Introduction

Antitrust enforcement in most jurisdictions, including the European Union and the United States, applies a consumer-welfare standard. This standard is administered separately from trade policy, keeping antitrust focused on consumer welfare and insulated from trade-policy considerations (Bradford and Chilton 2021). As trade costs fall, lowering costs through offshoring becomes more profitable for domestic incumbents and increases their support for free trade. At the same time, if domestic producers are subject to intense import competition, domestic consolidation may raise their demand for protectionism.

Antidumping (AD) and countervailing duty (CVD) petitions are the most common ways domestic producers seek trade protection. The investigating authority subjects imports to tariffs if it determines that they are sold at less than fair value or subsidized by a foreign government, and that the domestic industry is materially injured by those imports. The mandate of the investigating authority can conflict with the competition authority’s consumer-welfare standard.Recognizing this tension, in 2022 the U.S. Senate Subcommittee on Competition Policy, Antitrust, and Consumer Rights asked the U.S. Government Accountability Office to review AD/CVD processes and domestic market competition considerations, particularly focusing on “how aspects of market competition factor into the AD/CVD process” (U.S. Government Accountability Office 2022). Despite this potential conflict, the trade-policy implications of mergers remain unexamined in clearance decisions. Across multiple jurisdictions, I identify high-profile horizontal mergers between domestic producers followed by tariff petitions. In many of the corresponding merger clearance decisions, imports are mentioned as a competitive constraint on the merging parties.

In this paper, I study how merger control affects domestic incumbents’ incentives to petition for tariffs. I develop a framework to estimate whether a proposed merger lowers or raises the profitability of tariffs to the merging parties, and hence their propensity to file. I apply this framework to the Whirlpool–Maytag merger in the U.S. appliance industry, specifying and estimating a structural model of demand and supply following Montag (2026), with which I quantify the effect of the domestic merger across counterfactual production-location scenarios. To examine the effect of cross-border mergers, I also quantify how a hypothetical merger between Whirlpool and LG changes the profitability of tariff petitions.

The paper’s central contribution is to identify and quantify a trade-policy channel through which mergers between domestic producers can harm consumers beyond direct market-power effects. Theoretically, when offshoring is unavailable, a domestic merger strictly raises the profitability of tariffs to the merging parties; when offshoring is available, the sign becomes ambiguous. Empirically, I find that for the Whirlpool–Maytag merger, offshoring is more profitable than petitioning for tariffs in years when import competition is weak, while petitioning becomes more profitable than offshoring in years when import competition is strong. Even holding the likelihood of tariffs fixed, the merger raises tariff-induced consumer harm by more than 10 percent; in total, the consumer harm through the trade-policy channel is of the same order of magnitude as the harm from unilateral market power. A hypothetical cross-border merger reduces the profitability of tariffs in the same market.

I specify a three-stage model to analyze the effect of mergers on the incentives to petition for tariffs. In Stage 1, the competition authority adjudicates a horizontal merger involving a domestic incumbent that faces import competition. In Stage 2, given the competition authority’s decision, the incumbent chooses among offshoring production to lower marginal costs, petitioning for tariffs to raise foreign rivals’ costs, or maintaining the status quo.Igami (2018) studies the relationship between import competition and offshoring in the Hard Disk Drive Industry, showing that offshoring is pro-competitive and benefits consumers. Stage 3 embeds a differentiated-demand, oligopolistic-supply model in which firms set prices and consumers choose products.

Most directly, the paper relates to a literature concerned with the stringency of merger control. Nocke and Whinston (2022) show that current concentration thresholds are too lax in the absence of large efficiency gains. Retrospective analyses of consummated mergers found mixed results (Ashenfelter et al. 2013; Kwoka 2015; Bhattacharya et al. forthcoming). Asker and Nocke (2021) and Shapiro and Yurukoglu (2026) review this literature and conclude that the estimated price effects vary widely and that the evidence is mixed. Breinlich et al. (2018) study optimal merger policy for international mergers in settings where multiple national agencies can block a transaction across jurisdictions.Horn and Levinsohn (2001) and De Stefano and Rysman (2010) develop models in which a country chooses the level of domestic concentration through merger policy and show that when firms are exporters, national authorities may prefer a level of concentration that is excessive from a global perspective. Montag (2026) extends the scope of merger analysis by studying how the Whirlpool–Maytag merger affects total domestic welfare (consumers and workers) when potential acquirers for Maytag differ in their offshoring plans.

I extend this literature by identifying a channel through which mergers affect consumers via trade policy: by changing the merging parties’ incentives to raise rivals’ costs through tariff petitions.Salop et al. (1984) discuss how tariffs can be used to raise the cost of rivals. Furthermore, I provide a quantitative framework that allows competition authorities to assess how a merger changes the profitability of tariffs for the merging parties and the resulting consumer harm. I show that imports impose weaker competitive discipline on a merged domestic incumbent than equivalent domestic rivals would: because imports can be restricted through trade remedies but domestic rivals cannot, the discipline-from-imports defense for merger clearance can be weaker than it appears.This limits the widely held presumption, formally examined by Neven and Seabright (1997), that trade liberalization can substitute for domestic competition policy. The stronger the competitive constraint from imports, the stronger the incentive to petition for trade remedies. Merger control should therefore discount the competitive constraint from imports when this can be restricted post-merger.

The paper also contributes to a literature on market structure and lobbying. Classic political-economy models predict that organized sectors obtain protection (Grossman and Helpman 1994; Goldberg and Maggi 1999), and firm size predicts participation and intensity (Bombardini 2008). Kang (2016) finds that while lobbying has a small effect on policy enactment, the returns to lobbying are high. Recent evidence indicates that consolidation raises lobbying across (Cowgill et al. 2024) and within industries (Moshary and Slattery 2024).

I contribute to this literature in two ways. First, I extend it beyond traditional political lobbying. AD/CVD petitions are a quasi-judicial channel, relatively insulated from direct political bargaining, that present incumbents with a distinct choice set (petition, offshore, or maintain the status quo). I show that mergers between domestic producers raise the profitability of tariffs for the merging parties. Second, I decompose the merger effect into two channels: the appropriation effect (the acquirer internalizes the benefits of tariffs to the target) and the strategic effect (the tariff raises the profit of the merged firm by more than the sum of standalone profits).While domestic producers could overcome the collective-action problem by petitioning together, in practice, they often do not. Bombardini and Trebbi (2012) find that firms in more concentrated industries are more likely to lobby on trade issues individually rather than through a trade association. Distinguishing them matters because the two channels respond differently to the merger structure.

A related literature studies how AD/CVD cases can raise market power (Nieberding 1999; Konings and Vandenbussche 2005; Pierce 2011; Rovegno 2013) and facilitate collusion (Staiger and Wolak 1989). Because dumping margins depend on foreign pricing, the option value of a petition can induce higher foreign prices even before a case is filed. Blonigen et al. (2013) find that binding quotas increased market power in the U.S. steel industry, whereas tariffs did not, which is consistent with strong domestic competition from minimill producers disciplining outcomes. Flaaen et al. (2020) show that the initial AD/CVD actions on large residential washers primarily induced tariff jumping, whereas the 2018 global safeguards raised U.S. washer prices.

I extend this literature to show that when there are few domestic competitors, tariffs can generate substantial consumer harm. Whereas the literature focuses on how protection changes competition, I focus on how mergers alter the likelihood and harm from tariffs.

The Draghi (2024) report argues that European economic growth requires scale economies, prompting calls to relax EU merger control.Even earlier, France and Germany urged approval of the Siemens/Alstom merger to create a “European champion” in rail equipment; the European Commission nonetheless blocked the transaction in 2019. The European Commission’s April 2026 Draft Merger Guidelines also emphasize the enablement of scale economies. My results do not imply that merger control should be tightened across the board. Instead, mergers between domestic firms facing strong import competition should be scrutinized more closely; cross-border consolidation, by contrast, can deliver scale economies without raising incumbents’ returns to tariff petitions.

