Pre-registration
Brazil's Nova Indústria Brasil (NIB) 2024 industrial-policy plan targets six "missions" (sustainable agribusiness, health complex, infrastructure, digital transformation, defence/ bioeconomy, decarbonisation) with R$300bn of subsidised credit + procurement preferences. The pre-registered claim is that targeted-mission sectors show measurably faster value-added growth (>2 pp annualised) than non-targeted control sectors over 2024-2026, AND that this within-Brazil sectoral divergence does not appear in Brazil's pre-NIB 2018-2022 baseline. The competing prediction is that the NIB outlay produces no detectable within-country sectoral divergence once commodity-cycle and macro controls are applied — i.e., classical industrial-policy- ineffectiveness in middle-income context.
Falsification criterion — what would disprove this
This hypothesis is considered falsified if:
Refuted if Callaway-Sant'Anna ATT on log_sectoral_value_added for NIB-targeted vs control sectors over 2024-2026 is less than 2 pp annualised at p < 0.10 clustered at sector level, OR if the same divergence appears in placebo windows pre-2024 (suggesting selection on pre-trends). Strongly refuted if ATT is negative or zero. Supported if ATT >= 2 pp annualised AND placebo-window ATT < 0.5 pp. Mixed if 1-2 pp annualised.
formal test & threshold
test: callaway_santanna_did_within_country_sector threshold: ATT >= 2 pp annualised log_VA growth AND placebo_pre_ATT < 0.5 pp
Method
- Template
did_callaway_santanna- Clustering
country- Sample
- 1 countries · 2018 – 2026
- Evidence type
- causal
Primary: Callaway-Sant'Anna DiD with NIB-targeted sectors as treated, non-targeted manufacturing + services sectors as control, all within Brazil. Pre-period 2018-2023 vs post-period 2024-2026. Secondary: event-study around 2024-Q1 NIB launch with sector × time interaction. Tertiary: synthetic-control across countries with Brazil as treated and other middle-income LatAm (Mexico, Argentina, Colombia, Chile) as donor pool, on the manufacturing-VA outcome.
Data
| Variable | Source | Transform |
|---|---|---|
sectoral_value_added_log outcome | ipeadata:pim_pf_setortier 2 ipeadata:pib_setorial_trimestraltier 2 | log_level_quarterly |
sectoral_employment outcome | ipeadata:caged_setortier 2 ilostat:employment_by_sectortier 2 | log_level_quarterly |
sectoral_investment_share outcome | ipeadata:fbcf_setorialtier 2 | pct_of_sectoral_va |
nib_targeted_sector_indicator treatment | constructed:binary = 1 for CNAE codes mapped to NIB missions; 0 for other manufacturing/services sectorstier 5 | binary |
nib_implementation_period treatment | constructed:binary = 1 from 2024-Q1 onwardtier 5 | binary |
commodity_terms_of_trade control | world_bank_wdi:TT.PRI.MRCH.XD.WDtier 2 | log_change |
brazil_policy_rate control | ipeadata:selic_anualizadatier 2 | level |
brazil_real_exchange_rate control | bis:WS_EERtier 2 | level |
us_policy_rate control | fred:FEDFUNDStier 1 | level |
● ready · ● pending · ● reconstruct-needed
Detailed result card
Result card — lula3_industrial_policy_2023_2026_reshoring_outcomes
Verdict: INCONCLUSIVE_DATA_PENDING — no outcome variable loaded
Pre-registration
- Claim: Brazil's Nova Indústria Brasil (NIB) 2024 industrial-policy plan targets six "missions" (sustainable agribusiness, health complex, infrastructure, digital transformation, defence/ bioeconomy, decarbonisation) with R$300bn of subsidised credit + procurement preferences. The pre-registered claim is that targeted-mission sectors show measurably faster value-added growth (>2 pp annualised) than non-targeted control sectors over 2024-2026, AND that this within-Brazil sectoral divergence does not appear in Brazil's pre-NIB 2018-2022 baseline. The competing prediction is that the NIB outlay produces no detectable within-country sectoral divergence once commodity-cycle and macro controls are applied — i.e., classical industrial-policy- ineffectiveness in middle-income context.
- Falsification rule: Refuted if Callaway-Sant'Anna ATT on log_sectoral_value_added for NIB-targeted vs control sectors over 2024-2026 is less than 2 pp annualised at p < 0.10 clustered at sector level, OR if the same divergence appears in placebo windows pre-2024 (suggesting selection on pre-trends). Strongly refuted if ATT is negative or zero. Supported if ATT >= 2 pp annualised AND placebo-window ATT < 0.5 pp. Mixed if 1-2 pp annualised.
Estimate (Callaway-Sant'Anna staggered DiD, TWFE approximation)
- Error: no outcome variable loaded
Variables resolved
Missing data
ipeadata:pim_pf_setor; ipeadata:pib_setorial_trimestral(outcome)ipeadata:caged_setor; ilostat:employment_by_sector(outcome)ipeadata:fbcf_setorial(outcome)constructed: binary = 1 for CNAE codes mapped to NIB missions; 0 for other manufacturing/services sectors(treatment)constructed: binary = 1 from 2024-Q1 onward(treatment)world_bank_wdi:TT.PRI.MRCH.XD.WD; imf_pcps:Primary(controls)ipeadata:selic_anualizada(controls)bis:WS_EER(controls)fred:FEDFUNDS(controls)
Generated by scripts/run_did_callaway_santanna.py at 2026-04-30T09:47:25+00:00
Strongest opposing argument
Every hypothesis ships with its charitable opposing argument. The framework earns credibility by handling objections at their strongest, not weakest.
Notes
Within-country difference-in-differences across sectors is the cleanest identification strategy because it controls for all Brazil-wide macro shocks. Sector-level value-added is available from IPEADATA and IBGE quarterly. Treatment is the binary indicator of "NIB-targeted sector" mapped from CNAE 2.0 sector codes to the NIB mission definitions. The middle-income context makes this distinct from the Korean/Taiwanese industrial-policy literature (which operated in low-income / catch-up settings); the Lula 3 NIB is closer to the post-2008 developmentalist-revival literature (Rodrik, Mazzucato).