“Buy local” campaigns in times of crisis: insights from reactance theory

Abstract

Purpose

Reactance theory is applied to investigate consumer responses to “buy local” campaigns initiated by government to counteract the effects of an economic crisis, using the COVID-19 pandemic as an illustrative context.

Design/methodology/approach

A conceptual model is developed, aimed at revealing the extent to which “buy local” campaigns – explicitly justified by the need to fight an economic crisis – are likely to lead to (a) compliance (i.e. support for local products/retailers) or (b) freedom restoration (i.e. support for foreign products/retailers). The model is subsequently tested on samples of German (N = 265) and Italian (N = 268) consumers.

Findings

“Buy local” campaigns are likely to generate reactance amongst consumers and such reactance can lead to both non-compliance and, albeit less so, freedom restoration outcomes. At the same time, consumer ethnocentrism acts as a countervailing influence by attenuating the effects of generated reactance and its undesirable outcomes.

Research limitations/implications

Psychological reactance theory offers a novel perspective for conceptually approaching the likely responses of consumers towards “buy local” campaigns and the empirical findings support the use of the theory in this context.

Practical implications

Policymakers seeking to encourage consumers to support the local economy during times of an economic crisis need to be aware that “buy local” campaigns may, against their intended communication goals, result in non-compliance as well as consumer responses in the opposite direction. Thus, the reactance-generating potential of such campaigns needs to be explicitly considered at the planning/implementation stage.

Originality/value

The findings confirm the relevance of reactance theory as a conceptual lens for studying the effects of “buy local” campaigns and have important implications for domestic/foreign firms as well as for policy makers seeking to encourage consumers to support the local economy during times of an economic crisis.

Citation

Matarazzo, M., Diamantopoulos, A. and Raff, A. (2023), "“Buy local” campaigns in times of crisis: insights from reactance theory", International Marketing Review, Vol. 40 No. 7, pp. 45-78. https://doi.org/10.1108/IMR-09-2022-0202

Introduction and background

“Swadeshi is that spirit in us which restricts us to the use and service of our immediate surroundings to the exclusion of the more remote” (Gandhi, 2009, p. 11) wrote Mahatma Gandhi on the Swadeshi movement (Eng.: made and sold or used in India (Cambridge Dictionary, no date)), a movement that aimed at achieving India's independence through, among other things, encouraging people to buy more local products and support local producers (Balasubramanian et al., 2021). Initiatives that try to get people to “buy local” [1] can seemingly be traced back to the Roman empire where “the call to prefer goods that had been produced within the realm made its appearance as a topic of mass media communication” (Kühschelm, 2020, p. 82). More recently, former US president Trump's “America First” campaign epitomizes such initiatives as illustrated in his 2017 inaugural address in which he vowed to “follow two simple rules: buy American, and hire American” (Mittal, 2017, p. 20).

“Buy local” appeals can be initiated and disseminated by different senders. Companies can incorporate such appeals in their advertising campaigns, for example, Walmart's “Buy American,” Plymouth's “Born in America,” Miller's “Made the American Way” and Chevrolet's “This is our country. This is our truck” campaigns (Ettenson et al., 1988; Bunkley, 2006; Zellner, 1992). Moreover, a “buy local” initiative can emerge from the population itself – like the Swadeshi movement in India described above – or can be set up by nongovernmental organizations, as illustrated by the United Auto Workers trade union that started the “Build Buy USA” campaign in 2017 seeking to increase national consumption of US-made articles “through the power of our wallets” or the “Buy New Zealand-Made” campaign jointly implemented by the New Zealand Manufacturers Federation and the New Zealand Council of Trade Unions (Garland and Coy, 1993). Last but not least, governments often encourage their citizens to buy domestic products (thus implicitly discouraging the purchase of foreign products) as illustrated by the “Buy Australian-Made,” “Buy Slovakian-made” and “Proudly South African” campaigns (Cameron and Elliott, 1998; Elliott and Cameron, 1994; Mtigwe and Chikweche, 2008; Saffu et al., 2010). Indeed, there has been a large increase in such government-led “buy local” campaigns in recent times given the succession of recessions the world has witnessed. Specifically, the global financial crisis (GFC) between the end of 2008 and the middle of 2009, resulted in the sharpest fall in output in recorded history and the most devastating since Second World War, with an adverse effect on international trade much more significant than during the Great Depression. Following the GFC, the global economy has continued to be characterized by imbalances and instability and has been exposed to systemic shocks, both exogenous like the COVID-19 pandemic and endogenous like the current Russian-Ukrainian conflict.

Irrespective of who initiates a “buy local” campaign, the aim of the latter is invariably “to encourage consumers to purchase locally made products in preference to imported goods” (Elliott and Cameron, 1994, p. 50) or, bluntly stated, to promote “ethnocentric consumption, favouring domestic products at the expense of foreign substitutes” (Pekkanen and Penttilä, 2021, p. 301, original emphasis). While the objectives of “buy local” campaigns have often been found to enjoy widespread support among domestic consumers (e.g. Cameron and Elliott, 1998; Saffu et al., 2010), concern has been voiced that they can backfire (Insead et al., 1991). Such concern seems to be well-justified as several studies in different countries have found that “buy-local” campaigns do not result in stronger preferences for and greater purchases of domestic products (e.g. see Ettenson et al., 1988 in the Fenwick and Wright, 2000 in New Zealand; Mtigwe and Chikweche, 2008 in South Africa; Saffu and Walker, 2006 in Ghana). Inevitably, such findings “seriously question the wisdom of extensive financed or nation-wide governmental support for made-in-our-country advertising and promotional campaigns” (Insch et al., 2017, p. 250).

In this paper, we argue that the theory of psychological reactance (Brehm, 1966; Brehm and Brehm, 1981) provides an explanation as to why – despite their best intentions – “buy local” campaigns may fail (Bunkley, 2006). Reactance theory argues that when personal freedom is reduced, eliminated, or threatened with elimination, a person will experience an (unpleasant) state of arousal (reactance) that induces attempts to recover or reestablish the lost or threatened behavior. The reason is that people have a natural predisposition towards preserving and restoring their personal freedoms and therefore “a threat to or loss of a freedom motivates the individual to restore that freedom” (Brehm and Brehm, 1981, p. 4). While some people may happily comply with the proposed behaviors of “buy local” campaigns, others might resist such influence attempts because they perceive them as threats to their freedom of choice. A greater threat leads to a greater magnitude of reactance and higher levels of reactance will more strongly prompt the person to reestablish the freedom that has been lost or threatened. This, in turn, may lead to lower attractivity of the proposed behavior and higher attractivity of the restricted or threatened behavior (Clee and Wicklund, 1980) [2].

Drawing on the above-mentioned theory, “buy local” campaigns may be perceived as an imposition on and a restriction of individual freedom, especially because they are related to shopping which is an integral part of people's daily life. As a result, consumers may start to view local purchase options as forced choices devaluing their attractiveness, because they see them as responsible for the threatened freedom. At the same time, consumers may become more attracted towards the threatened behavior (i.e. buying foreign goods), “consistent with the maxim, ‘the forbidden fruit is the sweetest’” (Loebnitz et al., 2022, p. 1046). Indeed, a recent study found that nationalistic appeals (suggesting that consumers should shun foreign brands for moral reasons) may, against their intended communication goal, increase the reputation of foreign brands (Bartikowski et al., 2021). In short, if reactance is at play, government “buy local” campaigns may potentially lead to the exact opposite of what was intended, namely a lower perceived desirability of national/domestic products and higher perceived desirability of foreign products. Having said that, both the magnitude of reactance and its effects are likely to be influenced by consumer ethnocentrism, namely “the beliefs held by […] consumers about the appropriateness indeed morality, of purchasing foreign-made products” (Shimp and Sharma, 1987, p. 280). Given the well-known biasing effects of consumer ethnocentrism in favor of domestic products and against foreign products (for relevant reviews, see Shankarmahesh, 2006; Myers, 2015; Balabanis and Siamagka, 2022), “buy local” campaigns may generate less reactance among ethnocentric consumers.

Against this background, the present study investigates the extent to which government “buy local” campaigns – explicitly justified by the need to fight the economic consequences of COVID-19 – are likely to be perceived as a threat to freedom and lead to (a) compliance (i.e. support for domestic product/retailers) or (b) freedom restoration (i.e. support for foreign products/retailers). Specifically, drawing on reactance theory, we develop and empirically test a conceptual model linking freedom threat to both cognitive and affective manifestations of reactance as well as capturing the impact of the latter on consumer compliance and defiance. Embedded in our model is also the level of consumer ethnocentrism which is expected to also influence reactance and its outcomes.

