Why Americans Voted for Trade Protectionism Again: A Review of Political-Economic Models

Karl Farmer

Department of Economics, University of Graz, Graz, Austria.

DOI: 10.4236/me.2025.164029                

Abstract

Since 2018, the USA has unilaterally imposed tariffs on imports of goods from abroad, especially from China. President Trump enacted since the inauguration for his second term further increases in tariffs, for imports from Mexico, Canada, China, und announced recently tariffs for imports from Europe. Why did a politician who explicitly advocated foreign trade protectionism become again president of the world’s most economically developed country? Why do US voters value the economic freedom of being able to buy cheaper foreign goods instead of more expensive domestic goods less than the rather dubious promise of making domestic industry great again? The paper answers these questions based on a review of political-economic models published since Trump’s first term as follows: Foreign trade protectionism has once again become politically powerful in the USA because, within the framework of international supply policy, US globalization losers could not be compensated, white lower middle-class men in the old industrial areas of the USA were tormented by status anxiety towards people of different colors and foreigners, about half of the US electorate no longer identified with the nation as a whole but only with their own social class, and after China joined the WTO, voters and party representatives who were in favor of free trade became nationalists and protectionist social conservatives. According to an economically liberal interpretation of natural law, however, free international trade does not contradict social conservatism.

Keywords

Protectionism, Social Identity, Identity Politics, Social Conservatism

1. Introduction

In 2018, during his first term, US President Donald J. Trump unilaterally imposed import tariffs on steel, aluminum, and a wide range of Chinese goods to protect US jobs. Since the inauguration for his second term, Trump enacted again tariffs for imports from Mexica, Canada, China and announced tariffs for imports from Europe. Why did the electorate in the world’s most economically developed country help again a politician to power who explicitly advocated foreign trade protectionism, i.e. the country’s protection of domestic industries from foreign competition? Why did almost half of US voters give up the economic freedom of being able to buy cheaper foreign goods instead of more expensive domestic goods for the rather dubious promise¹ of making domestic industry great again? Have own-interested economic decisions become the frightening ghost against which a supposedly powerful president promised protection?

That trade protectionism became politically powerful again with Trump’s election as the 45th and again as 47th US President is worrying for most academic economists and only partly understandable. Worrying because the imposition of US import tariffs reverses the 70-year trend of worldwide falling import tariffs and accepts global welfare and growth losses. This “new protectionism” is partly understandable because after tariff cuts, some domestic sectors of the economy (e.g. the US steel and automotive sectors) could no longer cope with foreign competition, companies went bankrupt and jobs were lost, while jobs in other sectors of the economy were not found immediately or at all. Representatives of industries with strong foreign competition then turned to politicians with a request for an increase in import tariffs and received these increases if the business representatives pledged (financial and ideological) support to the government (Grossman and Helpman, 1994).

However, this explanation is only partially satisfactory. As in 2016, Trumps foreign trade policy rhetoric during the 2024 campaign was aimed at broader sections of the electorate than just members of individual economic sectors. Trump’s messages fundamentally question the advantages of the freedom in international economic relations achieved so far—through the multilateral world trade order—by pointing to the “melting down of industrial cores” and the associated loss of US jobs. Although opinion polls before 2016 show a clear majority of the US population to be in favor of free trade, by voting for Trump in 2016 and again in 2024, about half of US electorate accepted the restriction of economic freedom to buy goods where they are cheapest.

The explanation of this decision against economic self-interest in the US voting booths in November 2016 and again in November 2024 has led political scientists (Mutz, 2018) and economists (Grossman and Helpman, 2018, 2021; Bonomi, Gennaioli and Tabellini, 2021; Gennaioli and Tabellini, 2018, 2019, 2024) to see not only the benefits from the consumption of material goods but also the psycho-social benefits from revived national pride and identification with a social group that is important for one’s own self-esteem as relevant to the decision. As explained below, economic and/or cultural shocks generate fears of social decline and cause individuals to identify with different social groups than before. Loss of material benefits because of an increase in import tariffs is accepted in exchange for the gain in benefits from warding off loss of status (“America great again”) and re-identifying with a social group that is only like one’s own and clearly different from an out-group. This re-identification has enabled Trump in 2016, as a political entrepreneur, to stoke anti-elite prejudices (“us” and “them”) (Grossman and Helpman, 2018: p. 20) and exploit the fears of loss of status among the white lower middle class in the US (Mutz, 2018) to achieve his goal of becoming president.

The electoral success of a decidedly protectionist presidential candidate (“I am a tariff man”) was also due to the prevailing international macroeconomic political environment. Since the liberalization of exchange rates in the early 1970s, the macroeconomic supply policy of Reagan and Thatcher and the liberalization of international capital movements in the 1980s, the scope for national fiscal policy to counteract negative globalization shocks has been significantly limited. Since the late 1980s, the main aim has been to maintain or to increase the international competitiveness of each country. This was not compatible with massive national intervention measures in favor of the losers of globalization (the “backward”). Trump took already 2016 the “inaction” of the previous presidents due to this “international supply policy” (Winkler, 2017) as an opportunity to advance trade policy interventions to demonstrate activity in the interests of those left behind.

The renewed foreign trade protectionism is not only to be seen as a (questionable) political reaction to increased foreign competition since the economic rise of East Asia. As will be shown below based on Grossman and Helpman (2018, 2021) and Gennaioli and Tabellini (2018, 2019, 2024), it is also cultural shocks (such as the decriminalization of abortion, the questioning of the traditional family) that make a policy of isolation from foreign countries understandable. Of the 81% of US evangelicals who voted 2016 for Trump, only a small proportion cited Trump’s anti-abortion campaign as a voting motive in post-election surveys. For most, however, Trump was still the lesser evil, which is an indication that this group of voters also wanted to prevent Clinton’s socially progressive policies. 2016 and again 2024, social conservatism only seems to be possible at the price of foreign trade protectionism. Free international trade seems to contradict social conservatism. But is this current policy-defining sacrifice of economic freedom for possibly higher moral goods such as the right to life from conception to natural death inevitable from a broader meta-economic perspective? That is the question that will be briefly addressed in Section 5 of this article.

In the next section the lost control of national macro policy after Reagan and before Trump and Trump’s economic nationalism are described in line with Winkler (2017). Section 3 is devoted to identity politics and protectionist trade policy by reviewing Grossman and Helpman (2018, 2021), while Section 4 deals with belief distortion, political division and protectionist social conservatism in line with Gennaioli and Tabellini (2018, 2019, 2024). Section 6 concludes.

2. Loss of Control of National Macro Policy and Trump’s Economic Nationalism

Adalbert Winkler (2017) analyzed the reasons for Trump’s 2016 election success primarily from a macroeconomic perspective. He first notes that Trump denounced during his first campaign the decline of the industrial cores in the USA and the stagnation of real incomes of the (lower) US middle class since the 1980s and used his slogan “Make America Great Again” as a reference to the “golden” 1950s, when the USA was not only militarily but also economically “great”. “Economic growth was high [back then] and there was full employment. Income and wealth were distributed much more equally than today, jobs were more secure and the opportunities for social and economic advancement were much greater than today.” (Winkler, 2017: p. 117) With his economic and foreign trade policies, Trump wanted to bring back the good times of the 1950s for economically “left behind” Americans. But is this possible if the presidents before Trump were unable to prevent the negative developments diagnosed above?

To this end, Winkler (2017) describes US economic policy in the 1950s, the failure of this policy in the early 1970s, the new consensus of an “international supply policy” (Winkler, 2017: p. 120), Trump’s departure from it towards economic nationalism, and the role of cultural factors in Trump’s political calculations.

Winkler attributes the economic boom of the USA in the 1950s to “national Keynesianism” with high top tax rates, expansive fiscal and monetary policy and extremely limited international capital mobility. In the fixed exchange rate system of Bretton Woods with politically adjustable exchange rate parities, the current account imbalances remained relatively small, in contrast to the large current account imbalances since the liberalization of international capital movements in the 1980s. The fixed exchange rate system collapsed in the early 1970s because the American government abused Keynesian budget policy as a cover for financing expensive social programs and the Vietnam War, which led to inflation and in 1971 to President Nixon’s abolition of the link between the US dollar and gold.

The failure of national Keynesianism in the face of inflation and external economic turbulence paved the way for the international supply-side policy mentioned above, which is not concerned with sufficient aggregate demand, as in Keynesianism, but with the aggregate supply of increasingly open economies. The aim of the international supply-side policy is to be internationally competitive, or to become so again. This is understandable for several reasons: Firstly, since 1973 the exchange rates of the currencies of the countries that were leading in the 1970s have been determined by supply and demand on the foreign exchange markets and are no longer fixed by the national central banks. Favorable national production costs (an important supply factor) increase the real value of the domestic currency. Secondly, technological progress in the post-war decades has led to a dramatic fall in transport and communication costs between the leading industrial countries (excluding the Eastern Bloc and Communist Asia), which gave countries with favorable supply factors greater advantages. Thirdly, the bad experiences that the countries of the “free” world had with Keynesian demand policy brought the advantages of a policy of supply management through lowering top tax rates, deregulating private economic activity and privatizing state-owned companies into the focus of economic policymakers inclined towards economic freedom (Reagan, Thatcher). Fourthly, the collapse of the Bretton Woods system with its strict restrictions on international capital movements paved the way for international capital movements based on purely economic calculations: a nationally controlled financial system became a globally market-controlled one. All this significantly limited the scope of action of national fiscal policy to ensure full employment and an equal distribution of income, leading to the loss of control of national macro-policy, which Winkler (2017: p. 123) regrets, following Rodrik (2011).

From then on, national production and employment developments were heavily dependent on the development of the global economy. Following the collapse of the Soviet Union, the fall of the Iron Curtain, European integration, China’s accession to the WTO and the opening of other populous Asian countries (such as India) to the global market, the number of labor providers increased by around 1.6 billion. In addition, the sharp reduction in international transport and communication costs worldwide has given companies in highly developed countries with high labor costs the opportunity to relocate production to cheaper countries abroad. This costs domestic jobs that were filled by relatively less qualified workers. In addition, technological progress accompanied by computerization, digitization and the worldwide web accelerated the demand for highly qualified workers and drove up their wages, while demand for less qualified workers declined and their pay stagnated despite economic growth. Finally, the development of internet-based platforms has increased the importance of intangible business investments over traditional investments in machinery, buildings and land and has enormously increased the profits of successful internet pioneers, while the incomes of low- and medium-skilled white Americans have declined or stagnated, something which Trump already in his 2016 campaign has publicly complained about.

For the US presidents before Trump, these were regrettable but unavoidable consequences of the now worldwide (apart from parts of Africa) division of labor between less and more developed countries. For the few critics of this international economic laissez-faire policy, such as Rodrik (2011), this was a policy failure that enabled Trump to win over the US losers of international supply-side policy with a new form of economic nationalism (“I am a tariff man”). “Massive restrictions on free trade and immigration are intended to give the nation state back its ability to act and thus its responsibility for economic development.” (Winkler, 2017: p. 125) For Trump’s lower-middle-class voters, the promised isolation from foreign countries was attractive because jobs would be retained (at least in the short term) in the sectors protected by tariffs², and less immigration from abroad would reduce the pressure on domestic wages for the less qualified. Both promised an improvement in the economic situation of Trump voters.