The remainder is structured as follows: Section 2 reviews measures to protect against import competition and their relation to merger control, Section 3 specifies the model, Section 4 describes the appliance industry, Section 5 details the empirical model and estimation, Section 6 presents the parameter estimates, Section 7 simulates counterfactuals, and Section 8 concludes.

Trade Protection and Merger Control

AD and CVD measures are the most commonly used trade-defense instruments worldwide. Since these are grounded in World Trade Organization (WTO) rules, the criteria and procedures for AD/CVD are codified in WTO agreements and apply across WTO members. Global safeguards (GS) are also WTO-authorized but used much less frequently. While the following discussion focuses on the institutional implementation in the United States, it should be understood as applying to other jurisdictions as well.

In the United States, the most commonly used alternative trade-defense tools are Section 232 actions (national security-based trade measures) and Section 301 actions (retaliatory trade measures). In the U.S., AD/CVD measures accounted for 97 percent of all trade actions initiated between 2002 and 2024 and in 2022 resulted in tariffs covering $37.4 billion of imports (Liu 2026). AD/CVD almost always originate from a petition filed by a domestic stakeholder. In contrast, Section 232 and Section 301 actions are initiated by the government and are not grounded in WTO authorized procedures. Based on interviews with practitioners, Liu (2026) reports that AD/CVD petitions remain the first tool of choice for domestic producers seeking protection from import competition.

AD duties are imposed on imports that are determined to be sold at less than fair value and that materially injure a domestic industry. Selling at less than fair value typically refers to a situation in which a firm sells a product at a lower price in the importing country than in its home market (Blonigen and Prusa 2016). If the exporter’s home market is deemed unsuitable for comparison, its sales price in a third country may be used instead. Since products destined for home and export markets often differ, defining the foreign-like product affords the Department of Commerce considerable leeway in AD cases. An alternative standard used in many AD cases is sales below cost. Although allocating fixed costs to products is notoriously difficult and standard economic theory shows that firms may rationally sell below average total cost (but above average variable cost), a price below average total cost is considered dumping.Blonigen and Prusa (2016) explain that although the U.S. Antidumping Act of 1916 was originally designed to protect domestic producers from predatory pricing, the required predatory intent was soon dropped from the law, and it has since become an ordinary protection tool.

While the U.S. International Trade Commission (USITC) may solicit downstream purchaser information during its investigations, AD/CVD laws do not allow the USITC to consider the economic effects of exporters’ behavior on downstream purchasers or on the national interest (U.S. Government Accountability Office 2022). In practice, the Department of Commerce determines whether the product is sold at less than fair value, and the USITC determines whether a domestic industry is materially injured by reason of the imports. In making the injury determination, the USITC cannot take into account any potential harm that AD/CVD duties may impose on downstream industries or consumers. This constraint lies at the core of the tension between trade law and competition law. While federal agencies, including the Department of Justice (DOJ) and the Federal Trade Commission (FTC), can submit statements of interest in AD/CVD cases, the DOJ has done so only once and promptly withdrew its statement. No other federal agency has submitted a statement in recent decades (U.S. Government Accountability Office 2022). WTO rules do not impose this constraint. Australia, Brazil, Canada, and the European Union all have some form of public-interest provisions that can lower or eliminate duties if doing so benefits downstream users.

AD is popular among domestic petitioners for several reasons. First, AD is a particularly effective instrument against import competition because it discourages exporters from competing aggressively: the lower the exporter’s price, the more likely a domestic rival can establish that the product is sold at less than fair value. Since the tariff rate increases with the exporter’s productivity, Ruhl (2014) shows that AD is particularly distortionary. Second, investigations last at most 18 months and the clear criteria and quasi-judicial framework make them predictable and more insulated from political interference (Blonigen and Prusa 2016). Third, while AD duties require periodic review, many remain in effect for decades.

CVD measures address cases in which imports are found to benefit from foreign subsidies. As with AD, the imports must materially injure, or threaten to materially injure, a domestic industry. Although the trade practices targeted by AD and CVD differ, the procedures and underlying concerns are often similar, and petitioners frequently seek protection under both measures simultaneously (Liu 2026).

Unlike AD/CVD, global safeguards can be imposed on fairly traded imports from all countries if a domestic industry is found to be seriously injured by a surge in imports; they do not require evidence of dumping or foreign subsidization. In the U.S., they are imposed at the discretion of the President for an initial duration of up to four years. They are therefore also more subject to the political process.

AD/CVD petitions occur frequently. According to data compiled by Bown et al. (2025), between 1980 and 2024 the EU initiated 440 AD, 77 CVD, and 7 GS investigations, while India initiated 517, 24, and 48, respectively. U.S. authorities initiated 747 AD, 453 CVD, and 15 GS investigations during the same time period. Between 2011 and 2021, 74 percent of AD/CVD petitions in the U.S. resulted in orders (U.S. Government Accountability Office 2022). At the same time, preparing a petition is costly, requiring legal counsel, expert economic analysis, participation in administrative hearings, as well as periodic sunset reviews. Practitioners interviewed by Liu (2026) estimate that the cost of a simple AD/CVD petition ranges between $1 million and $3 million, and can be substantially higher for complex cases involving multiple products or origin countries.

Petitions for trade remedies are typically initiated by domestic firms or industry associations that claim injury from foreign competition. Liu (2026) reports 789 petitioner appearances in AD/CVD cases between 2002 and 2024, representing 528 unique entities. Of these, 14 were labor unions, 217 were trade associations or coalitions, and the remaining 297 (56%) were individual domestic producers. In many cases, only a subset of domestic producers participate in a petition. Although 55 percent of petitioners are connected to the steel or chemicals industries, petitions arise in many tradable-goods sectors.

While any domestic producer may petition for trade remedies, smaller producers often cannot meet the statutory industry-support thresholds required for filing and therefore cannot petition alone. According to the WTO Antidumping Agreement’s industry-support thresholds, a petition is deemed “on behalf of the industry” if its supporters account for at least 25 percent of total domestic production of the domestic like product and more than 50 percent of the production of those expressing a view. When these thresholds are met, Commerce may initiate without polling producers, thereby reducing procedural frictions and the risk of standing challenges (United States Code 2025b, 2025a; Code of Federal Regulations 2025; U.S. International Trade Commission 2015).

To understand whether mergers are associated with trade-remedy petitions, I assemble a non-representative sample of nine horizontal mergers affecting 16 antitrust markets, where the merging parties have domestic production capacity in the relevant antitrust markets and where AD/CVD petitions were filed in at least one antitrust market within five years post-merger.The mergers span aluminium extrusions in Australia; thermoplastic resins in Brazil; oil country tubular goods in Canada and the United States; continuous filament glass fibre, stainless steel, and graphic paper in the European Union; flat-rolled carbon steel in India; and appliances in the United States.

A more systematic descriptive analysis is constrained by data availability. While data on AD/CVD petitions is publicly available and aggregated by Bown et al. (2025), information about market shares and the number of domestic producers at the antitrust-market level is generally not disclosed. For mergers that competition authorities scrutinize more heavily, redacted versions of this information sometimes appear in the published decisions.This is especially true for the European Commission, which publishes detailed merger decisions that include lengthy discussions of market definition for Phase II merger investigations. Crucially, the same producer-share information is not available for the “control markets” needed to construct a comparison group.Section 4.3 takes an alternative approach, exploiting variation across U.S. appliance markets where some experience a fall in the number of domestic producers and others do not.