We intentionally chose the need to fight the economic consequences of COVID-19 as an illustrative context for “buy local” campaigns because of two reasons. First, such campaigns became commonplace in response to the economic crisis caused by COVID-19 which resulted in a worldwide drop of output by 4.3% in 2020 (thus three times the losses produced by the GFC in 2009; United Nations, 2021). Many governments around the world appealed to the solidarity of their people by asking them to shift their purchasing behavior in favor of local products. For example, in March 2020, the Italian foreign minister, Luigi Di Maio, appealed to Italians to buy (and eat) “Made in Italy” products (Leali et al., 2020). This appeal was even launched on social networks (#IoComproMadeInItaly) to support those who were at work during the crisis (the message was to buy “Made in Italy” by posting the photo of favorite Italian products in social media). In the same month, during a press conference, the Portuguese economic minister urged consumers to buy Portuguese products (Borges, 2020). Similarly, in June 2020, the French agriculture minister, Didier Guillaume, tried to sensitize his compatriots to the country of origin of the articles they buy and emphasized the importance to prioritize French products (Poingt, 2020). Similar appeals were made by government leaders in several other countries such as Austria (Szigetvari, 2020), Germany (Schöneberg, 2020) and South Africa (Krugersdorp News, 2020).

Second, the necessity of a “buy local” campaign to fight the economic consequences of COVID-19 is arguably more likely to be accepted by consumers than if such campaigns were driven by xenophobic and/or nationalistic motives (e.g. the “America First” campaign associated with former US president Donald Trump). In this context, the restoration of the home economy to pre-crisis levels is a rationale more likely to be embraced by different consumers both for the sake of their own economic well-being and because the pandemic was supposed to induce a spontaneous “we are all in this together” attitude, resulting in expected stronger support for local companies. We thus apply reactance theory under “conservative” conditions, that is, in a setting in which “buy local” campaigns are less likely to be seen as a major freedom threat resulting in high levels of reactance. Moreover, to enhance generalizability, we test our model in two country settings (Germany and Italy) that differ on pertinent background characteristics expected to impact both the magnitude of the reactance generated by a “buy local” campaign and the extent of compliance/resistance with the latter.

Our intended contribution is three-fold. First, we demonstrate the usefulness of reactance theory as a theoretical framework for studying “buy local” campaigns and offering an explanation as to why such campaigns may fail. Second, we highlight, for the first time, the role of consumer ethnocentrism as an important construct impacting reactance and its outcomes. Third, from a managerial and policy making perspective, our findings offer empirically-based insights to decision makers regarding the potential success of “buy local” campaigns and the consumer segments most likely to (not) respond in line with the campaigns' aims.

Theory of psychological reactance

Nature of reactance

When a person does directly the opposite of what he/she has been told to do, one can refer to the theory of psychological reactance to explain this behavior (Brehm, 1966). The theory posits that such a behavioral outcome would result from reasserting one's freedom, the latter being defined “as a belief that one can engage in a particular behavior” (Brehm and Brehm, 1981, p. 35), whereby “behavior” subsumes any performable act such as (not) choosing an alternative, (not) having an opinion about a particular matter, or (not) doing something (Brehm, 1966). Reactance theory has been used widely in marketing and consumer research to study issues as diverse as framing of marketing communications (Loebnitz et al., 2022), the impact of social norms on consumer behavior (Melnyk et al., 2022), revenge buying (Gupta and Mukherjee, 2022) and social media activism (Almazyad et al., 2023).

Psychological reactance plays a particularly important role in the context of persuasive communication for explaining and predicting the success and failure of messages and campaigns. Taking a particular stance or performing a specific behavior may pose a threat to the freedom of the receiver, as the person feels pressured toward a certain attitude or a particular action (Clee and Wicklund, 1980) and is therefore motivated to resist the communication or even to do the opposite (Ratcliff, 2021). Brehm and Brehm (1981) define a threat as “any force on the individual that makes it more difficult for him or her to exercise the freedom” (p. 30) and offer a classification of threats by differentiating between impersonal events, including laws, shortages in materials, accidents; self-inflicted threats such as a choice situation where choosing any option threatens the freedom to select other option(s); and social influence attempts (where the focus of the current study lies) and which includes commands, persuasion, and bribes.

A persuasive communication attempt will typically elicit, on the one hand, reactance forces that motivate one to reassert the freedom and, on the other hand, compliance forces that motivate one to act in accordance with it (Brehm and Brehm, 1981; Clee and Wicklund, 1980). The strength of these opposing forces is highly idiosyncratic and crucially dependent on the situation and the subject. For designers of persuasive communications, such as “buy local” appeals by governments, it therefore becomes essential to determine the relative strength of these forces in order to optimize the effectiveness of the communication.

Determinants of reactance

Regarding the determinants of reactance, first, the importance of the threatened freedom will be critical in determining the amount of reactance caused by the threat to this freedom (Miron and Brehm, 2006). Thus, a persuasive communication threatening an unimportant freedom can lead to “overt compliance” (Brehm and Brehm, 1981, p. 96), whereas a threat to a freedom of moderate to high importance can cause reactance effects that surpass the forces which motivate compliance and thereby lead to “boomerang” outcomes (i.e. doing the opposite of what has been suggested) (Brehm and Brehm, 1981). The importance of the freedom also sets a maximum of possible arousable reactance and is a product of its unique capacity to satisfy a specific need and the magnitude of the need (Brehm and Brehm, 1981). If, for example, a particular product has no instrumental value to satisfy a need, restricting access to it would enact little or no reactance. Similarly, a choice situation between, for example, two options where access to one is made more difficult, will only lead to substantial reactance (and a subsequent increase in attractivity of the restricted option) if both options were considered to be attractive before (Brehm and Brehm, 1981). Needless to say, that individual perception of available behavioral freedoms can vary widely. As Brehm and Brehm (1981) put it, “there will be considerable variation in people's beliefs concerning their existing freedoms” (p. 22) and “cultural patterns will contribute heavily to the specific freedoms that individuals within a given context perceive themselves to possess” (p. 28). Hence, different people will come to highly idiosyncratic conclusions about which behaviors to consider as freedoms both in general and in specific situations. In any case, only if there is a perceived freedom, it can be threatened and evoke reactance.

The magnitude of reactance is also a function of the characteristics of the threat. For example, in a choice situation, the magnitude of the threat increases with the proportion of freedoms being threatened (Brehm and Brehm, 1981). Consequently, reactance should be higher if, for example, 50% rather than 10% of the choices are eliminated or threatened (Brehm and Brehm, 1981). Similarly, the magnitude of the threat increases with an increase in the number of freedoms being threatened (Brehm and Brehm, 1981). Additionally, the perceived implication for future threats presented by a current threat can enhance reactance to this threat (Wicklund, 1974).

In the realm of persuasive communication, the magnitude of the threat also depends on how something is communicated and whether forceful or acquiescent language is being used; the former naturally causes more reactance than the latter (Miller et al., 2007; Quick and Stephenson, 2008; Zhang and Sapp, 2013). Importantly, and regardless of communication features, the mere presence of persuasive communication naturally implies a threat to freedom, as any message which has the objective to elicit a change in behavior can threaten the freedom of the receiver (Rains, 2013). When considering the fact that the perceived intent to persuade already increases the magnitude of the threat, it becomes clear how carefully a persuasive communication (such as a “buy local” campaign) has to be designed to minimize any perception of a freedom threat and account for its inherent reactance-inducing element.

Furthermore, of potential importance is the social power of the sender as determined by the latter's authority, prestige, expertise and trustworthiness, all of which add to the influence pressures elicited by a persuasive communication and thereby can increase the magnitude of the threat and eventually the reactance outcomes (Brehm and Brehm, 1981). For example, Heinemann et al. (2008) found that a political reform communicated by experts elicited more reactance than when its message was delivered by laypeople or celebrities.