Winkler (2017) also points out the importance of socio-cultural factors for Trump’s election victory over Clinton. The core of this explanation for the rise of economic nationalist politicians is the rejection of socially liberal values, as embodied by the socially progressive Democratic presidential candidate. Even if the combination of economically liberal and socially liberal positions is not necessary, it is precisely this that is used by power-hungry politicians to promote and implement economically illiberal policies. Because with state protection of domestic economic sectors against cheaper foreign competition and a restrictive immigration policy, it is easier to keep the promise of protecting traditional (conservative) values in parts of the population. In addition, socially conservative policies promise immediate economic benefits for white US men: women concentrate again on their role as mothers and housewives, abortion is again prosecuted, marriages are only accepted as a heterogeneous sexual union and are legally (financially) preferred, immigration is strictly controlled for reasons of national security, the shortages on the labor markets change in favor of the white US men who have come to kiss the hand. They are needed again as “breadwinners” because they are less displaced by the “others”. They can be proud of their country again, exactly what Trump wants with his slogan “Make America great again” and was implemented during Trump’s first term and is in the process of being implemented with renewed vigor during his second term.

For Winkler (2017), it was this combination of socio-cultural and economic factors that explains Trump’s 2016 victory and the defeat of the socially liberal elite. And it is no exaggeration to claim that his 2024 victory can be also referred to this combination of socio-cultural and economic factors, even if economic circumstances as consumer price inflation have changed. For President Biden and his team, the decline of American industry due to the use of labor-saving technologies and the relocation of production to cheaper countries abroad, as well as the poorly controlled immigration from abroad, was “unavoidable” because it was the result of global economic developments. Trump promised those “left behind” by this development an alternative: to make up for the national loss of control due to international supply policy by isolating themselves from foreign countries.

Finally, Winkler (2017) mentions the global financial crisis of 2007/2008 and the subsequent major recession in 2009 as reinforcing the acceptance of economic nationalist policies. The near collapse of the global financial system and the subsequent losses in real income and growth, particularly in the highly developed countries, have firstly shaken confidence in the market-economy system, even among those income groups that had always benefited from it. Secondly, in the face of the impending collapse of the financial system, politicians have resorted to “rescue measures” that fundamentally contradict the principles of neo-liberal economic policy. The billions of dollars in state bank rescue and the trillions of dollars in unconventional monetary policy may have prevented the collapse, but the political elite and the media can no longer fully support the relatively free global market economy. In return for the interventionist state’s domestic economic retreat through tax cuts and deregulation, restrictions on international economic freedom are accepted, as was the case in the USA in 2016 and is again in 2024.

3. Identity Politics and Protectionist International Trade Policy

Winkler (2017) attributed the return of protectionist US international trade policy to the inaction of the US governments before Trump to cushion the loss of US industrial jobs through a supranational full employment and redistribution policy. This did not happen in the pre-Trump era and will not happen soon. So, the only option left is domestic redistribution, but with Erich Weede (2018: p. 99) it is doubtful whether “domestic redistribution in favor of the losers of free trade is sufficient to cushion the political consequences of ‘creative destruction’ (in Schumpeter’s sense).” Even if doubtful, Trump’s voters seem to believe that their political favorite’s promise will materialize.

3.1. Loss of Status and the Preference for Protectionist International Trade Policy in the 2016 Campaign

This is especially true if it was not the individual job and income losses of those “left behind” by globalization that determined the voting decision for Trump and against Clinton in 2016, as Diana Mutz (2018) shows in a comparative empirical analysis of voter opinion polls before the 2012 and 2016 presidential elections. Rather, it was fears of losing the high social status of white men with Christian beliefs in the Northeast and Midwest of the USA (“rust belt”), which made them more likely to vote for Trump rather than Clinton in the 2016 election.

The feeling of being threatened with one’s achieved status generates defensive reactions. The first is that the status quo ante and hierarchical social and political arrangements become more attractive. “Thus, conservatism surges along with a nostalgia for the stable hierarchies of the past.” (Mutz, 2018: p. 331) The second defensive reaction consists in defending one ‘s own dominant group, emphasizing conformity of group members with group norms, and other social devaluing groups (“us” and “them”). People feel good when they belong to the dominant group, and to maintain this state, when threatened, inexpensive measures are taken such as voting for politicians who promise to eliminate the status threat. According to Mutz (2018), two threats to perceived group status were particularly strong in the USA: First, the feeling of being in decline as white people compared to their fellow citizens of color who are rising the list. Second, white Americans felt threatened by growing dependence on foreign countries. This feeling is confirmed by economic data that the USA will soon no longer be the largest economy in the world and will have to give way to China and India in the not-too-distant future. In addition, around half of the US population believes that trade with foreign countries (especially with China) creates jobs abroad at the expense of American jobs. Foreign trade is seen as a zero-sum game (=what one wins, the other loses), a view that poses a threat to US dominance.

In 2016, firstly, significantly more voters than in 2012 voted for the more conservative Republican Party and secondly, in 2016, noticeably more voters than in 2012 evaluated international trade negatively. In 2016, the gap between the average voter’s opinion on foreign trade and the perceived position of the presidential candidate was significantly smaller for Trump than for Clinton. Mutz (2018: p. 334) concluded from this, “that it was change in how candidates positioned themselves on status threat-related issues… that predicted increasing support for the Republican candidate in 2016.” And: “[T] he candidates’ changing relative positions on trade account for the greatest net impact on Trump support.” (Mutz, 2018: p. 336) It was not so much the economic problems in the old US industrial areas, but the feelings of status threat among the previously dominant section of the population that decided the 2016 presidential election. Trump’s 2016 election slogans were closer to the positions of status-threatened Americans than those of his competitor.

3.2. Social Identification and Trade Policy Preferences

These results from the empirical study by Mutz (2018) point to the importance of group status in the formation of political preferences. This hint is considered by prominent economic theorists Grossman and Helpman (2018, 2021) in which they incorporate psycho-social components into political-economic models of party competition. The aim of the model analysis by Grossman and Helpman (2018, 2021), which is now presented in an overview, is to explain why a tariff on foreign goods (import tariff) is in the interest of both the voters and the political party or candidate who promotes the tariff.

Grossman and Helpman (2021) begin by stating that the increased preference among voters for a protectionist foreign trade policy in highly developed countries in recent years can no longer be explained by political lobbying by special interest groups such as industries threatened by imports, as was the case in the 1980s and 1990s. Much broader sections of the population are nowadays opposed to dependence on foreign countries and imports from cheaper countries. To explain this change in opinion, which Mutz (2018) empirically documented for the USA, a closer look at changes in the policy preferences of broader sections of the electorate is necessary. It turns out that voters’ trade policy preferences are not based solely on material self-interest, but also on the well-being of members of those social groups with which the person in question identifies. This is not universal altruism, but rather particular altruism, in that individuals only care about the well-being of those people who appear like them. Changes in social identification triggered by economic and cultural shocks can then trigger unexpected leaps in foreign trade policy.

Social psychologists define social identity as “the individual’s knowledge that he belongs to certain social groups together with some emotional and value significance to him of the group membership.” (Tajfel, 1974: p. 31) The theory of social identity “studies how identification with a group influences individual perceptions and behavior.” (Gennaioli and Tabellini, 2018: p. 5). The theory of social identity is related to the theory of self-categorization (Turner et al., 1987) which “studies primarily which of the many possible groups (e.g. occupational, religious, etc.) an individual identifies with.” (Gennaioli and Tabellini, 2018: p. 5) Both theories assume that individuals categorize their social reference groups according to self-esteem which they derive from identification with a particular group. The individuals in the groups with which they identify do not have to be “accepted”, nor do the members identified as the reference group have similar views to the individual. Rather, the individual compares himself with a prototypical reference group member and derives satisfaction from the status that the reference group enjoys in society. The reference groups as social categories differ according to nationality, race, ethnicity, gender, occupation, religion and can become significant (salient) depending on the political, economic and cultural context.

According to Akerlof and Kranton (2000), an individual derives utility not only from the consumption of material goods, but also from self-esteem, which increases when the individual conforms to the behavioral norms of a social reference group. Shayo (2009) models this idea in such a way that an individual experiences a utility gain when he associates with a group with high social status but loses utility if it identifies with a group whose prototypical member is far removed from one’s own assessment of a social category. Shayo (2009) defines social identity equilibrium as a state in which individuals categorize themselves within a set of salient identity groups, depending on the actions and action outcomes of others, with the actions induced by politicians and the policy environment and their outcomes determining individuals’ identity decisions.

3.2.1. Model Economy

Grossman and Helpman (2021) apply a variant of social identity equilibrium to the Heckscher-Ohlin model of international trade (see e.g. Farmer and Schelnast, 2021: pp. 283-295) in products from two industrial sectors between two small countries that consider the world market prices of their products to be externally determined. In each country there are only two factors of production, namely higher and lower skilled labor. These are used to produce one product for export and another that competes with a product imported from abroad. The production of the export good requires relatively much higher skilled labor, while the import good uses relatively much lower skilled labor. The better qualified individuals count themselves as the upper class of the country’s society, but are in the minority, and identify themselves with the country’s “elite”. The less qualified majority identify with the (lower) middle class. In addition, and not exclusively, members of both the elite and the middle class can identify with the broader social category of the “nation” (e.g. “Americans”). Like Shayo (2009), Grossman and Helpman (2021) assume that the non-material benefit from individual identification with a social group of a certain category increases with the presumed status of this group. The authors measure group status based on the standard of living (wage income) of the average group member. The psychological costs of identification, on the other hand, increase with the distance between the individual’s own economic and cultural characteristics and the prototypical reference group member. This is where the cognitive dissonance comes into play, which occurs when one is asked to identify with someone who is different from oneself.

As for the political system, Grossman and Helpman (2021) assume two parties (as is typical for the USA) with fixed ideological concepts. Individuals in their function as voters differ in their preference for one of the two parties. In electoral competition, parties use foreign trade policy, namely the level of an ad valorem tariff on the imported product, to compete for votes. An individual votes for a party if the preference for the proposed trade policy of that party outweighs the preference for the ideological conception of the rival party. In social identity equilibrium, both parties converge on the level of the import tariff.

3.2.2. Social Identities and Populism

For the purposes of this paper, it is crucial whether the inclusion of social identities and the self-categorization of individuals change foreign trade policy. First, it must be noted that in the market economy of a small country without social identification described above, the welfare-maximizing tariff is always zero, i.e. completely free foreign trade is welfare-maximizing. Economic freedom at home and between home and abroad is best for the common good of this country. This changes when individuals take the benefits and costs of social identification into account in their utility calculations. To make the why and how understandable, Grossman and Helpman (2021: p. 1104) introduce an identification regime that represents a list of all social reference groups with which the inhabitants of the small country identify. In the simple model with two types of individuals, there are only four identification regimes: Higher-skilled individuals can identify with the elite as a social group or not, or they can identify with the nation or not. Similarly, lower-skilled individuals can identify with the middle class or the nation, or neither, nor both.

