Whistleblower laws and exposed corruption in the United States

Abstract

This research creates a unique internet-based measure of awareness about state-level whistleblower laws and provisions to examine their effects on observed corruption in the United States. Are whistleblower laws complementary or substitutes for other, more direct, corruption control measures? Placing the analysis within the corruption literature, the findings show that greater whistleblower awareness results in more observed corruption and this finding holds across specifications. Internet awareness about whistleblower laws seems relatively more effective at exposing corruption than the quantity and quality of state whistleblower laws themselves.

Keywords:

• corruption
• whistleblower
• law enforcement
• internet
• United States

JEL Classification:

• K4
• H7

Acknowledgements

We are grateful to Mike Naretta for assistance with the initial version of this article, to a referee and to seminar participants at the University of Naples ‘Parthenope’, TOBB University of Economics and Technology, Western Illinois University and the Bank of Finland. An earlier version was circulated as a BOFIT Discussion Paper #9/2013.

Notes

¹ See Phillips & Cohen LLP, http://www.all-about-qui-tam.org/fca_history.shtml#leg . The legislation was amended in 1943 and again in 1986 when it was strengthened to improve the financial incentives to report wrongdoing and to make it easier to prosecute such activity.

² See Congressional Research Service (2007): http://www.fas.org/sgp/crs/natsec/RL33918.pdf

³ For an overview of Federal Whistleblower Laws, see the US Department of Labor website: http://www.dol.gov/dol/compliance/comp-whistleblower.htm

⁴ See National Conference of State Legislatures, http://www.ncsl.org/issues-research/labor/state-whistleblower-laws.aspx

⁵ Whistle-blowing is also a useful instrument of corporate governance. Dyck et al. (2008) have noted that in the corporate world, multiple constituencies are interested in learning about whistleblowing provisions. 

⁶ http://www2.illinois.gov/gov/whistleblower/Pages/default.aspx

⁷ However, effective whistleblower laws must also contain provisions to prevent overzealous and false exposures.

⁸ Goel et al. (2012) have employed a similar approach at capturing internet corruption awareness across countries, while Goel (2014, forthcoming) measures the availability of anti-smoking internet resources across US states.

⁹ Focusing on the behaviour of business firms, Søreide (2008) notes that market structure, efficiency and institutional considerations might affect business firms’ reluctance to expose corruption. Further, Friebel and Guriev (2012) argue that in the corporate world, higher management offer bribes to potential whistleblowers to keep them quiet.

¹⁰ Additionally, for states bordering large media markets, it is not clear which key newspaper would make more sense – e.g., in the case of Virginia, does one use the Washington Post or the Richmond Times-Dispatch?

¹¹ At the time of this study, 2009 data on corruption convictions were the most recent available.

¹² Still, as mentioned above, the available data series does not include corruption prosecutions at the state level and does not make fine distinctions between petty and grand corruption.

¹³ There is some recognition in the broader literature about the role of the internet in deterring various crimes (Katz and Rice, 2002; Andersen, 2009). For studies linking the media and corruption, see Brunetti and Weder (2003), Di Tella and Franceschelli (2011), Freille et al. (2007), Goel et al. (2012), Reinikka and Svensson (2005) and Vaidya (2005).

¹⁴ In preliminary analysis, we also included various measures of the gross state product originating from the federal government in a state (i.e., from the defence and civilian sectors). These variables were generally not statistically significant and were dropped from the final models presented below. 

¹⁵ Corruption prosecutions from state district attorneys or attorneys general are not available and hence are excluded from the analysis. Such prosecutions are estimated to be around 20% of the total (Corporate Crime Reporter, 2004).

¹⁶ Note that given the nature of the internet searches, the data for internet hits is more current than the secondary data that are the latest available, albeit with somewhat of a lag. Thus, there is some mismatch in the time periods of some of the variables. However, many of the web postings that the searches generate have been available on the internet for quite a while. Further, the various robustness checks reported confirm the overall validity of our main findings (see Section ‘Robustness checks’).

¹⁷ We also collected data from Microsoft’s Bing search engine, but the data were identical to the data collected from Yahoo. This is logical since Microsoft and Yahoo recently began an alliance where Yahoo search is powered by Bing (http://yahoobingnetwork.com/en-apac/home).

¹⁸ Also, it may return results of people with the name Virginia.

¹⁹ We used the same technique to remove references to West Virginia when searching for Virginia and vice versa.

²⁰ As the reader will notice, technical limitations constrain the extent of refinements we can perform. For instance, there is likely to be some duplication in webpages in some cases due to cross postings at mirror sites and there seems no efficient way to weed them out. Also, postings in nonEnglish are not covered. However, this is unlikely to be a big issue for states across the United States.

²¹ This was the result of an internet search conducted after the main data were collected. 

²² To gain some additional insights, we also incorporated some components of the index upon which the Whistleblower Law Strength variable was constructed in preliminary analysis (see for details, http://www.peer.org/assets/docs/wbp2/overview.pdf). Not surprising, given the results presented in Table 3, many of these disaggregated measures had insignificant impacts on exposed corruption. However, states that required an investigation by state auditor or other investigative entities of whistleblower disclosures (5 states in all) were found to have a deterring effect on corruption. Details are available upon request. 

²³ Using similar measures of corruption, Alt and Lassen (2012, forthcoming) have recently shown that greater prosecutorial resources result in more corruption convictions in the United States.

²⁴ Given its nature, corrupt activity can have a cause and effect relationship with many variables (see Lambsdorff, 2006). To mitigate reverse feedbacks from other determinants in equation (1), we include the beginning of 2000–2009 period values for Income, URBAN, GSPstate and LawEnforce (see Table 1).

²⁵ Additional data for both instruments were obtained from the Statistical Abstract of the US.

²⁶ We also experimented with using the number of internet users (normalized by population) in a state as an instrument (in combination with the number of public library computers and land area). The relevant tests showed that the set of instruments reported in the paper were a better fit. Details are available upon request.

²⁷ The fact that WhistleblowerBroad and WhistleblowerNews searches were conducted about six months after the Whistleblower internet searches also alleviates concerns about one incident at a point in time disproportionately increasing internet activity. One should also bear in mind the possibility that some web posting are taken down over time.

²⁸ Complete details are available upon request.

²⁹ WhistleblowerNews (again normalized by population) had a pairwise correlation of 0.67 with Whistleblower and of 0.77 with WhistleblowerBroad.

³⁰ Again, complete details are available upon request.