Finally, one's ex ante position on a topic could be an important predictor of reactance (Brehm and Brehm, 1981). As reactance is often evoked to explain unexpected failures of persuasion (Shen and Dillard, 2005), it is not surprising that both pro- and counter-attitudinal communications can cause reactance (Brehm and Brehm, 1981; Smith, 1979; Zhao, 2017). The mechanism through which counter-attitudinal communication increases reactance is based on the discrepancy between one's own position and the position advocated (Brehm and Brehm, 1981). In contrast, the mechanism through which pro-attitudinal communication increases reactance is by threatening one's freedom to decide on one's own (Zhao, 2017). Independent of the valence of one's position, the importance to hold that position can be a predictor of reactance (Brehm and Brehm, 1981); therefore, a person who, for example, places considerable importance to holding a neutral position would probably experience equivalent amounts of reactance to communications arguing in any direction, as all directions are then supposedly of counter-attitudinal nature.

Outcomes of reactance

Regarding the consequences of reactance, faced with a persuasive communication that is perceived as a threat to freedom, a person can be expected to upgrade the attractivity of the threatened behavior and simultaneously degrade the attractivity of the proposed alternative behavior (Clee and Wicklund, 1980). Reasserting one's freedom can take different forms and be reflected in an increased willingness to behave opposite to the threat, perform a related opposing behavior, or vicariously assert the freedom by associating with others who engage in restoring the threatened freedom (Brehm and Brehm, 1981; Quick and Stephenson, 2007). More specifically, Quick and Stephenson (2007) distinguish between three different freedom restoration motivations, which have been coined “boomerang” (the motivation to do the opposite of what has been suggested), “related boomerang” (the motivation to do something which is opposed to the idea of the received communication) and “vicarious boomerang” (associating with people who oppose the threatening communication).

State vs trait reactance

At this point, it is necessary to draw a distinction between reactance as a state phenomenon (as discussed so far) and reactance as a trait phenomenon. Despite being originally conceptualized only as a state phenomenon, Brehm and Brehm (1981) noted that different personalities could differ in their perception of available freedoms and threats to freedoms, paving the way to the study of reactance also as a trait phenomenon (Shen and Dillard, 2005). This has led to the well-accepted notion that people can indeed vary in their propensity to experience state reactance thus transcending actual situational characteristics (Merz, 1983; Hong and Ostini, 1989; Hong and Faedda, 1996). Specifically, all other things being equal, individuals with high trait reactance experience greater state reactance in response to persuasive appeals (Quick and Stephenson, 2008). This implies that trait reactance needs to be accounted for when empirically investigating state reactance in the context of persuasive communication.

Conceptual model and hypotheses

State-of-the-art reactance theory “characterizes reactance as a process consisting of three components: a threat to freedom (antecedent), an attempt to reinforce freedom (outcome), and an intervening psychological response. Measuring or manipulating freedom threats is thus essential to a theory-consistent examination of reactance” (Ratcliff, 2021, p. 1047). In line with this approach, we follow Matarazzo and Diamantopoulos' (2022) recent guidelines on conceptualizing reactance as both negative cognitions (e.g. source derogation) and negative affect (e.g. anger) resulting from a perceived threat to freedom and leading to the motivation to reassert the freedom (see Figure 1). Characterizing reactance as a mediating psychological variable (Brehm and Brehm, 1981) comprising both negative cognitions and negative affect “affords a more comprehensive evaluation (Dillard and Shen, 2005) and may give researchers a better ability to detect reactance across varied individual and contextual factors” (Ratcliff, 2021, p. 1049). In this context, the model specification in Figure 1 explicitly acknowledges that “while a freedom threat can be expected to trigger reactance, the nature of the latter need not be uniform for all individuals, in that for some it may be primarily manifested in negative cognitions and for others in negative affect; moreover, the corresponding impact on attitudinal and behavioral outcomes may be different” (Matarazzo and Diamantopoulos, 2022, p. 4). Our model also explicitly takes into account the potential influence of consumer ethnocentrism as a further driver of reactance and its outcomes since ethnocentric consumers are more likely to identify with and embrace the aims of “buy local” campaigns (Balabanis and Diamantopoulos, 2004; Pekkanen and Penttilä, 2021; Saffu et al., 2010).

Our conceptual model is entirely consistent with the premises of reactance theory in that the government's “buy local” campaign is expected to be perceived as a freedom threat by consumers. This perceived threat to freedom is subsequently anticipated to result in (state) reactance captured by both negative cognitions (e.g. disagreement with the message, derogation of the source) and negative affect (i.e. anger, hostility, irritation). The link between perceived threat and reactance has been repeatedly established in the context of different persuasive appeals to either engage in a certain behavior such as flossing, regular exercise, sunscreen usage and organ donation (Dillard and Shen, 2005; Miller et al., 2007; Reinhart et al., 2007; Quick and Stephenson, 2008) or to abstain from certain behaviors such as binge drinking or cigarette smoking (Dillard and Shen, 2005; LaVoie et al., 2017). We thus hypothesize that:

H1.

Perceived threat will positively impact reactance as captured by (a) negative cognitions, and (b) negative affect associated with the “buy local” campaign.

In considering the impact of perceived threat on reactance as captured by H1, the potentially moderating role of trait reactance needs to be taken into account. Such a moderating role appears – according to prior research – to be context-specific with some studies revealing a positive interaction between perceived threat and trait reactance on (state) reactance (e.g. Dillard and Shen, 2005; Quick and Stephenson, 2008) and others failing to identify a moderating influence of trait reactance (LaVoie et al., 2017). In the current context, while it is intuitive to argue that the threat to the personal buying freedom brought about by the “buy local” campaign may be rationalized as being “temporarily” necessary to overcome the economic crisis due to COVID-19, one would expect less compliance motivation among individuals scoring high on trait reactance. Such individuals have a high need for autonomy and low appreciation of social norms (Dowd et al., 1994) and this is likely to “fuel” the impact of perceived threat on (state) reactance. We thus hypothesize that:

H2.

Trait reactance positively moderates the effect of perceived threat on reactance as captured by (a) negative cognitions, and (b) negative affect associated with the “buy local” campaign.

In addition to perceived threat, consumer ethnocentrism is also expected to impact reactance but in the opposite direction (i.e. negatively). Consumer ethnocentrism is the key construct in our model that informs about the likely ex ante position of a respondent regarding the topic of the “buy local” campaign and whether the latter communicated a rather pro-attitudinal or counter-attitudinal message for the recipient. As noted earlier, an individual's ex ante position on an issue can influence the generation of reactance (Brehm and Brehm, 1981). In this regard, the campaign's rationale to “buy local” because of the country's economic situation due to the COVID-19 crisis aligns perfectly with the consumer ethnocentrism construct, as the motive of buying local to protect the home economy is central to the latter (Shimp and Sharma, 1987; Shankarmahesh, 2006; Siamagka and Balabanis, 2015). As a result, consumer ethnocentrism is expected to act as a countervailing force to the reactance generated by the perceived threat relating to the “buy local” campaign. We thus hypothesize that:

H3.

Consumer ethnocentrism will negatively impact reactance as captured by (a) negative cognitions, and (b) negative affect associated with the “buy local” campaign.

Shifting attention to the outcomes of reactance, previous research shows that it leads to a lower behavioral intention for the advocated behavior (Dillard and Shen, 2005; Miller et al., 2007) and to a higher motivation to do the restricted or “opposite” behavior (Quick and Stephenson, 2007, 2008). We thus anticipate that negative cognitions and affect will negatively impact compliance (i.e. motivation to buy domestic products and support domestic retailers) and positively influence freedom restoration (Quick and Stephenson, 2007), with the latter comprising both “boomerang effects” (i.e. motivation to buy foreign products and support foreign retailers) and “related boomerang” effects (i.e. motivation to purchase from online marketplaces). We thus hypothesize that:

H4.

Compliance will be negatively related to reactance as captured by (a) negative cognitions, and (b) negative affect associated with the “buy local” campaign.

H5.

Freedom restoration will be positively related to reactance as captured by (a) negative cognitions, and (b) negative affect associated with the “buy local” campaign.

Lastly, we expect consumer ethnocentrism to also impact the outcomes of reactance in terms of motivations to buy domestic vs foreign products and support domestic vs foreign retailers. Given that “at the core of CE [consumer ethnocentrism] is the moral obligation of people to support and protect the domestic economy by buying domestic products” (Balabanis and Siamagka, 2022, p. 746), consumer ethnocentrism can be expected not only to reduce reactance (as postulated by H3) but also have an impact on its outcomes in terms of purchase behavior. Specifically, we hypothesize that:

H6.

Consumer ethnocentrism will be (a) positively related to compliance, and (b) negatively related to freedom restoration as outcomes of the “buy local” campaign.

