Finds documents with both search terms in any word order, permitting "n" words as a maximum distance between them. Best choose between 15 and 30 (e.g. NEAR(recruit, professionals, 20)).
Finds documents with the search term in word versions or composites. The asterisk * marks whether you wish them BEFORE, BEHIND, or BEFORE and BEHIND the search term (e.g. lightweight*, *lightweight, *lightweight*).
The article 'Framing Layoffs: Media Coverage, Blame Attribution, and Trade-Related Policy Responses' delves into the intricate dynamics of media coverage, blame attribution, and policy responses following the 2018 General Motors plant closures. It examines how media portrayals, influenced by politicians and corporations, shape public perceptions of responsibility and policy preferences. The study reveals that media frames significantly impact blame attribution and trade policy preferences, with specific frames shifting blame away from corporations and towards government policies. The survey experiment conducted shows that while media coverage does not significantly alter support for government assistance programs, it does influence attitudes toward trade policies. The findings highlight the strategic use of trade policy narratives by politicians to mitigate blame and shape public opinion. The article concludes by emphasizing the importance of understanding these dynamics in an increasingly global economic landscape.
AI Generated
This summary of the content was generated with the help of AI.
Abstract
Who is blamed when factories close or when there are mass layoffs? Whether it be the closing of an auto plant or the threatened off-shoring of the Carrier furnace factory, media reports frequently incorporate justifications—or frames—that provide context about the closure or layoffs. We conduct an analysis of media coverage of factory layoffs in the United States and Canada, and find that the most common frames often include foreign competition and trade policy, changing market conditions, or exogenous shocks, such as the pandemic. We argue that such frames alter who the public holds responsible and thus affects the public’s preferred policy responses. We test the effect of media frames on the public’s blame attribution and subsequent policy preferences using a survey experiment about General Motors factory closings. The results from a sample of almost 6,000 respondents in the US and Canada show that the public is quick to shift blame to the government, reducing blame to the company, and shifting attention to particular government responses. We find that the most common media frames significantly shift support for trade policy in both countries, but have no effect on domestic public assistance programs such as unemployment benefits or retraining and education programs. Notably, most treatment effects are similar across ideological types. The results hold practical implications in terms of the incentives of politicians to promote specific explanations of factory closings and theoretical implications in terms of moderating the highly partisan expectations within the current literature on economic blame attribution.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
In late 2018, General Motors (GM) announced the planned closure of five production plants in the United States and Canada, igniting a Twitter storm from President Trump and attempts by both the government and GM to attribute blame away from themselves. Preventing US factory closings had been a prominent plank of the President’s campaign. Early in his Presidency, Trump had claimed personal responsibility for negotiating with United Technologies to reverse the plan by subsidiary Carrier to move furnace production from Indiana to Mexico, which would have cost 1400 American jobs. Standing on the Carrier factory floor in Indianapolis, President Trump declared that his administration’s policies would save more jobs by saving American manufacturing: “These companies aren’t going to be leaving anymore,” he said. “They’re not going to be taking people’s hearts out.”1
At the time, Trump lashed out at GM, blaming GM and its CEO Mary Barra for failing to keep the factories open. Union leaders agreed and called GM’s calculations callous, citing concessions made during GM’s bankruptcy proceedings in 2011 and 2015. While Prime Minister Justin Trudeau’s Tweets were more measured in expressing disappointment in GM,2 the Canadian press excoriated GM and both current and former governments for the factory closing. In contrast, GM and CEO Barra generally framed plant closings in terms of adjusting to new market conditions and cutting costs,3 although Trump claimed that, behind closed doors, Barra blamed the auto unions.4 Other politicians, such as Ohio Senator Sherrod Brown, jumped into the fray, citing Trump’s failure to negotiate a strong enough replacement to NAFTA; and some industry experts cited the increased costs of Trump’s new tariffs, which cost GM $493 million in the first quarter of 2018 alone.5
Advertisement
This circle of finger-pointing became amplified in media coverage. While some outlets carried the bare bones news of the announcement, most incorporated Trump’s, GM’s, or unions’ framing to varying degrees. Local press—such as the Detroit Free Press—included greater detail about workers’ and the unions’ responses (UNIFOR in Canada and the UAW in the United States). In contrast, financial and industry news sources tended to focus on GM’s future savings and the role of government policies. More politically-focused outlets highlighted the arc between Trump’s initial promises to save American jobs to the closure of iconic factories. Canadian papers similarly covered not only the local job losses, but also the national political backlash against both the Liberal Party in power at the time of the factory closing announcement and also the Conservative Party for decisions during the 2008 bailout.6
Does it matter who the public blames for the loss of these jobs? The absolute number of lost jobs in each case, while devastating for the individuals and their communities, was nationally relatively small: 2900 in Ontario (Oshawa Assembly), 1,877 in Michigan (Detroit-Hamtramck Assembly and Warren Transmission Operations), 1618 in Ohio (Lordstown Assembly), and 310 in Maryland (Baltimore Operations). During the fourth quarter of 2018—at the time of the announcement—US private-sector establishments had gross job gains of 7.7 million and gross job losses of 6.9 million, according to the Bureau of Labor Management.7 In Canada, a net gain of 109,100 jobs (8000 in manufacturing) in the fourth quarter of 2018 offset the 2900 expected Canadian job losses from the Oshawa Assembly closure.8
Yet, for those not directly involved, factory closings can serve as an assessment of a politician’s economic competence. The tradition of politicians standing in front of a closed factory promising to do better than prior administrations is longstanding. During the 2008 Presidential campaign, against a backdrop of the closed mill his father had worked at, John Edwards blamed “bad government and corporate greed.”9 For an incumbent President, such closures can become fodder for retrospective voting—i.e., when voters reward or punish the incumbent for a past record—and as such create an incentive for politicians to shift blame from government and especially their own policies. Corporations similarly face incentives to be viewed by investors as competent managers and thus frame layoffs as reasonable responses to external forces such as changes in the market and government policies. The recent Covid-19 pandemic and resulting layoffs have heightened the salience of these issues, as debates have continued over the expected role of government relative to the responsibilities of corporations in economic downturns. In just one month—April 2020—US manufacturing employment dropped by 10% (1.3 million), and overall non-farm employment dropped by 20.5 million according to the Bureau of Labor Statistics,10 and the number of firms decreased approximately 6% compared to 2019.11 How the public attributes blame for such layoffs plays an important role in shaping the political consequences and policy responses to mass layoffs.
