Do US tariffs protect US jobs?

Do tariffs benefit the national economy? More particularly, does the imposition of steel tariffs (taxes on imported steel) in the United States (US) generate employment growth? The effects of steel tariffs continue as a prominent policy question in the US since at least 2016. The ‘evidence’ used will largely determine one’s answer to these questions.

What evidence should we use to answer this question? For economics students and those untrained in quantitative analytical techniques, knowing which set of ‘evidence’ is most reliable is challenging, to say the least. 

The aim of this blog is to educate economic students to understand the strengths and weaknesses of three different economic analytical techniques. The context of this evaluation is the relationship between US tariffs and employment. Each piece of analysis uses a different methodology: Computable General Equilibrium, partial equilibrium model, and difference-in-difference. 

Current policy context

Upon taking the US Presidential office for a second term, President Trump quickly outlined his ‘America First’ trade policy. President Trump’s office released the following statement in January 2025:

“Americans benefit from and deserve an America First trade policy.  Therefore, I am establishing a robust and reinvigorated trade policy that promotes investment and productivity, enhances our Nation’s industrial and technological advantages, defends our economic and national security, and — above all — benefits American workers (emphasis added), manufacturers, farmers, ranchers, entrepreneurs, and businesses” (The Whitehouse, 2025a). 

One aspect of this America First trade policy is ‘Made in America’. In announcing a sweeping set of tariff measures (April 2025), The White House directly stated that US workers would benefit through “better paying American jobs”.

“‘Made in America’ is not just a tagline—it’s an economic and national security priority of this Administration. The President’s reciprocal trade agenda means better-paying American jobs (emphasis added) making beautiful American-made cars, appliances, and other goods” (The Whitehouse, 2025b).

In support of this April 2025 announcement, The White House used the findings of ‘research’ to support and validate the tariff policy. 

“A 2024 economic analysis found that a global tariff of 10% would grow the economy by $728 billion, create 2.8 million jobs, and increase real household incomes by 5.7%” (The Whitehouse, 2025b).

and

“A 2023 report by the U.S. International Trade Commission that analyzed the effects of Section 232 and 301 tariffs on more than $300 billion of U.S. imports found that the tariffs reduced imports from China and effectively stimulated more U.S. production of the tariffed goods, with very minor effects on prices (The Whitehouse, 2025b).

The claim of economic growth, job creation, and real (nominal wage – inflation) wage increases read like a magic bullet for improved living standards. However, for those more experienced in economic analysis, such claims are unrealistic, at best. 

Evidence 1: Computable General Equilibrium (CGE) model

The “2024 economic analysis” (Ferry, 2024) cited by the White house does not appear to be publicly available on the internet. The journal (Empirical Economic Letters) that published the analysis is ranked in the lowest of four quality categories (A* – C) by the Australian Business Deans Council (ABDC, 2025). The modelling technique used (according to the GTAP, Purdue University website) is a Computable General Equilibrium (CGE). Such models are widely used in industry, but infrequently used in academia. A CGE model is deterministic in nature. That is, the relationship between variables are ‘hard coded’ with little or no room for variability or randomness. If these relationships don’t accurately reflect any given changes to an economy at the time, then estimates may be inaccurate. A few interconnected, inaccurately measured variables, can throw out estimates by some margin. Without access to the model parameters and the magnitude of variables changes on outcomes, CGE models remain ‘black boxes’ – opaque to outside observers. The reliability of CGE estimates are generally low.

Evidence 2: partial equilibrium model

The modelling technique used by the US International Trade Commission in their “2023 report” employs a more transparent approach – partial equilibrium model. This modelling technique estimates market prices and quantities. It is less deterministic, but assumes that all market prices outside of the market of concern are independent (exogenous). While this simplifies the analysis, such an assumption is strong. A failure of this assumption to reflect reality, will lead to systematically biased estimates.

As an aside, what the White House failed to mentioned was that the “2023 report” concluded that the net effect of the steel and aluminum tariffs enacted in 2018 was a USD 2.2 billion net reduction in production (US International Trade Commission, 2023). Steel production increased USD 1.3 billion by 2021, but downstream industries produced USD 3.5 billion less over the same period due to the section 232 tariffs. 

Evidence 3: Difference-in-Difference

Finally, two recent academic publications (in journals that have ABDC rank of A*) model steel tariffs introduced in 2002 (Lake and Liu, 2025) and 2018-19 (Flaaen and Pierce, 2024) on US employment. The analytical methodology used in these analyses is the most rigorous and uses the most detailed data. A Difference-in-Difference (D-in-D) approach controls for two sets of differences as a means of identifying the causal effect of a change in policy on groups i) directly affected by the policy change and ii) not directly affects by the policy change. In the case of the analysis by Lake and Liu (2025), the first difference is the pre and post time difference (subscript t), and the second difference is exposure to steel tariffs (subscript c). The equation below outlines the intuition behind the regression analysis.

