Dealing with Ethical Dilemmas: A Look at Financial Reporting by Firms Facing Product Harm Crises

Coca-Cola’s 1999 PHC in Europe provides a vivid illustration of the potential consequences of a PHC (Mitroff and Silvers 2010, pp. 7–9). Following production problems at bottling plants in France and Belgium, more than 200 consumers ended up ill after drinking the firm’s products and McDonald’s even stopped serving Coke in its restaurants. A major product recall ensued as the governments of several countries banned the firm’s products. Beyond its health effects on consumers and the economic costs to the firm, the PHC ultimately cost Coca-Cola’s CEO, Ken Ivester, his job less than 2 years after his appointment. A 20-year Coca-Cola veteran with an impeccable record, Mr. Ivester was perceived to have mishandled the crisis, thus exposing the firm to additional economic and reputational costs. The Coca-Cola PHC also illustrates that the range of stakeholders affected by a PHC can be wide, encompassing consumers, management, shareholders, governments, regulators, suppliers, and employees (if plants are shut down or production is stopped).

The implicit claims normally have no legal standing, and hence they can be breached by either party. However, Bull (1987) argues that there are forces that prevent firms from breaching the implicit claims. Rather, firms have incentives to build their reputation as having the ability to fulfill the implicit claims because such reputation ultimately determines the trade terms between firms and their stakeholders (Cornell and Shapiro 1987; Titman 1984).

Since the advent of the Sarbanes-Oxley Act of 2002 and following the global financial crisis (2007–2009), stakeholders from various arenas have become increasingly critical of the ethics underlying certain financial reporting practices that were previously deemed as “business as usual” (e.g., Cryan and Theriault 2012; Sherman and Young 2016).

Bowen et al. (1995) illustrate that customers care about firms’ accounting numbers including earnings. One example in their paper shows that LeCie advertised that customers can trust the company because its hard drives are backed by $400 million in assets and thus the company is a reliable source of high-quality machines and components. Moreover, media usually pays much attention to and publicizes firms’ earnings performance after PHCs. For example, Samsung hit the news headlines because of the Note 7 recall, but subsequent news reports specifically mentioned its soaring fourth quarter earnings (Yahoo 2017).

Available at: http://​faculty.​washington.​edu/​pdemerj/​data.​html, accessed March 15, 2019.

Managers can anticipate a product recall several months before its public announcement (Chen et al. 2009; Gao et al. 2015; Gokalp et al. 2016). Thus, even if a PHC occurs towards year-end, managers may engage in earnings management during the year before its announcement. For this reason, we focus on annual estimates rather than more granular quarterly estimates of earnings management. Nevertheless, we hand collected data on recall dates for a subsample of crisis firms in our main analysis. Analyzing 64 firms with fourth quarter product recalls, we find that their quarterly discretionary accruals exhibit an increasing trend up to the third quarter and a slight decrease in the fourth quarter, which is consistent with managers manipulating accruals even before they announce the product recalls in the fourth quarter.

As robustness checks, we use discretionary accruals derived from modified Jones (1991) model and performance-adjusted modified Jones (1991) model as described in Kothari et al. (2015) and find consistent results.

We use the two-stage regression with an IV to alleviate the concern that unobservable factors drive both a firm’s likelihood of experiencing a PHC and its earnings management behavior (Tucker 2010; Lennox et al. 2012). In Sect. 5.1, we conduct additional robustness tests using propensity score matching and entropy balancing to further mitigate the endogeneity concern arising from observable differences in firm characteristics between crisis and non-crisis firms.

We also check the VIF of other control variables and none of them have VIF larger than three.

Our results are not sensitive to the cut-off of difference in propensity scores. Our results stay the same if the cut-off is set to be 0.1 or 0.01.