Empirical studies

Research design

Following Uncles and Kwok (2013, p. 1399, added emphasis), we designed our empirical investigation with in-built differentiated replication, whereby replication is seen as “an integral component of the initial study”[3]. Specifically, we first estimated the model in Figure 1 in Germany (Study 1) and then did the same in Italy (Study 2) to test its robustness and stability. Thus, Study 2 serves as a conceptual replication of Study 1, the aim being to answer the question: “[t]o what extent are the sign, significance, and effect size of original results robust with respect to changes in the stimuli, settings, participant characteristics, contexts and time of the study?” (Lynch et al., 2015, p. 335). The selection of Germany and Italy as research settings sought to shed light into the arousal and effects of reactance associated with a “buy local” appeal in two comparably developed countries which, however, were differentially affected by the COVID-19 crisis and also vary in terms of the confidence and trust citizens have towards their government as well as in terms of the level of consumer ethnocentrism.

It is important to note that our investigation is not comparative in nature. This is evident from our hypotheses which do not involve postulating and testing for differences regarding how German and Italian consumers respond to “buy local” campaigns aimed at fighting the economic consequences of COVID-19. Rather, the aim of our research is to demonstrate the relevance and applicability of a particular theory (i.e. reactance theory) for systematically studying “buy local” campaigns in general. Thus, both the specific context (COVID-19 crisis) and the countries chosen are illustrative. However, following good methodological practice to ensure robustness/replicability, we used two separate studies conducted in different countries to see whether our reactance-based model is stable under different conditions (hence the choice of Germany and Italy). In this context, both selected countries have highly developed economies with Germany ranking 4th and Italy 7th in terms of their GDP in 2020 (OEC, 2023). Moreover, Germany and Italy respectively occupied the 3rd and 7th places in terms of total exports and the 3rd and 10th places in terms of total imports (OEC, 2023), indicating that Germans and Italians can choose from a wide variety of domestic and foreign products and that the product choice of consumers in both countries would be severely restricted if their governments urged them to “buy local.”

One key difference between the two countries relates to the severity of the COVID-19 pandemic. Italy was arguably the first Western democracy being faced with COVID-19 (Belligoni, 2021) and, in March 2020, it was the country with the most cases after China (Saglietto et al., 2020). The shocking images of the temporarily overloaded hospitals in Italy went around the world and served as a cautionary tale urging other countries (like Germany) to react promptly and take decisive action. Although Italy eventually succeeded in limiting the spread of the virus and the infection and death rates in Italy and Germany developed similarly (Stewart, 2022), Italy still measured more deaths (in both relative and absolute terms) connected to COVID-19 (de Best, 2022) at the time our studies were conducted (April 2021). In this context, the greater severity of the COVID-19 crisis in Italy might act as a doubly-edged sword: on the one hand, one could speculate that Italian consumers might be more receptive to measures – including economic ones – aimed at combatting the effects of the pandemic then German consumers. This might be the case because the “buy local” request might be perceived as more legitimate and perceived legitimacy of a request can decrease reactance (Zhang and Sapp, 2013). On the other hand, the greater severity of the pandemic inevitably led to more health- and economy-related freedom restrictions in Italy and such threats would be further exacerbated by the additional freedom threat brought about by a “buy local” campaign instigated by government.

A second important difference between the two countries concerns the relationship between the government and the people. As already mentioned, the source of a persuasive communication can have considerable influence on the amount of reactance that is elicited by it (Heinemann et al., 2008; Song et al., 2018) and there are substantial differences between Germany and Italy regarding the confidence and trust that ordinary citizens have in their governments. According to the Global Trustworthiness Index based on 19,570 respondents from 28 countries, surveyed between April and May 2021, 12% (9%) of Germans (Italians) consider politicians trustworthy while 50% (67%) consider them untrustworthy (Ipsos, 2021). Furthermore, 42% (51%) of Germans (Italians) find government ministers untrustworthy (Ipsos, 2021). A similar picture is painted by Hensel et al. (2020) who asked over 100,000 respondents in 58 countries between late March and early April 2020, about their attitudes towards governments' responses to the COVID-19 crisis (among other things); while 34% of German respondents stated that they “strongly trust” the government only 10% of Italians did so [4]. Notably, Liu et al. (2022) found that the number of confirmed COVID-19 cases decreased trust in the government, whereas the implementation of containment measures against the virus and economic support measures increased trust in the government. Bearing in mind the lower trust in government in Italy versus Germany as well as the greater severity of the pandemic in Italy noted earlier, the reactance potential of a “buy local” campaign would, ceteris paribus, appear to be greater in Italy than in Germany. Having said that, the two countries have also been found to differ substantially with respect to the level of consumer ethnocentrism, with Italy scoring much higher on the CETSCALE (Shimp and Sharma, 1987) than Germany (Balabanis and Siamagka, 2022) [5]. Prior research has found ethnocentrism to be “associated with a kind of prosociality, in which the country's interests take precedence over a person's self-interest” (Siamagka and Balabanis, 2015, p. 69). This suggests that a “buy local” campaign may be perceived more positively in Italy thus potentially leading to less reactance.

Study 1

Data collection

Consistent with prior reactance research employing direct measurement of reactance (e.g. Quick and Stephenson, 2007, 2008), we employed a one-group, posttest-only design to empirically investigate the relationships in Figure 1 [6]. Specifically, 265 German ordinary consumers (44.9% female; Mage = 40.8, SD = 13.01) participated in an online survey conducted on the Clickworker crowdsourcing platform (see www.clickworker.de) [7]. Crowdsourcing platforms offer several benefits including convenience and comparatively low cost as well as the ability to reach participants based on specific criteria that align with the research question and target population (Berinsky et al., 2012; Casler et al., 2013; Stewart et al., 2017). In light of these benefits, data collection with crowdsourcing platforms has been increasingly used in consumer behavior studies; for example, 43% of all studies in the 2016 volume Journal of Consumer Research have been conducted with the crowdsourcing platform Amazon Mturk and the trend has been increasing (Goodman and Paolacci, 2017). Importantly, after integrating findings from different studies on the reliability and validity of survey answers of crowdsourcing workers, Goodman and Paolacci (2017, p. 201) conclude that “there is no evidence that the efficiency gains of crowdsourcing come at the expense of data quality.”

Procedure and measures

Participants were first informed that the study's goal was to obtain their opinions on issues relating to consumer behavior. Respondents were also reassured that there were no right or wrong answers, that their provided data would be treated anonymously, and that they could take their time in responding to the questions. Subsequently, respondents were first exposed to a scenario, then perceived threat was explicitly measured, then state reactance was assessed (both in terms of negative cognitions and negative affect) and, finally, reactance was linked to different compliance and freedom restoration outcomes. This “chain” of measurement is typical in reactance studies and is necessary in order to ensure that the scenario does indeed result in perceived threat; in the absence of the latter, it is not possible to claim that observed negative cognitions/affect actually represent reactance (Ratcliff, 2021; Rosenberg and Siegel, 2018).

Bearing the above in mind, study participants were first presented with a short text describing the impact of COVID-19 on the German economy. The text closed with a note on the “buy local” campaign of the government stating that, as part of the economic recovery efforts, the German economic ministry (Bundesministerium für Wirtschaft und Energie) has been running an intensive communication campaign to encourage people to support German retailers and buy more German products to help the domestic economy reach pre-crisis levels as quickly as possible.