We consider how this fight to attribute blame resonates with the public and influences preferences for related welfare and trade policies. Prior research has shown the media’s ability to influence preferences when factory closings are directly related to trade policies (Guisinger, 2017). Yet, despite insights provided by Iyengar and Simon (1993), in much of this trade policy focused work, the intermediate step of adjudicating blame is generally provided, assumed, or ignored. In this paper, we focus on the decision pathway: first, the conditions under which the public will assign blame and, second, the extent to which blame-shifting influences preferences for domestic and international policies to mitigate the effects of factory closings. In doing so, we draw together two distinct literatures—blame attribution and individual-level policy preferences. We argue that the common frames provided in newspaper reports of factory closings serve to influence the public’s attribution of blame as well as their preferences for policies to mitigate the effects of factory closures.
Advertisement
Blame Attribution and Policy Outcomes
Factory closings lie at the intersection of government and market responsibility. Shifting expectations of the role of the government and commercial actors (Hacker, 2006) have created the opportunity for blame deflecting, and the media offers a forum to do so. In this process, we argue that the media and politicians both have incentives to focus their attention on mass layoffs and factory closings. It is well known that the media tends to focus on negative events (Fogarty, 2005; Ju, 2008), as is evident in coverage of economic news (Brutger & Strezhnev, 2022; Guisinger, 2017). Factory closings certainly fit the definition of being a “negative event,” and so we expect that the media is likely to report on such stories. Furthermore, as noted above, politicians frequently focus on mass layoffs and factory closings in their public statements and campaigns, drawing additional attention to these events. Given that communication scholars have found that the media’s coverage often follows politicians’ agendas, and vice versa (Bennett, 1990; Green-Pedersen & Stubager, 2010; van der Pas et al., 2017), we expect that the media’s attention on negative events, combined with politicians’ desire to politicize such events, will lead to significant press coverage of factory closings.
The composition of the media coverage may vary, depending on how the media functions. On one hand, if the media passively reports politicians’ preferred frames about factory closings, we would expect to see coverage dominated by government officials diverting blame away from themselves, and directing it toward the companies initiating layoffs. On the other hand, if the media plays a more independent role in selecting what they report, even if they are influenced by politicians, then we would expect to see a wide variety of frames that would include the perspectives of both politicians and other actors. We assess the composition of blame attribution in the media by examining the case of the General Motor’s factory closings. Specifically we consider whether the media coverage provided a full distribution of frames, consistent with the media playing an independent role, or instead converged on politicians’ preferred frames. As we discuss in greater detail below, we find that media coverage of the closings provided a variety of frames blaming different actors for the closings, which is consistent with the media playing a relatively independent role in selecting how to report on the closings.
Drawing on the work of Iyengar and Simon (1993), we expect that media coverage of factory closings increases the relevance of the closings, and that the context provided may shift the public’s blame attribution and preferred policy responses. For media coverage to shape public attitudes, members of the public must be directly exposed to the media, or the content of news coverage must be disseminated indirectly through community and social networks. As Kertzer and Zeitzoff (2017) summarize, public opinion is shaped by a multitude of factors that include both bottom-up processes and top-down elite cues. Though our study is not designed to adjudicate the various mechanisms that shape public opinion, we argue that the media plays an important role in both top-down and bottom-up processes. In a top-down framework, the media is an important conduit that transmits cues from elites, which can shape public attitudes. In a bottom-up process, members of the public form opinions based off of the information they receive from their peer and social networks, in addition to the media and elites, and this information is then evaluated through individual lenses that vary with various dispositions and moral frameworks (Kertzer et al., 2014). What is consistent across both models is that members of the public who are not directly affected by the events in question must learn something about these events if they are to form an opinion about them. In the case of factory closings, those who live in close proximity to the factory may learn about the closing without the help of the media, but most members of the public are likely to learn about a factory closing through the media or from members of their network who learned about the closings through the media.
Attributing Blame
Building from the current attribution literature, theoretically we focus on two components of blame attribution: first, the role of media in shaping the breadth of blame attribution and second, the differential effects of these frames across individuals (focusing on political ideology, nationality, and news consumption).
Much of the prior literature on economic blame attribution focuses primarily on the extent to which the public hold the current government responsible for economic conditions. Scholars of retrospective voting (e.g., Fiorina (1981); Healy and Malhotra (2013); Lewis-Beck and Stegmaier (2000)) assume that voters link economic outcomes to incumbent politicians, and much evidence suggests that voters in general do link overall economic performance to a country’s incumbent leaders (Hibbing & Alford, 1981; Karyotis & Rüdig, 2015).
However, economic outcomes are seldom the result of a single cause. Hellwig et al. (2008) find that individuals can assign blame across a wide spectrum of actors, including businesses and markets as well as the government. In comparison to general economic conditions, we expect that specific economic events—such as factory closings—offer an even more constrained set of actors for the public to assess and attribute blame. Following Rudolph (2003), we assume that individuals simplify their assessment by first considering whether they primarily blame private or public sector actors before determining what types of response they would support. Identifying responsible actors and adjudicating between them is an information-intensive task, and individuals may differ in their ability to assess the responsibility of one or multiple actors. Gomez and Wilson (2001) propose that voters’ ability to attribute responsibility for personal and national economic conditions depends on their political sophistication, measured by their political knowledge base. Relatedly, Hellwig et al. (2008) consider the role of education in broadening individuals’ blame attribution beyond proximate sources such as the government to distal forces such as national and international markets and businesspeople. While recognizing the same informational issue that could hinder the attribution process, we argue that media coverage provides the frames that allow people to assign blame to actors beyond the government, regardless of their political sophistication or education levels.12
We argue the media provides contextual information about mass layoffs, which is an important factor in allowing individuals to adjudicate blame (Stiers, 2021). In studying blame attribution for the US governments’ failures after Hurricane Katrina, Malhotra and Kuo (2008) found that while respondents’ uninformed response was to blame the opposite party, when provided information about job titles, respondents partitioned blame according to job responsibilities instead. Similarly, in a case where the intersection of states’, international organizations’, and non-governmental organizations’ responses to the Darfur crisis created uncertainty, Ecker-Ehrhardt (2010) argues that the media coverage served to allocate responsibility and promote specific policies. In the case of mass layoffs, because of the consequences for government-based solutions, we consider how the media’s identification of involved actors and their role influences whether or not individuals attribute primary blame to the government or other actors.
