IESET.
Hypotheses·growth·taiwan_itri_frontier_capability_effect

Taiwan's ITRI-led semiconductor strategy (ITRI founding 1973, UMC spinoff 1980, TSMC spinoff 1987, continued state-industry co-investment through the 1990s-2000s) produced a frontier-capability industry that market-led alternatives in comparable-income economies did not generate over the same 40-year window.

Tested against donor pool of Asian and Latin American economies with similar 1970s per-capita GDP and absent coordinated state technology strategy.

PARTIALengine/runs/taiwan_itri_frontier_capability_effect

partial — One primary held but not the other (or magnitudes fell short of the strong-form threshold). TFP gap +87 log-pts (threshold +20, rank 1); GDPpc gap +26 log-pts (threshold +50, rank 4).

confidence cueThe result is useful, but not decisive. Treat it as a clue, not a settled conclusion.

policy briefMixed or noisy

In ordinary language

Over a long period, do more market-oriented institutions translate into higher income or productivity, once the comparison looks beyond a single success story?

plain answer

The evidence is suggestive but not decisive. One primary held but not the other (or magnitudes fell short of the strong-form threshold).

why it matters

Growth claims can look convincing in single success stories. This test asks whether the pattern survives a broader comparison.

how the test works

It compares 9 country or place units from 1970 to 2019, using a synth did design.

what was measured
What changed
  • Taiwan itri treated
What we checked
  • Log income pc cost-of-living adjusted
  • High tech exports pct
  • Economic complexity index
what this does not prove

A single test is not the whole truth. It narrows the claim under a specific sample, time period, and method. Strong policy conclusions need the pattern to survive nearby tests, alternative data, and serious objections.

verification

No evidence packet has been generated yet.

Results

engine/runs/taiwan_itri_frontier_capability_effect
Loading chart…

Who has skin in the game — schools predicting on this

1 school list this hypothesis as a test of their position. The chips below are school-level scoreboard outcomes, not a second hypothesis verdict.

hypothesis verdict vs scoreboard outcome

The banner verdict judges this hypothesis as written. The scoreboard asks whether each school's polarity-corrected prediction was right. Raw status is not a school win: SUPPORTED supports schools that needed SUPPORTED, but refutes schools that needed REFUTED.

Structuralist / Developmentalistwins if SUPPORTEDpartial/no-call

Pre-registration

pre-registered
first-spec commit 4c8ce8e · 2026-07-18T22:11:21Z

Taiwan's ITRI-led semiconductor strategy (ITRI founding 1973, UMC spinoff 1980, TSMC spinoff 1987, continued state-industry co-investment through the 1990s-2000s) produced a frontier-capability industry that market-led alternatives in comparable-income economies did not generate over the same 40-year window. Tested against donor pool of Asian and Latin American economies with similar 1970s per-capita GDP and absent coordinated state technology strategy.

Falsification criterion — what would disprove this

set before the run · honoured after

This hypothesis is considered falsified if:

PRIMARY 1 (capability proxy): SUPPORTED if Taiwan's cumulative log-change in PWT TFP (rtfpna), 1973 -> 2019, exceeds the donor-pool mean by at least +0.20 log-points (~22% extra TFP) AND Taiwan ranks #1 or #2 in the 9-country sample. REFUTED if Taiwan's TFP gain is below the donor-pool mean. PARTIAL otherwise. PRIMARY 2 (output proxy): SUPPORTED if Taiwan's cumulative log-change in real GDP per capita (PWT rgdpe / pop), 1973 -> 2019, exceeds the donor-pool mean by at least +0.50 log-points (~65% extra income) AND Taiwan ranks #1 in the 9-country sample. REFUTED if Taiwan's gain is below the pool mean. PARTIAL otherwise. Joint verdict: SUPPORTED only if BOTH primaries pass. REFUTED only if BOTH primaries refute. PARTIAL otherwise. INFORMATIVE: also report the gap to a 'market-led-only' subpool excluding KOR (which is itself a developmentalist case). METHOD_VALID: PWT 10.x covers TWN and at least 6 of 8 donors over the 1973-2019 window. If donor coverage falls below this gate, emit inconclusive.

formal test & threshold
test:      twn_vs_donor_pool_cumlog_gap_pwt_1973_to_2019
threshold: PRIMARY 1: tfp_gap_vs_pool >= +0.20 log-pts AND tfp_rank_twn <= 2. PRIMARY 2: gdppc_gap_vs_pool >= +0.50 log-pts AND gdppc_rank_twn == 1.

Method

Template
synth_did
Clustering
none
Sample
9 countries · 1970 – 2019
Evidence type
associational

Taiwan treated 1973-2023; donor pool of comparable-1970s-income economies without coordinated state technology strategy (PHL, MYS, BRA, MEX, ARG). Outcomes: global foundry market share, semiconductor patent stock, and process-node-leadership indicators. Tests whether ITRI strategy produced above-counterfactual capability buildup.

