Our Methodology

This page explains exactly how we calculate tariff costs, where our data comes from, and what assumptions we make. If you disagree with our methodology, we want to hear from you.

Data Sources

Data Pipeline

TariffTax operates a multi-stage data pipeline that transforms raw government data into the estimates you see on the site:

1. Ingest: Raw data files are pulled from government APIs and bulk downloads (Census, USITC, BLS, USDA). Each source is fetched on its own schedule — daily for tariff rates, monthly for trade flows.
2. Normalize: HTS codes, country names, and industry classifications are standardized across sources. We map 10-digit HTS codes to consumer product categories using a custom concordance table.
3. Calculate: Tariff revenue, household costs, state impacts, and distributional effects are computed from the normalized data. All calculations are deterministic and reproducible.
4. Validate: Output estimates are cross-checked against published figures from the Tax Foundation, Yale Budget Lab, and Peterson Institute. Anomalies trigger manual review.
5. Publish: Final data is exported as JSON files served from /public/data/ and consumed by the site's React components.

Complete Data Source List

We draw from 15 primary data sources:

1. US International Trade Commission (USITC) — Harmonized Tariff Schedule
2. Census Bureau / USA Trade Online — import/export values
3. USITC DataWeb — detailed trade statistics and duties collected
4. BLS Consumer Expenditure Survey — household spending by category and income
5. BLS Consumer Price Index — price change tracking
6. BLS Current Employment Statistics — employment by industry and state
7. USDA Foreign Agricultural Service — agricultural trade data
8. Congressional Budget Office — revenue projections and economic estimates
9. Federal Reserve / FRED — exchange rates, industrial production, economic indicators
10. Federal Register — tariff proclamations, executive orders, exemptions
11. Tax Foundation — tariff revenue estimates and economic modeling
12. Budget Lab at Yale — household cost estimates and distributional analysis
13. Peterson Institute for International Economics — trade policy analysis
14. World Trade Organization — global tariff and trade flow data
15. UN Comtrade — international merchandise trade statistics

Household Cost Estimation

Our estimates of tariff costs per household follow this methodology:

1. Identify tariff rates: We compile the current tariff rate for every HTS code, including base MFN rates, Section 301 (China) tariffs, Section 232 (steel/aluminum) tariffs, IEEPA tariffs, and any applicable anti-dumping or countervailing duties.
2. Calculate revenue by product category: Using Census Bureau trade data, we calculate the total tariff revenue generated by each broad product category (clothing, electronics, autos, food, etc.).
3. Estimate pass-through rates: Based on academic research (Amiti, Redding & Weinstein 2019; Fajgelbaum et al. 2020; Cavallo et al. 2021), we apply pass-through rates — the percentage of the tariff that is ultimately borne by consumers. For most product categories, research indicates near-complete pass-through (95-100%).
4. Allocate to households: Using BLS Consumer Expenditure Survey data, we allocate tariff costs to households based on their spending patterns. We divide total tariff-driven price increases by the number of US households (~131 million) to produce per-household estimates.
5. Validate against external estimates: We cross-reference our estimates with those published by the Tax Foundation, Budget Lab at Yale, Peterson Institute for International Economics, and National Retail Federation. Our figures typically fall within the range of these published estimates.

Income Distribution Analysis

Our distributional analysis (who pays what share of tariff costs by income level) uses BLS Consumer Expenditure Survey data, which breaks down spending by income quintile. Because lower-income households spend a higher share of income on consumption — and a higher share on heavily tariffed categories like clothing and footwear — tariffs are regressive. We calculate effective tariff burden as:

Tariff Burden Rate = (Household tariff cost ÷ Household pre-tax income) × 100

State-Level Estimates

State-level tariff impact estimates incorporate three factors:

1. Consumer cost: State median income and consumption patterns determine the per-household tariff burden.
2. Export exposure: State export data from the Census Bureau identifies how much each state's exports are affected by retaliatory tariffs.
3. Industry exposure: BLS employment data by state and industry identifies states with high concentrations of tariff-affected industries.

Tariff Rate Methodology

We report two types of tariff rates throughout the site, and it's important to understand the difference:

• Simple average tariff rate: The unweighted mean of all tariff lines. This treats a rarely-imported product the same as a high-volume one. Useful for understanding the breadth of tariff coverage but can be misleading about actual economic impact.
• Trade-weighted average tariff rate: Each tariff line is weighted by its share of total import value. This better reflects what importers (and ultimately consumers) actually pay. A 50% tariff on $100M of imports matters more than a 50% tariff on $1M. This is our primary metric.

For country-specific rates (e.g., “China faces a 145% effective tariff”), we calculate the trade-weighted average across all HTS codes for that country's imports, summing base MFN rates, Section 301 tariffs, Section 232 tariffs, IEEPA tariffs, and any AD/CVD duties.

State Impact Score

Each state receives a composite impact score (0–100) based on three equally weighted components:

1. Consumer burden (33%): Per-household tariff cost as a percentage of state median income. States with lower incomes bear a proportionally higher burden.
2. Export exposure (33%): The share of state GDP derived from exports to countries imposing retaliatory tariffs. States like Louisiana (petrochemicals) and Washington (aerospace, agriculture) score high.
3. Industry concentration (33%): Employment concentration in tariff-sensitive industries (manufacturing, agriculture, mining) relative to the national average. States with heavy manufacturing dependence score higher.

Each component is normalized to a 0–100 scale using min-max normalization across all 50 states plus DC. The final score is the simple average of the three components. A score of 80+ indicates severe exposure; below 40 indicates relatively low impact.

Limitations and Assumptions

• Pass-through timing: Tariff costs don't reach consumers instantly. There's a lag of weeks to months as inventory turns over. Our estimates reflect the steady-state cost, not the initial transitional period.
• Substitution effects: Some consumers switch to untariffed domestic alternatives. This reduces the direct tariff cost but may increase costs in other ways (domestic products are often more expensive). Our estimates account for partial substitution based on estimated elasticities.
• Dynamic effects: We don't fully capture second-order effects like reduced business investment, supply chain restructuring, or macroeconomic slowdown caused by tariffs. Our estimates are therefore likely conservative.
• Exemptions and exclusions: Tariff policy includes numerous exemptions that change frequently. We track published exemptions but may lag unpublished administrative decisions.

Update Schedule

• Tariff rates: Updated within 24-48 hours of any Federal Register notice or executive order changing rates
• Trade data: Updated monthly when Census Bureau releases new trade statistics
• Household cost estimates: Recalculated monthly with updated trade data
• Analysis articles: Published as developments warrant; updated when material changes occur

Corrections Policy

We take accuracy seriously. If you find an error in our data, calculations, or analysis, please email info@thedataproject.ai. All corrections are published transparently with a note on the corrected page.

Academic References