Before Brazil’s so-called Lava Jato (“Car Wash”) Operation, almost every attempt to prosecute high-level corruption in Brazil failed. Many cases were never investigated or prosecuted, but even in those cases where prosecutors started investigations, identified crimes, and brought charges, appeals courts ended up nullifying the proceedings, often before trial, on technical grounds for failure to comply with procedural rules (see, for example, here, here, here, and here). The result was a culture of impunity, in which grand corruption thrived. The Lava Jato Operation has been hailed as a historic breakthrough not only because of the breadth of the corruption it uncovered, but also because the convictions secured by prosecutors had, by and large, been affirmed on appeal. Unfortunately, there are troubling signs that the Brazilian judiciary is reverting to its old ways. Last October, for example, the Brazilian Supreme Court issued a procedural ruling  concerning the sequence of closing arguments that the Court held required the nullification of two Lava Jato convictions (so far), and may end up doing more widespread damage. The larger issue here, though, is the double-standard that Brazilian appellate courts seem to have embraced: adopting an (excessively) stringent and unforgiving view of even minor technical procedural noncompliance in corruption cases involving elite defendants, while at the same time relying (properly) on “harmless error” doctrines to excuse similar sorts of procedural noncompliance in cases involving other sorts of crimes, such as drug trafficking.

Last October’s ruling illustrates this double standard in action. That case involved the order in which the parties present their (written) closing arguments at the end of a criminal trial. Brazilian criminal procedure law provides that in complex cases, after the production of evidence before the judge, the prosecution must present written closing arguments in five days, and the defendants must present their closing arguments five days after that. The rules as written do not distinguish between different types of defendants—in particular, although some of the defendants in a particular case may have chosen to cooperate with the prosecution and give evidence against their co-defendants, the criminal procedure laws don’t say anything about different timing of closing arguments for cooperating and non-cooperating defendants. In the Lava Jato cases—a very large number of which involve cooperating defendants—the trial courts had been following the sequence laid out in the statute: Five days after the close of the evidence-production phase, the prosecution would submit its written closing statements, and five days after that, all the defendants (both cooperators and non-cooperators) would submit their closing statements. The Brazilian Supreme Court held that this sequence deprived the non-cooperating defendants of their constitutional rights. The Court reasoned that a cooperating defendant becomes an assistant of the prosecution, and therefore the Constitution requires that the cooperating defendants make their closing arguments before the non-cooperating defendants, and that any trial that failed to follow this sequence was subject to retroactive annulment and remand to redo the closing argument stage. The delay in punishing corruption, especially in the slow moving and burdensome Brazilian criminal procedure system, is a problem by itself. But the remand may cause additional damage, leading to revocation of pre-trial detentions and freezing of assets, which are subject to time limits, as well increasing the probability that the statute of limitations will run out before the new closing arguments are completed.

Some might find the Brazilian Supreme Court’s ruling defensible, though that is by no means obvious (especially since in many other democratic countries, such as the United States, the prosecution speaks last). But it’s hard to see how noncompliance with this rule could justify nullifying a conviction in an ordinary case, and even harder to justify invoking this newly-announced rule to nullify convictions in trials that have already been concluded before the rule was announced. It’s difficult to imagine how changing the order in which written closing statements are submitted to the judge could make any difference to the outcome in a typical case, or that it would threaten notions of fundamental fairness. And indeed Brazil, like other countries, has a concept of “harmless error”: Article 563 of the Brazilian Criminal Procedure Code expressly states that procedural errors do not justify nullifying a conviction if the error is harmless, and this harmless error doctrine is regularly invoked in other kinds of cases. But here is precisely where we see a troubling double standard: Brazilian appeals courts (including especially the Supreme Court) invoke the harmless error rule much more rarely in corruption cases involving elite defendants than in ordinary crime cases.

Consider, for example, how the Brazilian courts have handled a seemingly similar controversy about the order in which the parties to a criminal trial make their presentations, but that has arisen in the context of drug crimes rather than corruption. As a general matter, the Brazilian Criminal Procedure Code states that defendants have a right to testify at the end of the evidence-collection phase of the trial, after the other witnesses have testified. However, a more specific provision of the statute on drug trafficking states that defendants accused of drug trafficking should testify at the beginning of the criminal proceeding, and trial judges have relied on this provision to take the testimony of drug trafficking defendants as the first step in the criminal proceeding. Defendants have argued that this violates their constitutional rights, and that they should have the opportunity to testify after the production of evidence against them. Since 2008, the Brazilian Supreme Court had been rejecting this constitutional challenge, holding that a defendant’s constitutional rights are not seriously affected by whether the defendant testifies first or last. Eight years later, though, in a different drug case, the Court held that the defendants should have testified at the end of evidence production. The Court did not declare unconstitutional the rule providing for taking defendants’ testimony first; the court only stated that the general rules placing the defendant’s testimony at the end of the proceeding are more compatible with the Constitution. Yet despite setting a new precedent in favor of the defendants, the Court did not nullify the conviction in the case before it; the Court held that the obligation to take the defendant’s testimony last should apply only prospectively. Even after that change the Brazilian Supreme Court has declined to invalidate proceedings in which drug traffickers’ testimony is taken first, stating that the inversion of the correct sequence is harmless error. So, convictions in drug trafficking cases are rarely nullified by Brazilian appeals courts.

 The contrast between the Brazilian Supreme Court’s forgiving approach to the sequencing of trial testimony in drug cases, and the Court’s stringent approach to the sequencing of closing arguments, is striking and hard to justify. This contrast may help to explain why, according to the last available official data (updated in June 2019), Brazilian prisons are full of drug traffickers (304,416 out of 772,254 inmates), but only a handful of corruption perpetrators (1,753 inmates total). Brazil’s criminal system is very harsh with respect to some categories of crimes and defendants, but with respect to others, it is too much lenient. When it comes to high-level corruption, investigations are conducted, evidence is gathered, people involved in corruption schemes are charged and sometimes convicted, yet all too often in Brazil, in the end everything is nullified by appeals courts based on seemingly minor procedural technicalities.

Brazilian judicial authorities need to change their approach to possible non-compliance with procedural rules in anticorruption cases, bringing the approach into line with how courts approach the same problem in the context of other crimes, such as drug trafficking. In particular, appeals courts should show greater deference to first instance judges. The decisions of first-instance courts should be entitled to a presumption of regularity and should not be reversed unless there was an abuse of discretion or clear error. With respect to procedural rules in particular, noncompliance should be grounds for annulment only when the abuse of discretion or clear error caused substantial harm to parties’ fundamental rights or actually affected the judgment. Furthermore, when an appeals court does create or announce a new procedural rule (or a new interpretation of what existing law requires), this new ruling should, at least in most cases, apply only prospectively, and should not be grounds to retroactively nullify convictions in cases that have already been tried. These changes would not only ensure equal treatment of all defendants regardless of the crimes with which they are charged, but would also ensure the effective enforcement of Brazil’s anticorruption laws.