H1
Individuals reading news frames highlighting non-governmental actors as the source of mass layoffs are less likely to blame the government than those reading frames highlighting government actions.
If H1 is true, news articles should lead their readers to greater convergence in the attribution of blame and, as discussed later, beliefs about policy responses. However, we must also consider whether the effects of frames on blame attribution may differ across individuals according to their political ideology, nationality, and news consumption.
Recent work by Bisgaard (2015) raises the question about the extent to which individuals use motivated reasoning when assigning blame for economic outcomes and in doing so further cement ideologically driven policy preferences. In the case of the GM factory closings, the media frequently highlighted the role of the Trump Administration’s tariff increases. If political ideology shapes blame attribution, then we would expect that conservatives in the US would be less likely than their liberal counterparts to accept negative information about the effect of Trump’s policies and blame the government for those policies.
H2a
Americans who identify as liberals will be more likely than conservatives to respond to frames blaming the high costs of tariffs by assigning blame to the government.
As alluded to in the previous hypothesis, we expect there to be differences in how Americans and Canadians respond to different frames about factory closings. Since GM’s factory closings occurred in both the US and Canada, we can evaluate both the similarities and differences in how citizens of each country respond to different media frames. The dimension where we have the strongest theoretical priors for divergent treatment effects is political ideology. Though we expect ideology to be a strong moderator in the United States, we do not expect ideology to have as strong of an effect in Canada. Unlike in the United States, Canadian conservatives would not associate the high costs of tariffs with their leader’s policies, and so we would not expect as strong of an ideological processing of the media frames as in the US.
H2b
We do not expect Canadian liberals to respond differently than conservatives when the media blames the high costs of tariffs for the factory closings.
Lastly, we also theorize about the influence of media coverage on different types of people, based on their level of engagement with the relevant news. When considering the effects of media framing, we would expect that media frames would be most influential on those who have the weakest priors about the events in question. For example, we would expect that if someone was already following the events related to GM’s factory closings, then reading another article about the factory closing would have relatively little effect on their opinion about the closings. By contrast, if someone had not been following the story, then we would expect the media’s framing to have a relatively larger effect on their view of the events.13
H3
Those who have actively been following the story of GM’s factory closings will have less of a response to additional news about the closings than those who have not been following the story.
Connecting Blame to Policy Preferences
We argue that by influencing which decision-makers are viewed as responsible, common frames should influence the public’s support for ameliorating government policies. As Iyengar (1994) noted, the media’s ability to frame responsibility influences the public’s understanding of the causes and solutions to social problems. Given our focus on factory closings and layoffs, and the repeated rhetoric blaming trade for America’s manufacturing decline, we focus on how blame attribution affects support for trade and related social policies. We theorize about the influence of blame attribution on support for common responses to economic volatility: trade policy, income support, and retraining. Each type of policy has been used extensively in the past by the US and Canada, with the goal of either assisting workers who have been laid off or to protect industries and workers from forces leading to layoffs.
Prior work suggests that media coverage of layoffs could be a double-edged sword. On one hand, when increased context shifts the blame away from those in need of government assistance, the affected group is more likely to receive sympathy (Kluegel & Smith, 2017; Harell et al., 2013). On the other hand, increased context may negatively affect support for national policy response. Iyengar and Kinder (2010) found that when the news focuses on negatively affected societal actors and individuals specifically, a more ‘vivid’ (their term) presentation can diminish support for a national response. In other words, the more readers know about the specifics, the less likely they are to view the problem as a general one to be solved by national-level policies.
In the case of the GM factory closings, we theorize that providing more context about factory closings, especially via frames that focus blame on GM, may serve to decrease, rather than increase, support for national policy solutions. Specifically, when GM or industry specific market conditions are blamed, we would expect the public to be less supportive of national government policies, since the problem is more likely to be viewed as limited to the company, as opposed to being a national issue.
H4
Individuals reading news frames highlighting non-governmental actors, or industry-specific market forces, as the source of mass layoffs are less likely to support government assistance via national policies than those reading frames highlighting government actions.
As with blame attribution, we expect that motivated reasoning will limit the influence of frames directly implicating the Republican Administration in power at the time. Consistent with H2a and H2b, we expect frames blaming the high costs of tariffs to have less of an effect on American conservative’s policy preferences, as opposed to American liberals, though we do not expect this same motivated reasoning to occur amongst Canadians, as noted in H2b.
H5
Americans who identify as liberals will be more likely than conservatives to respond to frames blaming the high costs of tariffs by supporting open trade policies.
Furthermore, when it comes to policy preferences, there are two competing reasons we may expect Americans and Canadians to respond differently to factory closings. First, since Canada has a much more robust social safety net than the US (Brown, 2016), we might expect that Canadians would be more likely than Americans to believe they already have a sufficiently strong social safety net. If so, we would expect Canadians’ preference for social safety net policies to be less responsive to blame attribution than their American counterparts. Conversely, it may be that Canadians’ greater experience with public programs, such as universal healthcare, have led them to expect the government to step in with new policies to support members of the public who are struggling, such as recently laid off workers. In this second scenario, we would expect Canadians’ preference for social safety net policies, such as unemployment and education programs, to respond more positively than Americans’, especially when exposed to frames that blame the government for factory closings. We do not have strong theoretical expectations about which of these effects is likely to be more influential, so we test the relative effects by examining the following hypotheses.
H6a
Because of extant welfare programs, the effect of blame attribution frames will affect Canadian’s policy preferences less than American’s policy preferences.
H6b
Because of support for extant welfare programs, Canadians will respond to non-company frames with increased support for welfare and education compared to Americans.
