What Were the Employment Impacts of the 2002 Bush Steel Tariffs?

This study investigates whether the 2002 Bush steel tariffs protected steel industry jobs or harmed employment in steel-consuming sectors. Using a generalized difference-in-difference approach with local labor market (commuting zone) data from 1993 to 2008, the authors examine employment outcomes based on regions’ exposure to steel production and consumption. They find the tariffs caused large, persistent employment declines in downstream steel-consuming industries but had no significant effect on steel industry jobs. The impact on steel-intensive manufacturing was approximately 75% the size of the well-known China shock’s effect on overall manufacturing employment, despite the tariffs being temporary.

Full Citation and Link to Article

Lake, James and Liu, Ding. “Local Labor Market Effects of the 2002 Bush Steel Tariffs.” American Economic Journal: Applied Economics (forthcoming). Available at: https://www.aeaweb.org/articles?id=10.1257/app.20220329

Extended Summary

Central Research Question

This paper investigates the labor market effects of the 2002 Bush steel safeguard tariffs. Specifically, it examines whether these tariffs succeeded in protecting employment in the domestic steel industry and what their broader consequences were for employment in steel-consuming (downstream) industries. Using a generalized difference-in-difference framework applied to U.S. commuting zones (CZs), the study asks: Did the temporary tariffs help preserve steel jobs, or did they instead cause significant employment losses in industries that use steel as a key input?

Previous Literature

The paper builds on three strands of economic literature. First, it contributes to studies on the downstream effects of trade protection, such as Cox (2022), who found persistent declines in exports and employment for downstream industries after the Bush steel tariffs. Second, it relates to work analyzing how local labor markets respond to trade shocks, including the well-known China shock literature (e.g., Autor et al. 2013, 2021) and research on labor market heterogeneity (Bloom et al. 2019; Charles et al. 2016, 2019). Third, it aligns with studies on the persistence of temporary trade shocks, such as Dix-Carneiro and Kovak (2017) on Brazilian tariff liberalization, and Juhász (2018), who studied the long-run effects of the Napoleonic trade blockade.

Despite widespread media and political attention to the Bush steel tariffs, no prior academic study had conducted a systematic, local labor market analysis of their employment effects. This paper fills that gap.

Data

The authors use annual county-level employment data from 1993 to 2008, which they aggregate to commuting zones to construct local labor market outcomes. Employment is measured across multiple sectors, with particular attention to manufacturing, steel-intensive manufacturing (defined as industries with high usage of steel inputs), and the steel-producing industries directly protected by the tariffs.

To assess exposure to the tariffs, the study constructs two main CZ-level measures:

1. Vulnerability (Vc): A commuting zone’s exposure to downstream effects, based on the share of local employment in industries that rely heavily on steel inputs.

2. Protection (Pc): A commuting zone’s exposure to upstream benefits, based on the share of local employment in steel-producing industries.

Input-output data from the 1997 BEA IO tables allow the authors to identify steel-using industries and measure how steel was embedded in intermediate inputs. Tariff data come from Presidential Proclamation 7529 and USITC sources. The study also controls for confounding shocks, including Chinese import competition (the “China shock”) and anti-dumping duties on steel.

Methods

The empirical strategy uses a generalized difference-in-difference framework. The authors model labor market outcomes (e.g., manufacturing employment share) as a function of commuting zone exposure to the tariffs (Vc and Pc), interacted with a full set of year indicators from 1993 to 2008. The specification includes CZ and state-year fixed effects, as well as time-varying controls for China shock exposure and steel anti-dumping (AD) duties. Robustness is checked using exposure-robust standard errors and a wide range of alternative specifications.

To construct the key treatment variables, the authors calculate per-worker tariff burdens for each industry based on pre-tariff import volumes and industry-specific steel usage. These are then aggregated to the CZ level using local employment weights. The identification relies on variation across CZs in their employment composition in industries more or less vulnerable to the tariffs.

Findings/Size Effects

The study produces four key findings:

1. Large negative effects on downstream employment. The tariffs had substantial negative effects on employment in steel-consuming industries, both in the broader manufacturing sector and in the most steel-intensive manufacturing subsectors. For example, by 2002, a CZ with one standard deviation higher vulnerability had a 0.133 standard deviation larger decline in its steel-intensive manufacturing employment share. This effect grew over time, reaching 0.185 standard deviations by 2008.

2. No employment gains in steel-producing industries. The analysis finds no evidence that CZs with more steel-producing jobs experienced relative employment gains. Rather, the steel industry continued its pre-existing employment decline. Even during the period when tariffs were in place, the steel-production employment share showed no upward deviation from its secular downward trend.

3. Persistent negative effects. The employment losses in downstream industries did not reverse after the tariffs were lifted in late 2003. Effects remained strong through 2008, five years after the policy ended. These persistent effects are consistent with a mechanism involving fixed plant entry costs: temporary shocks can cause plant exits that are not reversed, leading to lasting employment declines.

4. Large relative size compared to the China shock. Although the China shock has received more attention, the Bush steel tariffs had similarly large local effects. The impact of Bush tariff exposure on the manufacturing employment share was about one-quarter the size of the China shock’s effect. More strikingly, the impact on steel-intensive manufacturing employment was roughly three-quarters that of the China shock on all manufacturing employment.

The study finds little to no evidence that displaced workers in steel-consuming industries were absorbed into other local industries. Non-steel-intensive manufacturing did not grow in vulnerable CZs, and while there was no clear increase in non-employment, the evidence suggests sluggish reallocation.

The authors also explore heterogeneity in effects based on other macroeconomic trends. CZs that experienced a housing construction boom or benefited from rising military expenditures in the 2000s showed weaker negative impacts, suggesting local resilience may be shaped by other simultaneous shocks. By contrast, auto sector-dependent CZs showed stronger negative impacts, consistent with the industry’s vulnerability to higher input costs and weak global competitiveness.

Conclusion

This study provides compelling evidence that the 2002 Bush steel tariffs, though intended to protect domestic steel jobs, produced substantial and lasting negative effects on downstream employment, without clear benefits for steel-producing workers. These effects persisted long after the tariffs were removed and rival the much-discussed China shock in their magnitude—particularly in steel-intensive manufacturing.

The findings challenge the conventional wisdom that tariffs on intermediate goods are effective job-saving tools. Instead, they highlight the unintended costs of protectionist trade policy, especially when inputs are widely used across the domestic supply chain. The evidence also suggests that temporary trade policy can cause permanent labor market damage due to firm exit and industry restructuring.

Importantly, the study underscores the need for policymakers to consider local economic structure and cross-industry input linkages before implementing protectionist policies. In this case, the 2002 tariffs not only failed to arrest the long-run decline of steel employment but also triggered broader employment losses elsewhere in manufacturing—many of which never recovered.