Following exposure to the “buy local” campaign, study participants completed established scales capturing the constructs in Figure 1. State reactance was operationalized in line with the theory by Brehm (1966), whereby an induction check of a perceived threat to freedom precedes the measurement of reactance, which itself consists of negative cognitions and negative affect (Dillard and Shen, 2005), and is the commonly used approach in communication research (Reynolds-Tylus, 2019; Ratcliff, 2021; Matarazzo and Diamantopoulos, 2022). Specifically, perceived threat was measured with a four-item scale (α = 0.895) adapted from Dillard and Shen (2005) (e.g. “The campaign tried to make a decision for me”; Likert-type format ranging from 1 (“strongly disagree”) to 7 (“strongly agree”)). Negative cognitions were captured by the three-item scale (α = 0.975) developed by Quick et al. (2015) and subsequently validated by Reynolds-Tylus et al. (2021) (e.g. “The thoughts you had about this message were [unfavorable/negative/bad]”; Likert-type format ranging from 1 (“strongly disagree”) to 7 (“strongly agree”)). Negative affect was operationalized by the four-item scale (α = 0.941) used by Dillard and Shen (2005) (“The campaign made me [irritated/angry/annoyed/aggravated]”; 5-point rating scale ranging from 0 “none of this feeling” and 4 “a great deal of this feeling”). Next, to measure the motivation elicited by activated (state) reactance to comply and restore freedom, Quick and Stephenson's (2007) Reactance Restoration Scale (RRS) was employed, whereby the respondent is presented with a statement such as, for example, “Right now I am … to [buy foreign brands]” and asked to indicated his/her position on a seven-point continuum on four semantic differential items (motivated-unmotivated, determined-not determined, encouraged-not encouraged and inspired-not inspired). We used this item structure to capture compliance as reflected in the motivation to buy domestic products (α = 0.948) and shop at domestic retailers (α = 0.937), as well as freedom restoration as reflected in the motivation to buy foreign products (α = 0.942), patronize foreign retailers (α = 0.954), and/or shop at online marketplaces (α = 0.946) such as Amazon, Zalando or Yoox. Finally, trait reactance and consumer ethnocentrism were respectively operationalized by Hong and Faedda's (1996) 11-item Hong Psychological Reactance Scale (α = 0.865) and Verlegh's (2007) short (five-item) version of Shimp and Sharma's (1987) original CETSCALE (α = 0.883). Demographic information on the respondents' gender, age and income was also obtained and these characteristics were used as control variables in the analysis. Full details on the measurement scales employed and their associated psychometric properties can be found in Appendix.

The questionnaire, as well as the experimental manipulation (i.e. the “buy local” campaign text), were first developed in English, subsequently translated into German and finally back-translated into English by two bilingual speakers following established literature guidelines (e.g. Behling and Law, 2000). To assess common method bias (CMB), we used the well-established marker variable procedure by Lindell and Whitney (2001). Specifically, we adjusted the bivariate (zero-order) correlations between the individual items measuring the constructs in our model by controlling for a variable that does not bear a conceptual link to any of the model constructs. This marker variable was represented by the statement “I act the same way at home that I do at school (or work)”, measured on a seven-point Likert format. Partial correlations analyses controlling for the marker variable, revealed 608 significant correlations (vs 607 zero-order significant correlations prior to adjustment). These results clearly indicate that CMB is not of concern in our study.

Analysis and results

We used Model 7 of the PROCESS routine (Hayes, 2017) with 5,000 resamples to test our research hypotheses and obtain 95% bias-corrected bootstrap confidence intervals (BCCIs) for the (conditional) indirect effects. As Figure 2 shows, the proposed conditional mediation model includes perceived threat (PTHREAT) as the independent variable (X), two parallel mediators M1 and M2 corresponding to negative cognitions (COG) and negative affect (AFF), respectively, trait reactance (TR) as the moderator (W) and each compliance or freedom restoration outcome (OUT) in turn as the dependent variable (Y). Furthermore – although not shown in Figure 2 for simplicity – consumer ethnocentrism (CET) and the control variables (gender, age, income) were also included in the model specification.

Given the model structure in Figure 2, support for H1 would be provided if the direct effects PTHREAT → COG (i.e. a₁₁) and PTHREAT → AFF (i.e. a₁₂) are positive and significant; H2 would be supported if the interaction effects (i.e. a₃₁ and a₃₂) are positive and significant; H4 would be supported if the direct effects of COG → OUT (i.e. b₁) and AFF → OUT (i.e. b₂) are negative and significant when OUT captures compliance; and H5 would be supported when the direct effects COG → OUT (i.e. b₁) and AFF → OUT (i.e. b₂) are positive and significant when OUT captures freedom restoration. Moreover, the indirect effects PTHREAT → COG → OUT (i.e. a₁₁b₁) and PTHREAT → AFF → OUT (i.e. a₁₂b₂) would reveal the link between the freedom threat and consumers' responses in terms of compliance and freedom restoration [8]. Finally, regarding the impact of CET, H3 would be supported if the direct effects CET → COG and CET → AFF are negative and significant and H6 would be supported if the direct effect CET → OUT is positive (negative) and significant when OUT represent compliance (freedom restoration). Table 1 summarizes the results of the PROCESS analysis for each outcome variable.

Consistent with hypotheses H1a and H1b, perceived threat is positively linked to both negative cognitions (β = 0.689, p < 0.001) and negative affect (β = 0.439, p < 0.001), which is fully in line with reactance theory (Ratcliff, 2021). Trait reactance also positively impacts negative cognitions (β = 0.495, p < 0.001) and affect (β = 0.229, p < 0.001). Regarding the hypothesized moderating role of trait reactance, the interaction term (perceived threat x trait reactance) on negative cognition is not significant, offering no support for H2a. However, the corresponding interaction term on negative affect is positive and significant (β = 0.107, p < 0.05) suggesting that trait reactance increases the amount of negative affect that one feels in response to the perceived freedom threat caused by the “buy local” appeal. Thus, H2b is supported.

Contrary to expectations, consumer ethnocentrism is not significantly linked to either negative cognitions or negative affect thus offering no support for H3a and H3b. Given that this is the first time that – to the best of our knowledge – consumer ethnocentrism has been linked to reactance, it is not clear whether this result is idiosyncratic to the current sample or whether it reflects a true absence of a relationship between consumer ethnocentrism and reactance. We will revisit this link in Study 2.

Focusing on the consequences of reactance, in line with H4a, negative cognitions are negatively related to compliance as captured by the motivation to buy domestic products (β = −0.163, p < 0.05) and patronize domestic retailers (β = −0.256, p < 0.01). However, negative affect is not significantly related to compliance, thus offering no support for H4b. Regarding freedom restoration, no “boomerang” effects can be noted as neither negative cognitions nor negative affect have any impact on the motivation to buy foreign products or patronize foreign retailers. However, negative affect has a positive impact (β = 0.297, p < 0.05) on the “related boomerang” motivation to buy from online marketplaces (most of which are foreign-owned anyway). Thus, no support can be offered for H5a and only partial support for H5b.

Consistent with H6a, consumer ethnocentrism is positively related to both compliance outcomes (buying domestic products (β = 0.382, p < 0.001) and patronizing domestic retailers (β = 0.306, p < 0.001)) thus fully supporting H6a. Moreover, ethnocentrism is negatively related to the motivation to buy foreign products (β = −0.152, p < 0.01) thus providing partial support for H6b.

Finally, of the control relationships, only age and income seem to have any effect on the model relationships. Specifically, older consumers appear to exhibit greater compliance as reflected in shopping at domestic retailers (β = 0.014, p < 0.05) while more affluent consumers are likely to display greater negative affect (β = 0.099, p < 0.05) and a stronger freedom restoration tendency in terms of buying foreign products (β = 0.201, p < 0.01).

Study 2

Data collection

We employed the same research design, stimuli [9], construct measures and statistical procedures as in Study 1 but carried out the study in a different setting (Italy). Specifically, 268 Italian ordinary consumers (37.3% female; Mage = 37.05, SD = 11.43) participated in an online survey again conducted on the Clickworker platform (see www.clickworker.de) [10]. The questionnaire, as well as the experimental manipulation were initially developed in English and subsequently translated into Italian using the same procedures as in Study 1; all construct measures exhibited high reliability (perceived threat, α = 0.924; negative cognitions, α = 0.930; negative affect = 0.956; motivation to: (1) buy domestic products, α = 0.921, (2) patronize domestic retailers, α = 0.915, (3) buy foreign products, α = 0.926, (4) patronize foreign retailers, α = 0.929, (5) shop at online marketplaces, α = 0.921; trait reactance, α = 0.797; consumer ethnocentrism, α = 0.943). As in Study 1, we applied the marker variable technique (Lindell and Whitney, 2001) to assess CMB; partial correlation analyses controlling for the same market variable as in Study 1, revealed 612 significant correlations (vs. 614 zero-order significant correlations prior to adjustment). Thus CMB is not of material concern in Study 2 either. Full measurement details and psychometric properties can be found in Appendix.

Results

The results of estimating our model on the Italian sample are summarized in Table 2.

In line with Study 1, perceived threat is positively linked to both negative cognitions (β = 0.692, p < 0.001) and negative affect (β = 0.534, p < 0.001) thus further supporting H1a and H1b. Trait reactance also has, as in Study 1, a positive influence on reactance both in terms of cognition (β = 0.380, p < 0.01) and affect (β = 0.264, p < 0.05). Similar to Study 1, the interaction term of perceived freedom threat and trait reactance on negative cognitions was not significant, offering no support for H2a. However, the corresponding interaction term on negative affect was positive and significant (β = 0.183, p < 0.01) thus supporting H2b. Unlike in Study 1, CET also has a direct negative impact on reactance arousal in terms of negative cognitions (β = 0.139, p < 0.01) but not in terms of negative affect. Thus H3a is supported in this sample, while H3b is not.