In many of these cases, the clearance decision explicitly cites import competition as a competitive constraint on the merging parties. Two of the sample cases illustrate the pattern. In its Owens Corning / Saint Gobain Vetrotex conditional approval, the European Commission cites customer reports claiming that imports are a viable alternative to domestic producers for certain types of rovings (European Commission 2007). Two years after the merger, a group of domestic producers, including Owens Corning, filed an AD petition on continuous glass fibre from China, which includes rovings. In the United States, the DOJ cleared the Whirlpool–Maytag merger on the grounds that foreign manufacturers posed a sufficiently large competitive constraint to prevent post-merger price increases (Department of Justice 2006). Once these foreign manufacturers gained significant market shares, Whirlpool filed for AD and CVD on large residential washers from Korea and Mexico.

While AD/CVD are sometimes discussed in merger cases, this is usually in the context of how they shape the competitive environment. For example, in its ArcelorMittal / Ilva decision, the European Commission writes “As the trade defence measures on certain [...] products cover some of the major steel producing and steel exporting countries, any assessment of the extent to which imports of HR, CR and HDG products may exert competitive pressure on EEA-based flat carbon steel producers and, in particular the merged entity post-Transaction, must be made in light of the situation as restored by anti-dumping duties” (European Commission 2018, 56). The Commission thus treats the trade-defense environment as an exogenous input to merger analysis, rather than considering how the merger itself may alter the incentives to petition for new measures.

Figure 1 plots the number of domestic producers that together account for at least \(80\%\) of domestic production (\(n^{80}_{jt}\)), as well as the average within-market deviation of the number of active AD/CVD cases from its pre-merger (\(t=-1\)) baseline, for the nine horizontal mergers across 16 markets.Appendix 9 reports the per-case concentration changes, restriction counts, and source documentation. It shows that while the cross-case mean number of major domestic producers decreases post-merger, the count of active AD/CVD investigations increases.

Although this relationship is purely descriptive and could be caused by other market trends affecting concentration and AD/CVD petitions, it shows that in some mergers that are heavily scrutinized by antitrust authorities, increases in domestic production concentration are followed by increases in trade-remedy petitions.

In summary, AD/CVD cases are pervasive across the economy. The administrative structure of the process makes petition outcomes predictable, and success rates are higher when the foreign competitor prices aggressively. Petitions are costly and are often filed by a single domestic firm. Filing a petition therefore requires the petitioner to expect sufficiently large increases in profits to offset the cost of petitioning. Finally, mergers between domestic producers are sometimes followed by trade-remedy petitions. The remainder of the paper develops a model of merger-induced petitioning informed by these facts, estimates it on the Whirlpool–Maytag case, and quantifies the resulting consumer-welfare consequences.

Stylized Model

I specify a stylized model illustrating the link between mergers and the demand for protectionism.

Setup

Consider a market with three firms. Firm 1 is a domestically producing incumbent, firm 2 is a domestic acquisition target, and firm 3 is a foreign rival. Each firm produces a single, horizontally differentiated product and sells exclusively in the domestic market. Demand is generated by a unit mass of consumers with heterogeneous preferences following a standard logit specification.

Each consumer derives utility from purchasing a single product or an outside good. Products differ in product-specific deterministic utility \(\delta_j\). The utility that a consumer obtains from product \(j \in \{1,2,3\}\) and from the outside option is \[\begin{equation} \tag{1} U_j = \delta_j - \alpha p_j + \varepsilon_j, \qquad U_0 = \varepsilon_0, \label{eq:sty_utility} \end{equation}\] where \(p_j\) denotes the price of product \(j\), \(\alpha > 0\) governs price sensitivity, and \(\varepsilon_j,\varepsilon_0\) are i.i.d. Type I extreme value.

Marginal cost depends on the location of production. Marginal cost is \(c_D\) for domestic production and \(c_F\) for foreign production, with \(c_D>c_F\). Initially, firms 1 and 2 produce domestically at \(c_D\), and firm 3 produces abroad at \(c_F\). Relocation is a firm-level decision: firm \(f\) can relocate all of its production at a fixed cost \(R_f>0\), paid once regardless of the number of products it relocates. \(R_2\) is assumed to be high enough such that firm 2 never relocates; \(R_1\) is finite. If firm 1 acquires firm 2, it controls both products and, paying \(R_1\) once, may relocate any subset.

Firm 1 may petition for tariffs on all imports at petitioning cost \(L>0\). If filed, the tariff is imposed with certainty. The level of the ad valorem tariff \(\kappa>0\) is exogenously determined by the trade commission and scales foreign marginal costs to \((1+\kappa)c_F\).

Stage 1: merger control

The domestic incumbent (firm 1) proposes a horizontal merger; let \(\mathcal{M}\) denote the proposed configuration. The competition authority (CA) applies a policy rule to decide whether to clear or challenge. The CA’s baseline rule clears the merger whenever the predicted change in consumer surplus from unilateral market-power effects, \(\Delta^{MP} CS(\mathcal{M})\), exceeds a policy threshold \(\bar{\Delta}\le 0\), the largest tolerated reduction in consumer surplus: \[\begin{equation} \tag{2} \Delta^{MP} CS(\mathcal{M}) \;\ge\; \bar{\Delta}. \label{eq:ca_rule} \end{equation}\] In settings where horizontal mergers can alter firms’ demand for trade protection, which translates into consumer surplus through tariffs and prices, there is an additional trade-policy channel of consumer surplus change, \(\Delta^{TP} CS(\mathcal{M})\), computed from the equilibrium petitioning, offshoring, and pricing decisions specified in Stages 2 and 3 below. If the CA evaluates mergers on consumer welfare regardless of channel, it should apply the threshold to the total effect, \(\Delta^{MP} CS(\mathcal{M}) + \Delta^{TP} CS(\mathcal{M})\).

Stage 2: petitioning and offshoring

Following the merger control decision, firm 1 decides whether to offshore production, petition for tariffs, or maintain its current production structure without petitioning. It chooses the option that maximizes static profits net of fixed costs. Firm 1 never optimally chooses to petition and offshore, because petitioning while offshoring would raise firm 1’s own costs through the tariff.

If the merger is cleared, firm 1 prices products 1 and 2 jointly and internalizes relocation and tariff effects across both products. If the merger is blocked, firm 2 remains a separate domestic single-product firm with cost \(c_D\) and never relocates (since \(R_2\) is sufficiently high).

Stage 3: pricing and demand

In the third stage, firms simultaneously choose prices in a Bertrand-Nash equilibrium and consumers make discrete purchase decisions.

Let \(s_j(\mathbf{p})\) denote the logit market share of product \(j\) implied by the utility specification above; \(s_0(\mathbf{p})\) is the outside share. Given realized marginal costs \(c_j \in \{c_D,c_F,(1+\kappa)c_F\}\) from Stage 2, each firm \(f\) chooses prices to maximize \(\sum_{j\in \mathcal{J}_f}(p_j-c_j)s_j(\mathbf{p})\), where \(\mathcal{J}_f\) is the firm’s product set. For a single-product firm \(j\), the standard logit FOC gives \[\begin{equation} \tag{3} p_j=c_j+\frac{1}{\alpha(1-s_j)}. \label{eq:sty_foc} \end{equation}\] If firm 1 is multiproduct post-merger, prices solve the following system of first-order conditions \[\begin{equation} \tag{4} \mathbf{p}=\mathbf{c}-\left(\frac{\partial \mathbf{s}}{\partial \mathbf{p}}\circ\Omega\right)^{-1}\mathbf{s}(\mathbf{p}),\qquad \frac{\partial s_j}{\partial p_k}=\begin{cases} -\alpha s_j (1-s_j),& j=k,\\ \alpha s_j s_k,& j\neq k, \end{cases} \label{eq:sty_pricing} \end{equation}\] where \(\mathbf{p}\), \(\mathbf{c}\), and \(\mathbf{s}(\mathbf{p})\) are the stacked vectors of prices, costs, and shares, \(\partial \mathbf{s}/\partial \mathbf{p}\) is the matrix of own- and cross-price share derivatives, \(\Omega\) is the ownership matrix with \(\Omega_{jk}=1\) if the same firm owns products \(j\) and \(k\), and \(\circ\) denotes the Hadamard (element-wise) product.