Data

VariableSourceTransform
log_gdp_pc_ppp
outcome
world_bank_wdi:NY.GDP.PCAP.PP.KDtier 2
log
high_tech_exports_pct
outcome
world_bank_wdi:TX.VAL.TECH.MF.ZStier 2
level
economic_complexity_index
outcome
owid:economic-complexity-indextier 2
level
patent_applications_resident
outcome
wipo:patent_applications_residenttier 2
log
tfp_index
outcome
pwt:rtfpnatier 3
level
taiwan_itri_treated
treatment
constructed:indicator = 1 for TWN, year >= 1973 (ITRI founded)tier 5
indicator
log_initial_gdp_pc
control
world_bank_wdi:NY.GDP.PCAP.KDtier 2
log
log_population
control
world_bank_wdi:SP.POP.TOTLtier 2
log
gross_capital_formation_pct_gdp
control
world_bank_wdi:NE.GDI.TOTL.ZStier 2
level

● ready  · ● pending  · ● reconstruct-needed

Detailed result card

Taiwan ITRI frontier-capability effect

Verdict: partial — One primary held but not the other (or magnitudes fell short of the strong-form threshold). TFP gap +87 log-pts (threshold +20, rank 1); GDPpc gap +26 log-pts (threshold +50, rank 4).

Summary

  • Treatment year: 1973 (ITRI founded). End of available data: 2019 (PWT 10.x).
  • Donor pool: KOR, PHL, MYS, BRA, MEX, ARG, THA, IDN (n=8); 'market-led-only' subpool excludes KOR.

Primary 1 — TFP cumulative log-change (PWT rtfpna):

  • Taiwan: +90 log-pts.
  • Donor-pool mean: +2 log-pts.
  • Market-led subpool mean (no KOR): -7 log-pts.
  • Gap vs pool: +87 log-pts (threshold for SUPPORTED: ≥20 log-pts AND TWN rank ≤ 2).
  • Taiwan rank in 9-country sample: 1.

Primary 2 — Real GDP per capita cumulative log-change (PWT rgdpe / pop):

  • Taiwan: +184 log-pts.
  • Donor-pool mean: +158 log-pts.
  • Market-led subpool mean (no KOR): +145 log-pts.
  • Gap vs pool: +26 log-pts (threshold for SUPPORTED: ≥50 log-pts AND TWN rank #1).
  • Taiwan rank in 9-country sample: 4.

Method

Single-treated-unit cumulative-gap test on PWT 10.x productivity indicators, 1973 (ITRI founding) to 2019 (last PWT year). The spec originally requested foundry market-share, semiconductor patent stock, and the Atlas Economic Complexity Index, but none of those series are present on disk in a Taiwan-inclusive form (WIPO ip_statistics_data_center has no TWN rows; OWID economic- complexity-index file is not present; WDI excludes Taiwan entirely). PWT is the only on-disk publisher with full TWN coverage 1955-2019 plus the donor pool. TFP is the closest available proxy for 'frontier capability' the spec asks about; real GDP per capita is the closest proxy for 'output of a frontier industry' that flows to the wider economy.

We do not run a synthetic-control fit. With one treated unit, 9-country donor pool, and only ~17 years of pre-period (1955- 1972), a synth fit on long-horizon outcomes is under-identified; the dispositive question is the magnitude of Taiwan's gap to a comparable-income peer mean, which the cumulative-log statistic answers directly.

Caveats

  • KOR is itself a developmentalist case (HCI drive, chaebol- state coordination), not a clean control. The 'market-led subpool' diagnostic strips KOR out so a reader can see both framings; the headline gap is to the full 8-country donor pool.
  • The treatment is 'ITRI strategy' but PWT productivity captures all sources of Taiwan's TFP catch-up — Cold War US technology transfer, returning diaspora engineers, the global pure-play foundry shift. The replication cannot isolate the ITRI channel.
  • PWT ends 2019, so the 2020s semiconductor boom (TSMC 3nm/N2, CHIPS Act spillovers) does not enter the test.
  • This is an associational descriptive comparison, not a causal identification. The verdict labels a magnitude pattern, not a causal estimate.

Data

  • pwt:rtfpna (TFP at constant national prices, 2017 = 1)
  • pwt:rgdpe (real GDP at chained PPPs, expenditure-side)
  • pwt:pop (population, millions)

Notes

Seeded from a developmentalist claim that Taiwan's ITRI-led semiconductor strategy produced a frontier-capability industry no comparable-income peer generated. Synth-DID with comparable-income donor pool; human review needed for foundry market-share threshold and process-node leadership measure.

Authored framework. Read the transparency note.