Of particular interest are situations in which a change in the political or economic environment leads to a change in the identification regime. A striking example is the change in voter preferences between 2012 and 2016 identified by Mutz (2018), which is usually interpreted as a resurgence of “populism”. Following Müller (2016: p. 40) populism is “a specific form of identity politics in which a group of voters rejects the elite members of society’s claims to moral legitimacy”. “Populists do not just criticize elites; they also claim that they and only they represent the true people.” (Ibid) Populist politicians support policies that they believe rightly disregard the interests of the “corrupt” segments of society while serving the interests of the “real” people (“little man”).

To depict the shift to populism in the model, Grossman and Helpman (2021: p. 1117) start with an identification regime in which the less qualified do not just identify as the (lower) middle class, but with the entire population of a country, including the more highly qualified. As a result of a populist revolution, the identification regime changes. The less qualified stop identifying with all their fellow citizens, but instead rather they equate the “nation” with their own reference group. The authors describe precisely the changes in the model parameters that cause such a change in the identification regime and examine the consequences for foreign trade policy in the social identification equilibrium. They can show that the new identification regime induces a discrete jump to a high import tariff.

3.2.3. Model Economy More Detailed

To better understand the logic behind these results, the model economy is described in more detail. Both the less and the more highly qualified derive material benefits from the consumption of export goods and other goods that can be produced domestically or imported from abroad (e.g. cars). The typical economic agent in the small open economy also derives benefits from identifying with those groups in society that are emotionally significant to the individual (psycho-social benefits). Psycho-social benefits have two components: one positive and one negative. On the one hand, an individual’s self-esteem increases when they see themselves as part of a valued group, and the higher the social status of the reference group, the higher the self-esteem increases. On the other hand, cognitive dissonance arises when an individual is asked to identify with a group that is very different from him or herself, i.e. the negative component of psychosocial utility is an increasing function of the difference between the individual’s material well-being and the status of a social reference group, which is either the “working class” (=low average income), the “elite” (=high average income) or the “nation” (=sum of working class and elite income weighted by the share of the population). A less or more highly qualified individual will identify with the social group where the total psycho-social utility (identification benefit minus identification costs) is not negative, otherwise he or she will not identify. Under certain assumptions, for each qualification level only two possibilities of identification remain: either with one’s own group or with the broad nation, or not. The four identification regimes are: (0,0) = neither the lower nor the higher qualified identify with the nation, (0,1) = the higher qualified do not identify with the nation, but the lower qualified do, (1,0) = the higher qualified identify with the nation, but the lower qualified do not, (1,1) = both the lower and the higher qualified identify with the nation.

The competition between the two parties with clearly distinguishable ideological positions for the votes leads to the result that policies are proposed which make the sum of the voters’ benefits as large as possible. The policy instrument chosen is an ad valorem tariff rate $t\ge 0$ which increases the domestic price $p$ of the good being imported by the tariff rate compared to the world market price $q$ : $p=\left(1+t\right)q$ . Which ad valorem tariff rate is the result of the political competition? As one can intuitively presume, the relationship between the domestic price of the imported good and the wage income of the less and more qualified is important for the optimal policy. The wages of the less and more qualified react to the domestic price of the imported good according to the famous Stolper-Samuelson theorem (see e.g. Farmer and Schelnast, 2021: p. 286): A higher price of the imported good due to the tariff causes the wages of the less qualified to rise more than proportionally to the price increase, i.e. the real wages of the less qualified increases while the real wage of the more highly qualified falls. A higher tariff therefore reduces wage inequality.

If you add up the individual total utility across both qualification groups, you get the political objective function that is maximized by choosing the ad valorem tariff rate under the constraint that all identity decisions are individually rational. If one or both qualification groups identify with the nation, the wage divergence between the more and less qualified appears in the political objective function (= total utility function). Then the aversion to wage inequality becomes an element of foreign trade policy. However, this is not because voters value fairness, but because identifying with the nation as a whole means paying a price for identifying with unequals. The aim is to keep this price as low as possible through tariffs.

3.2.4. Tariff Rates under Identification Regimes and Populist Revolution

Following Grossman and Helpman (2021: p. 1112, 1118), Figure 1 below shows the optimal tariff rates (domestic prices) that politicians choose under the four identification regimes and why there is a sharp increase in the tariff rate under populism.

Figure 1. Welfare-maximizing domestic prices in the four identification regimes and “populist” revolution. Source: Figure 1 and 2 in Grossman and Helpman (2021: p. 1112, 1118). Own adaptation.

In Figure 1, the domestic price of the imported good is plotted on the horizontal axis and the total utility level dependent on it is plotted on the vertical axis. The figure shows four strongly drawn and one broken, inverted U-shaped curves, which show the course of the total utility function for the four identification regimes (0,0), (0,1), (1,1), (1,0), (1,1) and (1,1)’. The leftmost curve, marked (0,0), corresponds to the identification regime in which neither the less qualified nor the more highly qualified identify with the nation. The next curve from left to right, marked (0,1), symbolizes the identification regime in which only the less qualified identify with the nation. The middle curve, marked (1,1), corresponds to the identification regime in which both the less qualified and the more highly qualified also identify with the nation. The curve furthest to the right represents the identification regime (1,0), in which only the more highly qualified identify with the nation.

In Figure 1, the global utility maximum is initially reached at the domestic price p(1,1) with the tariff rate $t\left(1,1\right)=\left[p\left(1,1\right)-q\right]/q$ . If politicians set this tariff rate, it can be shown that the two groups of voters agree with the prevailing identification regime in their own interest. This also applies to domestic prices p(0,1) and p(1,0), and also to free trade policy, for which $t\left(0,0\right)=0$ and therefore $p\left(0,0\right)=q$ . If the global utility maximum is reached at a tariff of zero, both qualification groups identify only with their own social group, but not with the nation. The optimal tariff is zero, as without social identification, although both the less and the more qualified identify with their own social group, i.e. they care about the welfare of the others in their own group. However, there is no protection of the domestic economy because the interests of the less qualified are exactly the opposite of the interests of the more qualified and their interests neutralize each other: while the less qualified prefer a tariff to free trade, the more qualified want a negative tariff (=import subsidy). If, on the other hand, the global utility maximum is reached at p(1,0)—here only the more highly qualified identify with the nation—then the utility function to be maximized contains, in addition to the aggregated material utility, a component that decreases with the wage difference. These are the psychological costs for the more highly qualified if they do not only identify with their peers. Because wage inequality decreases with higher tariffs, the tariff rate at the point of maximum utility is greater than zero, and even greater than the domestic price p(1,1), if the proportion of the population that is more highly qualified is not too large. The argument is similar for the identification regime (0,1). Citizens therefore vote for foreign trade protection if identification (via the nation) with the other socioeconomic group causes psychological (dissonance) costs.

Now there is a change in the self-identification of the less qualified, such that the less qualified, who had previously identified with the entire nation, now only identify with the prototype of the working class. The identification regime changes from (1,1) to (1,0). The more qualified are branded as a “corrupt elite” with whom the lower middle class can no longer identify. The introduction of labor-saving technologies (IT, robots and AI) or a lower world market price due to cheap foreign competition (individually or together) trigger this change in the identification regime, which induces a discrete jump in the optimal tariff rate.

Before the regime change, the aggregate utility curve reaches a global maximum at the domestic price p(1,1). If, for example, a lower world market price due to cheap foreign competition prevails, the (1,1) curve shifts downwards and slightly to the right, as shown by the dashed curve (1,1)’ below the (1,1) curve in Figure 1. The high point of the (1,1) curve is only just above the high point of the (1,0) curve. After the (1,1) curve shifts downwards, the high point of the (1,0) curve becomes the global maximum. Instead of a small increase in the tariff rate when the global maximum would be on the (1,1)’ curve, the policy now increases the tariff rate abruptly from $t\left(1,1\right)=\left[p\left(1,1\right)-q\right]/q$ to $t\left(1,0\right)=\left[p\left(1,0\right)-q\right]/q$ in order to take account of the change in the social identification of the less qualified , and this when the proportion of the more highly qualified in the population is not too large. There are two opposing forces at work: on the one hand, the group of less qualified people no longer gains any benefit from identifying with the nation, which reduces the redistributive benefit of a tariff increase, and on the other hand, the psychological dissonance costs of identifying with “different people” are eliminated, which makes protection less costly. Protection increases when the last effect is stronger than the first, and this is exactly the case in Figure 1.

3.2.5. Resume on Social Re-Identification and Trade Policy

From a purely economic point of view, protectionist foreign trade policy in a small, open economy is irrational because it runs counter to the self-interest of the average citizen. This is no longer the case if individual self-interest is not limited to the benefit from consuming goods. As a social being, a rational voter is guided not only by the likely effects of an election program on her material benefit, but also by psychological factors such as pride, social acceptance and self-esteem. She or he chooses those policies that make her or him emotionally satisfied, but not necessarily rich. Social psychology teaches that emotional satisfaction and self-esteem come in part from the fact that the individual sees himself as a member of a social reference group. The individual seeks confirmation from others and wants to belong. The individual identifies with fellow human beings whom he values, he experiences satisfaction when the reference person is successful, discomfort when the person one identifies with is not successful. It is therefore natural to see the well-being of others with whom one identifies as part of one’s own satisfaction and to support policies that benefit the people one identifies with as well as oneself. Identity politics is then the logical consequence of this way of thinking. Protectionist foreign trade policy as part of this identity politics maximizes the material and psycho-social benefits of the average citizen in the small open economy³.

Identification is entirely voluntary and cannot be forced from the outside. To the extent that identity reflects self-categorization, it will respond to economic and cultural change. Because identity influences political preferences, changes in the identification regime influence political outcomes. Social identification occurs in discrete steps (jumps): one is part of an in-group or part of an out-group. Strong changes in political preferences, called the “populist revolution”, have been observed in the USA and other rich countries since 2015. Populism is understood as identity politics, where individuals no longer identify broadly with the nation but more closely with their own reference group. The other "elite" group is branded as corrupt and the less qualified no longer see the elite as a legitimate source of national pride. There is a shift in the identification regime from broad to narrow, regardless of whether this is caused by technological progress or globalization. If the proportion of the elite in the population is not too high, foreign trade protection is introduced abruptly, even if cheaper imports are not the trigger.

4. Belief Distortion, Political Division and Protectionist Social Conservatism

Like Grossman and Helpman (2021), Gennaioli and Tabellini (2018, 2019), examine the mechanisms behind inter alia the increased voters’ preference in the USA and several European countries for protectionist social conservatism in recent years. In contrast to Grossman and Helpman (2021), in their political-economic models, identification is not status-related but influences policy through voters’ beliefs. As political science shows, voter attitudes often have the property of being distorted and exaggerating (fearmongering) facts. Alesina et al. (2018) show that both US and EU citizens overestimate the actual number of immigrants living in the country by several times, with the overestimation being significantly higher among the politically conservative population.

