At last Month’s Chatham House conference on Combating Global Corruption, much of the discussion focused on how to create incentives for corporations to uncover and voluntarily disclose violations of foreign anti-bribery laws like the U.S. Foreign Corrupt Practices Act (FCPA). This is important, because as I noted in last week’s post, most FCPA violations are revealed because of self-disclosures, rather than government or media investigation. During the conversation, a distinguished lawyer (whom I cannot identify by name, because of the Chatham House Rule) made the following argument: Although the U.S. Department of Justice claims to give corporations credit for self-disclosure of FCPA violations, “a careful examination of the evidence reveals” that self-disclosure does not result (on average) in any reduction in penalties.
This is not the first time I’ve heard the claim that quantitative studies have shown no beneficial effect of self-disclosure. But that claim is false. Self-disclosure of FCPA violations may, or may not, result in significant penalty reductions; the quantitative evidence (at least that available in public sources) is inconclusive–it does not “reveal” anything one way or the other.To the best of my knowledge, the oft-repeated “no benefit from self-disclosure” claim derives from two sources. (I’m only considering sources that attempt quantitative data analysis; lots of other sources look at a handful of cases, mostly concluding that there may be a benefit from self-disclosure but it’s hard to tell how big it is.)
The first source is a student thesis by Bruce Hinchey, subsequently published in the Public Contract Law Journal in 2011, which examined FCPA actions between 2002 and 2009 and compared the ratio of penalties paid to bribe size for both self-disclosing and non-self-disclosing firms. He found that the ratio was not any smaller for the self-disclosers—indeed, it was somewhat larger. Critics of DOJ’s FCPA enforcement program took this finding and ran with it, but thinking carefully about this for 30 seconds should be enough to realize that this finding tells us essentially nothing about whether there’s a real benefit from self-disclosure.
The reason is that the pool of companies that choose self-disclosure may well be systematically different from the pool of companies that choose not to disclose voluntarily—and the differences may not just be the size of the bribes. Other factors that might influence both the self-disclosure decision and the ultimate sanctions include the strength of the evidence, the egregiousness of the violations, the company’s own history of violations, the size of the firm, etc. Hinchey’s study doesn’t even attempt to control for any of this, let alone do so convincingly. To conclude, from this study, that there’s no benefit to self-disclosing FCPA violations would be akin to concluding from the fact that people who go to the hospital have a higher mortality rate than people who don’t that going to the hospital increases your risk of death.
The other source, an unpublished working paper by the NYU law professors Stephen Choi and Kevin Davis, suffers from similar problems. The impact of self-disclosure on penalty size is not Choi & Davis’s main concern, but in one of their many regressions they include self-disclosure and find that (when controlling for the firm’s market capitalization and the size of the bribe) voluntary disclosure does have a mild negative correlation with penalty size, but this correlation is not statistically significant. To ascribe significance to this finding one would have to believe that, controlling for firm size and bribe size, the characteristics of self-reported violations and the characteristics of non-self-reported violations are, on average, indistinguishable. But again, we have lots of reasons to suspect that might not be true—firms might be more likely to voluntarily disclose violations that, for example, involve particularly willful or high-level or widespread conduct (independent of the monetary size of the bribes involved), or where the firm has been in FCPA trouble before, or where the evidence of wrongdoing is particularly strong.
None of this is to say that self-disclosure in fact does lead to significant credit. The DOJ claims this is the case, and there are plenty of anecdotes to support that position. There are also plenty of anecdotes involving firms that voluntarily disclosed and got hit with huge penalties. The average benefit of voluntary disclosure is still very much an open question. But it’s sloppy and irresponsible to make sweeping claims to the effect that “careful examination of the evidence reveals” that self-disclosure has no benefit, when a careful examination of the evidence reveals no such thing.
This is quite thought-provoking. Two quick questions/reactions:
(1) It seems like the selection-bias issues make this a really difficult question to address empirically. Have there been any micro-level/case study-style analyses of the interactions between firms and the DOJ that shed light on the utility of disclosure?