The consumer surplus change between two equilibria is the compensating variation (Small and Rosen 1981): \[\begin{equation} \tag{5} \Delta CS=\frac{1}{\alpha}\Big[\log\big(1+\textstyle\sum_{j}\exp(\delta_j-\alpha p_j^{\text{after}})\big)-\log\big(1+\textstyle\sum_{j}\exp(\delta_j-\alpha p_j^{\text{before}})\big)\Big]. \label{eq:sty_cs} \end{equation}\]

Mergers, offshoring, and trade protection

I now analyze the firms’ strategic choices in light of the merger decision and the availability of trade-policy instruments. I focus on how a merger between firms 1 and 2 affects firm 1’s incentive to offshore production versus petition for trade protection, and how these choices interact with market structure and consumer welfare. Proofs can be found in Appendix 10.

Let \(\pi^{r}_{f,\mathcal{C}}\) denote firm \(f\)’s variable profit in regime \(r \in \{\text{off},\text{pet},\text{sq}\}\) chosen by firm 1, under merger configuration \(\mathcal{C} \in \{\mathcal{S},\mathcal{M}\}\). \(\Pi^{r}_{f,\mathcal{C}}\) denotes total profit, net of the relevant fixed cost. When the merger is cleared, firm 1 owns products 1 and 2; when blocked, it owns only product 1. For the merged entity I drop the firm subscript and write \(\pi^{r}_{\mathcal{M}}\) and \(\Pi^{r}_{\mathcal{M}}\).

A higher tariff raises foreign costs only, shifts shares toward firm 1, and increases its markups; offshoring leaves foreign costs unchanged and lowers firm 1’s own costs. There is a unique \(\kappa^*\) at which firm 1 is indifferent, above which it petitions and below which it offshores. The petitioning payoff is monotonically increasing in \(\kappa\) while the offshoring payoff is independent of \(\kappa\), giving a unique crossing \(\kappa^*\).

To see how this threshold \(\kappa^*\) evolves with the competitiveness of the foreign rival, comparative statics of \(\kappa^*\) in \(\delta_3\) are most transparent in a two-product reduction (firms 1 and 3 only).

As \(\delta_3\) rises, the foreign product becomes more appealing. The relative effect on firm 1’s profits differs by regime. Under petitioning, the tariff has two opposing effects: by handicapping the foreign rival it shrinks the rival’s share and insulates firm 1, but by making firm 1 the dominant domestic producer it widens firm 1’s markup and leaves more profit exposed to a more appealing rival. When the duty is large enough to nearly eliminate the rival, the first effect dominates and petitioning insulates firm 1 more than offshoring, so \(\kappa^*(\delta_3)\) falls; when the foreign rival remains strong even under the duty, the second effect dominates and offshoring insulates more, so \(\kappa^*(\delta_3)\) rises.This is not true for any demand system. Under CES demand with monopolistic competition (where each firm takes the price index as given), markups are constant and the price of firm 1 does not react to the quality of product 3. In a finite-firm CES oligopoly, markups depend on market shares and some strategic interaction is restored, but it remains weaker than under logit demand.

I first compare firm 1’s gain from petitioning for a given \(\kappa\), with and without the merger, against a status-quo baseline.

The appropriation effect captures the fact that the merger internalizes a petitioning externality: firm 1 can now appropriate the rents from tariff protection that would otherwise accrue to firm 2. The strategic effect captures the fact that the merger raises how much firms 1 and 2 jointly gain from tariffs. Intuitively, the merged firm has already internalized domestic competition, so the foreign rival constitutes a proportionally larger part of its competitive environment; weakening the foreign firm via the tariff therefore benefits the merged entity more than the standalone firms. The aggregative-games framework of Nocke and Schutz (2018, 2025) underpins this result.

While I rely on logit demand, the result that the merger increases firm 1’s gains from tariffs is true under mild assumptions on demand (downward-sloping demand, substitutability across products, and standard regularity conditions). In particular, while the strategic effect is zero under CES demand with monopolistic competition, the appropriation effect, and thus the overall effect, remain positive.

Next, I switch the baseline from status quo to offshoring.

The strategic effect reflects two competing forces. First, by Proposition 3, the tariff raises the merged entity’s profit by strictly more than the sum of standalone profits. So the tariff gain is larger with the merger, and increasingly so as \(\kappa\) rises. Second, offshoring lowers costs, and the merger uniquely enables offshoring of product 2. So the offshoring gain is also larger with the merger, independently of \(\kappa\). The strategic effect is the difference between these two forces. When the cost gap \(c_D - c_F\) is small, the threshold \(\kappa^{SE}\) is low and even moderate tariffs make the strategic effect positive. When the offshoring option is sufficiently valuable, the tariff gain falls short at every duty level and the strategic effect is negative throughout: protection, however high, cannot compensate the merged firm for the cost savings it forgoes.

Even if a merger makes petitioning relatively more attractive (i.e., decreases \(\kappa^*\)), firm 1 will not petition for tariffs if the tariff rate \(\kappa\) that the trade commission sets in the event of a petition is below the post-merger \(\kappa^*\).

Finally, I consider how the consumer harm from a given tariff changes with a domestic merger.

The marginal consumer harm from the tariff is affected by the merger through two channels. The merger raises the foreign share \(s_3\) (higher domestic prices shift demand to the foreign product). The merger also concentrates domestic production into a single firm. Pooling the domestic firms into one raises the strategic-complementarity multiplier that grows with each firm’s market share. The merger increases consumer harm regardless of \(s_3\).

Foreign-rival response in a two-period model

To assess the robustness of these results, I consider a two-period extension in which firm 3 can respond to the tariff by relocating production to the domestic market. Firm 1 maximizes the discounted sum of profits \(\Pi_{1} = \Pi_1^{t=1} + \beta \Pi_1^{t=2}\), where \(\beta \in [0,1]\).

In the first period, marginal costs are determined as in the static model. In the second period, if a tariff is in place, firm 3 may choose to pay a fixed cost \(R_3\) to relocate. A necessary condition for firm 3 to relocate to \(D\) is that the tariff is high enough that \(c_D < (1+\kappa) c_F\). I focus on the case where \(R_3\) is small enough that tariff jumping is firm 3’s best response.

Tariff jumping strictly reduces the incentive to petition relative to offshoring: it erodes the future rents from protection while leaving the offshoring payoff unchanged. However, petitioning can remain optimal even with perfect patience (\(\beta = 1\)) and certainty of tariff jumping in \(t=2\), provided the short-run windfall from the tariff is sufficiently large. The appropriation and strategic effects from Propositions 3 and 4 operate in full during the period of tariff protection.

In the second period, the comparison across merger regimes is more complex: tariff jumping equalizes the foreign rival’s costs, but the merger also changes the offshoring counterfactual by enabling relocation of product 2 and alters equilibrium pricing through joint ownership. The net second-period effect is parameter-dependent, but the first-period mechanism through which the merger increases the incentive to petition remains intact.

Cross-border mergers

The previous results highlighted how a merger between domestic producers can create demand for protectionist policies and magnify their harm to consumers. A natural question is whether cross-border mergers generate the same forces.