Analysis of Media’s Coverage of GM Factory Closing
To understand how the media portrays factory closings and the information environment that the public is exposed to, we conducted an analysis of news coverage in the United States and Canada. Our analysis of the GM factory closings lets us examine which frames the media chooses to present in its coverage. We focus our media analysis on the 100 days after GM closed the Lordstown and Oshawa plants. We used NexisUni to search for articles in national, regional, and local papers in the US and Canada, which we then manually coded to identify to what extent blame attribution for the closing occurred, and what types of blame attribution were most common. This coding provides the material for a rich descriptive analysis of the media environment in both countries. Details about the search queries and coding procedures are provided in section 1 of the appendix.
We find that blame attribution is extremely common in the media coverage of factory closings. More than 89 percent of the articles we analyze blamed at least one of the following: the government, General Motors, or market forces. In Fig. 1 we break down the media’s blame attribution by country and the target of the blame. In both countries, market forces are the most frequently cited cause of the factory closings, though GM and the government also receive substantial blame. While similar proportions of articles in each country blame the government (26 percent in the US and 24 percent in Canada), the US press was more than twice as likely to blame GM than the Canadian press. This analysis demonstrates that blame attribution is a major component of articles about factory closings, and the public is repeatedly exposed to competing frames about the causes of factory closings.14
Fig. 1
Blame Attribution in News Articles.
Note: Fig. 1 displays the percentage of articles about GM factory closings that mention specific causes for the closures. The search query for articles was limited to the day of the respective factory closing (Lordstown and Oshowa), up until 100 days later. Articles may mention more than one cause, which is why the percentages add to more than 100
When we dive deeper into the reported causes of GM’s factory closings, we find that the specific government policies that are most often blamed relate to trade and tariffs. Within our sample, 20 percent of US articles and 25 percent of the Canadian articles about the factory closings mention trade policy. Somewhat surprisingly, we find that the arguments presented about trade policy’s effects on factory closings are roughly split between blaming higher tariffs, which raise the costs of producing cars, and blaming lower tariffs, which increase international competition. Eight percent of the US articles and 7 percent of the Canadian articles portray lower tariffs and liberalized trade as a cause of factory closings. In almost equal proportions, 9 percent of the US articles and 10.5 percent of the Canadian articles identify tariffs and their role in raising production costs for GM as a cause of factory closings.
Experimentally Testing Media’s Coverage Effect on Blame Attribution
To test how media coverage shapes public perceptions of accountability and policy responses to mass layoffs, we employ a survey experiment fielded on diverse national samples in the United States and Canada.15 The advantage of using a survey experiment is that it allows us to randomly assign different media frames to respondents, so we can measure the causal effect of different justifications for mass layoffs and the effect of those justifications on blame attribution and policy preferences.
We fielded our study with the survey firm Dynata in the summer of 2020. In an effort to ensure data quality, we follow the advice of Burleigh et al. (2018), blocking respondents from participating if they were located outside of our sample country (US or Canada) or were flagged for using a Virtual Private Server (VPS). We also checked the quality of respondents based on a free response question prior to our study. Respondents who wrote gibberish or who entered text that was unresponsive to the prompt were deemed to not be paying attention and were dropped from the sample. This process yielded a sample of about 6000 respondents who consented to the research, passed our quality checks, and completed our study.16
To test the effects of media coverage on blame attribution and policy preferences related to mass layoffs, our study first informed all respondents that they would “read a news report about developments in the auto industry and then be asked your opinion on the situation.” Each news report was based on actual media coverage and public statements that had been reported in the news. Respondents were randomly assigned to either the control condition or one of four treatment conditions. For all conditions, the news report included a bold headline announcing “General Motors to close US and Canadian plants.”17 The headline was followed by an image of the Lordstown, Ohio General Motors plant for the US respondents and a picture of the Oshowa, Ontaria General Motors plant for Canadian respondents.18 The full text of the experiment is displayed in Fig. 2.
Fig. 2
Survey Instrument.
Note: Expected job losses were higher at Oshawa (2900) than Lordstown (1618), but we used the floor of “at least 1500” to ensure our treatments were consistent across the US and Canadian surveys. In the tariff treatments, the text shown specifies “Someone familiar with the decision noted...” In the full study, the information provider was randomly assigned from a variety of potential cue givers, which the authors analyze elsewhere. Because the treatments are presented in the style of a newspaper article, the treatment is “bundled” in that the paragraphs broadly alter the theoretical concept of interest, though there are additional subtle differences that may extend beyond the treatment
After reading the news story, respondents were presented with a bulleted summary of the story and were then asked to answer a series of questions about the layoffs. To assess blame attribution, we asked respondents “which of the following most closely resembles your thoughts about why the factories are closing?” Respondents could select from “Government policies failed,” “General Motors management failed,” or “Other reasons.” We also asked respondents about their attitudes toward various types of government policies, which allows us to measure how different media frames of the layoffs shape public support for policy responses. Specifically, we asked “Do you favor or oppose [the United States / Canada] reducing its barriers to trade?” Response options were on a five-point scale from “strongly favor” to “strongly oppose” with higher values corresponding to supporting reduced barriers to trade. This scale forms the dependent variables “Support for Trade.” We also asked respondents “Which of the following do you believe should be available to laid off G.M. workers?” Respondents were provided a list of options and were asked to check “yes” or “no” for whether each should exist. The responses were aggregated into two variables measuring support for different types of government assistance programs: “Support for Wage Support” and “Support for Education and Training.” The first measures support for wage-supplement policies, specifically unemployment benefits or wage supplements.19 The second measures support for retraining and education programs. Each set of responses was summed and rescaled from zero to one for ease of interpretation, with zero representing low support and one representing high support. Taken together, these measures let us evaluate how different justifications for the auto layoffs affect public blame attribution, and subsequently how those justifications alter support for trade and government social safety net policies.