In line with Study 1 and in support of H4a, negative cognitions are negatively related to compliance as captured by the motivations to buy domestic products (β = −0.175, p < 0.05) and to patronize domestic retailers (β = −0.159, p < 0.05); negative affect is, again, not significantly related to compliance, thus offering no support for H4b. Also in line with Study 1, negative cognitions have no impact on any aspect of freedom restoration thus offering no support for H5a. However, negative affect positively impacts the motivation to buy from online marketplaces (β = 0.548, p < 0.001) as also observed in Study 1 and, in addition, the motivation to patronize foreign retailers (β = 0.271, p < 0.05). These results are in line with H5b.

Regarding the impact of consumer ethnocentrism, the results fully support H6a and H6b. Specifically, and consistent with Study 1, ethnocentrism increases compliance as reflected in buying domestic products (β = 0.402, p < 0.001) and supporting domestic retailers (β = 0.355, p < 0.001), and decreases freedom restoration behaviors as reflected in buying foreign products(β = −0.341, p < 0.001), supporting foreign retailers (β = −0.352, p < 0.001) and purchasing from online marketplaces (β = −0.182, p < 0.01).

Lastly, regarding control variables, male consumers are more likely to display compliance by buying domestic products (β = 0.393, p < 0.01) whereas younger consumers are more likely to display reactance in terms of both negative cognitions (β = −0.014, p < 0.05) and negative affect (β = −0.014, p < 0.01). Table 3 summarizes the hypothesis-testing results of both studies.

Discussion and implications

Since the 2008–2009 global financial crisis, economic instability caused by geopolitical tensions seems to be the new normal. The COVID-19 pandemic was an exogenous systemic shock causing unprecedented economic consequences which the world economy had to suffer. Although the pandemic-induced health crisis has eased considerably due to the invention and widespread administration of vaccinations, the associated economic crisis is still far from over. On the contrary, in view of the current world situation characterized by the Ukraine war and inflation, the World Bank predicts a “period of feeble growth and elevated inflation […] [which] raises the risk of stagflation” (World Bank, 2022). To counter this crisis-ridden future, governments have been implementing a variety of economic measures, including “buy local” communication appeals. The present study applied reactance theory to investigate the potential consequences of such “buy local” initiatives in two different country settings (Germany and Italy) and offer insights into the reactance-generating potential of such initiatives as well as consumers' likely responses in terms of compliance and freedom restoration. In doing so, it also explicitly introduced consumer ethnocentrism as a construct impacting reactance and considered reactance outcomes specific to purchase behavior (i.e. buying domestic/foreign products and patronizing domestic/foreign retailers as well as online marketplaces). Importantly, the current investigation applied reactance theory under conservative conditions as captured by a threat to freedom that is seemingly well-justified since the intention of the “buy local” campaign by government was to support the home economy during times of a (major) crisis. Several theoretical and managerial implications emerge from our findings, as discussed below.

Theoretical implications

Our investigation contributes to the body of literature on “buy local” campaigns by offering a theoretical explanation as to why such campaigns may not be successful even if (a) consumers are aware of the campaign and its objectives, and (b) the campaign is a response by government to fight a major crisis (such as that caused by the COVID-19 pandemic). While failures of “buy local” campaigns to achieve their desired objectives have been repeatedly documented in the past (e.g. see Ettenson et al., 1988; Fenwick and Wright, 2000; Mtigwe and Chickweche, 2008), there have been limited theoretical insights as to the reasons behind such failures. By applying reactance theory, we offer a theoretical lens for approaching “buy local” initiatives and empirically demonstrate that such initiatives – even if justified on the need to fight the economic consequences of a major crisis (such as COVID-19) – are not risk-free. More specifically, government-driven appeals that merely solicit consumers to support domestic products and retailers in order to mitigate adverse economic consequences without providing any further rationale (e.g. an appeal to quality, innovativeness, or design) are likely to be perceived as a threat to freedom and thus lead to reactance. Such reactance is likely to be higher for consumers scoring high on trait reactance, a result which is fully in line with previous research (e.g. Dillard and Shen, 2005). The implication of this is that trait reactance must be controlled for when investigating the impact of perceived threat on state reactance, otherwise the latter effect is likely to be overestimated.

Trait reactance is also likely to amplify the impact of perceived threat on the affective dimension of reactance manifestation. This finding, on the one hand, confirms prior research showing that the moderating role of trait reactance is situation-specific (e.g. Dillard and Shen, 2005; Quick and Stephenson, 2008; LaVoie et al., 2017) and, on the other hand, highlights the importance of distinguishing between cognitive and affective aspects when modeling reactance (Matarazzo and Diamantopoulos, 2022). Importantly, perceived threat, trait reactance, their interaction and consumer ethnocentrism explain 55.2% (53.1%) of the variance in negative cognitions and 47.8% (50.0%) of the variance in negative affect in the German (Italian) sample, indicating strong effect sizes (Cohen, 1988).

Activated state reactance was expected to simultaneously lead to a devaluation of the proposed behavior and a revaluation of the restricted behavior (Brehm et al., 1966; Clee and Wicklund, 1980; Laurin et al., 2012), respectively reflected in negative motivations to buy domestic products and support domestic retailers and positive motivations to buy foreign products and patronize foreign retailers. Our findings are largely in line with these expectations of reactance theory and further underline the importance of distinguishing between the cognitive and affective dimensions of reactance as they seem to impact different outcomes. Specifically, the former dimension discourages compliance as demonstrated by the negative links between negative cognitions resulting from the perceived threat to freedom and the motivation to buy domestic products and support domestic retailers, which was observed in both our studies. In contrast, negative affect seems to mainly impact freedom restoration and, in particular, “related boomerang” effects as reflected in the motivation to buy from online marketplaces (also observed in both studies). Interestingly, in the current context, reactance does not seem to lead to “boomerang” effects, that is, exactly the opposite than what was intended. Reactance did not fuel the motivation to buy foreign products in either country setting and an increased motivation to patronize foreign retailers was only observed in the Italian sample. A possible reason for the absence of strong “boomerang” effects could be the fact that consumers, while objecting to the “buy local” campaign and perceiving the latter as a freedom threat, may be more tolerant in their responses due to (reluctantly) accepting the necessity of such a campaign in light of the COVID-19 crisis.

Our investigation also contributes to reactance theory by identifying – for the first time – consumer ethnocentrism as an important construct impacting reactance and its outcomes. However, its effect seems to be context-specific. More specifically, ethnocentrism has no impact on the arousal of reactance in the German sample, however, it significantly reduces reactance in the Italian sample. Moreover, while ethnocentrism encourages compliance in both countries, its impact on freedom restoration outcomes is much more prevalent in Italy than Germany (reducing not only the motivation to buy foreign products but also the motivation to patronize foreign retailers and shop at online marketplaces). These differences on the role of consumer ethnocentrism may reflect differences in terms of the perceived legitimacy of a “buy local” appeal in Germany and Italy in light of the different severity of COVID-19 in these countries. Given that – as outlined in the Research Design section – Italy was more severely hit by COVID-19 than Germany, the perceived legitimacy of a “buy local” appeal would be greater in Italy. This line of argument is supported by the fact that Italy saw considerably greater decreases in household income (−2.2% vs −0.1%) and GDP (8.5% vs 5.2%) compared to Germany in 2020 (OECD, 2021). Given that the “buy local” appeal was specifically intended to tackle the economic problems caused by COVID-19, it would be expected that Italian (German) consumers are likely to perceive it as more (less) legitimate. Such differences in perceptions, in turn, could qualify the impact of ethnocentrism on reactance and its outcomes. In case of a severe economic crisis, highly ethnocentric individuals may be motivated to actively suppress negative reactions towards a “buy local” campaigns; this seems to have happened in Italy. However, in the absence of a significantly felt economic crisis, this motivation may not be activated among highly ethnocentric consumers; this seems to have occurred in Germany. In short, it seems that the effects of ethnocentrism are not stable but bound to the economic well-being of the home country.

