Brazil and the Bloodsuckers

The initial clues appeared in Rondonia, a wild-west state sandwiched between Bolivia to the south and Amazonas to the north. More often than not, the ambulances cost more than they were supposed to. In the municipality of Ouro Preto do Oesto, funds intended for a much larger vehicle purchased a modest Renault Trafic, which, as a van more suited to plumbers than emergency responders, didn’t have space for the necessary equipment. In Sao Miguel do Guapore, the local government spent an extra $4,000 on its ambulance.

Corruption of this nature is especially hard to track because it is shallow as well as broad. Wrongdoers skim off the top at a merely marginal rate, such that the level of greed levied on any single transaction is never quite enough to cause alarm. This helps a scheme become sticky. As a rule of thumb, where corruption is widespread, it tends to remain so, since many of those involved share incentives to turn on whistleblowers. (Conversely, where corruption is rare it struggles to get off the ground, because anyone tempted to commit a fraud or demand a bribe knows that, if discovered, there’s little to hold over unveilers’ heads.)

Even so, those first few costly ambulances in Rondonia soon began cropping up in the reports of auditors from Brazil’s national anti-corruption agency, the CGU (Controladoria-Geral da Uniao, or Comptroller General of Brazil), in 2003 and 2004. That was when groups of the agency’s staff — think of them as “audit SWAT teams” — first began fanning out across the country to inspect municipal finances. They’re still at it today, and they’ve proven a strikingly effective tool for picking up on the kind of hard-to-detect corruption exemplified by the ambulance scandal.

Armed with notepads, calculators, and digital cameras, each inspection team typically spends a fortnight checking how federal government cash has been spent by local governments. Where the auditors go is utterly random, which makes the policy unique and powerful. This is because random means representative. Just like scientists aiming to deduce the laws of nature from carefully designed experiments, the nerds in Brazil’s anti-corruption agency have been building up a general picture of the country’s everyday corruption from randomly selected samples of municipalities.

The dodgy ambulance peddlers are among the most notorious of their many busts. “The auditors found the problem in one municipality, then in another, then another,” says Yves Zamboni, an economist who served eight years as advisor to the anti-corruption agency’s head. “Then another guy in another state found the same thing. And then in another state. And another. The central office put these data together, and realized how big and organized the scheme was.”

Prosecutors in Mato Grosso, a state about one-and-a-half times the size of France and home to the capybaras and caimans of the world’s largest tropical wetland, found the audit team reports to be interesting reading. The prosecutors had been investigating several companies in the city of Cuiaba whose names cropped up whenever an overpriced ambulance was supplied to Acre, another state that had asked Mato Grosso to investigate the matter. The same names littered the audit teams’ reports from Rondonia and many other states as well. So the federal police started wiretapping.

In the end, investigators determined that the ambulance scheme reached into 19 of Brazil’s 26 states. It involved white-collar gangsters directing parliamentary aides to request federal budget amendments that opened up the flow of cash to city governments where mayors were willing to cooperate. In those municipalities, a common roster of front companies won bids to provide ambulances at about 120 percent of the market price. The profits were then pumped around the system, and kickbacks circulated to all involved. About 300 mayors are thought to have gained from the scheme, many of whom were implicated by the audit teams’ evidence. In 2006, when the police were done, a parliamentary commission recommended opening impeachment proceedings against 69 federal deputies and three senators. The media dubbed the whole episode the “Scandal of the Bloodsuckers” – the latter word being a synonym, in Brazilian Portuguese, for the leech, the hardy worm that drains small amounts of blood from its mostly unsuspecting hosts.

Out of 175 countries, Brazil ranks 69th on Transparency International’s corruption perceptions index — middling, but hardly insignificant. Corruption is thought to reduce Brazil’s GDP by 1.35 to 5 percent, many tens of billions of dollars. In the 1990s, before the bloodsuckers, there was the “Scandal of the Dwarfs,” involving seven short-statured members of the Congressional Budget Committee who swapped patronage for payoffs. In 2005 came the mensalao, a scheme of monthly payments for congressional votes that threatened former President Luiz Inacio Lula da Silva’s re-election. Revelations of bribes for contracts at Petrobras have left the partly government-owned oil giant shut out of bond markets for many months. Even Lula, the world’s most popular democratically elected head of state when he left office in 2010, now faces allegations of improper influence peddling.

