VoxEU Column International trade Inflation Monetary Policy

Unfriendly friends: Trade and relocation effects of the US Inflation Reduction Act

3 Jul 2023

The adoption of the US Inflation Reduction Act in 2022 raised concerns among trading partners about the discriminatory nature of domestic content requirements in the law. This column provides a model-based assessment of the spillover effects of the main provisions of the Act on global production and trade. It estimates substantive effects in the US sectors directly affected by the Act as well as non-negligible losses in specific EU sectors. The policies to boost the green transition need to carefully balance attracting investments and innovations with the risks of retaliatory measures and domestic security priorities.

Authors

Maria Grazia Attinasi

Lukas Boeckelmann

Baptiste Meunier

The Inflation Reduction Act (IRA) significantly steps up the US efforts in the fight against climate change and aims to increase the US ability to attract key green technologies. To these ends, it foresees significant financial incentives (i.e. tax credits for the purchase of electric vehicles (EVs) and investments in renewable energy equipment) conditional on specific domestic content requirements.¹ The IRA substantially revised the US law on tax credits for EVs; to qualify for it, final assembly must take place in North America and a significant proportion of the components and minerals in the battery must be manufactured or assembled in North America (components) or in a country with which the US holds a free trade agreement (minerals).² For investments in renewable energy development the IRA implements an additional tax credit if a certain proportion of the components are produced in North America. By discriminating against products manufactured outside the USMCA (US, Mexico, and Canada trade Agreement), the IRA provisions could adversely affect third countries via trade and relocation channels.

Modelling the spillover effects

We quantify the potential economic effects of the Inflation Reduction Act using the Baqaee and Farhi (2023) model for a sample of 41 countries (or country groups) and 30 sectors.³ This model allows us to estimate the non-linear effects of higher trade barriers and to consider the endogenous reaction of producers and consumers in an inter-connected world economy.⁴ The focus of the analysis is on two IRA provisions: the tax credit for the purchase of EVs and the bonus for investment in renewable energy equipment, and their related content requirements. These measures are modelled as a trade cost shock that decreases the price of EVs and renewable energy equipment for US consumers if these are bought from USMCA producers.⁵ In addition, a second set of shocks simulates the impact of domestic content requirements for intermediate inputs used by USMCA producers of EVs, batteries, and renewable energy equipment. This is simulated by imposing an iceberg trade cost shock to imports from suppliers outside the USMCA of batteries, critical minerals, power components, steel, and iron.⁶

In order to reflect the growing significance of the green sectors, we rescale the magnitude of the abovementioned shocks using the International Energy Agency (IEA) projections for 2030.⁷ In addition, as the speed of the green transition remains uncertain, we use three different sets of assumptions from the IEA namely: (1) the ‘conservative’ scenario, based on stated policies as of 2022, (2) the ‘accelerated’ scenario, which also includes announced policies and is more optimistic concerning the speed of the green transition, and (3) the ‘net zero’ scenario, based on policies required to abide by the goal of net-zero carbon emissions by 2050 and entails more radical policy measures.⁸ Furthermore, standard trade elasticities (i.e. measuring the ease of substituting suppliers across countries), might not adequately reflect the emerging nature of the green sectors given that, compared to more mature production lines, producers may find it easier to divert new investments abroad. Therefore, for the accelerated and net zero scenarios we apply higher trade elasticities than estimated in the literature.⁹ Likewise, as it might be hard to find alternative suppliers for rare critical minerals and high-end technologies, the simulations in the accelerated and net zero scenarios assume lower than standard elasticities of substitution across products (measuring the ease of substituting across different inputs for production).¹⁰ Finally, a main objective of the IRA is to make the US the preferred location of new investments, generating technology-investment driven productivity growth. This would increase productivity in the US while having the opposite effect in the rest of the world, widening inter alia the productivity gap between EU and US. We account for such a productivity impact of green investments by adding an exogenous total factor productivity (TFP) shock to the simulations.¹¹ We find that these productivity gains would increase the comparative advantage of US producers relative to the rest of the world, thus magnifying the relocation and trade effects. The quantification of trade and relocation effects in the remainder of this column refer to long run effects with a time horizon of 2030.