Results from the Survey Experiment
We proceed through our analysis in three steps. First, we report the blame attribution from the control condition, which allows us to establish a clear baseline of how the public perceives the factory closing and associated layoffs. We then report the effects of our treatments for the full sample (H1), followed by a brief examination of whether respondents’ ideology influences their response to the treatments, whether American and Canadian publics respond similarly (H2a and H2b), and whether respondents’ self-reported consumption of news about the auto layoffs moderates the treatment effects (H3). Lastly, we analyze how the treatments shape public preferences for trade and social safety net policies (H4), conditional on ideology (H5) and nationality (H6a/b). Our results show that public blame is significantly shaped by media frames. Somewhat surprisingly, we find that ideology does not moderate most of our treatment effects, nor does nationality; however, ideology does shape how Americans respond to the high costs of tariffs, which were associated with former President Trump’s trade policy. Furthermore, we find that the public’s support for trade policy also shifts with different media frames, though we do not find that the media frames affect support for government safety net policies, regardless of ideology or nationality.
Analysis of the control group shows that the public is divided over whom to blame for auto layoffs. As shown in Fig. 3, we find that 18.3 percent of respondents blame the government, 50.6 percent blame General Motors, and 31.1 percent blame something else. These results are relatively similar across the US and Canadian respondents, though slightly more Canadian respondents blame GM (55.2%) than American respondents (45.6%). Although the blame is split, in both countries General Motors receives the highest share of the blame from the public. As a reminder, the control simulates a Reuters or Associated Press news report providing only information about the factory closing and number of employees laid off. Thus the baseline captures respondents blame attribution without additional media framing.
Fig. 3
Baseline blame attribution—from control group.
Note: Fig. 3 displays the proportion of respondents who attribute responsibility for the factory closing to the government, General Motors, or “other” in the control condition
We now turn to the focus of our study, which is how media coverage affects blame attribution for the auto layoffs (H1). The average treatment effects for the full sample are presented in Fig. 4. The most striking results are for the two tariff treatments. Both the Tariff-Competition and Tariff-Costs treatments result in more respondents blaming the government. The largest treatment effect in our study is from the Tariff-Costs treatment, which results in 27 percent more of the respondents blaming the government (\(p<0.01\)), but the Tariff-Competition treatment also has a strong effect, increasing blame to the government by about 16 percentage points (\(p<0.01\)). In each of these treatments it appears that the mention of government tariff policies and the argument that they have hurt the auto industry—regardless of whether this is due to increased international competition from lower tariffs or due to higher costs of materials like steel—resonates with the public and shifts blame to the government. The clear beneficiary from both of these framings is General Motors, with 18 percent fewer respondents blaming them in the Tariff-Costs treatment (\(p<0.01\)) and 7 percent fewer in the Tariff-Competition treatment (\(p<0.01\)).
Consistent with the results from the tariff treatments, we find that any of our explanations about layoffs appear to help General Motors receive less blame. The treatment effects on public blame of GM are all negatively signed and, with the exception of Market Conditions, each is significant at \(p<0.01\). While the public does not generally think well of large corporations (Public Affairs Council, 2015), our results show blame can be easily diverted from the company conducting layoffs to the government or others through media messaging. This suggests that additional coverage of mass layoffs, which almost always includes an explanation for why companies are conducting layoffs, is likely to reduce blame to the company while shifting the public’s ire to other factors.
Fig. 4
Framing treatment effects.
Note: Fig. 4 displays the treatment effects on the proportion of respondents blaming the government, General Motors, or “other” for the full sample. The figure contains the average treatment effects and 95 percent confidence intervals from a series of OLS regressions run for each blame variable. The baseline proportions of blame attribution in the control are: 0.183 blame the government, 0.506 blame GM, and 0.311 blame “other”. The full regression results are displayed in section 6 of the appendix
Perhaps the most surprising results from our study are the effects of the Pandemic treatment. Like the tariff treatments, the Pandemic treatment reduces blame to GM, but it has no effect on blame to the government. This means that the public is not holding the government responsible for auto layoffs attributed to the pandemic, and yet the public is, at least partially, willing to let GM off the hook, with 12 percent fewer respondents blaming GM in the pandemic treatment (\(p<0.01\)). This suggests that the government is somewhat insulated from public blame due to layoffs associated with the pandemic.
For the preceding results, we also analyzed whether American and Canadian respondents reacted differently to the treatments and whether liberals and conservatives responded differently (H2a and H2b). We tested for these heterogeneous treatment effects by interacting the treatments with an indicator for whether the respondent was in the US study and interacting our treatments with a seven-point measure of ideology on the liberal-conservative spectrum as well as a trichotomous measure of liberals, moderates, and conservatives in case moderates behave differently (Theodoridis, 2017; Klar & Krupnikov, 2016).20 The full results testing for differences between American and Canadian respondents are reported in section 7 of the appendix. The only significant interaction is for the Tariff-Costs treatment on the measure of blame for GM. The effect of the Tariff-Costs treatment was negative and significant for both the US and Canadian samples, with the effect among Canadians resulting in 23 percentage points fewer respondents blaming GM, but the effect of the Tariff-Costs treatment was significantly smaller among Americans, with the treatment resulting in 12 percent fewer respondents blaming GM in the US sample. Other than this one difference, no other treatments had significantly different effects for Canadians or Americans.21
The full results for our ideology interactions are reported in §8 of the appendix. The only interaction that reaches significance is ideology and the Tariff-Cost treatment, where we find that more conservative respondents are less likely to blame the government when layoffs are attributed to the high costs of tariffs (\(p < 0.01\)). Conservative respondents are also somewhat more likely to blame GM in this treatment condition (\(p < 0.08\)). These results are driven by American conservatives, but not Canadians, which is consistent with H2a and H2b, as shown in §8 of the appendix. These findings suggest that US conservatives may be less likely to blame the government for the high cost of tariffs that were imposed by former President Trump.
To test our third hypothesis, we consider whether our media framing treatments have differential effects among those that are, or are not, following the news. Specifically, we asked respondents “Were the GM plant closings a story that you had followed in the news?” In our sample, 37 percent of respondents reported that they had been following the story about GM plants closing. As noted in H3, we expect that respondents who had actively been following the story would have stronger priors about who was responsible for the layoffs, and thus we would expect our treatments to have smaller effects among those already following the story. Consistent with our expectations, the effect sizes are generally smaller amongst those following the news, which is seen in Fig. 5 by comparing panel (a) to panel (b). Even though the differences in effect sizes between those following and not following are in the expected direction, they are not significantly different for most treatments, as shown in section 9 of the appendix. However, the differences are significant for the Pandemic treatment, with those not following the news significantly more likely to blame “other” and those following the news somewhat more likely to blame GM. Although we observe some separation between the previously informed and the uninformed, overall the treatments move both types of respondents, suggesting that when observed in the real world, frames influence both relatively uninformed and relatively informed readers.