To analyze this, I extend the setup to four firms. Firm 4 is a second foreign producer with marginal cost \(c_F\), acquired by firm 1 in a cross-border merger \(\mathcal{X} = \{1,4\}\). Firms 2 and 3 remain independent. I impose the following institutional constraint: if the merged entity petitions for tariffs on foreign imports, it must relocate product 4’s production from \(F\) to \(D\) at cost \(R_1\), raising its marginal cost from \(c_F\) to \(c_D\). The rationale is that a firm cannot credibly petition for duties on foreign-produced goods while itself importing the same goods, since AD/CVD standing requires the petitioner to represent domestic production. Offshoring remains available only for product 1 (product 4 is already produced abroad), again at cost \(R_1\).

The contrast with the domestic merger under the status quo baseline (Proposition 3) is stark. In the domestic case, both the appropriation and strategic effects are strictly positive. The merger internalizes a positive externality, since the domestic target benefits from the tariff. In the cross-border case, the target is a victim of the tariff. Internalizing its profit change means internalizing a loss. The appropriation effect flips sign, and the merged entity must additionally bear the relocation cost \(R_1\) to petition credibly. The strategic effect in the cross-border case conflates two channels — joint pricing under the merger and the variable cost restructuring from relocating product 4 — so it does not isolate the pure joint-pricing surplus as in Proposition 3, and its sign can go either way.

When the baseline is offshoring instead of status quo, the same decomposition structure applies. The appropriation term becomes \(\pi^{\text{pet}}_{4,\mathcal{S}} - \pi^{\text{off}}_{4,\mathcal{S}}\), which is ambiguous in sign: when \(\kappa\) is small, firm 4 may prefer the petitioning world (firm 1 remains a weak competitor at \(c_D\)); when \(\kappa\) is large, the direct cost penalty on firm 4 dominates. The strategic effect is also ambiguous in sign, and the relocation cost \(R_1\) is still strictly positive. By contrast, in the domestic merger (Proposition 4), the appropriation effect is strictly positive.

While the overall effect of a cross-border merger on petitioning incentives remains parameter-dependent, the decompositions suggest that cross-border mergers are less likely to increase the demand for protectionism. I quantify these channels for a specific empirical application in Section 7.

Institutional Setting, Data, and Descriptive Evidence

To study the interplay between mergers, concentration, and the demand for trade protection in a concrete setting, and to illustrate how the trade-policy channel of mergers could be incorporated into merger policy, I focus on the household appliance industry.

Household Appliance Industry

In 2000, the import share for most major appliances in the EU and U.S. was below 10 percent. By 2018, it exceeded 30 percent for most categories and approached 50 percent for some, such as clothes washers, dryers, and refrigerators.

European manufacturers such as BSH and Electrolux had established a presence in the U.S. by the 1990s, and U.S. firms such as Whirlpool were similarly active in Europe. However, these firms produced locally rather than exporting across regions. The U.S. market saw new entry from LG and Samsung in the mid-2000s, and from Haier, which first attempted to acquire Maytag in 2005 and later entered successfully by acquiring GE Appliances in 2016. European markets experienced a similar pattern, with entry from Arçelik and Vestel (Turkey), followed by LG and Samsung (Korea), and later Haier and Hisense (China).

In 2006, Whirlpool, the leading U.S. appliance manufacturer, acquired Maytag, its main domestically producing rival. The Department of Justice cleared the merger on the grounds that foreign manufacturers posed a sufficiently large competitive constraint to prevent post-merger price increases (Department of Justice 2006).

The rise in import share reflects both foreign entrants producing abroad and domestic incumbents shifting production overseas. Some incumbents offshore part of their previously domestic output; others relocate all of it.

Although product market concentration increased modestly across most markets, the key variation between markets lies in the decline of major domestic producers. By 2015, the U.S. market for clothes washers and bottom-mount refrigerators had only two domestic producers remaining, compared to at least four for EU washers and U.S. dishwashers. The markets with few domestic producers are also those where Whirlpool filed for AD/CVD.

While the U.S. petition for bottom-mount refrigerators was ultimately unsuccessful,See U.S. International Trade Commission, Investigation Nos. 701-TA-477 and 731-TA-1180-1181, 2012. large residential washers (LRWs) were subject to multiple rounds of tariffs. Preliminary antidumping duties were first imposed on imports from Korea and Mexico in 2012, followed by antidumping duties on washers produced in China in 2016, and culminating in a global safeguard in 2018 (Flaaen et al. 2020).

Data

The primary data source is the TraQline household survey, described in detail in Montag (2026). TraQline surveys approximately 600,000 U.S. households annually on major appliance purchases, including product characteristics, prices, second-choice brands, the retailer, and household demographics. I observe survey responses for 2005–2015. The product scope includes refrigerators, dishwashers, clothes washers, dryers, and freestanding ranges. I define products as brand-retailer-characteristic combinations, using brand identity and retailer as proxies for unobserved differentiation. I define a market as a product category in a given country or region, for example U.S. clothes washers.

For the descriptive analysis, I extend the market share series through 2023, using OpenBrand data provided by Dewey Data (OpenBrand 2022).TraQline has been part of OpenBrand since 2024. This extended dataset does not include non-price product characteristics other than brand, so the structural analysis cannot be extended beyond 2015.

To measure product market concentration in the European washer industry, I use washing machine sales for most European countries between 2000 and 2018 from Gesellschaft für Konsumforschung.

To measure the number of major domestic producers by market and year, I combine production data from Appliance Magazine, Euromonitor, and hand-collected information on production locations for the years 2000 through 2023, subject to availability. For U.S. clothes washers in 2005–2015, I use hand-collected production location data from Montag (2026).

Finally, I compute import shares for each market and year using trade data from the USITC and COMTRADE.

Descriptive analysis

The stylized model in Section 3 suggests that mergers between domestic producers are particularly likely to generate demand for trade protection, implying that petitions should be more common in markets with few domestic producers.

To examine which market characteristics are associated with domestic producers petitioning for trade protection, I estimate a linear probability model at the market-year level for the household appliance industry. The dependent variable is an indicator for whether a petition for AD/CVD or global safeguards was filed in a given market and year. Whirlpool filed AD/CVD petitions on bottom-mount refrigerator imports into the U.S. from Korea and Mexico on March 30, 2011 and on large residential washers into the U.S. from Korea and Mexico on December 30, 2011.The results are insensitive to moving the filing date of the washer petition from 2011 to 2012. The analysis is descriptive and does not claim to identify causal effects.

The linear probability model relates petition incidence to three key market-level variables: the import share, the degree of domestic product market concentration, and the number of domestic producers. I estimate the parameters of the following specification: \[\begin{equation} \tag{12} \mathbb{1}\{ \text{Petition}_{it} \} = \beta_1 \, \text{impshare}_{it} + \beta_2 \, \text{prodmkthhi}_{it} + \beta_3 \, \text{domprod}_{it} + \delta_i + \gamma_t + \varepsilon_{it}, \label{eq:lpm} \end{equation}\] where \(i\) indexes product markets and \(t\) denotes years. The model includes market fixed effects \(\delta_i\) and year fixed effects \(\gamma_t\).

The outcome \(\mathbb{1}\{ \text{Petition}_{it} \}\) is an indicator for whether a trade remedy petition was filed in market \(i\) in year \(t\). The variable \(\text{impshare}_{it}\) measures the import share in the market, \(\text{prodmkthhi}_{it}\) is the Herfindahl-Hirschman Index based on each producer’s (foreign and domestic) sales share in the domestic market, and \(\text{domprod}_{it}\) is the number of domestic producers or an indicator equal to one if there are more than two domestic producers.