(2) It seems to me that even if the DOJ is reducing penalties on account of self-disclosure, that’s no guarantee that many companies will have the right incentives to disclose. First, the reduction in penalties anticipated by companies would have to be large enough to offset the expected value to companies of not disclosing and possibly not receiving a penalty at all. Second, even if disclosing has a positive expected value, there might be other factors counseling against disclosure. For example, a company’s stock price might go down after a disclosure and it might open itself up to shareholder litigation (and it’s not clear that these consequences would be significantly less severe after a self-disclosure than after a non-disclosed violation comes to light).
On your first question, to the best of my knowledge the answer is no. In addition to the two attempts at quantitative analysis discussed in my post, there are a few papers (or law firm reports) that look at a handful of case studies, but they look at the publicly available information about what happened — the details of the interactions between DOJ/SEC and the self-disclosing companies is almost always confidential.
On your second question, you’re completely right. Self-disclosure involves a complex cost-benefit analysis by the firm (and its lawyers). Companies are not legally required to self-disclose FCPA violations, though DOJ/SEC encourage disclosure (at least of serious matters); when deciding whether to disclose, firms will take into account all the factors you mention, and more. The reason I wanted to counter the “no benefit from disclosure” claim is that, to the extent it’s believed, it might shift some decisions at the margin, possibly without a good reason — which could be bad both for the government and for the firm.
I think Sam does a wonderful job teasing apart the cost-benefit analysis facing companies. I also agree that overly conclusions should never be drawn from statistical analysis. However, I wonder whether practically “[t]he average benefit of voluntary disclosure is still very much an open question.” If a company calculates the expected outcome of self-disclosure, the examples where companies have self-disclosed only to get big fines would weigh against disclosing.
Uncertainty created by the lack of transparency weighs down any incentive effect DOJ may have intended by publicly claiming self-disclosure would lead to reduced penalties. To this end, it might be helpful for DOJ to publish the counterfactuals— i.e. what the companies would have paid if they hadn’t self-disclosed— to silence the critics once and for all.
This is an important post! A reaction that picks up on Sam’s point: has there been a study that examines differences between companies that self-disclose and others? Such a study might find that self-disclosing companies often count the federal government as a large client, or rely on government approval to begin a project at the time of disclosure
Not to my knowledge – that’s an interesting suggestion, and it would be great if someone looked in more detail at whether self-disclosing companies differ in some systematic ways from companies that do not self-disclose but nonetheless end up getting hit with FCPA penalties.
That still wouldn’t solve the main data problems, though:
* We don’t observe those firms that don’t self-disclose and never end up the subject of an investigation.
* We don’t observe those firms that do self-disclose but the government decides to take no action (with the exception of one case – Morgan Stanley in China – declinations are not public.
* We don’t observe many of the ways that the individual cases involving self-disclosing and non-self-disclosing firms may differ, such as the strength of the evidence, the pervasiveness of the corruption, the degree of cooperation with the government, etc.
All these factors make it very, very hard to draw any firm conclusions from comparing penalty magnitudes in the cases that do result in a public settlement, even if we try to control for a range of factors, such as the ones you mention. That’s not to say that your suggestion isn’t interesting and worth pursuing — it’s just to say that I don’t think it would get us much further in answering the original question.
This is perhaps a more general comment about using statistical methods to analyse FCPA enforcements. While I am proponent of using data, this post highlights an over-arching danger of generalising conclusions from a limited sample. Most of us are not well versed in the nuances of statistical methods and hard numbers gives the perception of impartiality. I agree with Matthew that a lot of the problems with empirical studies in this area cannot be easily solved without more access to DOJ’s FCPA data. To this end I think it is in DOJ’s interest to make more disclosures, at least for things they want to incentivise companies to do.
I think it’s really interesting how the quantitative data is inconclusive. The issues raised in your post makes me wonder a few things:
1. If the costs of self-disclosure differ between violations enforced by the DOJ and violations enforced by the SEC
2. I understand that many companies end up settling with the DOJ to avoid a FCPA prosecution. I would assume that if a company self discloses, it would end up settling anyways. From a pragmatic perspective, the company ends up saving money by settling in terms of legal resources. Is there any information indicating if companies pay higher fines if they are prosecuted and found guilty than companies that end up settling? If there is a difference, then that could suggest that self disclosures could be beneficial.