Fig. 5
Treatment effects for those (Not) Following the News.
Note: Fig. 5 displays the treatment effects on the proportion of respondents blaming the government, General Motors, or “other.” Panel a consists of those who self-reported they were following the auto story and panel b consists of those who self-reported they were not following the auto story. Each effect is calculated running a separate model for that dependent variable. For those following the auto story, the proportion blaming each option in the control is 0.211 blaming government, 0.550 blaming GM, and 0.239 blaming other. For those not following the auto story, the proportion blaming each option in the control is 0.165 blaming government, 0.479 blaming GM, and 0.356 blaming other
Support for Trade and Government Assistance Policies
We now turn to the question of how media frames affect support for policy choices. We begin by testing whether the media frames affect respondents’ support or opposition to reducing barriers to trade, income assistance, and retraining (H4). In the main text, we focus on trade policy since politicians repeatedly turn to trade policy as a tool to protect American workers and foreign competition is often blamed for the demise of US production.
The results of our study show that both tariff treatments shift support for trade policy, and they do so in opposite directions, as expected.22 By contrast, neither the Market Conditions or Pandemic treatments have any effect on support for trade, as we discuss below. The results for “Support for Trade” are displayed in Fig. 6.23 The Tariff-Competition and Tariff-Costs treatments generate mirror image shifts in preferences, although they differ slightly in strength of significance. The Tariff-Competition treatment, which emphasizes that reducing tariffs increases foreign competition, significantly decreases support for trade on the five-point scale (\(-\)0.09, \(p=0.05\)). The effect in substantive terms is a 3 percentage point decline in support for liberal trade policies. By contrast, we find that the Tariff-Costs treatment has a similarly sized, but positive, effect on the five-points scale for supporting trade liberalization (0.08, \(p=0.08\)). The substantive effect of the Tariff-Costs treatment is a 5.1 percentage point increase in support for liberal trade policies (\(p=0.02\)). When we include controls for individual-level factors known to shape attitudes toward trade, the effects of both the tariff treatments are significant with p-values less than 0.05.24 Neither effect size would support an interpretation that a single news article dramatically transforms individuals’ preferences for trade. But rather, in light of other numerous determinants of preferences, including current partisanship on trade, they support an interpretation that cumulative interactions with such frames may over time influence individuals’ preferences.
Fig. 6
Effects on support for trade Fig. 6 displays the treatment effects on the five-point trade support measure, where higher values indicate more support for reducing barriers to trade. The average trade support score in the control condition is 3.480, on the scale from 1 to 5. The figure contains the average treatment effects and 95 percent confidence intervals from OLS regressions. The full regression results are displayed in §10 of the appendix
Additionally, we examine the general effects of media frames on support for worker-focused government safety net policies, such as wage-support programs or retraining and education programs. We do not find that the treatments have a significant effect on support for other government policies, as shown in section 11 of the appendix. In fact, we find remarkably stable null results, which are also robust to inclusions of a broad range of controls that are likely to shape attitudes toward government assistance programs, as shown in section 12 of the appendix.25 We do find that that respondents differ in their baseline support for policies based on whether they were in the US or the Canadian sample, with Canadians more likely to support training and education programs. As before, we test for heterogeneous treatment effects by ideology (H5) and nationality (H6). The treatments did not have differential effects in the US and Canada (Appendix, §7) or across the ideological spectrum (Appendix, §15). While conservatives show less support for non-trade related assistance in the form of income support and retraining, conservatives were no more or less influenced by the media coverage than liberals.
In summary, while media frames influenced blame attribution, the rollover effect in terms of policy preferences were weak for all but the change in preferences for trade liberalization. Notably, the treatment effects were similar regardless of education or ideology, suggesting that individuals responded similarly to the information. Some prior literature (Hellwig et al., 2008; Rudolph, 2003) assumed that general knowledge or political sophistication initially influenced blame attribution for overall economic conditions. In the case of a specific event, such as a factory closing, we find little evidence that education as a proxy for general knowledge or political sophistication mattered in the attribution of blame. Instead, we found evidence that how media coverage framed the event influenced blame attribution, and to a lesser extent policy preferences, and that these effects were similar regardless of education level.
Similarly, while prior scholarship shows that ideology and partisanship appears to shape overall assessments of economic conditions and related attribution of responsibility, assessments of blame for a specific economic event were less conditioned by ideology. As much of the literature has focused either on general economic conditions or specific crisis events, our findings provide insights into how voters evaluate smaller-scale but commonplace economic events that could shape their perception about the suitability of government assistance.
Conclusion
Just days after GM’s announced closure, the online news magazine Politico exhorted it’s readers “Don’t just blame Trump for GM’s Layoffs—Blame GM.”26 Our analysis of media coverage in the US and Canada found that media outlets framed factory closures in terms of government, corporate, or market responsibility; and in turn, our analysis of our survey experiment shows that the variation in the framing used helps shape who readers blame for the factory closing and thus their preferred policy responses. We found that commonly occurring frames used by media sources—particularly the Tariff-Competition and Tariff-Costs frames—not only shifted blame, but also had significant effects on policy preferences.
Our results also suggest that the public has more entrenched views toward domestic government assistance programs, such as unemployment benefits and worker training programs, and is more likely to shift their trade policy preferences in response to media frames. While shifting the blame to the government for trade policies resulted in matching shifts in preferences for trade policies, no frame influenced preferences for government assistance among affected individuals. It is possible that the high levels of support for government welfare programs were influenced by the Covid-19 pandemic, though our results corroborate pre-pandemic studies. Like Di Tella and Rodrik (2020) we found that although support for direct worker assistance is high, the frames presented in media coverage of factory closings resulted in larger shifts in trade policy preferences than support for government assistance programs.27
The divergent effects of media framing on support for trade policies versus other policy responses to mass layoffs have a number of important implications. First, it suggests that politicians are acting strategically when they repeatedly turn to trade policy as a policy response to factory closings or layoffs. We find that the public is quick to blame the government when media reports draw connections between trade and layoffs in the manufacturing sector, as they frequently do. The results also show that the public is swayed by such reporting, which shifts their attitudes toward trade policy. Thus, politicians who are concerned about being blamed for poor government policies, as retrospective voting models predict, can respond to public preferences by publicizing their trade policies as benefitting domestic workers and companies.