4.1. Identity, Beliefs, and Political Conflict

Gennaioli and Tabellini’s (2018, 2019)⁴ model approach focuses on belief distortions and (trade) policy consequences and thus complements Grossman and Helpman (2021). “Why do losers from free trade become nationalistic, hate immigrants and turn socially conservative? Why do they vote for platforms that include policies that seem to run counter some of their interests, such as tax cuts or unsustainable budget deficits?” (Gennaioli and Tabellini, 2018: p. 2) In response to these questions, the authors develop an economic theory approach that explains why voter attitudes reflect political identities, and economic and/or cultural shocks can induce major changes in political preferences across a wide range of areas. The central idea included is “that voter’s beliefs are shaped by endogenous social identities. Individuals routinely identify with a group of people or a party with similar values and interests. Once they do so, they ‘depersonalize’: they attenuate individual specificities, slanting their beliefs toward the ‘prototypical’ or ‘distinctive’ group member.” (Ibid). Social identities are not immutable. Significant changes in the economic and social environment can trigger identity change. As a result, attitudes change and become distorted into new, correlated dimensions, amplifying the effects of economic or cultural shocks.

Economic globalization, data-driven technological progress and cultural change are creating new and serious conflicts of interest that are reshaping social identities. “The more disadvantaged members of society see themselves as more like a nationalist or socially conservative group, whose distinctive features are the sector of employment or the region of residence, rather than income. The distorted beliefs associated with these new identities dampen demands for traditional redistributive policies and give rise to a reassessment of political demands over many issues such as trade protection, control of immigration, less progressive civil rights.” (Gennaioli and Tabellini, 2018: p. 3).

4.1.1. Model Economy in More Technical Terms

In their revised and extended model (Gennaioli and Tabellini, 2019: pp. 28-31), the two authors assume, firstly, that there are different groups in society that differ in terms of income, trade openness and cultural disposition. Each voter feels more like a group whose interests are closer to their own and where the conflict between the in-group and the out-group is greater. Secondly, as soon as an individual identifies with a social group, his or her attitude tends towards stereotypical group attitudes, the position that is most recognizable from the outside.

Like Grossman and Helpman (2021), Gennaioli and Tabellini (2019: pp. 28-31) consider a small open economy with a continuum of size 1 and two traded goods: $x$ represents the exported good, and $m$ is the imported good. The price of the exported good is 1, the price of the imported good is $p$ . As in Grossman and Helpman (2021), there is an ad valorem tariff $t$ such that $p=\left(1+t\right)q$ whereby $q$ is the world market price of imports. In addition to the tariff rate, the government has as a second instrument: a distorting income tax rate $\tau$ . Revenue from both instruments is used to finance a public good $g$ . Finally, there is a third instrument denoted as $r$ which represents civil rights.

Individuals differ according to where their income comes from. Firstly, everyone earns a random and taxable income from employment in the export sector of $1+\tilde{\epsilon }.$ The average of $\tilde{\epsilon }$ captures the individual’s income $\epsilon$ . Secondly, everyone receives tax-free income from the supply of a factor that can also find employment in the import-substituting sector, with a probability that varies from individual to individual and depends on a parameter that indicates the intensity of competition in the import-competing sector. The income is 1 if employed in the export sector, and $p$ if employed in the import competing sector. The probability of employment in the import sector is $\sigma \left(1-\tilde{\eta }\right)$ , whereby $\sigma \in \left(0,1\right)$ captures exposure to foreign competition, while $\tilde{\eta }$ is a random variable with individual-specific mean $\eta$ . An individual with higher $\eta$ is more employable in the export sector. Thus, the expected income of type $\eta$ from the specific input is $1+\sigma \left(1-\eta \right)\left(p-1\right)$ . The average value of $\eta$ is zero, so that aggregate output in the import competing sector is $\sigma$ .

The utility of a voter of type $i=1,\mathrm{...},I$ is: $w^i=x^i+U\left(m^i\right)+\nu g-\left(\kappa /2\right){\left(r-\tilde{\psi }\right)}^2$ , where $U(.)$ is a quadratic utility function, $\nu g$ is the utility of public consumption $g$ , with $\nu >1$ and large, $\kappa >0$ and $\tilde{\psi }$ is individual i’s exposure to culture. The typical individual i derives material benefit from the consumption of private export and import goods and public goods (infrastructure). The third element in the utility function is the preference for socially liberal civil rights with $\kappa$ denoting the utility coefficient of this preference. i’s budget constraint, based on expected income $y^i$ , demands: $x^i+p{m}^i=y^i$ . The type of voter is represented by the stochastic vector of individual income, trade openness and cultural progressivism: $\left(\epsilon ,\eta ,\psi \right)$ . The typical individual derives material benefit from the consumption of private export and import goods and public goods (infrastructure). The third element in the utility function is the preference for socially liberal civil rights.

Individuals are unsure about the consequences of an extension of civil rights, and their socio-political views may be subject to stereotypes. The three individual random variables income, openness to foreign trade and social progressivity $\left(\epsilon ,\eta ,\psi \right)$ may be uncorrelated, but following Enke’s (2018) cultural predisposition for “communal values”, then social conservatism and opposition to foreign trade are positively correlated $\rho >0$ , while there is only a weak correlation with income.

These assumptions lead Gennaioli and Tabellini (2019: p. 3) to two main observations: (1) Identification makes attitudes towards specific issues more extreme, which fuels political conflicts. For example, a voter who identifies with the group of poor working women will exaggerate the suffering of poverty because she focuses on the distinguishing feature of this group, poverty. The opposite is true for a voter who identifies with the rich. Therefore, expectations about one’s own ability to earn income become more polarized and the redistribution conflict becomes more intense compared to the world without identification. (2) Economic shocks can change identification by causing polarization to occur in new dimensions while decreasing along previous dimensions. For example, a globalization shock (populous China opens to the world market) can cause individual interests to accumulate differently than before: instead of poor-rich, they become for-or-against globalization. Political conflict about redistribution decreases but the conflict about globalization becomes more intense.

4.1.2. Three-Dimensional Political Conflict

As the results of the presidential elections in the USA in 2016 and again in 2024 show, this two-dimensional conflict is joined by a third dimension: namely a “culture war” between the socio-politically “progressives” and the socio-politically “conservatives”. It is this three-dimensional political conflict that Gennaioli and Tabellini (2018: pp. 23-24) examine in a probabilistic election model of a small open economy like that of Grossman and Helpman (2021), but with the difference that individual income, attitudes towards foreign trade and the chances of success of progressive cultural policy are random.

In the small open economy under consideration, identification takes place in the three dimensions of income, threat from foreign trade and social progressiveness. In the first dimension, the voter is poor or rich, in the second free trade or nationalist, in the third socially conservative or socially progressive, with the boundaries between the respective pairs being historically given. Moreover, the number of socially conservative and low-income voters is in the majority.

If an individual identifies himself through income class, then his in-group is poor or rich, and so is his out-group. The same applies to identification along the trade and culture dimensions. The individual chooses a dimension of conflict that divides society into in-groups and out-groups. A voter who identifies with low-income earners considers all high-income earners to be an out-group, regardless of how they identify themselves. Similarly, a nationalist considers all free traders and a social conservative all progressives to be out-groups. According to the theory of self-categorization, an individual identifies along the dimension in which his ingroup conflict is small, compared to the inter-group conflict. This conception of similarity and identification thus includes two types of conflict: that between the individual and his in-group and between his in-group and the out-group in the chosen dimension. Gennaioli and Tabellini (2019: p. 17) measure the conflict between an individual and his in-group by the loss of utility that occurs when the voter deviates from his individually optimal tax rate, tariff rate and cultural policy instrument towards the instrument combination of his in-group. The inter-group conflict is captured by the loss of utility that the in-group suffers by switching to the out-group. Adding the two losses of utility gives the relative distance between the individual and his in-group. The individual feels like the in-group and easily identifies with it if the relative distance is small. This is the case when the conflict with the in-group is small compared to the inter-group conflict. The parameter $\lambda >0$ represents the strength of the conflict between the groups compared to agreement with the in-group and, if it is large enough, acts as a driving force for the identification of all individuals with only one of the three dimensions. Two other parameters, namely $\gamma >0$ and $\alpha >0$ , control the model results. $\gamma$ measures the importance of the conflict over foreign trade compared to the income conflict and $\alpha$ the severity of the conflict over culture compared to the income conflict.

4.1.3. Changing Identities and Correlation between Trade Protection and Social Conservatism

A larger $\gamma$ promotes identification along the trade dimension and exacerbates the trade policy conflict. $\gamma$ increases when individuals as employees in the import-competing sector feel stronger foreign competition, when the price elasticity of import demand decreases and the welfare costs of income taxation increase. With a larger $\alpha$ cultural policy conflicts increase. $\alpha$ becomes larger when the benefit costs of progressive cultural policy and income taxation increase. If the low-income, nationalists and social conservatives are in the majority and is $\alpha >\gamma$ , all individuals identify along the lines of culture, otherwise, for $\alpha <\gamma$ along the lines of trade. In the first case, the “culture war” increases, in the second case the trade dispute increases in comparison to the “class struggle” (over income) (Gennaioli and Tabellini, 2019: p. 30).

Economic and cultural change, i.e. higher $\gamma$ and $\alpha$ , change identities. The correlation between trade openness and social progressivity (depicted by coefficient $\rho >0$ ) is also important for the direction of identification. In fact, a stronger correlation leads to cultural and trade identification dominating over income-based identification. In addition, with a stronger correlation, identities are more responsive to $\gamma$ and $\alpha$ shocks. Highly correlated dimensions are an efficient source of identification: they reduce conflicts across dimensions and increase the individual feeling of similarity.

Historically determined clusters of values are therefore prone to become vehicles of identification and political action, and thus of polarization, replacing class-based identities. To put it differently, trade and technology shocks may have increased the relevance of pre-existing fault lines within traditional political groups defined on the left versus right dimension. Socially conservative poor voters, who traditionally identified with left wing groups despite their social conservatism, are now attracted by nationalism because it appeals to both their trade preferences and their cultural views, and vice versa for voters with opposite political features. As this happens, traditional income or class-based conflict wanes and is replaced by new political cleavages over correlated dimensions (Gennaioli and Tabellini, 2018: p. 25).

The “depersonalization” of individual attitudes occurs when individuals overvalue stereotypical group attitudes. The parameter $\theta >0$ measures the importance of stereotypes in distorting individual means of income, trade openness, and culture. If $\theta$ is high, identification-based attitudes deviate strongly from individual means (without identification).

Correlation between trade and culture changes the consequences of identification: The low-income, latently globalization-critical and socially conservative US voter understates his expected income before China’s accession to the WTO in 2001 due to identification with the low-income, while he is in the middle of society in terms of trade and cultural policy. China’s accession to the WTO, the relocation of labor-intensive US production to China and the mass import of cheap Chinese products to the USA let $\gamma$ increase, which triggers a change in identity from income-based to trade-based. The previously tolerant towards foreign trade policy neutral voter becomes a nationalist and, because of the positive correlation between trade aversion and social conservatism, a social conservative. The income prospects of the typical voter are no longer distorted, so he or she demands less redistribution. The gap between in-groups and out-groups within trade and cultural policy is widening, if also $\theta$ is increasing, i.e. the importance of stereotypes rises. This increases polarization and conflict between different views of the world.

