12 to 8.14	+0.71	2.05	44
Health System Capability	Human Capital	13.47	14.14 14.12 to 14.18	+0.67	1.87	44
Fiscal Capacity & Public Finance	Financial Strength	15.09	15.57 15.54 to 15.60	+0.48	3.04	105
Energy Infrastructure	Strategic Infrastructure	13.42	13.88 13.81 to 13.97	+0.46	1.14	44
Advanced Manufacturing	Strategic Infrastructure	15.31	15.77 15.73 to 15.83	+0.46	1.51	44
Government Effectiveness & Public Administration	Governance & Integrity	14.69	15.11 15.08 to 15.14	+0.42	1.68	44
Innovation & Commercialisation	Production & Innovation	11.70	12.08 12.06 to 12.11	+0.38	1.90	44
Population Health	Human Capital	13.03	13.38 13.37 to 13.39	+0.35	1.36	44
Economic Output & Productivity	Production & Innovation	13.15	13.42 13.40 to 13.45	+0.27	2.29	44
Semiconductors & Advanced Computing	Critical Technology	8.42	8.58 8.56 to 8.61	+0.16	1.66	44
Industrial & Sectoral Base	Production & Innovation	15.02	15.07 15.06 to 15.09	+0.05	1.81	44
Nuclear Deterrence	National Security	0.00	0.00 0.00 to 0.00	0.00	0.00	0
Financial System Depth & Access	Financial Strength	15.42	15.07 15.02 to 15.09	−0.35	1.77	44
Investment Attraction & Projection	Trade & Investment	15.56	15.20 15.17 to 15.23	−0.36	2.17	44
Research Talent & Knowledge Creation	Human Capital	10.47	10.02 9.95 to 10.07	−0.45	0.00	0
Joint Command, Projection & Sustainment	National Security	12.00	11.40 11.38 to 11.43	−0.60	0.00	0
Resource Infrastructure	Strategic Infrastructure	15.48	14.80 14.76 to 14.84	−0.68	0.00	0
Water, Sanitation & Waste Infrastructure	Strategic Infrastructure	13.11	12.41 12.40 to 12.43	−0.70	0.00	0
Technical & Vocational Skills	Human Capital	12.44	11.74 11.73 to 11.74	−0.70	0.00	0
Maritime Power	National Security	12.82	12.10 12.08 to 12.11	−0.72	0.00	0
Lifelong Learning & Reskilling	Human Capital	10.77	10.02 10.00 to 10.04	−0.75	0.00	0
Cyber & Electromagnetic Operations	National Security	10.49	9.73 9.68 to 9.78	−0.76	0.00	0
Intelligence & Assessment	National Security	12.08	11.31 11.27 to 11.36	−0.77	0.00	0
Early Childhood Development	Human Capital	14.00	13.21 13.19 to 13.24	−0.79	0.00	0
Force Generation & Mobilisation	National Security	12.46	11.67 11.66 to 11.68	−0.79	0.00	0
Air Power	National Security	9.97	9.17 9.15 to 9.19	−0.80	0.00	0
Land Power	National Security	12.00	11.19 11.15 to 11.22	−0.81	0.00	0
Space, Aerospace & Defence Infrastructure	Strategic Infrastructure	9.86	9.00 8.95 to 9.04	−0.86	0.00	0
Agriculture & Food Infrastructure	Strategic Infrastructure	14.12	13.20 13.18 to 13.24	−0.92	0.00	0
Social Infrastructure	Strategic Infrastructure	13.81	12.88 12.85 to 12.90	−0.93	0.00	0
Strategic Strike & Missile Defence	National Security	7.49	6.53 6.44 to 6.60	−0.96	0.00	0
Foundational Education	Human Capital	14.19	13.19 13.17 to 13.22	−1.00	0.00	0
Space Operations	National Security	6.37	5.30 5.22 to 5.37	−1.07	0.00	0
Justice System & Fundamental Rights	Governance & Integrity	10.41	9.31 9.29 to 9.33	−1.10	0.00	0
Transparency, Oversight & Accountability	Governance & Integrity	10.30	9.20 9.19 to 9.22	−1.10	0.00	0
Internal Security & Order	Governance & Integrity	13.02	11.91 11.87 to 11.94	−1.11	0.00	0
Political Institutions & Civic Participation	Governance & Integrity	10.15	9.04 9.02 to 9.06	−1.11	0.00	0
Culture, Language & Education Diplomacy	Information & Influence	12.94	11.74 11.66 to 11.81	−1.20	0.00	0
Artificial Intelligence	Critical Technology	8.36	7.11 7.09 to 7.12	−1.25	0.00	0
Connectivity, Cyber & Trust Technologies	Critical Technology	8.73	7.47 7.46 to 7.48	−1.26	0.00	0
Diaspora & Global Communities	Information & Influence	13.44	12.15 12.14 to 12.17	−1.29	0.00	0
Development Cooperation & Global Assistance	Information & Influence	10.10	8.75 8.75 to 8.76	−1.35	0.00	0
Quantum Technologies	Critical Technology	5.24	3.86 3.85 to 3.87	−1.38	0.00	0
Robotics & Autonomous Systems	Critical Technology	6.77	5.37 5.35 to 5.41	−1.40	0.00	0
Space Technologies	Critical Technology	3.53	2.12 2.12 to 2.13	−1.41	0.00	0
Structural Transformation & Complexity	Production & Innovation	13.67	12.17 12.15 to 12.18	−1.50	0.00	0
Alliances & Security Cooperation	Information & Influence	12.68	11.17 11.14 to 11.22	−1.51	0.00	0
Sensing, Timing & Navigation	Critical Technology	6.95	5.41 5.33 to 5.47	−1.54	0.00	0
Advanced Materials & Manufacturing	Critical Technology	9.37	7.77 7.76 to 7.78	−1.60	0.00	0
Sovereign Capital & External Financial Power	Financial Strength	12.76	11.10 11.08 to 11.12	−1.66	0.00	0
Diplomatic Statecraft & Global Networks	Information & Influence	12.91	11.23 11.21 to 11.26	−1.68	0.00	0
Digital & Services Trade	Trade & Investment	13.06	11.37 11.32 to 11.40	−1.69	0.00	0
Financial Regulation & Integrity	Financial Strength	14.08	12.07 12.01 to 12.15	−2.01	0.00	0
Global Value Chain Position	Trade & Investment	16.09	13.99 13.97 to 14.04	−2.10	0.00	0
Biotechnology & Health Technologies	Critical Technology	8.60	6.48 6.46 to 6.49	−2.12	0.00	0
Information Environment & Strategic Communication	Information & Influence	12.42	10.25 10.22 to 10.29	−2.17	0.00	0
Trade Policy & Market Access	Trade & Investment	15.09	12.92 12.81 to 12.98	−2.17	0.00	0
Economic Leverage & Dependence	Trade & Investment	13.96	11.73 11.68 to 11.80	−2.23	0.00	0
Labour & Employment	Production & Innovation	13.05	10.42 10.41 to 10.43	−2.63	0.00	0