Variables:

y_ct – employment across time and across local geographies.

B_c – measures of local geography exposure to use of steel as a production input and protection of steel industry.

Post_t – indicates if time is post 2001 (period of analysis is 1998-2003).

Greek letters – 𝞪, 𝞪₁, 𝞪₂, 𝞫 – are parameter estimates from the regression model.

The interaction between the two D-in-D differences (denoted by subscripts c and t) enables the estimation of the aggregate effect while controlling for possible effects caused by each difference. The above equation outlines the basic structure used by Lake and Liu (2025). Additional levels of analysis are applied to control for other real-world factors (i.e. anti-dumping protections introduced, retaliatory tariffs and downstream US industry effects). 

Based on the 2002 steel tariffs, Lake and Liu (2025) conclude that the temporary 2002-03 steel tariffs had negative net employment effects in the US. 

“Our main result is that the Bush steel tariffs have statistically and economically significant effects on employment in industries relying on inputs (emphasis added) hit with the tariffs. Moreover, these effects did not reverse themselves once the Bush steel tariffs were removed at the end of 2003. Instead, they persist until at least 2008. And the effects are economically large” (Lake and Liu, 2025, p.326). 

“In contrast to the effects on downstream employment, we find no evidence (emphasis added) of increased employment in the ­steel-producing industries protected by the Bush steel tariffs. Ultimately, our analysis emphasizes the costs of the tariffs on intermediate inputs and downplays the benefits of the tariffs for protected industries” (Lake and Liu, 2025, p.327). 

In a similar vein, Flaaen and Pierce (2024) analyse of the employment effect of President Trump’s first terms steel tariffs.

“We find that tariff increases enacted since early 2018 are associated with relative reductions in U.S. manufacturing employment (emphasis added) and relative increases in producer prices. In terms of manufacturing employment, rising input costs and retaliatory tariffs account for the negative relationship, and the contribution from these channels more than offsets a small positive effect from import protection” (Flaaen and Pierce, 2024, p.4).

“In terms of economic significance, we find that shifting an industry from the 25th percentile to the 75th percentile in terms of exposure to each of these channels of tariffs is associated with a relative reduction in manufacturing employment of 2.7 percent, with the positive contribution from the import protection effects of tariffs (0.4 percent) more than offset by the negative effects associated with rising input costs (-2.0 percent) and retaliatory tariffs (-1.1 percent)” (Flaaen and Pierce, 2024, p.4).

Conclusion

Not all forms of economic analysis are equal. As evidenced across the three forms of economic analysis presented above, the fewer stringent assumptions required by a model and the more realism allowed by a model, ceteris paribus, the more reliable the results. In the case of the impact of US tariffs on employment, the results from the CGE and D-in-D analysis are very different – opposing in fact. As indicated by the quality of the journals which published the research, the D-in-D analysis provides more reliable estimates. As such, economics students and policy makers should be aware of the assumptions and constraints of models they use to support policy recommendations. 

References

ABDC (2025). ABDC Journal Quality List. https://abdc.edu.au/abdc-journal-quality-list/ (Accessed: 27.11.2025)

Flaaen, A. & Pierce, J. (2024). Disentangling the Effects of the 2018-2019 Tariffs on a Globally Connected U.S. Manufacturing Sector. The Review of Economics and Statistics. doi: https://doi.org/10.1162/rest_a_01498

Ferry, J. (2024). Global 10% tariffs on U.S. imports would raise incomes and pay for large income tax cuts for lower/middle class. Coalition for a Prosperous America. https://prosperousamerica.org/global-10-tariffs-on-u-s-imports-would-raise-incomes-and-pay-for-large-income-tax-cuts-for-lower-middle-class/ (Accessed: 25.11.2025)

Lake, J., & Liu, D. (2025). Local labor market effects of the 2002 Bush steel tariffs. American Economic Journal: Economic Policy, 17(4), 292-330. https://pubs.aeaweb.org/doi/pdfplus/10.1257/pol.20220472 (Accessed: 24.11.2025)

The Whitehouse (2025a). America First Trade Policy. January 20, 2025. https://www.whitehouse.gov/presidential-actions/2025/01/america-first-trade-policy/  (Accessed: 25.11.2025)

The Whitehouse (2025b). Fact Sheet: President Donald J. Trump Declares National Emergency to Increase our Competitive Edge, Protect our Sovereignty, and Strengthen our National and Economic Security. April 2, 2025. https://www.whitehouse.gov/fact-sheets/2025/04/fact-sheet-president-donald-j-trump-declares-national-emergency-to-increase-our-competitive-edge-protect-our-sovereignty-and-strengthen-our-national-and-economic-security/ (Accessed: 25.11.2025)

USITC (2023). Economic Impact of Section 232 and 301 Tariffs on U.S. Industries. Publication No. 5405. Investigation No. 332-591. https://www.usitc.gov/publications/332/pub5405.pdf (Accessed: 25.11.2025)