In contrast, in the 1980s the US population only identified with low-income or high-income, the other two dimensions of identification were only latent. When the conflict over the cultural dimension became apparent in the 1990s, it encouraged the adoption of new cultural identities. Overall, this accelerated social polarization because culture by its nature encompasses broad areas. This and the fiercer import competition from China also exacerbated the conflict between winners and losers of open borders and gave rise to new identities in terms of place of residence and employment sector, which are correlated with the appreciation of local places and regions.

4.2. Political Demand for and Supply of Protectionist Social Conservatism

As explained by Gennaioli and Tabellini (2018, 2019) and presented in the subsection before, US voters attach increasing importance to cultural issues while conflict over redistribution has declined despite rising income inequality. “Something similar has happened on the supply side. In their propaganda, US parties attach growing importance to cultural issues relative to economic ones.” (Gennaioli and Tabellini, 2024: p. 2) Moreover, voters have realigned across parties: less educated and poor white people increasingly voted Republican, while the opposite is true for top income earners and highly educated voters. (Ibid)

Gennaioli and Tabellini (2024) follow Bonomi, Gennaioli and Tabellini (2021) in modeling the political demand for protectionist social conservatism while also dealing with the political supply of it. Cultural conflict in the USA has intensified as less educated, culturally conservative workers face increased import competition from developing countries, especially China (Autor et al., 2020). Low trade barriers since China’s accession to WTO in 2001 increased the salience of the educational and cultural divide and led to identity shifts already before Trumps first term. New facts show that US voters in regions more exposed to the China shock have become more anti-immigrants (if religious) and demand less redistribution (if poor) than in the past. In addition, congressmen in these regions adopted a more conservative rhetoric, particularly if Republicans (Gennaioli and Tabellini, 2024: p. 5).

4.2.1. Extended Model Economy

To provide a political-economic model explanation of these facts, Gennaioli and Tabellini (2024: pp. 22-26) consider a small open economy consisting of several districts indexed by $z=1,\mathrm{...},Z$ . In each district there are two sectors, export $x$ and import $m$ , with international prices 1 and $q$ respectively. Voters earn their taxable income $1+{\epsilon }^i$ in the export sector. A distortionary tax rate $\tau$ on his income is applied to finance a national public good $g$ . The national government sets a social policy parameter $r$ as in the previous subsection. A smaller $r$ means more conservative social policy.

New is the assumption that voters also earn a non-taxable income from two units of labor that can be employed in either sector, with voter- and district-specific probability. Let ${\eta }_z^{ij}$ be the probability that type $ij$ in district $z$ is employed in the import sector. Half districts are non-exposed: $z=n$ . Here, no voter earns import-sector income: ${\eta }_n^{ij}=0$ . Half districts are exposed: $z=e$ . A voter type $ij$ is generally characterized by the income-culture profile $\left({\epsilon }^i,{\psi }^j\right)$ whereby there are two cultural types: Progressive $P$ , and Conservative $C$ , with ${\psi }^P=\psi$ and ${\psi }^C=-\psi$ . Higher $\psi >0$ signifies more cultural disagreement. Moreover, there are two economic types: Upper class $U$ and Lower class $L$ , with ${\epsilon }^U=\epsilon$ and ${\epsilon }^L=-\epsilon$ , whereby $\epsilon >0$ captures economic inequality. The population is equally split into four types: upper class and progressive $ij=UP$ , upper class and conservative $ij=UC$ , lower class and progressive $ij=LP$ , and lower class and conservative $ij=LC$ .

Conservative voters in exposed district $z$ earn import-sector income, with equal probabilities across classes, ${\eta }_e^{UC}={\eta }_e^{LC}=\eta >0$ , while progressive voters earn no import-sector income: ${\eta }_e^{UP}={\eta }_e^{LP}=0$ . Therefore, aggregate domestic production of the imported good is $\eta /2$ . With rising $\eta$ conservative voters in exposed districts are more exposed to imports. That conservatism and import exposure are positively correlated means that less skilled/less educated workers are both more conservative (lower ${\psi }^j$ ) and more exposed to imports (higher ${\eta }_e^{ij}$ ).

The utility function of voter $ij$ in district $z$ reads as follows:

$u_z^{ij}=x_z^{ij}-\frac{1}{2}{\left(\varpi -m_z^{ij}\right)}^2+vg-\frac{\kappa }{2}{\left(r-{\psi }^j\right)}^2,$

where $x_z^{ij}$ and $m_z^{ij}$ denote private consumption of the exported and imported goods and $v>1$ , $\varpi >0$ .

The expected disposable income of voter $ij$ in district $z$ is:

$I_z^{ij}\left(\tau ,t\right)=\left(1+{\epsilon }^i\right)\left(1-\tau \right)-{\tau }^2/2+2\left[\left(1-{\eta }_z^{ij}\right)+\left(1+t\right)q{\eta }_z^{ij}\right]$ .

Taking the government budget constraint $g=\tau +T\left(t\right)$ into account, voter’s expected welfare function reads as follows:

$W_z^{ij}\left(\tau ,t,r\right)=I_z^{ij}\left(\tau ,t\right)+S\left(t\right)+v\left(\tau +T\left(t\right)\right)-\frac{\kappa }{2}{\left(r-{\psi }^j\right)}^2,$

where $T\left(t\right)=tq\left(\widehat{m}-\eta /2\right)$is aggregate tariff revenue in terms of the export good, $\widehat{m}=\varpi -\left(1+t\right)q$ is optimal consumption of imports, and the consumer surplus is$S\left(t\right)=-\left(1/2\right){\left(\varpi -\widehat{m}\right)}^2-\left(1+t\right)q\widehat{m}$ .

The first-order conditions (FOCs) for maximal $W_z^{ij}\left(\tau ,t,r\right)$ are as follows:

${\tau }^{ij}=v-\left(1+{\epsilon }^i\right),r^{ij}={\psi }^j,t_z^{ij}=\widehat{t}+\frac{2{\eta }_z^{ij}}{q\left(2v-1\right)}\text{with }\widehat{t}=\frac{\left(\varpi -q\right)\left(v-1\right)-v\eta /2}{q\left(2v-1\right)}.$

To ensure $\widehat{t}>0$ , it is assumed that $\left(\varpi -q\right)\left(v-1\right)>v\eta /2$ , and to have for all $t$ positive tariff revenue $\varpi -q\left(1+t\right)>\eta /2$ is assumed. The FOCs unambiguously show that voter $ij$ in district $z$ prefers a higher tariff $t_z^{ij}$ with higher exposure ${\eta }_z^{ij}$ . Moreover, less progressive voters, lower ${\psi }^j$ , demand a more conservative social policy, lower $r$ . Finally, richer voters, higher ${\epsilon }^i$ , demand less redistribution, lower $\tau$ , due to their greater tax burden.

4.2.2. Social Identity in Technical Terms

In line with social identity theory, a voter has several potential identities, defined by income, occupation, religion, race etc. As mentioned above, a typical voter in the present model can identify with his or her class, $i=U,L,$ or cultural group, $j=C,P$ . The identity group of voter $ij$ will be denoted by $\iota \left(ij\right)=i,j$ —often denoted by $\iota$ for short. In principle, “the voter identifies with the group that is most salient and to which she feels more similar.” (Gennaioli and Tabellini, 2024: p. 14). The salience of ingroup $\iota$ is formalized by its policy conflict with outgroup $-\iota$ , measured by the welfare loss by the average ingroup when moving from her ideal policy $\left(r^{\iota },{\tau }^{\iota },t^{\iota }\right)$ to the average out-group $\left(r^{-\iota },{\tau }^{-\iota },t^{-\iota }\right)$ . The contrast between in-group and out-group now takes the form:

$\Delta \left(\iota ,-\iota \right)=\frac{\kappa }{2}{\left(r^{\iota }-r^{-\iota }\right)}^2+\frac{1}{2}{\left({\tau }^{\iota }-{\tau }^{-\iota }\right)}^2+\frac{q^2\left(2v-1\right)}{2}{\left(t^{\iota }-t^{-\iota }\right)}^2$ .

A voter’s dissimilarity from his group is equal to:

${\Delta }_z^{ij}\left(\iota \right)=\frac{\kappa }{2}{\left({\psi }^{\iota }-{\psi }^j\right)}^2+\frac{1}{2}{\left({\epsilon }^{\iota }-{\epsilon }^i\right)}^2+\frac{2{\left({\eta }^{\iota }-{\eta }_z^j\right)}^2}{\left(2v-1\right)}.$

Using $t_z^{ij}=\widehat{t}+\frac{2{\eta }_z^{ij}}{q\left(2v-1\right)}$, we get: $t^{\iota }-t^{-\iota }=\frac{2\left({\eta }^{\iota }-{\eta }^{-\iota }\right)}{q\left(2v-1\right)},{\eta }^U={\eta }^L=\frac{\eta }{4},{\eta }^C=\frac{\eta }{2},{\eta }^P=0.$ Remembering ${\psi }^P=\psi ,{\psi }^C=-\psi ,{\epsilon }^L=-\epsilon ,{\epsilon }^U=\epsilon$ , $\Delta \left(C,P\right)=2\kappa {\psi }^2+\frac{{\eta }^2}{2\left(2v-1\right)}$ and $\Delta \left(L,U\right)=2{\epsilon }^2$ . Under class identity, in exposed respective non-exposed districts we obtain:

$\begin{array}{l}{\Delta }_e^{\iota P}\left(\iota \right)=\frac{\kappa }{2}{\psi }^2+\frac{{\eta }^2}{8\left(2v-1\right)}\text{ and }{\Delta }_e^{\iota C}\left(\iota \right)=\frac{\kappa }{2}{\psi }^2+\frac{9{\eta }^2}{8\left(2v-1\right)},\\ {\Delta }_n^{\iota j}\left(\iota \right)=\frac{\kappa }{2}{\psi }^2+\frac{{\eta }^2}{8\left(2v-1\right)}\text{, }\iota \text{=}L,U\text{and }j=C,P.\end{array}$

Under cultural identity, in exposed respective non-exposed districts we have:

${\Delta }_z^{iC}\left(C\right)=\frac{1}{2}{\epsilon }^2+\frac{{\eta }^2}{8\left(2v-1\right)}\text{ and }{\Delta }_z^{iP}\left(P\right)=\frac{1}{2}{\epsilon }^2,\text{for }i=U,L\text{and }z=e,n.$

Voter $ij$ identifies with the most salient ingroup $\iota$ , economic or cultural, provided he or she feels similar enough to it. Voter $ij$ solves:

$\iota \left(ij\right)=\arg \underset{\iota \in \left\{i,j\right\}}{\underbrace{\max }}\Delta \left(\iota ,-\iota \right)-\lambda {\Delta }^{ij}\left(\iota \right),$

where $\lambda \ge 0$ is the relative weight attached to dissimilarity.