A quantification of trade and relocation effects

Losses of global trade are limited, ranging from 0.2% (conservative) to 0.9% (net zero) (Figure 1, Panel A). This is explained by the fact that the trade shock originates from one country only (the US), which does not affect its largest trade partners (Mexico and Canada) and is limited to only three sectors. Moreover, the IRA induces a re-allocation of trade flows rather than a destruction thereof. On the other hand, sectoral losses can be substantial in sectors targeted by the IRA, notably in the electrical and optical equipment sector where global trade losses reach 6% in the net zero scenario (Figure 1, Panel A).¹² The rest of this column focuses on the effects in this sector.

US trade partners are concerned not only by the prospective loss of access to the US market, but also by reduced possibilities to divert their production to other destinations. China would lose between 10% (conservative) and 50% (net zero) of its exports to USMCA in electrical and optical equipment, and the EU would lose between 10% and 45% (Figure 1, Panel B). However, a compounding issue would be the absence of trade diversion. As shown in Figure 2 (Panel A) for the EU, while production losses are foremost due to lower exports to the USMCA given the IRA-induced trade barriers, exports to other countries are also lower– reflecting second-round effects, mostly in the form of excess supply in the sector and lower demand for upstream products.

Figure 1 Impact of the Inflation Reduction Act on trade (percentage change from steady state)

Sources: Baqaee and Farhi (2023), ADB IO table, ECB staff calculations
Notes: Non-linear impact simulated through 25 iterations of the log-linearised model. Electric vehicles (EV) are a sub-sector of ‘transport equipment’, EV batteries and renewable energy equipment of ‘electrical and optical equipment’, and processed rare Earth minerals of ‘basic and fabricated metals’. USMCA = US, Mexico, and Canada (trade) Agreement; RCEP = Regional Comprehensive Economic Partnership, a trade agreement among 15 Asia-Pacific countries; ROW = rest of the world. RCEP does not include China.

Figure 2 Impact of the Inflation Reduction Act on production (electrical and optical equipment, percentage and percentage points change from steady state)

Sources: Baqaee and Farhi (2023), ADB IO table, ECB staff calculations
Notes: Non-linear impact simulated through 25 iterations of the log-linearised model. USMCA = US, Mexico, and Canada (trade) Agreement; RCEP = Regional Comprehensive Economic Partnership, a trade agreement among 15 Asia-Pacific countries; ROW = rest of the world. RCEP does not include China.

The IRA entails large relocation of production capacities towards the US. Relocation effects are measured by production changes shown in Figure 2 (Panel B). The US would gain from positive relocation effects, increasing production by 6% to 30% in electrical and optical equipment. Relocation effects would also be positive for Mexico and Canada (3% to 19%); in contrast, the IRA entails production losses for the EU (between -0.5% and -3%) and China (between -1% and -5%). In the rest of the world, larger production losses are faced by countries with higher exposure to the USMCA such as Malaysia and Vietnam (up to -18% and -13% production losses respectively in the net zero scenario).¹³ These relocation effects translate into changes of market share.¹⁴ Focusing on the US domestic demand of electrical and optical equipment, Figure 3 (Panel A) shows that that US producers gain nine percentage points in market share, mostly at the expenses of China (-6 percentage points).¹⁵

While the rest of the world loses, the IRA would benefit the US through additional output and lower strategic dependence vis-à-vis China. Reported in nominal terms, the IRA would entail by 2030 the relocation to the US of $280 billion worth of annual output across all sectors (accelerated scenario) mainly at the expenses of China (-$210 billion) and, to a smaller extent, the EU (-$70 billion).¹⁶ While these represent a relatively small share of total output (i.e. a 0.7% increase in US output and a -0.2% and -0.4% decrease in EU and Chinese output respectively) sectoral effects are more sizeable (+15% for electric and optimal equipment production in the US, and -1.6% and -2.4% in the EU and China respectively). Output gains/losses would be larger in the net zero scenario for the US and China respectively (Figure 3, Panel B).¹⁷ At the global level, summing across all effects (positive in USMCA, negative elsewhere), the IRA translates into net annual output losses concentrated mostly in the green sectors, thus suggesting that the IRA could slow the green transition at global level.

Figure 3 ‘Winners’ and ‘losers’ of the Inflation Reduction Act

Sources: Baqaee and Farhi (2023), ADB IO table, ECB staff calculations
Notes: Non-linear impact simulated through 25 iterations of the log-linearised model. Panel a) refers to the accelerated scenario; red and green bars indicate respectively negative and positive impacts. USMCA = US, Mexico, and Canada (trade) Agreement; RCEP = Regional Comprehensive Economic Partnership, a trade agreement among 15 Asia-Pacific countries; ROW = rest of the world. RCEP does not include China. Panel b) states the US dollar change (in billions) from steady state expressed in 2021 levels.