The data span the years 2000 through 2023 and include five product markets: U.S. clothes washers, U.S. clothes dryers, U.S. dishwashers, U.S. bottom-mount refrigerators, and EU clothes washers. Each of these markets is observed at annual frequency; however, data are not available for all product markets in all years, resulting in an unbalanced panel.

The descriptive estimates of equation (12) in Table 1 indicate that higher product market concentration, as measured by the HHI, is not positively associated with petition filing. If anything, the association is negative or indistinguishable from zero. In contrast, the number of domestic producers is strongly and negatively associated with the likelihood of a petition. In particular, markets with two or fewer domestic producers are substantially more likely to see a filing. This pattern may reflect that petitions typically arise only after most domestic competitors have already exited. Alternatively, it may indicate that petitioners expect greater benefit from trade protection when fewer domestic firms remain to share the resulting market expansion. The following sections evaluate this second channel quantitatively in the case of the U.S. clothes washer market.

Empirical Model and Estimation

The stylized model in Section 3 showed that assessing whether a merger harms consumers through the trade-policy channel requires estimating the merging parties’ variable profits under different merger, tariff, and production-location scenarios. To estimate these objects, I specify a model of demand and supply tailored to the U.S. washer market.The empirical model closely follows Montag (2026).

Consumer Demand

Let utility for household \(i\) from purchasing product \(j\) be: \[\begin{equation} \tag{13} u_{ijt} \;=\; x_{jt}\beta \;+\; \sigma^{\text{FL}} \nu_i^{\text{FL}} x_{jt}^{\text{FL}} \;-\; \alpha_i\, p_{jt} \;+\; \xi_{jt} \;+\; \varepsilon_{ijt}, \qquad \alpha_i \equiv \exp(\alpha + \kappa_\alpha \iota_i), \label{eq:emp_utility} \end{equation}\] where \(x_{jt}\) is a vector of observed non-price characteristics, \(x_{jt}^{\text{FL}}\) is a front-loader indicator, \(\nu_i^{\text{FL}} \sim \mathcal{N}(0,1)\) is a random taste draw that captures heterogeneous preferences for front-loaders, \(\iota_i\) is income, and \(\varepsilon_{ijt}\) is an idiosyncratic shock drawn from a Type I extreme value distribution.

The utility of the outside good is normalized to zero. Each consumer purchases the single product, or the outside good, that yields the highest utility.

Given the distributional assumptions, the market share of product \(j\) is \[\begin{equation} \tag{14} s_{jt}(\mathbf p) \;=\; \int \frac{\exp\!\big(\delta_{jt} + \mu_{ijt}\big)} {\,1 + \sum_{k \in J} \exp\!\big(\delta_{kt} + \mu_{ikt}\big)} \, dP(\iota_i,\nu_i), \label{eq:emp_share} \end{equation}\] where \[\delta_{jt}=x_{jt}\beta+\xi_{jt}, \qquad \mu_{ijt}=\sigma^{\text{FL}} \nu_i^{\text{FL}} x_{jt}^{\text{FL}}-\alpha_i p_{jt}.\]

Demand is estimated by combining aggregate and household moments as outlined in Berry et al. (2004). I split the demand parameters into a linear component \(\theta_1\), the coefficients \(\beta\) on product characteristics, and a nonlinear component \(\theta_2 = (\sigma^{\text{FL}}, \alpha, \kappa_\alpha)\). The nonlinear parameters are estimated by the method of simulated moments using three moment conditions: two match household-level moments between the simulated and observed data, and the third is the orthogonality condition implied by excluding the real exchange rate, a cost shifter, from utility. With as many moments as parameters, \(\theta_2\) is exactly identified. Conditional on the estimate \(\hat\theta_2\), the linear parameters \(\theta_1\) are obtained by ordinary least squares. The data, estimation procedure, and moment conditions are the same as in Montag (2026). I refer readers to that paper for further details.

Marginal Costs and Pricing

Let \(j \in J_{ft}\) denote a product offered by firm \(f\) in market \(t\) with price \(p_{jt}\) and marginal cost \(mc_{jt}\). The firm’s variable profit is: \[\begin{equation} \tag{15} \pi_{ft} = \sum_{j \in J_{ft}} (p_{jt} - mc_{jt}) \cdot s_{jt}(\mathbf{p}) \cdot M_t, \label{eq:emp_profit} \end{equation}\] where \(s_{jt}(\mathbf{p})\) is the market share of product \(j\) as a function of all prices \(\mathbf{p}\), and \(M_t\) is market size.

Markups are pinned down by the derivatives of market shares with respect to prices. The Bertrand-Nash equilibrium prices solve: \[\begin{equation} \tag{16} \mathbf{p} = \mathbf{mc} - \left( \frac{\partial \mathbf{s}}{\partial \mathbf{p}} \circ \Omega \right)^{-1} \mathbf{s}, \label{eq:emp_pricing} \end{equation}\] where \(\Omega\) is the ownership matrix and \(\circ\) denotes the Hadamard product.

Let \(c(j)\) denote the country of origin of product \(j\); baseline (tariff-exclusive) marginal cost is \[\begin{equation} \tag{17} mc_{jt} \;=\; \lambda_{1f(j)} r_{f(j)t} \;+\; \lambda_2 w_{c(j)t} \;+\; \lambda_{3j} m_t \;+\; \omega_{jt}, \label{eq:mc} \end{equation}\] where \(r_{f(j)t}\), \(w_{c(j)t}\), and \(m_t\) are input prices for capital, labor, and materials; and \(\omega_{jt}\) is a product-level marginal-cost shock realized after production and sourcing decisions. Offshoring modifies \(c(j)\) and thus affects the labor-cost component.

Alternatively, the incumbent may petition for an ad valorem tariff \(\kappa>0\) on imports from an origin set \(O\). Tariff-origin pairs are indexed by \((\kappa,O)\).I assume that petitions always lead to tariffs. This does not affect the sign of the comparison between merger and no-merger petitioning incentives: a common petition success probability \(\rho<1\) multiplies each firm’s expected petitioning premium by \(\rho\), while the filing cost \(L\) cancels in the merger-vs-no-merger difference. The decomposition into appropriation and strategic effects is therefore preserved, with both terms scaled by \(\rho\). Tariffs modify marginal costs multiplicatively: \[\begin{equation} \tag{18} mc_{jt}^{(\kappa,O)} \;=\; \bigl[1 + \kappa \cdot \mathbb{1}\{c(j)\in O\}\bigr]\; mc_{jt}. \label{eq:mc_tariff} \end{equation}\]

Marginal costs are recovered by inverting firms’ first-order pricing conditions using observed market shares and prices. To estimate how marginal costs change with input costs, I estimate: \[\begin{equation} \tag{19} \displaystyle mc_{jt} = FE_f + \gamma_1 RER_{c(j)t} + \gamma_2 x_{j} +\omega_{jt} \,. \label{eq:mc_reg} \end{equation}\]

Firm fixed effects \(FE_f\) capture differences in capital intensity across firms. The real exchange rate \(RER_{c(j)t}\) is a product-level cost shifter capturing local wage and nominal exchange rate fluctuations. The non-price characteristics \(x_j\) capture material cost differences across products, while \(\omega_{jt}\) denotes transitory marginal-cost shocks.

Trade-Policy Channel of Mergers

With the tools to estimate firm profits under different merger, offshoring, and tariff scenarios, I can now connect the empirical model to the propositions in Section 3. The model shows that assessing how a merger changes the merging parties’ incentives to petition for tariffs requires estimating the appropriation and strategic effects in equations (7) and (9).