A progressive voter chooses cultural identity if:

$2\kappa {\psi }^2+\frac{2{\eta }^2}{4\left(2v-1\right)}-\frac{\lambda {\epsilon }^2}{2}>2{\epsilon }^2-\lambda \left[\frac{\kappa {\psi }^2}{2}+\frac{{\eta }^2}{8\left(2v-1\right)}\right],$

which reads after rearranging:

${\eta }^2>4\left(2v-1\right)\left({\epsilon }^2-\kappa {\psi }^2\right).$

A conservative voter in a non-exposed district chooses cultural identity if:

$2\kappa {\psi }^2+\frac{2{\eta }^2}{4\left(2v-1\right)}-\lambda \left[\frac{{\epsilon }^2}{2}+\frac{{\eta }^2}{2\left(2v-1\right)}\right]>2{\epsilon }^2-\lambda \left[\frac{\kappa {\psi }^2}{2}+\frac{{\eta }^2}{8\left(2v-1\right)}\right],$

which after rearranging reads as follows:

${\eta }^2\left(\frac{4-3\lambda }{4+\lambda }\right)>4\left(2v-1\right)\left({\epsilon }^2-\kappa {\psi }^2\right).$

A conservative voter in an exposed district chooses cultural identity if:

$2\kappa {\psi }^2+\frac{2{\eta }^2}{4\left(2v-1\right)}-\lambda \left[\frac{{\epsilon }^2}{2}+\frac{{\eta }^2}{2\left(2v-1\right)}\right]>2{\epsilon }^2-\lambda \left[\frac{\kappa {\psi }^2}{2}+\frac{9{\eta }^2}{8\left(2v-1\right)}\right],$

which after rearranging reads:

${\eta }^2\left(\frac{4+5\lambda }{4+\lambda }\right)>4\left(2v-1\right)\left({\epsilon }^2-\kappa {\psi }^2\right).$

As a next step, it is apt to study identity switches. For this sake, define $\underset{\_}{\eta }\equiv 2{\left[\left(4+\lambda \right)\left(2v-1\right)\left({\epsilon }^2-\kappa {\psi }^2\right)/\left(4+5\lambda \right)\right]}^{1/2}$ and $\overline{\eta }\equiv 2{\left[(2v-1)({\epsilon }^2-\kappa {\psi }^2)\right]}^{1/2}$ with $\overline{\eta }>\underset{\_}{\eta }.$ If ${\epsilon }^2>\kappa {\psi }^2$ and $\eta \approx 0$ , none of the inequalities exposed above hold: all voters identify with their income class. If $\eta$ increases and lies within the interval $\left(\underset{\_}{\eta },\overline{\eta }\right)$ , conservative voters in exposed districts switch to cultural identity, all other voters remain income-class identified. If $\eta$ rises above $\overline{\eta }$ , but ${\eta }^2\left(4-3\lambda \right)/\left(4+\lambda \right)<4\left(2v-1\right)\left({\epsilon }^2-\kappa {\psi }^2\right)$ , conservative voters in exposed districts and all progressive voters switch to cultural identity, conservative voters in non-exposed districts remain income-class identified. If $\eta$ increases above $\overline{\eta }$ and ${\eta }^2\left(4-3\lambda \right)/\left(4+\lambda \right)>4\left(2v-1\right)\left({\epsilon }^2-\kappa {\psi }^2\right)$ , all voters switch to cultural identity.

These considerations bring Gennaioli and Tabellini (2024: p. 24) to the following:

Proposition: Assume that ${\epsilon }^2>\kappa {\psi }^2$ and $\lambda >4/3$ . There is a threshold $\underset{\_}{\eta }>0$ such that, if $\eta <\underset{\_}{\eta }$ , all voters identify with their income-class, while if $\eta >\underset{\_}{\eta }$ , conservative voters in exposed districts switch to cultural identity. Conservative voters in non-exposed districts are all-income class identified. The identity of progressives depends on $\eta$ , but it is the same in all districts.

Thus, greater import exposure raises the salience of cultural conflict, since it “heightens conflict over trade policy between conservatives versus progressives... Conservatives demand a restrictive trade policy only in exposed districts. As $\eta$ rises, they feel more like the average conservative in-group, who also demands more protection, than to their (income-)class. The opposite happens in non-exposed districts, where conservatives do not lose from trade. Here, higher $\eta$ makes non-exposed conservatives less like their cultural group.” (Gennaioli and Tabellini, 2024: p. 24).

4.2.3. Conservative Voters’ Political Demand for Restrictive Social Policy

In line with social psychology, identity distorts beliefs by “depersonalization”: the typical voter moves his or her opinions toward those that are stereotypical of the in-group, namely which are more frequent in the average in-group $\left({\epsilon }^{\iota },{\psi }^{\iota }\right)$ compared to out-group $\left({\epsilon }^{-\iota },{\psi }^{-\iota }\right)$ . The beliefs of voter $ijz$ when he or she identifies with group $\iota$ read as follows:

$\begin{array}{l}{\epsilon }_{\iota }^{i\text{}jz}={\epsilon }^{ijz}+\theta \left({\epsilon }^{\iota }-{\epsilon }^{-\iota }\right),\\ {\psi }_{\iota }^{i\text{}jz}={\psi }^{ijz}+\theta \left({\psi }^{\iota }-{\psi }^{-\iota }\right),\iota =i,j,\end{array}$

where ${\epsilon }_{\iota }^{ijz}$ respective ${\psi }_{\iota }^{ijz}$ is the belief of voter $ij$ from district $z=e,n$ under the identity regime $\iota$ and $\theta =\chi /\left(1-2\chi \right),\chi <1/2$ with $\chi \ge 0$ capturing the strength of stereotyping. This has consequences for policy preferences. The optimal solution for voter $ijz$ who identifies i.e. with his or her income-class $\iota =\epsilon$ is:

$\begin{array}{l}{\tau }_{\epsilon }^{Ljn}={\tau }_{\epsilon }^{Lje}=\epsilon \left(1+2\theta \right),{\tau }_{\epsilon }^{Ujn}={\tau }_{\epsilon }^{Uje}=-\epsilon \left(1+2\theta \right),j=C,P,\\ r_{\epsilon }^{iPn}=r_{\epsilon }^{iPn}=\psi ,r_{\epsilon }^{iCn}=r_{\epsilon }^{iCn}=-\psi ,i=U,L.\end{array}$

If voter $ijz$ identifies with his or her culture, his or her policy demands are:

$\begin{array}{l}{\tau }_{\psi }^{Ljn}={\tau }_{\psi }^{Lje}=\epsilon ,{\tau }_{\psi }^{Ujn}={\tau }_{\psi }^{Uje}=-\epsilon ,j=C,P,\\ r_{\psi }^{iPn}=r_{\psi }^{iPe}=\psi \left(1+2\theta \right),r_{\psi }^{iCn}=r_{\psi }^{iCe}=-\psi \left(1+2\theta \right),i=U,L.\end{array}$

Demands in a policy domain, by each voter type, do not differ across exposed and non-exposed districts within a given identity regime. Assume that at time $t=0$ all voters identify with their income-class $\epsilon$ : Then voter types exhibit identical demands across districts, and so do average demands: $r^{n0}=r^{e0}=\left(1/2\right)\psi -\left(1/2\right)\psi =0$ and ${\tau }^{n0}={\tau }^{e0}=\left(1/2\right)\epsilon \left(1+2\theta \right)-\left(1/2\right)\epsilon \left(1+2\theta \right)$ = 0, where $q^{z0},z=n,e$ and ${\tau }^{z0},z=n,e$ are the average policy demands in district $z=n,e$ at time $t=0$ . In other words: in the baseline, all districts are identical. Suppose now that exposure to trade increases to $\left({\eta }_{Ce},{\eta }_P\right)$ . Then, only conservative voters in exposed districts switch to culture: $r^{e1}-r^{e0}=\left(1/2\right)\psi$ $-\left(1/2\right)\psi \left(1+2\theta \right)=-\psi \theta$ , while $r^{n1}-r^{n0}=0$ , ${\tau }^{e1}-{\tau }^{e0}=0$ and ${\tau }^{n1}-{\tau }^{n0}=0$ . The demand for less progressive cultural policy in exposed districts is concentrated among conservative voters. For $j=C$ , the change in $r$ is $2\psi \theta$ in $z=e$ and $0$ in $z=n$ . For $j=P$ , there is no change within any district and hence no differences across districts. Moreover, while the average demand for redistribution does not change within and across districts, it drops in exposed districts compared to non-exposed districts if one condition on lower class voters hold: $\Delta {\tau }_e^L\equiv {\tau }_{\epsilon }^{Le,1}-{\tau }_{\epsilon }^{Le,0}=-\epsilon \theta <\Delta {\tau }_n^L\equiv {\tau }_{\epsilon }^{Ln,1}-{\tau }_{\epsilon }^{Ln,0}=0$ . Assume that exposure to trade rises to $\eta >{\eta }_P\text{but }{\eta }^2\left(4-3\lambda \right)/\left(4+\lambda \right)<4\left(2v-1\right)\left({\epsilon }^2-\kappa {\psi }^2\right)$ . Then, also progressive voters switch to culture, but not conservative voters in non-exposed districts. $\Delta r_e\equiv r^{e1}-r^{e0}=\left(1/2\right)\psi \left(1+2\theta \right)-\left(1/2\right)\psi \left(1+2\theta \right)=0$ , $\Delta r_n=r^{n1}-r^{n0}=\left(1/2\right)\psi \left(1+2\theta \right)-\left(1/2\right)\psi =\left(1/2\right)\psi ,\text{while }\Delta {\tau }_e\equiv {\tau }^{e1}-{\tau }^{e0}=0$ and $\Delta {\tau }_n\equiv {\tau }^{n1}-{\tau }^{n0}=0$ . Also in this case, the lower-class demand for progressive social policy in exposed districts is concentrated among conservative voters, and there is a reduction in the demand for redistribution by lower class voters across exposed and non-exposed districts: $\Delta {\tau }_e^L\equiv {\tau }_{\epsilon }^{Le,1}-{\tau }_{\epsilon }^{Le,0}$ $=-2\epsilon \theta <\Delta {\tau }_n^L\equiv {\tau }_{\epsilon }^{Ln,1}-{\tau }_{\epsilon }^{Ln,0}=-\epsilon \theta$ . These results lead Gennaioli and Tabellini (2024: p. 25) to the following:

Proposition (Voters’ political demand). A trade shock, i.e. a higher $\eta$ , causing some voters to switch to cultural identity, affects exposed versus non-exposed districts as follows:

a) Conservative voters demand more conservative social policy, $\Delta r_e^C<\Delta r_n^C$ , while progressive voters are unaffected, $\Delta r_e^P=\Delta r_n^P$ . Thus, average demand for progressive social policies drops, $\Delta r_e<\Delta r_n$ .

b) The demand for redistribution drops for the lower class, $\Delta {\tau }_e^L<\Delta {\tau }_n^L,$ and rises for the upper class, $\Delta {\tau }_e^U>\Delta {\tau }_n^U$ , leaving average demand for redistribution unchanged.

Hence, foreign-imports exposed conservatives switch to cultural identity and demand a more restrictive social policy in comparison to non-exposed districts where identity does not change⁵. “The identity switch also de-polarizes redistributive conflict in the exposed districts, because exposed conservatives move away from their class.” (Gennaioli and Tabellini, 2024: p. 25).