What’s next?

The stylised analysis presented in this column shows that policies aimed at boosting the green transition, if coupled with measures erecting barriers against trade, would distort sectoral trade flows and lead to a reallocation of production in the green sectors mainly in response to financial incentives and not necessarily according to a country’s comparative advantage. Should the main US trading partners take retaliatory measures as part of their response to the IRA, this would likely cause further harm, as increased domestic production would come at the cost of a slower green transition including by hindering the international diffusion of knowledge in green technologies. Therefore, from a purely economic perspective, the first best response to climate change remains international cooperation; after all climate change is a global challenge that requires global solutions.

At the same time, it should be recognised that the re-evaluation of domestic security priorities is an important element in shaping domestic policies. The green transition, and the technological progress it entails, can usher an industrial and technological revolution with far reaching economic implications in terms of consumption possibilities, job creation, technological progress, and new strategic dependencies. Therefore, countries have an incentive to attract investments that will give them a competitive edge in the key sectors of the future (Terzi 2022). In light of these considerations, the estimated economic effects presented in this column likely represent a lower bound for both winners and losers.

Authors’ note: This column reflects the opinions of the authors and not necessarily those of the European Central Bank.

References

Atalay, E (2017), “How Important Are Sectoral Shocks?”, American Economic Journal: Macroeconomics 9(4): 254–280.

Attinasi, M G, L Boeckelmann and B Meunier (2023), “Friend-shoring global value chains: a model-based analysis”, Economic Bulletin Box 2, European Central Bank.

Bachmann, R, D Baqaee, C Bayer, M Kuhn, A Löschel, B Moll, A Peichl, K Pittel and M Schularick (2022), “What if? The economic effects for Germany of a stop of energy imports from Russia”, ECONtribute Policy Brief No 28/2022.

Baqaee, D and E Farhi (2023), “Networks, Barriers, and Trade”, Econometrica, forthcoming

Cai, J, N Li and A M Santacreu (2022), “Knowledge Diffusion, Trade, and Innovation across Countries and Sectors”, American Economic Journal: Macroeconomics 14(1): 104–145.

Costinot, A and A Rodriguez-Clare (2014), “Trade theory with numbers: quantifying the consequences of globalization”, Handbook of International Economics 4, 197.

Goes, C and E Bekkers (2022), “The impact of geopolitical conflicts on trade, growth, and innovation”, Staff Working Paper No. ERSD-2022-9, World Trade Organization.

