«13TH INTERNATIONAL PUBLIC RELATIONS RESEARCH CONFERENCE “Ethical Issues for Public Relations Practice in a Multicultural World” Holiday Inn ...»
This study analyzed the crisis communication strategies employed by the 10 largest banks in America during the onset of the 2008 financial crisis. The study examined the news releases of the these banks from July 1, 2008, through December 31, 2008, a six-month period during which the Dow Jones Industrial average dropped nearly 23 percent, Americans’ net worth fell nearly 18 percent, and the Consumer Confidence IndexTM slid more than 13 percent (Franco, 2008;
Hagenbaugh, 2009; Shell, 2009). News stories in the Wall Street Journal and the hometown newspaper of each bank’s headquarters city were also analyzed. The purpose of analyzing news coverage in these newspapers was not to correlate specific news releases with subsequent news articles, but rather to gauge the cumulative effect of the banks’ news releases on the overall tone of their news coverage during this volatile time. Wall Street Journal provided national perspective, while the banks’ hometown newspapers provided local and regional perspective.
This study was designed to yield lessons about managing reputation and maintaining public confidence in the midst of an industry-wide crisis. Beyond practical application for those who work in public relations, this study was also designed to evaluate two existing bodies of scholarly research – image repair theory (IRT) and situational crisis communications theory (SCCT). IRT, introduced in 1991 and originally called image restoration theory, focuses on the message strategies an organization might take when faced with a crisis (Benoit, 1997). What constitutes appropriate, effective message discourse varies by crisis, according to this model.
SCCT, introduced in 2002, focuses on the range of crises that an organization may face but does not include a category within which the 2008 financial crisis easily fits (Coombs & Holladay, 2002). The 2008 financial crisis was an industry-wide, interdependent issue, and SCCT does not explicitly address the concept of an industry-wide crisis. The study was designed to help determine whether the SCCT applies to such industry-wide situations.
Fundamentally, there does not seem to be a single crisis communications model that adequately addresses both the types of crises an organization may face (SCCT) and the restoration strategies an organization may wish to employ when facing a crisis (IRT). This study strived to test both models and the applicability of each when planning for, or facing, a crisis.
2008 financial crisis in America (Hamilton, 2009). However, no scholars or critics have yet provided systematic evidence or quantitative analysis of such crisis strategies or media coverage.
This study, then, is a longitudinal assessment of banks’ media relations strategies and newspaper coverage during a six-month period in 2008 when the financial crisis in America grew in scope and impact. As the crisis escalated, one would have expected the nation’s largest banks to express increasing concern for victims of the crisis. The situational crisis communications theory (SCCT) maintains that organizations must – in order to protect reputation – demonstrate concern for victims, particularly as the crisis escalates and the possibility of reputational damage increases (Coombs & Holladay, 2002). Therefore, SCCT provided a relevant theoretical framework from which to analyze the 2008 financial crisis.
In developing SCCT, its authors integrated crisis types from several previous models into a purportedly more cohesive family of crises (Coombs & Holladay, 2002). Their work ultimately resulted in a list of 13 potential crises that an organization might face. These crises are grouped into clusters, including the victim cluster, in which the organization is also a victim of the crisis, such as a natural disaster; the accidental cluster, in which the organizational actions leading to the crisis were unintentional; and the preventable cluster, in which the organization knowingly placed people at risk, took inappropriate actions, or violated a law or regulation (Coombs, 2006).
Aside from natural disaster, though, none of the 13 crisis types in the SCCT model represent industry-wide, interdependent situations. Because SCCT is used to develop recommendations for various crisis response strategies, it is important that the theory holds up when applied to an industry-wide situation such as the 2008 financial crisis, not just situations in which the pressure or spotlight is on a single organization.
It could be argued that some banks took inappropriate lending and underwriting actions, and that, as a result, all banks became drawn into the 2008 financial crisis. In this way, the preventable cluster within the SCCT model could be expanded from actions hurting not just a solitary organization but an entire industry of related organizations.
SCCT includes a family of crisis response strategies that the authors of SCCT contend are more precise than those used in older models, such as image repair theory (IRT), because they have been empirically tested (Coombs, 2006). However, the testing consisted of a single study in which 78 undergraduate students were asked to evaluate 10 different crisis response strategies and indicate their degree of agreement with each strategy. It is difficult to accept that a single study that does not consider real-world outcomes should hold more weight than nearly 20 years of case study analysis provided by researchers using IRT (Benoit, 2000). Therefore, response strategies outlined in IRT will be evaluated in this study instead of the strategies outlined in SCCT.
IRT, which was introduced 11 years before SCCT, established that an organization’s image is a subjective impression formed through one’s direct experience with an organization and/or interpretation of the organization relayed by a third party (Benoit et al., 1991). Crucially for this study, third-party interpretation consisted of media coverage, specifically articles in Wall Street Journal and the banks’ hometown newspapers.
The fundamental argument of IRT, relative to the practice of public relations, is that a crisis situation – or reputational risk – happens only when an offensive act has occurred and the organization is believed to be responsible for that act (Benoit, 1997). The financial crisis that developed in the second half of 2008 had wide-sweeping effect on the financial security and stability of the American public, and the nation’s banks, as a collective body, were held partially responsible for questionable lending, underwriting, and hedging strategies. In this way and 662 according to IRT, each of the nation’s 10 largest banks should have faced reputational risk in 2008.
Employment of IRT begins when an organization acknowledges a perceived crisis in some way. Typically, this is a strategic move made by the organization to announce or pursue corrective action, or to prevent litigation. Response strategies from the IRT model are grouped into five categories: denial, evasion of responsibility, reducing offensiveness of the event, corrective action, and mortification or apologizing (Benoit, 1997). These strategies represent a continuum of responses ranging from simply acknowledging the crisis to taking full accountability.