4.2.4. Political Supply of Restrictive Social Policy after a Trade Shock

There are two parties $\xi =D,R$ (as in US), “left” D and “right” R, each fielding a candidate in each district $z$ , who is fully trusted by only some voters. Candidates try to maximize their vote share in their district through simultaneously announcing policy platforms $Y_{z\xi }=\left({\tau }_{z\xi },r_{z\xi },t_{z\xi }\right),\xi =R,D$ . Moreover, they elicit propaganda $a_{\iota z\xi }$ for each identity group $\iota$ of connected voters. A candidate may engage in economic and cultural propaganda at strictly convex costs $\left(c/2\right)\left(a_{\iota z\xi }^2+a_{-\iota z\xi }^2\right),c>0$ . Party $R$ is connected to the upper class and to social conservatives, party $D$ is connected to the lower class and social progressives. Voters not connected to party$R$ are the lower class and progressive types, $ijz=LP$ while voters not connected to party $D$ in district z are the upper class and conservative types, $ijz=UC$ . $0<\alpha <1/4$ of voters not connected to party $\xi$ does not believe its policy promises.

In line with Persson and Tabellini (2000), Gennaioli and Tabellini (2024) assume probabilistic voting. Let $W^{ijz}\left({\widehat{Y}}_{\xi }^{ijz}\right)$ denote the expected welfare of voter $ijz$ if party $\xi$ wins the election, with ${\widehat{Y}}_{\xi }^{ijz}$ being the policy vector which voter $ij$ in district $z$ expects if party $\xi$ wins. Then, voter $k$ of type $ijz$ votes for $R$ if:

$W^{ijz}\left({\widehat{Y}}_R^{ijz}\right)-W^{ijz}\left({\widehat{Y}}_D^{ijz}\right)\ge {\tilde{\delta }}^k$,

where ${\tilde{\delta }}^k$ is a voter-specific popularity shock favoring party $D$ . It is uniformly distributed with mean 0 and density $\Phi$ . Then, in each district $z$ , each party $\xi$ solves the following problem:

$\underset{\left\{a_{\psi z\xi },a_{\epsilon z\xi },{\tau }_{z\xi },r_{z\xi },t_{z\xi }\right\}}{\text{max}}{V}_{z\xi }=\underset{\left\{a_{\psi z\xi },a_{\epsilon z\xi },{\tau }_{z\xi },r_{z\xi },t_{z\xi }\right\}}{\text{max}}\left(1/4\right){\displaystyle \sum _{ij}^{}{\pi }_{\iota z\xi }^{ij}}-\left(c/2\right){\left(a_{\psi \text{}z\xi }\right)}^2-\left(c/2\right){\left(a_{\psi \text{}z\xi }\right)}^2$

where $a_{\rho z\xi }$ is persuasion effort by party $\xi$ in district $z$ toward its in-group voters identified along dimension $\varsigma =\epsilon ,\psi$ , by taking into account that ${\chi }^{\iota }=\chi +a_{\varsigma z\xi }$ if $\iota =U,C$ and $\xi =R$ or if $\iota =L,P$ and $\xi =D$ , where $\iota$ is the group a voter of type $ij$ in district $z$ identifies with. Party propaganda is formalized as a costly effort to change ${\chi }^{\iota }$ for the group to which the party is connected. Party $R$ is connected to conservative (C) and upper class (U) groups, thus it affects ${\chi }^C$ and ${\chi }^U$ . Similarly for party $D$ . As a consequence, the distortion parameter $\theta$ turns out to be group-specific as follows: ${\theta }^C={\chi }^C/\left(1-{\chi }^C-{\chi }^P\right)$ and ${\theta }^P={\chi }^P/\left(1-{\chi }^P-{\chi }^C\right)$ .

A voter of type $ij$ in district $z$ votes for party $\xi$ with probability:

$\begin{array}{c}{\pi }_{\iota z\xi }^{ij}=\left(1/2\right)+\left(\Phi /2\right)[\kappa \left({\widehat{r}}_{z\overline{\xi }}-{\widehat{r}}_{z\xi }\right)\left({\widehat{r}}_{z\overline{\xi }}+{\widehat{r}}_{z\xi }-2r_{\varsigma z}^{ij}\right)+\left({\widehat{\tau }}_{z\overline{\xi }}-{\widehat{\tau }}_{z\xi }\right)\left({\widehat{\tau }}_{z\overline{\xi }}+{\widehat{\tau }}_{z\xi }-2{\tau }_{\varsigma z}^{ij}\right)\\ +\phi \left({\widehat{t}}_{z\overline{\xi }}-{\widehat{t}}_{z\xi }\right)\left({\widehat{t}}_{z\overline{\xi }}+{\widehat{t}}_{z\xi }-2t_{\varsigma z}^{ij}\right),\end{array}$

where in ${\pi }_{\iota z\xi }^{ij}$ index $\iota$ refers to the in-group of voter $ij$ when the identity regime is $\varsigma =\epsilon ,\psi ,\phi =q^2\left(2v-1\right)$ , $t_z^{ij}$ is the preferred tariff of voter $ij$ in district $z$ and $\overline{\xi }$ denotes the competitor of party $\xi$ . The FOC for party $\xi$ in district $z$ yield:

$\begin{array}{l}{r}_{z\xi }={\displaystyle \sum _{ij}^{}{\alpha }_{\xi }^{ij}}{\psi }_{\varsigma z}^j,{\tau }_{z\xi }=-{\displaystyle \sum _{ij}^{}{\alpha }_{\xi }^{ij}}{\epsilon }_{\varsigma z}^i,t_{z\xi }={\displaystyle \sum _{ij}^{}{\alpha }_{\xi }^{ij}}{t}_z^j,\\ \frac{\partial V_{z\xi }}{\partial a_{\varsigma \xi z}}=\frac{1}{4}{\displaystyle \sum _{ij}^{}\Phi \left[\kappa \left({\widehat{r}}_{\xi }^{ij}-{\widehat{r}}_{\overline{\xi }}^{ij}\right)\right]}\frac{\partial r_{\iota }^{ij}}{\partial {\theta }_{ij}}+\left({\widehat{\tau }}_{\xi }^{ij}-{\widehat{\tau }}_{\overline{\xi }}^{ij}\right)\frac{\partial {\theta }^{ij}}{\partial a_{\varsigma \xi z}}-a_{\varsigma \xi z}=0,\varsigma =\epsilon ,\psi .\end{array}$

Notice that $t_{zR}>t_{zD}\text{for }z=e\text{and }{t}_{zR}=t_{zD}\text{for }z=n,$ i.e. party $R$ proposes a strictly larger tariff than party $D$ in exposed districts.

If $\eta$ increases to the point that conservative voters in exposed districts switch to culture, the party platforms in exposed districts become $r_{eR}^{*}=-\left(1/\left(1-\alpha \right)\right)\psi {\theta }_{\psi e}-\left(\alpha /\left(1-\alpha \right)\right)\psi ,$ $r_{eD}^{*}=-\left(1/\left(1-\alpha \right)\right)\psi {\theta }_{\psi e}+\left(\alpha /\left(1-\alpha \right)\right)\psi \left(1+2{\theta }_{\psi e}\right),{\tau }_{eR}^{*}={\tau }^o-\left(\alpha /\left(1-\alpha \right)\right)\epsilon \left(1+2{\theta }_{\epsilon e}\right),$ ${\tau }_{eD}^{*}={\tau }^o-\left(\alpha /\left(1-\alpha \right)\right)\epsilon {\theta }_{\epsilon e}$ . The divergence between party $D$ and party $R$ is: $r_{eD}^{*}-r_{eR}^{*}=$ $\left(2\left(\alpha /\left(1-\alpha \right)\right)\right)\psi \left(1+{\theta }_{\psi e}\right),r_{nD}^{*}-r_{nR}^{*}=\left(2\left(\alpha /\left(1-\alpha \right)\right)\right)\psi ,$ ${\tau }_{eD}^{*}-{\tau }_{eR}^{*}=\left(2\left(\alpha /\left(1-\alpha \right)\right)\right)\epsilon \left(1+{\theta }_{\epsilon e}\right),$ ${\tau }_{nD}^{*}-{\tau }_{nR}^{*}=\left(2\left(\alpha /\left(1-\alpha \right)\right)\right)\epsilon \left(1+2{\theta }_{\epsilon n}\right)$ . Thus: $r_{eD}^{*}-r_{eR}^{*}>\left(r_{nD}^{*}-r_{nR}^{*}\right)$ . Moreover, ${\tau }_{eD}^{*}-{\tau }_{eR}^{*}$ < (${\tau }_{nD}^{*}-{\tau }_{nR}^{*}$ ). These results lead Gennaioli and Tabellini (2024: p. 26) to the following

Proposition (Political supply). A rise in trade exposure $\eta$ which causes some voters to switch from class to cultural identity induces the following effects in exposed compared to non-exposed districts:

a) Candidates from both parties announce more conservative social policies in exposed districts, but especially party $R$ candidates.

b) Party $R$ candidates in exposed districts propose higher tariffs than party $D$ candidates. In non-exposed districts both party candidates announce equal tariffs.

c) Party $D$ candidates announce a less redistributive policy while party $R$ propose the same or a more distributive policy such that the divergence in $\tau$ decreases.

These results depend on persuasion effort through stereotypes. In exposed districts parties engage in symmetric economic persuasion $a_{\epsilon eR}^{*}=a_{\epsilon eD}^{*}=a_{\epsilon e}^{*}>0$ . Gennaioli and Tabellini (2024: Online Appendix, 8) show that $a_{\epsilon n}^{*}>a_{\epsilon e}^{*}>0$ and thus ${\theta }_{\epsilon e}<{\theta }_{\epsilon n}$ which means the trade shock causes economic stereotypes to fall in exposed districts. Moreover, $a_{\psi eR}^{*}>0>a_{\psi eD}^{*}$ which means that party $R$ candidates in exposed districts fuel conservative stereotypes while party $D$ candidates reduce progressive stereotypes. The effects of trade exposure are that party $R$ increases conservative propaganda and both parties decrease class propaganda (Gennaioli and Tabellini, 2024: p. 26).

5. Finally: Social Conservatism and Trade Protection from a Natural Law, Economic-Liberal Perspective

The exchange of material for psycho-social benefits and the cultural predisposition for local “community values” make it understandable why, after a phase of social and foreign trade liberalism, social conservatism and foreign trade protectionism became politically powerful. However, the question remains whether the white, evangelical Christians, over 80 percent of whom voted in 2016 for Trump, who advocated conservative Christian social policy and foreign trade protection, only cast their vote to strengthen their self-esteem, as has been suggested so far. They probably also voted for Trump and against Clinton so that someone in the political arena would stand up for the highest moral goods, such as the right to life from conception to natural death. This is not just a local community value, but according to natural law a universal value that is inextricably linked to the nature and dignity of every human being (including the unborn). There are therefore goods that must not be subjected to a psycho-socially expanded utility calculation. Christians are probably more capable of understanding the inviolability of human life from conception to natural death than non-Christians, but according to natural law all people are capable of “of distinguishing reasonable and unreasonable choices and better and worse value judgments.” (Malhoney, 2004: p. 560).