When considering how the public shifts blame, it is worth noting that our measurement of blame attribution required respondents to identify who they blamed the most. While this allows us to evaluate who the public views as most responsible for the factory closings, it does not account for the potential nuance of people potentially blaming multiple actors. We view it as a productive avenue for future research to examine how people distribute blame across multiple actors, and whether primarily blaming one, or multiple, actors influences preferences for policies or politicians.
Notably, our findings were surprising in their consistency across the American and Canadian studies. In each country we find that the public alters their trade preferences in response to frames shifting blame for factory closings where as their views toward government assistance programs remain relatively unaffected by the same frames. Even the Pandemic treatment did not generate increased demand for wage-support policies. This result, combined with the finding that the government does not receive additional blame when layoffs are attributed to the pandemic, suggests that the government and individual politicians are somewhat insulated from blame and public pressure for increased government assistance from reports of factory closings and layoffs during the pandemic.
Importantly, our results show evidence that non-partisan news generates more convergence than divergence in opinion. In contrast to Bisgaard (2019), we found little evidence that the news stories prompted ideological divergence in blame attribution. Similarly, the effects of media framing on blame attribution were generally similar for those with high and low levels of education. Some prior literature (Hellwig et al., 2008; Rudolph, 2003) assumed that general knowledge or political sophistication initially influenced blame attribution for overall economic conditions. By contrast, our results demonstrate that for a specific event, like factory closings, media frames move those with and without college degrees. The consistency of our findings across the US and Canada, and various subpopulations, suggests that the nationality of the parent company may matter little when people are considering factory closings. Given the increasingly global nature of supply chains, and the relative lack of importance placed on corporate headquarters, the consistency of responses across countries suggests an important calculation by voters that should be explored more deeply and in other contexts.
Acknowledgements
We thank those who provided feedback on earlier drafts of this work at IPES, GRIPE, ISA, APSA, Harvard IR Speaker Series, and University of Pennsylvania’s Browne Center. We also thank Leonardo Baccini, Madison Brown, Thomas Chadwick, Stephen Chaudoin, Emma Guard, Sam Hall, Alistair Johnston, Regan Kin, David Kline, and Katie Lee for their feedback and/or research assistance.
Declarations
Conflict of interest
The authors have no conflict of interest to declare that are relevant to the content of this article.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Indianapolis Star, October 30, 2020. https://www.indystar.com/story/news/politics/2020/10/30/trump-campaigned-saving-jobs-carrier-what-its-like-there-now/6010437002/
In GM’s announcement, Barra said “The actions we are taking today continue our transformation to be highly agile, resilient and profitable, while giving us the flexibility to invest in the future. We recognize the need to stay in front of changing market conditions and customer preferences..” (November 26, 2018).
Donald Trump “Just spoke to Mary Barra, CEO of General Motors about the Lordstown Ohio plant. I am not happy that it is closed when everything else in our Country is BOOMING.... She blamed the UAW Union.. ” Twitter. March 17, 2019.
For example, Joe Warmington, “GM Oshawa closure proof government bailouts don’t work: O’Leary" Toronto Sun. November 28, 2018 https://torontosun.com/news/local-news/gm-oshawa-closure-proof-government-bailouts-dont-work-oleary.; Tonda MacCharles "GM plant closure stirs questions about massive 2009 government bailout," Toronto Star. November 27, 2018 https://www.thestar.com/politics/federal/2018/11/26/gm-plant-closure-stirs-questions-of-massive-2009-government-bailout.html.
Bureau of Labor Statistics, US Department of Labor, The Economics Daily, 9 out of 13 industries had net job gains in the fourth quarter of 2018 on the Internet at https://www.bls.gov/opub/ted/2019/9-out-of-13-industries-had-net-job-gains-in-the-fourth-quarter-of-2018.htm (visited January 22, 2021).
Statistics Canada. Table 14-10-0287-01 Labour force characteristics, monthly, seasonally adjusted and trend-cycle, last 5 months. Statistics Canada. Table 14-10-0022-01 Labour force characteristics by industry, monthly, unadjusted for seasonality (x 1000).
Bureau of Labor Statistics, US Department of Labor. News Release. May 8, 2020 (corrected September 23, 2020). www.bls.gov/news.release/archives/empsit_05082020.pdf.
It is also possible that respondents who are more economically vulnerable would respond differently to the treatments than those who are more economically secure. Similarly, those with higher levels of education (or political sophistication) may respond differently (Hellwig et al., 2008) We test for such heterogeneous treatment effects in sections 13 and 14 of the appendix, using multiple measures of economic vulnerability and a measure of whether the respondent obtained a college degree.
We analyze whether media outlets with the largest circulation report differently than smaller/regional outlets, though we do not find significant differences, as reported in §2 of the appendix.
This study was not pre-registered, though we make sure to report both significant results (blame attribution and trade support) and insignificant results (support for government assistance programs) to ensure transparency in the research process.
These results are generally consistent with findings from Di Tella and Rodrik (2020) who find that members of the public often look to trade policy as a response to economic shocks.
For those who are interested in how we think about the mediating effect of blame attribution on support for government policies, we discuss this in section 16 of the appendix.
Controls are reported in Fig. 6, which include factors that shape attitudes toward trade such as education, gender, income, age, and ideology (Mansfield & Mutz, 2009; Scheve & Slaughter, 2001).
These results are in contrast to the findings of Bacciniet al. (2022), who find that in the European context politicians are rewarded for advocating for redistribution during tough economic times.
Jaime Lincoln Kitman, “Don’t Just Blame Trump for GM’s Layoffs-Blame GM,” November 29, 2018. https://www.politico.com/magazine/story/2018/11/29/generalmotorstrump-layoffsfactoriesautoindustry222696.