Let \(\pi_{f,t,\mathcal{S}}(\kappa,O;\ell)\) denote the variable profits of firm \(f\) in year \(t\) under separate ownership (\(\mathcal{S}\)), and \(\pi_{t,\mathcal{M}}(\kappa,O;\ell)\) the variable profits of the merged entity (\(\mathcal{M}\)), which owns both merging firms’ products. Profits are evaluated at tariff-origin pair \((\kappa,O)\) and pre-petition production-location regime \(\ell\in\{\text{off},\text{sq},\text{dom}\}\) (offshore, status quo, domestic). When evaluating profits under a tariff petition, I take the with-tariff regime for the incumbent to be domestic production, i.e., \(\ell=\text{dom}\). For simplicity, in the remainder I denote the acquirer as \(f=1\) and the acquisition target as \(f=2\).

Then the appropriation effect of a merger can be written as \[\begin{equation} \tag{20} \text{Appropriation}_t(\kappa, O, \ell) = \pi_{2,t,\mathcal{S}}(\kappa,O;\text{dom}) - \pi_{2,t,\mathcal{S}}(0,O;\ell) \label{eq:emp_appropriation} \end{equation}\]

and the strategic effect can be written as \[\begin{equation} \tag{21} \begin{split} \text{Strategic}_t(\kappa, O, \ell) = \Bigl[ & \pi_{t,\mathcal{M}}(\kappa,O;\text{dom}) - \pi_{1,t,\mathcal{S}}(\kappa,O;\text{dom}) - \pi_{2,t,\mathcal{S}}(\kappa,O;\text{dom})\Bigr] \\ -\; \Bigl[ & \pi_{t,\mathcal{M}}(0,O;\ell) - \pi_{1,t,\mathcal{S}}(0,O;\ell) - \pi_{2,t,\mathcal{S}}(0,O;\ell)\Bigr] \,. \end{split} \label{eq:emp_strategic} \end{equation}\]

Finally, I quantify the consumer-surplus effect of a given tariff. For a tariff-origin pair \((\kappa,O)\), the change in consumer surplus under ownership structure \(m\in\{\mathcal M,\mathcal S\}\) is (Small and Rosen 1981) \[\begin{equation} \tag{22} CS^{m}(\kappa,O) \;=\; \int \frac{1}{\alpha_i} \Bigg[ \ln\!\Big(\sum_{j=0}^J e^{V_{ij}^{(\kappa,O;\,m)}}\Big) -\ln\!\Big(\sum_{j=0}^J e^{V_{ij}^{(0,O;\,m)}}\Big) \Bigg]\, dP(\iota_i,\nu_i), \label{eq:emp_cs} \end{equation}\] where \(V_{ij}^{(\kappa,O;\,m)}=\delta_{jt}+\mu_{ijt}^{(\kappa,O;\,m)}\) uses the equilibrium prices implied by \(m\) and \((\kappa,O)\).

For a given tariff-origin pair \((\kappa,O)\), the merger-induced change in consumer surplus from a tariff is \[\begin{equation} \tag{23} \Delta CS^{\mathcal M}(\kappa,O) \;:=\; CS^{\mathcal M}(\kappa,O)\;-\;CS^{\mathcal S}(\kappa,O), \label{eq:emp_dcs} \end{equation}\] so \(\Delta CS^{\mathcal M}(\kappa,O)<0\) indicates that the merger amplifies the consumer harm from a tariff.

Parameter Estimates

Table 2 summarizes the demand estimates. Column (1) shows that the real exchange rate is a strong instrument for price. The estimates in Column (4) imply that higher-income households are less price-sensitive (the income coefficient \(\kappa_\alpha\) is negative) and that households differ in their taste for front-loaders (the unobserved-taste dispersion \(\sigma^{\text{FL}}\) is positive and significant). The average own-price elasticity in the full mixed-logit model is \(-2.54\) at the product level.

Figure 2 displays the distribution of estimated marginal costs across all products.

Notes: Histogram of estimated marginal costs (deflated to 2012 dollars) across all products in the sample.

Finally, Table 3 quantifies how marginal costs depend on labor costs (captured by the deflated RER), product characteristics, and brand, retailer, and year fixed effects. As labor costs increase, the estimated marginal cost increases. Furthermore, top-loaders with an agitator have lower marginal costs than high-efficiency top-loaders and front-loaders.

Quantifying the Trade-Policy Channel for Whirlpool

In this section, I quantify the components of the trade-policy channel for Whirlpool’s domestic acquisition of Maytag, assessing how the merger affected the profitability of petitioning across the different rounds of tariff actions observed between 2011 and 2018. To contrast this with the trade-policy channel of a cross-border merger, I perform the same analysis for a hypothetical merger between Whirlpool and LG.

Trade-Policy Channel of a Domestic Merger

To assess how acquiring Maytag affected Whirlpool’s incentives to petition for tariffs, I quantify the appropriation and strategic effects (equations (20) and (21)) for different tariff-origin scenarios. I also estimate the corresponding consumer surplus effect \(\Delta CS^{\mathcal M}(\kappa,O)\) defined in equation (23).

While I observe realized tariff outcomes and relocation responses by LG and Samsung, the simulations do not incorporate this ex post information. Ex ante, petitioners cannot perfectly predict final tariffs or rivals’ immediate relocation strategies; for instance, preliminary AD margins on LRW imports from China were substantially revised downward between preliminary and final determinations (LG: \(49.88\%\rightarrow 32.12\%\); Samsung: \(111.09\%\rightarrow 52.51\%\)). I instead simulate uniform ad valorem tariffs of \(\kappa=50\%\) applied to three origin groups, \[O \in \{\text{Korea+Mexico},\ \text{China+Korea+Mexico},\ \text{Global}\},\] which mirror the historical sequence (2011 Korea and Mexico; 2015 China; 2018 global safeguards).AD/CVD petitions on imports from Korea and Mexico, as well as China, were filed in 2011 and 2015, respectively. Final duty determinations were made in 2013 and 2016, respectively.

For each calendar year \(t\), I recompute the Bertrand-Nash pricing equilibrium under explicit production-location assumptions. Under a tariff \((\kappa,O)\), Whirlpool and Maytag reshore any remaining foreign washer production to the U.S. in year \(t\), while all other firms’ production locations are held at their year-\(t\!-\!1\) configuration. This mirrors the domestic incumbents’ ex ante decision problem: rivals’ locations are expected to persist in the near term, and securing protection is anticipated to require reshoring by the petitioner.

I compare tariff scenarios to two no-tariff baselines. In the status-quo baseline, Maytag and Whirlpool’s production locations remain at where they were in year \(t\!-\!1\). In the incumbent-offshoring baseline, Whirlpool and Maytag additionally offshore front-loader production to Mexico in year \(t\) (top-loaders are not offshored).Top-loader offshoring is never observed in the data. All rivals’ production locations always remain at their locations in year \(t\!-\!1\).

Figure 3 reports the profitability of petitioning relative to its no-tariff alternative for Whirlpool in each year, with and without the Maytag acquisition, under two baselines. Panel A shows the premium of petitioning over the status quo, \(\pi^{\text{pet}}-\pi^{\text{sq}}\). It becomes immediately clear that if the alternative is the status quo, then petitioning is highly profitable in variable profit terms and the merger increases its profitability. However, since the status quo is dominated by offshoring, the relevant comparison to petitioning should also be offshoring.

Panel B shows the premium of petitioning over offshoring, \(\pi^{\text{pet}}-\pi^{\text{off}}\). Two patterns emerge. First, the premium rises as import competition intensifies, from about $20 million in 2005 to about $95 million in 2015 absent the merger. As the combined market share of LG and Samsung grows over the period, a tariff that raises foreign costs becomes more valuable relative to lowering Whirlpool’s own costs through offshoring. Second, the merger shifts the premium upward in every year. The gap between the two lines is the merger’s effect on the incentive to petition, which I decompose into an appropriation and a strategic effect below.