It was Lula’s government that introduced the audit teams. When he won his first term in 2003, the CGU was only two years old. It had a shoestring budget and about 5,570 municipal councils to keep tabs on, and random selection was a way to stretch its meager resources fairly and effectively. It meant that nobody could accuse the agency’s bosses of picking their targets through political calculation. Psychologically, randomness also has a subtext — “It could be you!” — that the policy’s architects thought might encourage mayors to start cleaning house even without being investigated (in much the same way that lotteries attract thousands of players, even though the odds of winning make buying a ticket economically irrational).

It was no accident that the CGU decided to employ Brazil’s national lotto company to operate its audit lotteries using the same revolving, ball-filled turbines that otherwise draw the winning numbers that make a handful of Brazilians fabulously rich. Zamboni sums it up: “They got the same machinery, the same people, the same process.” The CGU’s audit lotteries are open to the public, attended by national figures like the heads of the armed services, and often covered in the press.

After each draw, the agency spends a week preparing the audit teams. This grace period provides enough time to gather an exhaustive list of fiscal transfers, dating back to 2001, from every federal government department to every selected municipality. It doesn’t help corrupt mayors who might try to rewrite cooked books in a hurry, because the CGU’s auditors travel far and wide within each municipality, measuring, counting, and interviewing to verify that federal records align with public goods on the ground. They check the dimensions of newly built toilet blocks. They wander around hospitals looking for equipment that should have been purchased. When the teams are done, every last item of a municipality’s dirty laundry is aired online. (The reports from audits of more than 2,000 lottery selections can be downloaded from the CGU’s website.)

But the details of the audit policy aren’t always well understood locally — neither by the officials to be investigated, nor by the citizens who stand to benefit. Shortly after Sao Francisco do Conde, in Bahia, was picked in the sixth draw, thieves broke into the municipal council’s offices and stole cell phones, diskettes, and documents relating to the activities of two former mayors — both of whom had been accused of misappropriating public funds. (Those behind the break-in were clearly laboring under the mistaken assumption that a CGU audit involves only going through records held in municipal government offices.) Over time, however, citizens have come to understand how the policy works, and how to look up dirt on their former and current mayors. Among their other benefits, the auditors’ reports have also helped to undermine the monopolies of information that political bosses still enjoy in areas where they hold sway over local media.

The CGU’s lotteries also offer rich pickings for academics. Since the municipalities are selected for audit through pure chance — and since the results of these audits are made publicly available – the process produces a rich trove of scientifically invaluable data. The audit teams’ reports offer a way to rigorously identify the causes and effects of corruption.

Consider a typical corruption scandal, perhaps involving allegations propagated by an official’s main rival, and supported only by a smattering of circumstantial evidence that appears conveniently during an election campaign. Who’s to say whether the accusations have merit? The findings provided by the audit teams are far more impartial — and, potentially, far more actionable. Moreover, the agency hires its staff through competitive exams, and pays the kind of salaries that don’t necessitate “freelance work” to make ends meet. One need only to compare Brazil’s random audits with the investigations of The Telegraph and other British newspapers that reported on misuse of MPs’ expense accounts starting in 2009. To be sure, that scandal may have been blown open by a fully independent source, but unlike the 12-year history of Brazil’s audit lotteries, the British probe was a short, sharp shock. This makes the lessons learned from it largely anecdotal: shocking, perhaps, but yielding no general insights into how corruption affects society and the economy, and how it sustains itself over time.

By contrast, the randomness of the audit team data is especially valuable because they can be used to separate correlation from causation. If you sample a population of things entirely randomly, there should be no systematic bias between the group of things that gets measured and those that don’t – which is why full disclosure of the audited municipalities’ misdeeds makes for a useful societal experiment.