Nearly 20 years of case study analysis by researchers using IRT has shown that organizations at fault should admit this fact. However, while people want to know whom to blame, it’s often more important to reassure the public that steps have been taken to address the crisis and to prevent another crisis of the same sort in the future (Benoit, 1997; Benoit et al., 1991). Acknowledging the crisis, then, serves as a critical starting point for image repair, according to the IRT model. One of the goals for this study was to determine how these tenets hold true for an industry-wide crisis like the 2008 financial crisis.
Research Questions and Hypotheses This study examined a six-month window in 2008 during which the financial crisis began to unfold. Who to blame was not an easy or simple question to answer; however, the media and consumers pushed for assurance that the crisis would be addressed and would not happen again.
How did the nation’s largest banks respond to these concerns, and how were response strategies reflected in news coverage? These are the broad questions this study sought to answer.
Fundamentally, this study is a test of image restoration and confidence-building by the nation’s largest banks during the 2008 financial crisis. For this study, image restoration focuses on two broad actions – acknowledging the crisis and expressing concern for consumers – banks might have taken in an attempt to bolster confidence during the onset of the financial crisis.
These definitions of image restoration are drawn from the original IRT model, as stated earlier (Benoit, 1997). With regard to news coverage, confidence-building is the sum impression created by mentions in the newspaper; that is, news articles that generally place a bank in a positive light are confidence-building, while those that put the bank in a neutral or negative light are not confidence-building. This approach is consistent with the good news and bad news coding protocol used in previous research evaluating Wall Street news coverage (Lowry, 2008).
The primary research questions addressed through this study were:
• RQ1: How do the image restoration strategies used by banks during a financial crisis relate to the tone of newspaper coverage these banks receive – positive (confidencebuilding), neutral (neither confidence-building nor confidence-eroding), mixed (confidence-building and confidence-eroding) or negative (confidence-eroding)?
• RQ2: How do newspapers generally treat the role of banks in a systemic financial crisis – as victimized by the crisis, as accidentally involved in the crisis, or as perpetuators of preventable actions that might have averted the crisis?
that did not acknowledge the crisis at all. Employing IRT as a framework, we would expect the media to more favorably interpret the reputation of a bank that works to pro-actively repair its image, no matter which repair route it takes, versus a bank that does not even acknowledge that a crisis is happening.
• Hypothesis 2: Banks that expressed concern for consumers regarding the financial crisis earned a higher percentage of confidence-building news coverage than banks that did not express concern for consumers. IRT contends that acknowledging a crisis is one way that an organization might restore its image. SCCT takes this argument a step further and maintains that organizations must – in order to protect reputation – demonstrate concern for victims, particularly as the crisis escalates and the possibility of reputational damage increases.
The independent variables for Hypotheses 1 and 2 were the response strategies exemplified by the banks’ news releases. The dependent variable for each hypothesis was the percentage of news coverage that was confidence-building (i.e. positive, as opposed to neutral, negative or mixed).
Method Content analysis is a commonly used method in mass communication research to test hypotheses derived from theory (Riffe et al., 2005). In this study, two goals were desired. On one hand, the study sought to evaluate tenets of SCCT and IRT, both of which are widely used models in the crisis communications fields. On the other hand, the study was designed to possibly expand the reach of the SCCT model, by categorically assigning definitions of an industry-wide crisis to one of three existing crisis clusters which, with the exception of natural disaster, have previously been applied only to organization-specific crises.
Population The population for this study included news releases and news coverage from the period July 1, 2008, through December 31, 2008. The financial crisis escalated on September 29, 2008, when the Dow Jones Industrial Average suffered its largest single-day point loss in history; thus, the span from July 1 through December 31 provided a benchmark of three months prior to the market drop and three months of activity following the drop.
News releases were limited to those issued by the nation’s 10 largest banks based on total U.S. deposits as of January 1, 2008. Those banks were Citicorp, Bank of America, JP Morgan Chase, Wachovia, Wells Fargo, Washington Mutual, HSBC Holding, U.S. Bancorp, Bank of New York, and Sun Trust. All releases for the period July 1, 2008, through December 31, 2008, were drawn from the banks’ websites. The population contained 592 releases.
News clips were limited to newspapers published in the headquarters city of each major bank, as well as the Wall Street Journal. The headquarters city newspapers and the corresponding banks included the New York Times (Citigroup, JP Morgan Chase, Bank of New York), Charlotte Observer (Bank of America, Wachovia), San Francisco Chronicle (Wells Fargo), Seattle Times (Washington Mutual), Minneapolis Star-Tribune (U S Bancorp), and Atlanta Journal-Constitution (SunTrust). (Note: HSBC Holding is headquartered in London, England, and, therefore, had no hometown American newspaper to include in the study.) News articles from these newspapers mentioning the banks were drawn using the America’s Newspapers database and the Wall Street Journal proprietary database, with the legal names of 664 each bank as the keyword search terms used in separate searches for each bank. The population included more than 6,000 articles.
Unit of Analysis For news releases, the unit of analysis was the full text of the release, including the headline and the body of the release. For newspapers, the unit of analysis was the headline and the full text of the article.
Sampling Method For news releases, no sampling was done. All 592 releases in the population were coded.
For newspapers, the large number of articles within the population required the use of stratified sampling.
An efficient stratified sampling method for a year of daily newspapers is two constructed weeks from that year (randomly selecting two Mondays, two Tuesdays, etc.) (Riffe et al, 2005).
Given that this study looked at only six months of coverage, a single constructed week (randomly selecting one Monday, one Tuesday, etc.) for each newspaper would have sufficed.