This comparison is in variable profits and abstracts from the fixed costs of each option. A \(50\%\) global tariff is a strong instrument, so the premium is positive throughout the sample. The figure therefore overstates how early petitioning becomes worthwhile. Petitioning requires paying the filing cost and reshoring production to establish standing, whereas offshoring, already under way by the late 2000s, entailed no further fixed cost. The duties actually sought in the first rounds were also smaller and narrower than a \(50\%\) global tariff. Net of these costs, the rising premium is consistent with Whirlpool turning to trade protection only as import competition strengthened, filing its first petition in 2011.

Figure 4 plots the appropriation and the strategic effect of acquiring Maytag for Whirlpool from a \(50\%\) global tariff on imports of large residential washers. The panels compare results against the status-quo baseline and the incumbent-offshoring baseline.

The merger increases the profitability of global import tariffs for Whirlpool. While both the appropriation and strategic effects increase the profitability of petitioning for trade protection, most of the trade-policy channel comes through the appropriation effect. The strategic effect is about five times smaller than the appropriation effect.

Figure 5 shows the merger-induced increase in consumer surplus losses from a \(50\%\) global tariff. The merger amplifies the consumer harm from tariffs. Depending on the year, the merger-induced increase in annual consumer harm from a global tariff exceeds $100 million under the status-quo baseline. These figures do not account for any change in the probability of petitioning.Results for tariffs on imports from Korea and Mexico or China, Korea and Mexico only are reported in Appendix 11.1.

Table 4 reports the estimates of the effect of introducing a \(50\%\) tariff for the different tariff scenarios in the year of filing (or the final year of the data in the case of global tariffs). The first column reports the tariff-induced change in profits for Whirlpool in the absence of the merger. The appropriation and strategic effects together give the merger-induced increase in Whirlpool’s profits from tariffs. In most scenarios, acquiring Maytag more than doubles Whirlpool’s profits from tariffs. This suggests that the merger substantially increased the likelihood of a tariff petition.

The final two columns report the consumer surplus loss from a \(50\%\) tariff without the merger and the additional loss attributable to the merger. Two observations stand out. First, the consumer surplus loss from tariffs is substantial. Even in the absence of the merger, a \(50\%\) global tariff on washers decreases consumer surplus by around $718 million if the alternative is offshoring. Second, the merger substantially increases the consumer harm from tariffs. For instance, in the global tariff scenario, the merger increases the consumer surplus loss from a \(50\%\) tariff by around $74 million if the alternative is offshoring.The total annual consumer surplus loss from the global import tariffs is of the same order of magnitude as the $841 million annual consumer surplus loss from the actual global tariffs implemented in 2018 estimated by Flaaen et al. (2020). They use a difference-in-differences design to estimate the price effect of the tariffs and multiply it by the total number of washing machines transacted pre-tariff. The environments differ in several respects: the actual tariff was \(20\%\) on the first 1.2 million washers and \(50\%\) only beyond that, and Samsung relocated production to the U.S. during their observation period.

The total effect of the trade-policy channel on consumers combines the harm induced by the increased likelihood of tariffs with the harm of the tariffs conditional on being imposed. Consider the 2015 global tariff relative to the offshoring baseline. The merger doubles Whirlpool’s profits from the tariff, making a petition and the subsequent tariff substantially more likely.

To illustrate the magnitude of this channel, suppose the merger makes a petition and the subsequent tariff a certainty. If this reflects a \(10\) percentage-point increase over the no-merger baseline, consumers face an extensive-margin expected loss of \(0.10\times\$718\text{M}\approx\$72\) million. Furthermore, because the post-merger tariff occurs with probability one under this assumption, consumers fully bear the $74 million in additional conditional harm. Adding these gives about $150 million per year in expected consumer harm through the trade-policy channel. For comparison, Montag (2026) estimates that the direct market-power related annual consumer harm from the Whirlpool–Maytag merger is $225 million for clothes washers. This shows that the magnitude of the trade-policy channel can be substantial.

Trade-Policy Channel of a Cross-Border Merger

Proposition 6 shows that in a cross-border merger, two of the three channels in the petitioning-premium decomposition unambiguously reduce the incentive to petition under the status-quo baseline, suggesting that cross-border mergers lower the demand for protectionism.

To illustrate this point, I repeat the previous simulations for a cross-border merger between Whirlpool and LG. To make this comparable to the domestic-merger simulations, I demerge all Maytag brands from Whirlpool across all years. I assume that Maytag always produces in the U.S., whereas in the absence of petitioning, Whirlpool and LG either keep their observed production locations or offshore wherever feasible. Since LG is only producing abroad throughout the sample and Whirlpool never produces top-loaders outside the U.S., the offshoring scenario simply moves Whirlpool’s front-loader production to Mexico.

Figure 6 repeats the petitioning-premium comparison of Figure 3 for the cross-border Whirlpool–LG merger. Whereas the domestic merger raises the premium, the cross-border merger lowers it and pushes it below zero, so the merged firm strictly prefers offshoring to petitioning. The benefits of petitioning are limited, since the tariff now protects the merged entity only against Samsung, while the costs are substantial. In particular, petitioning requires the merged Whirlpool–LG to relocate LG’s production to the U.S. and absorb the associated cost increase. Because the premium is measured in variable profits and omits this relocation cost, it understates how unattractive petitioning is for the merged firm.

Table 5 reports the corresponding point estimates and bootstrap confidence intervals for all three tariff scopes. Appendix Figure A.7 decomposes the merger’s effect into the appropriation effect, now negative, and the strategic effect, and the results for the Korea+Mexico and China+Korea+Mexico tariffs are reported in Appendix 11.2.

Figure 7 shows that the consumer harm from tariffs is consistently larger in the absence of a Whirlpool–LG merger. Although tariffs always reduce consumer welfare, it is a priori unclear whether the cross-border merger increases or decreases this harm. The results show that competition between Whirlpool and LG is particularly valuable if Whirlpool offshores its front-loader production and LG remains independent. The offshoring baseline tariffs are therefore more harmful without the merger.

Conclusion

This paper shows that domestic mergers can harm consumers through a trade-policy channel that operates over and above the merger’s direct market-power effects.

I develop a three-stage model in which a competition authority adjudicates a merger, the domestic incumbent chooses among petitioning for tariffs, offshoring, or maintaining the status quo, and firms compete in prices. When offshoring is unavailable, a domestic merger strictly raises the profitability of petitioning through two channels: an appropriation effect, whereby the acquirer internalizes the trade-protection rents that would otherwise accrue to the domestic target, and a strategic effect, whereby joint pricing amplifies the merged firm’s gains from weakening foreign rivals. When offshoring is available, the incumbent can raise profits by lowering its own costs rather than raising its rivals’, and the sign of the merger’s effect becomes ambiguous. Cross-border mergers generate the opposite forces: internalizing the foreign target’s losses from tariffs and bearing the cost of relocating its production reduce the incentive to petition.

Using the U.S. washer industry as a case study, I find that the choice between offshoring and petitioning turns on the strength of import competition. When it is weak, the merging parties prefer to lower their own costs through offshoring; when it is strong, the merger makes it more profitable to raise their foreign rivals’ costs through tariffs. This is exactly when imports are most likely to be a pivotal competitive constraint in a merger clearance decision. A hypothetical cross-border merger, by contrast, reduces the profitability of petitioning and does not increase consumer harm from tariffs.

These findings have implications for merger control. In markets where imports can be restricted through trade remedies, competition authorities should place less weight on import competition as a constraint on the merged entity, since the merger itself can undermine that constraint. Mergers that reduce the number of domestic producers to one or two warrant particular scrutiny, whereas cross-border mergers can deliver scale economies without raising the incumbents’ returns to tariff petitions.

References