If this is true, it is probably not entirely unreasonable to assume that the typical Trump voter also voted against economic freedom without recognition of higher values. The fact that there is no general agreement with natural law today is, as Samuel Gregg (2018: p. 2) rightly points out, not a strong argument against a liberal order based on natural law. “On the contrary, it underscores the imperative of ... illustrating how natural law can supply a coherent basis for liberal institutions that liberal ideology can’t.” The liberal ideology we are talking about here means complete individual autonomy without recognition of higher moral goods, and freedom from coercion by third parties (the state) alone. If the vote for Trump was motivated by protest this empty freedom, it was legitimate in the author’s view. Unordered freedom had rightly become a chimera to be abandoned.

This leads to the question of whether an “ordered” freedom (Gregg, 2003) includes or excludes foreign trade protectionism. In agreement with Dominican Francisco de Vitorio and Hugo Grotius, according to Gregg (2013: p. 90), ordered freedom also excludes protectionism because people have a natural right to free exchange and free trade. This is not an absolute right, but the burden of proof lies with those who oppose free trade and support foreign trade protection. Therefore, in contrast to the currently politically dominant position, social conservatism and free trade are compatible with natural law, which is accessible to all people. This allows on it hope that itself by “explaining natural law to those many intelligent people of good will who have never encountered natural law or—more commonly—have been exposed to caricatures of it [and] defending natural law against critics” (Gregg, 2018: p. 3) the unpleasant connection between social conservatism and foreign trade protectionism again resolves.

6. Conclusion

This review article deals with the question why Americans voted again for a presidential candidate who presented himself as a social-conservative and as a tariff man to the US electorate already in 2010. The author reviews recent political-economic models which provide microeconomic-psychological explanations of the switch of American voters and politicians to social-conservatism, economic nationalism and foreign trade protection.

Commencing, however, with macroeconomics, Winkler (2017) explains Trump’s 2016 election success by his turning away from the international supply policy of Trump’s presidential predecessors and Trump’s turning towards restrictions on free international trade and free immigration to give the nation state back its ability to act. Winkler (2017) points out also socio-cultural factors as the rejection of progressive social-liberal values for Trump’s election victory over Clinton.

Turning to (microeconomic-) psychological answers Mutz (2018) firstly attributes Trump’s 2016 election victory to white Christian men in the Northeast and Midwest of USA fearing a loss of social status. Secondly, Mutz’s (2018) comparative empirical analysis of voter opinion polls shows that, in 2016, noticeably more voters than in 2012 evaluated international trade negatively, being threatened by growing dependence on foreign countries.

The importance of group status in the formation of political preferences is taken up by Grossman and Helpman (2018, 2021) to explain within a Heckscher-Ohlin model of international trade why an import tariff is in the interest of both the voters and the candidate who promotes the tariff. The clue is that voters’ preferences are not based solely on material self-interest but also on the well-being of members of those social groups with which the person in question identifies. The psycho-social component combines positive feelings derived from pride in the status of the in-group and a dissonance cost borne from identifying with the out-group. In line with the change in US voter preferences between 2012 and 2016 empirically identified by Mutz (2018), Grossman and Helpman (2021) explain Trump’s introduction of import tariffs in 2018 by a change in the identification regime where the less qualified of US voters stopped identifying with the nation but rather, they equate the “nation” with their own reference group. While this reduces the redistributive benefit of a tariff increase, the benefit of a higher tariff rises since the psychological dissonance costs of identifying with “different people” are eliminated. Less qualified prefer a disruptive tariff increase if the latter effect is larger than the former which is certainly true when the higher qualified are branded as corrupt elite during a populist campaign.

Complementing Grossman and Helpman (2021), Gennaioli and Tabellini (2018, 2019) focus on the American losers of relatively free international trade and explain why they become nationalistic, hate immigrants and turn socially conservative. Assuming both that voters identify with the social group whose interests are closest to theirs and that features the strongest policy conflict with out-groups, and identification causes voters to slant their beliefs of self and others toward group stereotypes, they explain observed distortions in political beliefs, actual polarization, and consequences of changing political cleavages as the effects of inter alia foreign trade shocks.

Moreover, Gennaioli and Tabellini (2024) based on Bonomi, Gennaioli and Tabellini (2021) model political demand for and supply of protectionist social conservatism because of cultural instead of income conflict becoming more salient in the USA during the past decade. While US voters in regions more exposed to the China shock have become more anti-immigrants if religious and demand less redistribution if poor than in the past, also congressmen in exposed districts adopted a more conservative rhetoric, particularly if Republican. Through their theory of changing dimensions of political polarization based on endogenous social identity and by adding parties spreading stereotypes to receptive voters, Gennaioli and Tabellini (2024) show why the cultural conflict between US⁶ voters and parties is growing, why despite rising income inequality the redistributive conflict is dampening, and why lower-class voters affected by the “China Shock” realigned from the left to the right.

Finally, in contrast to the former and currently politically dominant position of the former and recent Trump administration, social conservatism and free trade are compatible with natural law, which is accessible to all people of good will.

Acknowledgements

The author is grateful to an anonymous referee whose comments helped to extend the scope of the paper.

NOTES

¹That this promise is indeed rather dubious, in any case in the long run, underlines Kotsunis (2025) by claiming that both the growth rate of world GDP will decline, and US GDP will suffer from recent tariff increases.

²Oxford Economics (2021) found that contrary to Trump’s first term promises to protect US industry and US employment from Chinese competitors, the US-China trade war cost the USA 245,000 jobs with a 70% decline in agricultural exports to China and a 0.8% decline in US manufacturing gross value added 2020 compared to the pre-Trump no-tariffs situation.

³This also applies in a large open economy like the USA, because the terms of trade effect in the large open economy does not affect the psycho-social benefit components.

⁴With co-author Bonomi, Gennaioli and Tabellini (2021) published a refined and expanded version of the (2019) paper albeit under the same title.

⁵Gennaioli and Tabellini (2024: pp. 27-31) provide persuasive survey evidence that respondents in US commuting zones more exposed to the China shock favor social conservatism.

⁶Danieli et al. (2024) and Colantone et al. (2022) show that the US experience with protectionist conservatism also compare with European countries seeing rising nationalism and protectionism.

Conflicts of Interest

There is no conflict of interests.

References

[1]	Akerlof, G. A., & Kranton, R. E. (2000). Economics and Identity. Quarterly Journal of Economics, 115, 715-753. https://doi.org/10.1162/003355300554881
[2]	Alesina, A., Stantcheva, St., & Miano, A. (2018). Immigration and Redistribution. NBER Working Paper No. 24733.
[3]	Autor, D., Dorn, D., Hanson, G., & Majlesi, K. (2020). Importing Political Polarization? The Electoral Consequences of Rising Trade Exposure. American Economic Review, 110, 3139-3183. https://doi.org/10.1257/aer.20170011
[4]	Bonomi, G., Gennaioli, N., & Tabellini, G. (2021). Identity, Beliefs, and Political Conflict. The Quarterly Journal of Economics, 136, 2371-2411. https://doi.org/10.1093/qje/qjab034
[5]	Colantone, I., Ottaviano, G., & Stanig, P. (2022). The Backlash of Globalization. In E. Help-man, G. Gopinath, & K. Rogoff (Eds.), Handbook of International Economics (pp. 405-477). Elsevier. https://doi.org/10.1016/bs.hesint.2022.02.007
[6]	Danieli, O., Gidrom, N., Kikuchi, S., & Levy, R. (2024). Decomposing the Rise of the Popu-list Radical Right. Working Paper.
[7]	Enke, B. (2018). Moral Values and Voting. NBER Working Paper No. 24268.
[8]	Farmer, K., & Schelnast, M. (2021). Growth and International Trade. An Introduction to the Overlapping Generations Approach (2nd ed.). Springer.
[9]	Gennaioli, N., & Tabellini, G. (2018). Identity, Beliefs, and Political Conflict. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3300726
[10]	Gennaioli, N., & Tabellini, G. (2019). Identity, Beliefs, and Political Conflict. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3418631
[11]	Gennaioli, N., & Tabellini, G. (2024). Presidential Lecture: Identity Politics. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.4410239
[12]	Gregg, S. (2003). On Ordered Liberty. A Treatise on Free Society. Lexington Books. https://doi.org/10.5771/9780739158401
[13]	Gregg, S. (2013). Tea Party Catholic. The Catholic Case for Limited Government, a Free Economy, and Human Flourishing. The Crossroad Publishing Company.
[14]	Gregg, S. (2018). Why “Liberalism” Needs Natural Law. https://www.lawliberty.org/2018/05/10/why-liberalism-needs-natural-law-integralism-burke-political-economy/
[15]	Grossman, G. M., & Helpman, E. (2021). Identity Politics and Trade Policy. The Review of Economic Studies, 88, 1101-1126. https://doi.org/10.1093/restud/rdaa031
[16]	Grossman, G., & Helpman, E. (1994). Protection for Sale. American Economic Review, 84, 833-850.
[17]	Grossman, G., & Helpman, E. (2018). Identity Politics and Trade Policy. NBER Working Paper No. 25348.
[18]	Kotsunis, T. (2025). Global Trade Wars Warning If Trump Embraces US Protectionism. https://www.pinsentmasons.com/out-law/news/trump-us-election-global-trade#:~:text=Ultimate-ly%2C%20a%20more%20protectionist%20US,he%20becomes%20president%20in%20January/
[19]	Malhoney, D. J. (2004). Review of “On Ordered Liberty”. Markets & Morality, 7, 559-561.
[20]	Müller, J. (2016). What Is Populism? University of Pennsylvania Press. https://doi.org/10.9783/9780812293784
[21]	Mutz, D. C. (2018). Status Threat, Not Economic Hardship, Explains the 2016 Presidential Vote. Proceedings of the National Academy of Sciences, 115, E4330-E4339. https://doi.org/10.1073/pnas.1718155115
[22]	Oxford Economics (2021). The US-China Economic Relationship. A Crucial Partnership at a Critical Juncture. https://www.oxfordeconomics.com/resource/the-us-china-economic-relationship-a-crucial-partnership-at-a-critical-juncture/
[23]	Persson, T., & Tabellini, G. (2000). Political Economics. Explaining Economic Policy. The MIT Press.
[24]	Rodrik, D. (2011). The Globalization Paradox: Democracy and the Future of the World Economy. Asean Economic Bulletin, 28, 420. https://doi.org/10.1355/ae28-3k
[25]	Shayo, M. (2009). A Model of Social Identity with an Application to Political Economy: Nation, Class, and Redistribution. American Political Science Review, 103, 147-174. https://doi.org/10.1017/s0003055409090194
[26]	Tajfel, H. (1974). Social Identity and Intergroup Behaviour. Social Science Information, 13, 65-93. https://doi.org/10.1177/053901847401300204
[27]	Turner, G., Hogg, M. A., Reicher, St. D., & Wetherel, M. S. (1987). Rediscovering the Social Group: A Self-Categorization Theory. Basil Blackwell.
[28]	Weede, E. (2018). Protectionism Instead of Free Trade Endangers Our Future. ORDO, 68, 91-102. https://doi.org/10.1515/ordo-2018-0005
[29]	Winkler, A. (2017). Macroeconomics and Populism. Wirtschaftsdienst, 97, 115-123. https://doi.org/10.1007/s10273-017-2093-6