In discussions of appropriate sanctions for corporations that engage in bribery, much of the conversation focuses on the appropriate penalty reduction for firms that self-disclose violations, cooperate with authorities, or both. Self-disclosure and cooperation are often lumped together, but they’re not the same: Plenty of targets of bribery investigations, for example, did not voluntarily disclose the potential violation, but cooperated with the authorities once the investigation was underway.

This gives rise to a problem that is both serious and seemingly obvious, but that somewhat surprisingly is hardly ever discussed.

The problem goes like this: Enforcement authorities want to encourage self-disclosure, and they want to encourage full cooperation with the investigation; they would like to do so (1) by reducing the sanction for firms that voluntarily disclose relative to those that don’t, and (2) by reducing the sanction for firms that fully cooperate relative to those that don’t. But if the minimum and maximum penalties are fixed (say, by statute or department policy or other considerations), and the penalty reductions necessary to induce self-disclosure and full cooperation, respectively, are large enough (cumulatively greater than the difference between the maximum and minimum feasible sanction), then adjusting sanctions to encourage self-disclosure may discourage full cooperation, and vice versa.

It’s easiest to see this with a very simple numerical example:

Suppose, given the facts of the case, the background law, and any other relevant external constraints, the maximum penalty that an enforcement agency could impose on a firm—which it will impose if the firm neither self-disclosed nor fully cooperated—is $100 million. Suppose next that the minimum penalty that the enforcement agency would impose is $20 million, and that this is the penalty the agency would impose on a firm that both self-disclosed and fully cooperated. The agency must decide what penalty it will seek to impose on a firm that did not self-disclose, but did fully cooperate with the investigation after the agency learned of the potential violation through other channels. (To keep things simple, I’ll put to one side those cases in which an agency self-disclosed but subsequently declined to cooperate to the agency’s satisfaction. That possibility would add an additional complication to the discussion but wouldn’t alter the main point.) Let’s call that penalty–the one the agency imposes on the non-disclosing but fully cooperative firm–FC.

The agency can set FC anywhere between $20 million and $100 million. The difference between $100 million and FC is the amount the agency offers to a firm to fully cooperate with an investigation, while the difference between FC and $20 million is the additional amount that the agency offers a firm to self-disclose a violation. Lowering FC strengthens the former incentive but weakens the latter, while raising FC has the opposite effect.

The problem should now be clear: There’s a direct trade-off, because the difference between the maximum and minimum feasible penalties creates an $80 million “budget” (in this example) that the agency can allocate as between encouraging two different kinds of helpful conduct: self-disclosure and cooperation. Furthermore, it might well be the case that in order to induce the optimal level of both self-disclosure and cooperation, the agency would need to offer incentives that exceed this “budget.” The agency can maximize self-disclosure by refusing to offer any discount to a firm that failed to self-disclose, whether or not it subsequently cooperates – but then non-self-disclosing firms have very little incentive to cooperate, and the agency might not be able to secure a conviction or settlement without expending massive resources. Conversely, the agency can maximize cooperation with open investigations if it offers to reduce the penalty of a fully cooperating firm from $100 million to $20 million – but  then the agency has no way to induce self-reporting, because a firm knows it can get the maximum possible penalty reduction by cooperating if it ends up getting caught.

This might all seem very abstract, but it has important implications for debates over whether enforcement agencies (such as the US Department of Justice (DOJ) and the UK Serious Fraud Office (SFO)) are too generous with firms that do not self-disclose but do subsequently cooperate with investigations. (Over the past couple of months, I’ve heard versions of that argument advanced both as part of a critique of the DOJ’s “FCPA Pilot Program” and of the SFO’s handling of the Rolls-Royce case.) The argument runs that because the difference between the penalty reduction for “self-disclose and cooperate” and for “cooperate (without self-disclosure)” is too small, corporations won’t see sufficient benefit in self-disclosure, meaning that too many corporations that discover possible violations will roll the dice, knowing that they probably won’t be caught and that if they are, they can subsequently cooperate and get a substantial chunk of the penalty reduction. That concern is correct, as far as it goes, but it overlooks the fact that if the enforcement agencies adjusted their policies by substantially reducing the benefit non-self-disclosing corporations can get from cooperation, then although more corporations may self-report, those that find themselves under investigation may be much less likely to cooperate fully with the investigation.

I have no idea whether the penalty reduction policies (insofar as we can tell what they are) of the DOJ, SFO, or other enforcement agencies strike the balance in the right way. My only point is here is that this trade-off should be recognized more explicitly. Unless the agency can induce full cooperation with a sufficiently small penalty reduction, then there may be no way to use penalty reductions to induce self-disclosure without simultaneously weakening incentives for full cooperation in other cases.