Terzi, A (2022), “A green industrial revolution is coming”, VoxEU.org, 28 June

Footnotes

The Inflation Reduction Act entails a broader set of legislative provisions including measures to finance the budgetary impact of the subsidies and tax credits (e.g. tax increases and funding for the Internal Revenue Service which is estimated to lead to a better tax collection). These broader measures, however, are outside the scope of our analysis.
For example, once applicable, the critical mineral requirement for electric vehicle batteries will be met if the percentage of the value of the critical minerals (e.g. lithium, nickel) in the vehicle’s battery that were extracted or processed in the US, or in any country with which the US has a free trade agreement in effect, is equal to or greater than 40% for a vehicle that is placed in service in 2023. This required percentage increases annually to 50% in 2024, 60% in 2025, 70% in 2026, and 80% after 2026. The Inflation Reduction Act also targets explicitly the inputs coming from so-called ‘foreign entities of concern’ (China, Russia, North Korea, and Iran) and stipulates that from 2025 onwards, the critical materials will no longer be allowed to come from any of these countries.
The sample includes all sectors of the economy, including manufacturing, services, construction, energy, mining, and agriculture.
e estimates are subject to model limitations and uncertainties given the complexity of the IRA legislation. Despite the sectoral granularity of the Baqaee-Farhi model, the sub-sectors targeted by the IRA cannot be singled out, hence model-based estimates reflect the characteristics of the parent sector, which can differ from those of the targeted sector. Requirements on domestic content requirements are modelled by applying a trade cost shock, though the actual working of such regulatory constraints might be more complex. Finally, fiscal policy is absent of the Baqaee-Farhi model and therefore tax credits are not offset by higher tax collection.
As the tax credits for renewables also target businesses, we apply the same shock on purchases of renewable energy equipment by the construction and utilities sectors. The magnitude of the shock is calibrated according to the IRA provisions and amounts to 11.5% for electric vehicles and 10% for renewable energy equipment. For electric vehicles, 11.5% represents the amount of tax credit under the Credits for New Clean Vehicles ($7,500) over the average price of an electric vehicle in the US ($65,000). For renewable energy equipment, 10% is the additional tax credit granted as part of the Investment Tax Credit if the equipment satisfies domestic content criteria.
The first shock changes relative prices for US consumers, who substitute towards subsidised USMCA products. This generates a positive demand shock for USMCA producers and a negative demand shock for producers in the rest of the world. At the same time, domestic content requirements incentivise USMCA producers to substitute towards USMCA intermediate inputs, also to the benefit of USMCA suppliers. This substitution channel re-allocates production from the rest of the world towards the USMCA. However, as USMCA inputs can be more expensive than foreign counterparts, this feeds into higher prices for USMCA products. These effects drive further global substitution effects across products, which finally generate the general equilibrium response.
Conservative and accelerated scenarios are taken from IEA (2022). “Global EV Outlook 2022” for electric vehicles, and IEA (2022). “Renewables 2022” for renewable energy equipment. The net zero scenario is based on IEA (2021). “Net Zero by 2050”.
For example, sales of electric vehicles in the US are expected to grow by a factor of six by 2030 in the conservative scenario but by a factor of ten and 13 under the accelerated and net zero scenarios, respectively.
Standard elasticities in the conservative scenario are estimates from Costinot and Rodriguez-Clare (2014). In the net zero scenario, trade elasticities for affected sectors (electrical and optical equipment, basic and fabricated metals, transport equipment) are multiplied by a factor which sets their value along the maximum estimates from Costinot and Rodriguez-Clare (2014). Trade elasticities in the accelerated scenario are medians of those for the conservative and the net zero scenarios.
In the conservative scenario, the elasticity of substitution across inputs is set at the median estimate from Atalay (2017). The accelerated scenario uses the lower bound estimates from Atalay (2017), also used in the ‘rigid’ set-up of Attinasi et al. (2023). The net zero scenario is based on the elasticity in the ‘stressed’ scenario of Bachmann et al. (2022).
The Baqaee-Farhi model does not entail endogenous productivity. The total factor productivity (TFP) shock is calibrated by mapping changes in production (obtained on simulations without TFP) into TFP effects, by using elasticities of investment to output and of TFP to investment. These TFP shocks are then applied as an exogenous shock enhancing productivity in the USMCA and decreasing it in the rest of the world. We empirically check that the welfare implications of these productivity shocks are of similar magnitude of the effects of endogenising productivity in trade models as in Cai et al. (2022) and Goes and Bekker (2022).
The higher effects on the electrical and optical equipment are explained by the fact that this sector includes batteries and renewable energy equipment, making it the primary target of the IRA. In addition, most electric vehicles bought in the US are already assembled in the US so the impact of domestic content requirement is lower for finished cars than at the earlier stages of the electric vehicle supply chain (notably batteries). Finally, rare earth materials represent only a very small fraction (around 1%) of basic and fabricated metals, muting the impact on the aggregate sector.
Malaysia and Vietnam are two countries where Chinese companies have installed solar panel manufacturing capacity focused on exports to the US.
They would also translate into gains/losses of employment, however in the baseline version of the Baqaee-Farhi model with flexible wages used here labour supply is exogenous.
The IRA also leads to moderately higher producer prices in the USMCA as producers are force to substitute away from cheaper foreign inputs. Producer prices in the US increase by between +0.6% and +1.3% for electrical and optical equipment. In contrast, prices are marginally lower in the EU and in China (around -0.1% in for electrical and optical equipment) as losses in production translate into lower demand for labour, and therefore lower wages.
While our estimates refer to the change in annual output in the new steady state by 2030, cumulative output gains and losses until 2030 are likely to be substantially higher than that. We refrain from quantifying such cumulative gains as the precise path to the final reallocation effects remain beyond the scope of this analysis.
According to some estimates of the costs of the tax credit on clean vehicles, it would appear that the gains in output for the US far exceed the costs of the tax credit. For example, the Committee for Responsible Federal Budget estimates the cost of the tax credit on clean vehicles is estimated at around $36 billion.