Without random sampling, studies tend to put forward tantalizing correlations that prove less than compelling analytically. For instance, firms whose stock prices rose and fell in synchrony with changes in the health of President Suharto — who was in charge of Indonesia for 31 years — were the same set of firms with whom the dictator was understood to enjoy especially close ties. But that correlation hardly proves that the firms were corrupt. Similarly, highly corrupt governments tend to rule countries with slow economic growth. But this pattern does not mean that their corrupt acts cause low growth — nor that stamping out corruption will hike GDP.

The first question that the audit teams’ findings have been used to answer is one central to the study and practice of democracy. If citizens suddenly learn of corrupt goings-on behind the scenes, do they change their opinions in sufficient numbers to vote the crooked incumbents out of office? Research by two economists, Claudio Ferraz and Frederico Finan, concludes that Brazilian democracy does function properly in this sense. Their finding might seem uninteresting and predictable, but in many countries troubled by corruption, democracy gets surprisingly poor reviews. (The old autocrat, by comparison, at least had a firm hand.) Ferraz and Finan’s number-crunching has also confirmed that politicians near the end of their careers are bigger rogues than those in their first term, and that the media matters. In municipalities with radio stations, citizens were more likely to throw the rascals out.

Other researchers who have dived into the audit data have identified weaknesses, not strengths, in Brazilian democracy. One of those is another economist, Fernanda Brollo, who has matched the audit results with records detailing how the central government has responded to evidence of rotten local administrations. She finds that equally corrupt mayors are not treated equally. Those affiliated with the Workers’ Party — the party of the president since the CGU lottery policy began — get off relatively lightly. Like the other naughty mayors, their municipalities’ flow of cash from the federal coffers is cut back, but then it starts up again much more quickly — presumably so the Workers’ Party can be seen punishing its members like everyone else, but limiting its political losses.

Then there’s the question that often brings a wry smile to Brazilian faces: What does a politician need to do to make up for his or her dodgy dealings and win an election anyway? Carlos Pereira and Marcus Andre Melo have tried to answer this question for elections in the northeastern state of Pernambuco. They call the problem “the paradox of unpopular corruption and popular corrupt politicians.” Pereira and Melo found that spending on lots of visible public goods helps unsavory mayors stay in office. This kind of thing is common the world over, but well-designed policies can put numbers on it, and provide a means to follow its rise and fall. In Brazil the idea is so well established that it is captured in the saying rouba mas faz (“he steals but he gets things done”) — a slogan associated with Ademar de Barros, a former mayor of Sao Paulo, and a regular candidate for the presidency in the middle of the last century.

In my own research I look for corruption’s unexpected traces in society. Like Ferraz and Finan, I’ve found that local media outlets play a crucial role. When corrupt acts are committed by a male mayor, the number of votes women candidates get increases — provided local broadcasters are available to disseminate an audit team’s findings. This might be because voters stereotype women as more honest than men, or they might just see them as likely outsiders to the shady networks that have been getting things done. When no local broadcasters are present, however, corruption tends to shut women out of political life, quelling both their eagerness to run for office and their electoral success. Perhaps those considering a run for office learn how power is managed behind the scenes from an audit report, and the women among them then downgrade their own estimations of how well they will perform in office — assuming, again, that female candidates have less access to established networks of power than the men.

The last time the audit teams fanned out across Brazil was in February of this year. That was the 40th audit lottery draw, and, as far as anyone following the lotteries knows, the CGU has no plans to stop soon. The policy has led to numerous local scoundrels being fined, impeached, or dismissed, and has provided a bird’s eye view of nationwide corruption that has upended even slyly organized schemes like the one based on the overpriced ambulances. But, so far, no other country has copied Brazil’s methods.

When I ask Zamboni if the audit-team approach will spread, he notes that it’s a big hit at conferences of policy wonks. “I’ve spoken to people from Turkey, from Germany, and Argentina,” he answers, after a pause. “But this kind of policy isn’t easy to sell. Politicians don’t like being selected by a random process.” Perhaps that’s no surprise. Gamblers at heart they may be, but those with a taste for power will take some persuading before they’re ready to implement an anti-corruption program they can’t control.

Photo credit: Ascom/CGU