Checks

The automated checks run on this experiment before anything was written about it. A warning is a candidate finding for the improvement loop.

passcredits_conservedlargest gap between credits in and credits accounted for: 4.5e-13
passscores_in_boundsscores range 0.000 to 20.000
passno_nans0 missing values in the trajectory and objectives
passprobe_replay_reproducesthe recorded moves played back reproduce the simulation to within 0.0e+00 points (10 simulations)
passledger_balancesthe edge ledger sums to the propagated change to within 9.8e-07 points
passmove_without_shocklargest one-year move in a group with no shock landing: 1.00 points; 0% of simulations have one above 3
warntier_jumplargest gain in one domain within five years: 5 tiers (30 simulations above 2)
passall_tier1_from_outside_top10starting rank 27; 0 simulations reach Tier 1 in all nine domains
passsaturation_high0% of simulations end with a group at 20 that did not start there (median 0 groups)
passsaturation_low0% of simulations end with a group at 0 that did not start there (median 0 groups)
passspend_concentrationmedian concentration of spending across groups (HHI) 0.044; most funded group hard.strategic_infrastructure.financial_infrastructure
passsame_actionthe most repeated move (group and mode) appears in a median 33% of rounds

Settings and versions

Experiment cycle-002.e4-mys. Every simulation records the model, graph and parameter versions, the scenario, the method and its seed, so it can be reproduced exactly.

Nation: Malaysia (MYS)
Position: A reconstructed 1996 position, aligned to the Index's scale
Method: Scripted. Actions fixed in advance by the experiment: for a backcast, the nation's historical pathway.
Scenario: Scripted events: Global funding seizure in year 1, Global funding seizure in year 12, Trade bloc rupture in year 22, Legitimacy rupture in year 24, Severe pandemic in year 24
Objective: Capability index, the primary objective the experiment is judged on; every other objective is reported beside it
Horizon: 30 rounds, 1996 to 2026
Simulations: 30 seeds, 1 to 30. Every method in the batch faces the same draws for the same seed.
Graph elasticities: Drawn for each simulation, uniformly within each edge’s bracket
Income: Base credits 429.1 a round, moving with the economic groups
Versions: Model 0.1.0, Capability Dependency Graph 0.2.0, parameters 0.1.0
Batch: Backcast 1996-2026 (cycle-002), cycle 2 of the improvement loop
Runtime: 4.3 seconds in all; 1028 microseconds a round

Not a forecast: Every figure is a distribution over the simulations of a structural model, reported to compare methods and test the model. None is a prediction of what a nation will do or become.
Structural parameters: Costs, decay, vulnerability and credits are structural seed values with low or medium confidence, set by domain with single-group exceptions. Five groups have no defensible value in a single-nation model and are flagged in the parameter tables.
Other nations are frozen: One nation is simulated at a time. Every other nation stays at its 2026 rating, so rank and tiers are measured against a still field; where the simulated nation rises past others their own tiers are not re-sorted.
Starting depth and mode are defaults: The position is the National Capability Index ratings. How institutionalised each capability is, and how much was built, bought or partnered, have no data behind them yet and start from structural defaults.
Credits are calibrated, not measured: Each nation’s income is set so that its position is affordable with headroom that rises with its fiscal capacity. It is not yet fiscal capacity as a share of GDP.