From a methodological perspective, our investigation also raises some concern regarding the trait reactance scale by Hong and Faedda (1996). Although the scale demonstrated acceptable reliability in both country samples, the item loadings were rather weak, resulting in low average variance (AVE) extracted values (see Appendix 2). Given that trait reactance positively influences state reactance and may also moderate the effect of perceived threat on reactance, it is important to measure it effectively in empirical applications. While the Hong and Faedda (1996) scale is the most widely used trait reactance measure in the literature, future researchers studying reactance in a marketing context, might benefit from considering other instruments, such as Dowd et al.’s (1991) scale.

Managerial implications

Previous research has noted that a “buy local” campaign “may do more harm than good and firms would be advised to promote their own brands” (Insch et al., 2017, p. 242). We concur with this recommendation since such campaigns do not necessarily provide a “protective shield” to domestic companies because of their potential to generate reactance. It would thus be unwise for domestic firms to assume that they will automatically benefit from such campaigns instigated by their government (see also Fenwick and Wright, 2000). Rather than relying on government-sponsored initiatives to buy/shop locally, domestic firms should focus on developing strong brands, signaling quality and value for money thus making their offerings attractive on their own merits rather simply capitalizing on their domestic origin. Conversely, foreign firms should not necessarily view “buy local” initiatives as major threats to their internationalization efforts; while such initiatives may not directly benefit foreign companies, they will not harm them either. The reactance that “buy local” campaigns will inevitably generate is likely to adversely affect compliance and thus not materially alter the playing field in favor of domestic firms. Thus continuing to do “business as usual” is perhaps the best option for foreign firms to take when confronted with “buy local” campaigns in international markets.

From a policy making perspective, a key implication of our findings is that the reactance arousing potential of any “buy local” campaign should be explicitly considered prior to its launch. One way of doing this is to conduct scenario-based pilot studies aimed at assessing the amount of reactance likely to be generated (in terms of both negative cognitions and affect) as well as the likely effects on pertinent outcomes. In this context, particular attention should be paid on identifying people's ex ante position regarding the appeal (or lack) of the planned campaign since the latter is likely to be differentially received and complied with by different consumer segments. While, as already noted, ethnocentric consumers constitute a segment generally expected to display low reactance and high compliance, other, less-obvious segments – such as environmentally-sensitive consumers – may behave in a similar manner [11]. In contrast, segments mainly consisting of cosmopolitan (Riefler et al., 2012), disidentified (Josiassen, 2011) or xenocentric consumers (Balabanis and Diamantopoulos, 2016) may be (much) more difficult to handle since they are positively disposed towards buying foreign products. For such segments, the launch of a “buy local” campaign may well constitute a major freedom threat leading to substantial reactance.

Regarding the actual execution of “buy local” campaigns, research by Song et al. (2018) shows that perceived source similarity can reduce reactance, meaning that the receiver of a communication will feel less reactance when seeing him/herself to be similar to the source of the communication. This finding can be implemented by policymakers when disseminating a “buy local” campaign in different ways. For example, the message itself could entail appeals to enhance perceived similarity by emphasizing the notion that “we are all in this together” thus arguably decreasing reactance (although surely more creative variations of the slogan would be warranted, given it was all too popular around the world during times of COVID-19 (see Jaradat, 2020; Atkinson, 2021; Knapp et al., 2021)). Alternatively, perceived similarity could be increased by increasing the proximity of the source to the receiver; perceived similarity should increase when changing the source from a country-, to a state/region-, to a city-government level thereby also decreasing the reactance potential of the campaign. Supporting this notion, Grimalda et al. (2021) found that, in the COVID-19 context, when given the possibility, people donated more to COVID-19 relief efforts benefitting the people on a local vs. on a country level.

Limitations and future research

There are several limitations in our study which offer opportunities for future research. First, there is an obvious need to further replicate our findings in other countries to establish their generalizability (Lynch et al., 2015). Both our studies have been conducted in Western individualistic country settings (Hofstede Insights, 2021), in which personal freedom and choice are highly valued (Markus and Schwartz, 2010). Given that an interdependent mindset might decrease reactance from a threat to a personal freedom (Jonas et al., 2009), especially when the threat originates in the in-group (Graupmann et al., 2012), there may be possible cultural differences in the elicitation of reactance from a “buy local” appeal by government. This calls for future studies that examine the reactance potential of such appeals in countries with an interdependent or collective orientation such as countries in East Asia. Alternatively, one could account for the possible effect of an interdependent mindset by including variables which measure culture at the individual level such as independent and interdependent self-construals (Markus and Kitayama, 1991) or akin concepts like the Schwartz values (Schwartz, 2012). For example, “self-direction” with its emphasis on autonomy and independence (Schwartz, 2012) might positively increase the perception of a threat to freedom, whereas “conformity” characterized by “restraint of actions, inclinations, and impulses likely to upset or harm others and violate social expectations or norms” (Schwartz, 2012, p. 6) might negatively influence reactance.

Second, in both our studies, we only measured general motivations to buy foreign/domestic products and patronize foreign/domestic retailers. Therefore, it is not clear how much variance reactance actually explains in the case of a buying decision in a concrete product category or a specific brand. Thus, future research should investigate the effects of “buy local” campaigns on different product categories and/or different kinds of brands (e.g. mass-market vs. luxury). In the same vein, previous research has revealed that the effects of consumer ethnocentrism vary considerably across product categories (Balabanis and Diamantopoulos, 2004; Balabanis and Siamagka, 2017). Thus, the protective effect that ethnocentrism is exerting against the effects of reactance emanating from “buy local” appeals might be stronger/weaker for certain product categories or brands, warranting future research on this issue too.

Third, in the current study, we extended the psychological reactance model by including consumer ethnocentrism as a further explanatory construct. However, other constructs from the international marketing literature that capture important consumer traits and dispositions (such as cosmopolitanism, xenocentrism or global consumption orientation – see Bartsch et al., 2016; Papadopoulos et al., 2018 for relevant reviews) could also conceivably impact reactance and/or its outcomes. Investigating the role of such constructs in the context of reactance would complement the insights furnished by the present study.

Last but not least, the role of guilt as an additional construct impacting consumer responses to a “buy local” campaign deserves research attention [12]. Specifically, guilt could be modeled as a moderator between reactance and compliance/freedom restoration outcomes (e.g. guilt will likely weaken the relationship between reactance and buying foreign products). Alternatively, guilt could itself be considered as a “downstream” outcome of motivations to buy local or foreign products (e.g. buying foreign products may result in higher levels of guilt).

Figures

Figure 1

Conceptual model

Figure 2

Estimated conditional mediation model

Table 1

Model estimation results (study 1)

	Parameter estimate (Standard error)
	Compliance		Freedom restoration
Direct effects	Domestic products	Domestic retailers	Foreign products	Foreign retailers	Online marketplaces
PTHREAT→ COGa	0.689 (0.054)***
PTHREAT→ AFFa	0.439 (0.039)***
TR → COGa	0.495 (0.114)***
TR → AFFa	0.226 (0.082)**
PTHREAT * TR → COGa	0.090 (0.067)
PTHREAT * TR → AFFa	0.107 (0.049)*
PTHREAT→ OUT	0.014 (0.069)	0.050 (0.071)	0.064 (0.070)	0.025 (0.078)	0.092 (0.090)
CET → COGa	−0.072 (0.057)
CET → AFFa	0.037 (0.041)
CET → OUT	0.382 (0.057)***	0.306 (0.058)***	−0.152 (0.057)**	−0.065 (0.064)	−0.070 (0.073)
COG → OUT	−0.163 (0.079)*	−0.256 (0.081)**	0.102 (0.080)	0.066 (0.089)	−0.034 (0.103)
AFF → OUT	0.080 (0.111)	0.071 (0.113)	0.050 (0.112)	0.182 (0.125)	0.297(0.144)*
Indirect Effects	95% BCCI	95% BCCI	95% BCCI	95% BCCI	95% BCCI
PTHREAT→ COG → OUTb	(−0.235, −0.006)	(−0.324, −0.046)	(−0.029, 0.170)	(−0.063, 0.144)	(−0.156, 0.102)
PTHREAT→ AFF → OUTb	(−0.066, 0.142)	(−0.080, 0.149)	(−0.075, 0.120)	(−0.023, 0.190)	(0.066, 0.257)
Control relationships
GENDER → COGa	0.200 (0.144)
GENDER → AFFa	−0.087 (0.105)
GENDER → OUT	0.003 (0.146)	−0.094 (0.148)	−0.021 (0.147)	0.174 (0.163)	0.053 (0.189)
AGE → COGa	−0.010 (0.060)
AGE → AFFa	−0.006 (0.004)
AGE → OUT	0.009 (0.006)	0.014 (0.006)*	−0.008 (0.006)	−0.006 (0.006)	0.004 (0.007)
INCOME → COGa	0.090 (0.064)
INCOME → AFFa	0.099 (0.047)*
INCOME → OUT	0.075 (0.064)	−0.034 (0.066)	0.201 (0.065)**	0.138 (0.072)	0.096 (0.083)

