by Jonathan Karpoff, Scott Lee, and Gerald Martin
Managers caught bribing foreign officials enhance their shareholders’ wealth unless they simultaneously engage in accounting fraud. Most firms apprehended for foreign bribery generally did not engage in fraud and after adjusting for regulatory penalties, their average ex ante net present value of their bribery-induced projects is positive. In a new paper, we develop a model that uses a comprehensive set of bribery enforcement to shed light on the how common foreign bribery is among US-listed firms with foreign sales.
We use a comprehensive database of all 143 enforcement actions for foreign bribery violations by publicly traded companies initiated by the U.S. Department of Justice (DOJ) and Securities and Exchange Commission (SEC) from 1978 through May 2013 to provide the first evidentiary answers to numerous questions about foreign bribery. Our empirical measures are guided by a five-equation framework that characterizes the relationships between the ex ante and ex post net present value of projects that involve bribery, the size of bribe payments, the costs to firms that face bribery-related enforcement action, and market reactions to revelations of bribery and news about alleged bribe-tainted projects. The framework has 12 parameters. We measure five of these parameters directly, estimate two others, and solve for the five remaining parameters using the five-equation framework.
Our data and empirical findings make several contributions regarding the prevalence of bribery, the characteristics of bribe-paying firms, the value of bribe-related projects, and the probability and costs of getting caught paying bribes. First, we provide descriptive information about the frequency of bribery enforcement actions, their distributions across firm size and industry, and the countries in which bribes are paid.
Second, we describe the characteristics of firms alleged to have paid bribes. The probability of paying bribes in any given firm-year is negatively related to the firm’s current operating performance and positively related to the firm’s size, leverage, and R&D intensity; the level of corruption within the firm’s industry (per Transparency International); the number of geographic segments in which the firm operates and the level of corruption in those markets (per the World Bank’s World Governance Institute Control of Corruption index); and an aggressive corporate culture (per Miles and Snow, 2003).
Third, firms’ share prices increase significantly (3.30% average) upon news of a project that subsequently is tied to bribe payments, and share prices decrease significantly (‑5.44% average) upon news that a firm engaged in bribery. Hence, capital markets integrate both the benefits of bribe-related projects and the costs of enforcement.
Fourth, firms caught bribing suffer significant direct penalties (i.e., legal costs, fines, penalties, and monitoring costs) that average 3.27% of firm value. Bribe-paying firms that avoid comingled fraud charges, however, suffer no significant loss in reputation. That is, the revelation that a firm engaged only in bribery does not appear to harm its contracting relationships with suppliers, investors, or customers. This implies that virtually all of the cost imposed on a bribe-paying firm that is caught is imposed via legal sanctions and not via market transactions. In this regard, the revelation of bribery is similar to the revelation that a firm violated an environmental regulation, and is not similar to revelations that a firm engaged in consumer fraud or financial misrepresentation.
Fifth, we develop a bribery detection model that uses signal detection theory to guide the tradeoff of Type I and Type II errors in classifying firms as engaged or unengaged in bribery. Our baseline estimate is that 22.9% of all Compustat-listed firms with foreign sales engaged in illicit bribery programs during at least one five-year period sometime between 1978 and 2010. We venture no opinion on whether this estimate seems high or low, but we do test the sensitivity of this estimate across time and across varying estimation assumptions. We find that estimates of the other central parameters of our five-equation framework—including the ex ante and ex post net-present value (NPV) of bribery-related projects and the size of the reputation loss for apprehended firms—are not highly sensitive to even large deviations from the baseline estimate.
Sixth, our estimates yield inferences about managers’ motives to engage in bribery. We find that, on average, the ex ante NPV of a bribe-related project is strongly positive even considering the cost of the bribe and the expected penalties if the firm is caught. This result implies that managers bribe in pursuit of profitable projects rather than for self-serving purposes.
Seventh, we measure the division of gains from bribe-related business and find that bribe recipients capture 17.8% of the value of the benefit for which bribes are paid, on average. This evidence is inconsistent with arguments that bribe recipients extract most of the surplus from bribe-related contracts, and suggests that the average ex ante NPV of bribe-related project is positive, with bribe paying firms transferring only a small portion of the project value to bribe recipients.
Eighth, we conclude that penalties are too small to deter bribery at the margin. Our baseline estimate that the probability of getting caught in any given year is 6.4%, implying that total penalties imposed on bribe payers would have to increase by 8.3 times to drive the average ex ante NPV to zero. On the other hand, if the penalties remain at historical levels, the probability of detection would have to rise to 52.8% to achieve the same objective. At current levels, bribe-tainted projects will continue to be profitable, at least on an ex ante basis.
Finally, we discover an important regularity in the size and nature of the costs imposed on firms that are caught bribing. A minority of bribery enforcement actions is comingled with charges of financial fraud. Firms without comingled fraud charges experience losses in share values that are only 1/13 of the losses suffered by bribe-paying firms that face fraud charges (the average loss is 2.6% compared to 33.1% for firms with fraud charges). The largest source of this difference is that firms with fraud charges experience large reputational losses—averaging 18.8% of share value—compared to negligible reputational losses for bribe-paying firms without fraud charges. Reputational losses measure the long-term impact on the firm’s operations and contracting with investors, suppliers, or customers. So, these results indicate that the revelation of bribery, by itself, has little long-term impact, as long as the bribery is not comingled with charges of financial fraud. Because non-fraud firms face relatively small costs even when they are caught bribing, the average ex post NPV of the bribe-related activity is non-negative for the majority of firms without comingled fraud charges (+0.4% of market capitalization). Thus, even for firms that are caught, the net benefits of their bribe-related projects are non-negative—as long as the firm avoids comingled fraud charges.
Further tests indicate that our results and inferences are robust to alternative assumptions used to derive our estimates of the probability of bribing and the probability of apprehension.
Foreign Bribery: Incentives and Enforcement
Karpoff, Jonathan M. and Lee, D. Scott and Martin, Gerald S., Foreign Bribery: Incentives and Enforcement (April 7, 2017). Available at SSRN.
Jonathan Karpoff is a Professor of Finance and the Washington Mutual Endowed Chair in Innovation at the Foster School of Business at the University of Washington. Scott Lee is the Lincy Professor of Finance at the Lee School of Business, University of Nevada. Gerald S. Martin is an Associate Professor at the Kogood School of Business, American University.