We thank a reviewer for noting this is also consistent with Colantone and Stanig (2018).
Baccini, L., Guidi, M., & Poletti, A. (2022). Policy preferences in tough times: Experimental evidence.
Bennett, W. L. (1990). Toward a theory of press-state relations in the United States. Journal of Communication,40(2), 103–127.CrossRef
Bisgaard, M. (2015). Bias will find a way: Economic perceptions, attributions of blame, and partisan-motivated reasoning during crisis. The Journal of Politics,77(3), 849–860.CrossRef
Bisgaard, M. (2019). How getting the facts right can fuel partisan-motivated reasoning. American Journal of Political Science,63(4), 824–839.CrossRef
Brown, T. K. (2016). How US welfare compares around the globe. BBC News. August 26.
Brutger, R., & Strezhnev, A. (2022). International investment disputes, media coverage, and backlash against international law. Journal of Conflict Resolution,66(6), 983–1009.CrossRef
Burleigh, T., Kennedy, R., & Clifford, S. (2018). How to screen out VPS and international respondents using qualtrics: A protocol. Available at SSRN 3265459.
Colantone, I., & Stanig, P. (2018). The trade origins of economic nationalism: Import competition and voting behavior in Western Europe. American Journal of Political Science,62(4), 936–953.CrossRef
Ecker-Ehrhardt, M. (2010). Aid organizations, governments and the media: The critical role of journalists in signaling authority recognition. Public Policy and the Mass Media: the Interplay of Mass Communication and Political Decision Making, edited by Sigrid Koch-Baumgarten and Katrin Voltmer: pp. 106–123.
Fiorina, M. P. (1981). Retrospective voting in American national elections. Yale University Press.
Fogarty, B. J. (2005). Determining economic news coverage. International Journal of Public Opinion Research,17(2), 149–172.CrossRef
Gomez, B. T., & Wilson, J. M. (2001). Political sophistication and economic voting in the American electorate: A theory of heterogeneous attribution. American Journal of Political Science,45, 899–914.CrossRef
Green-Pedersen, C., & Stubager, R. (2010). The political conditionality of mass media influence: When do parties follow mass media attention? British Journal of Political Science,40(3), 663–677.CrossRef
Guisinger, A. (2017). American opinion on trade: Preferences without politics. Oxford University Press.CrossRef
Hacker, J. S. (2006). The great risk shift: The assault on American jobs, families, health care and retirement and how you can fight back. Oxford University Press.
Harell, A., Soroka, S., & Iyengar, S. (2013). Race, gender and support for the welfare state: A comparative experimental approach. In: Annual Meeting of the Canadian Political Science Association, Victoria BC.
Healy, A., & Malhotra, N. (2013). Retrospective voting reconsidered. Annual Review of Political Science,16, 285–306.CrossRef
Hellwig, T. T., Ringsmuth, E. M., & Freeman, J. R. (2008). The American public and the room to maneuver: Responsibility attributions and policy efficacy in an era of globalization. International Studies Quarterly,52(4), 855–880.CrossRef
Hibbing, J. R., & Alford, J. R. (1981). The electoral impact of economic conditions: Who is held responsible? American Journal of Political Science,25, 423–439.CrossRef
Iyengar, S. (1994). Is anyone responsible?: How television frames political issues. University of Chicago Press.
Iyengar, S., & Kinder, D. R. (2010). News that matters: Television and American opinion. University of Chicago Press.CrossRef
Iyengar, S., & Simon, A. (1993). News coverage of the gulf crisis and public opinion a study of agenda-setting, priming, and framing. Communication Research,20(3), 365–383.CrossRef
Ju, Y. (2008). The asymmetry in economic news coverage and its impact on public perception in South Korea. International Journal of Public Opinion Research,20(2), 237–249.CrossRef
Karyotis, G., & Rüdig, W. (2015). Blame and punishment? The electoral politics of extreme Austerity in Greece. Political Studies,63(1), 2–24.CrossRef
Kertzer, J. D., Powers, K., Rathbun, B. C., & Iyer, R. (2014). Do moral values shape foreign policy preferences? Journal of Politics,76(3), 825–40.CrossRef
Kertzer, J. D., & Zeitzoff, T. (2017). A bottom-up theory of public opinion about foreign policy. American Journal of Political Science,61(3), 543–558.CrossRef
Klar, S., & Krupnikov, Y. (2016). Independent politics. Cambridge University Press.CrossRef
Kluegel, J. R., & Smith, E. R. (2017). Beliefs about inequality: Americans’ views of what is and what ought to be. Routledge.CrossRef
Lewis-Beck, M. S., & Stegmaier, M. (2000). Economic determinants of electoral outcomes. Annual Review of Political Science,3(1), 183–219.CrossRef
Malhotra, N., & Kuo, A. G. (2008). Attributing blame: The public’s response to Hurricane Katrina. The Journal of Politics,70(1), 120–135.CrossRef
Mansfield, E. D., & Mutz, D. C. (2009). Support for free trade: Self-interest, sociotropic politics, and out-group anxiety. International Organization,63(3), 425–457.CrossRef
Public Affairs Council. (2015). Attitudes about big business and small business. Public Affairs Pulse Survey. Accessed December 18, 2020.
Rafael, D. T., & Rodrik, D. (2020). Labour market shocks and the demand for trade protection: Evidence from online surveys. The Economic Journal,130(628), 1008–1030.CrossRef
Rudolph, T. J. (2003). Who’s responsible for the economy? The formation and consequences of responsibility attributions. American Journal of Political Science,47(4), 698–713.CrossRef
Scheve, K. F., & Slaughter, M. J. (2001). What determines individual trade-policy preferences? Journal of International Economics,54(2), 267–292.CrossRef
Stiers, D. (2021). Political information and retrospective voting. West European Politics,44(2), 275–298.CrossRef
Theodoridis, A. G. (2017). Me, myself, and (I), (D), or (R)? Partisanship and political cognition through the lens of implicit identity. The Journal of Politics,79(4), 1253–1267.CrossRef
van der Pas, D. J., van der Brug, W., & Vliegenthart, R. (2017). Political parallelism in media and political agenda-setting. Political Communication,34(4), 491–510.CrossRef