Note(s): *p < 0.05, **p < 0.01, ***p < 0.001, ^a parameter estimate is identical across all models. ^b Indirect effects reported for TR = 0

Key: PTHREAT: Perceived Threat, COG: Negative Cognitions, AFF: Negative Affect, OUT: Compliance or Freedom Restoration Outcome, TR: Trait Reactance, CET: Consumer Ethnocentrism

Source(s): Authors own creation

Table 2

Model estimation results (study 2)

	Parameter estimate (Standard error)
	Compliance		Freedom restoration
Direct effects	Domestic products	Domestic retailers	Foreign products	Foreign retailers	Online marketplaces
PTHREAT→ COGa	0.692 (0.047)***
PTHREAT→ AFFa	0.534 (0.038)***
TR → COGa	0.380 (0.132)**
TR → AFFa	0.264 (0.108)*
PTHREAT * TR → COGa	0.115 (0.072)
PTHREAT * TR → AFFa	0.183 (0.059)**
PTHREAT→ OUT	−0.120 (0.064)	−0.064 (0.063)	0.079 (0.069)	0.063 (0.071)	−0.066 (0.078)
CET → COGa	−0.139 (0.047)**
CET → AFFa	−0.059 (0.038)
CET → OUT	0.402 (0.046)***	0.355 (0.046)***	−0.341 (0.051)***	−0.352 (0.052)***	−0.182 (0.057)**
COG → OUT	−0.175 (0.079)*	−0.159 (0.078)*	0.002 (0.086)	−0.026 (0.089)	−0.155 (0.098)
AFF → OUT	0.145 (0.960)	0.144 (0.094)	0.201 (0.104)	0.271 (0.107)*	0.548 (0.118)***
Indirect effects	95% BCCI	95% BCCI	95% BCCI	95% BCCI	95% BCCI
PTHREAT→ COG → OUTb	(−0.248, 0.031)	(−0.240, 0.008)	(−0.114, 0.108)	(−0.134, 0.091)	(−0.248, 0.035)
PTHREAT→ AFF → OUTb	(−0.034, 0.188)	(−0.026, 0.186)	(0.004, 0.216)	(0.039, 0.260)	(0.152, 0.458)
Control relationships
GENDER → COGa	−0.130 (0.144)
GENDER → AFFa	0.061 (0.118)
GENDER → OUT	0.393 (0.141)**	0.121 (0.139)	−0.010 (0.154)	0.077 (0.158)	0.206 (0.174)
AGE → COGa	−0.014 (0.065)*
AGE → AFFa	−0.014 (0.053)**
AGE → OUT	−0.001 (0.006)	−0.003 (0.006)	−0.005 (0.007)	0.006 (0.007)	−0.003 (0.008)
INCOME → COGa	0.040 (0.062)
INCOME → AFFa	0.005 (0.051)
INCOME → OUT	−0.064 (0.061)	0.011 (0.060)	0.083 (0.067)	−0.033 (0.068)	0.014 (0.075)

Note(s): *p < 0.05, **p < 0.01, ***p < 0.001, a parameter estimate is identical across all models. b Indirect effects reported for TR = 0

Key: PTHREAT: Perceived Threat, COG: Negative Cognitions, AFF: Negative Affect, OUT: Compliance or Freedom Restoration Outcome, TR: Trait Reactance, CET: Consumer Ethnocentrism

Source(s): Authors own creation

Table 3

Summary of hypothesis-testing results

Hypothesis		Study 1 (Germany)	Study 2 (Italy)
H1a	Perceived threat will positively impact negative cognitions	Supported	Supported
H1b	Perceived threat will positively impact negative affect	Supported	Supported
H2a	Trait reactance positively moderates the effect of perceived threat on negative cognitions	Not supported	Not supported
H2b	Trait reactance positively moderates the effect of perceived threat on negative affect	Supported	Supported
H3a	Consumer ethnocentrism will negatively impact negative cognitions	Not supported	Supported
H3b	Consumer ethnocentrism will negatively impact negative affect	Not supported	Not supported
H4a	Compliance will be negatively related to negative cognitions	Supported for all outcomes	Supported for all outcomes
H4b	Compliance will be negatively related to negative affect	Not supported for any outcome	Not supported for any outcome
H5a	Freedom restoration will be positively related to negative cognitions	Not supported for any outcome	Not supported for any outcome
H5b	Freedom restoration will be positively related to negative affect	Not supported for motivation to buy foreign products Not supported for motivation to buy from foreign retailers Supported for motivation to buy from online marketplaces	Not supported for motivation to buy foreign products Supported for motivation to buy from foreign retailers Supported for motivation to buy from online marketplaces
H6a	Consumer ethnocentrism will be positively related to compliance	Supported for all outcomes	Supported for all outcomes
H6b	Consumer ethnocentrism will be negatively related to freedom restoration	Supported for motivation to buy from foreign products Not supported for motivation to buy from foreign retailers Not supported for motivations to buy from online marketplaces	Supported for all outcomes

Source(s): Authors own creation

Notes

1. In this paper, the term “buy local” is used in a generic sense and refers to any initiative aimed at encouraging the consumption of products originating in one's home (i.e. domestic) country. Other similar terms used in the literature are “buy domestic,” “buy national,” “buy home-made” or “buy COUNTRY (e.g. American).”

2. For state-of-the-art reviews on reactance theory, see Rains (2013), Quick et al. (2013), Steindl et al. (2015), Rosenberg and Siegel (2018), Ratcliff (2021) and Amarnath and Jaidev (2021).

3. According to Uncles and Kwok (2013, p. 1399), “this viewpoint puts the onus on the original investigators to start the processes of replication […] Researchers should not delay the task to an unspecified future time, nor delegate the process to other researchers.”

4. These differences are further corroborated by the Global Corruption Barometer revealing that 10% of German versus 27% of Italian respondents suspected “that most or all people” in their government are involved in corruption (Transparency International, 2021a). Similarly, the Corruption Perception Index, ranks Germany on the 10th and Italy on 42nd place, thus also suggesting more perceived corruption in Italy than in Germany (Transparency International, 2021b).

5. In fact, Italy has the second highest CETSCALE score (M_ITALY = 4,753) across 57 countries and the highest score amongst 24 European countries (see Appendix 1 in Balabanis and Siamagka, 2022).

6. Early research on reactance compared an experimental group exposed to a threat to freedom with a control group with no such exposure. Differences in compliance or freedom restoration outcomes were then attributed to reactance, that is, it was inferred that such differences were due to psychological reactance but the latter was not explicitly measured. However, more recent research employs direct measures of perceived threat to freedom. This makes the use of a control group redundant and enables the use of one-group, posttest-only design to investigate the generation of reactance and its impact on relevant outcomes (see Quick and Stephenson, 2007, 2008). This is also the approach followed in the current investigation, whereby an induction check of a perceived threat to freedom (explicitly measured) precedes the (explicit) measurement of reactance and its outcomes (Reynolds-Tylus, 2019; Ratcliff, 2021).

7. Originally, 300 respondents took part in the survey but 35 were excluded following data quality checks. The latter involved exceeding a defined time minimum (190 s), elimination of duplicate answers, and passing an attention check involving a seemingly ordinary item which instructed respondents to select the “Neutral” option.

8. Following Zhao, Lynch and Chen (2010), when estimating the indirect effects, we also estimated the relevant direct effect (i.e. c' in Figure 2).

9. The Italian economic ministry (Ministero dell’ Economia e delle Finanze) was selected as the initiator of the “buy local” campaign in Study 2.

10. Originally, 307 respondents took part in the survey but 39 were excluded following the same data quality checks as in Study 1 (see endnote 7 above).

11. The authors would like to thank an anonymous reviewer for suggesting that environmental sensitivity/activism could also impact consumer responses to “buy local” campaigns.

12. The authors would like to thank an anonymous reviewer for this further research suggestion.

Appendix Constructs and psychometric properties