Crime rates and welfare payments in Brazil: a rancorous debate

It’s not often that a World Bank-financed academic paper authored by Phds from Stanford, Berkeley, and the University of Chicago causes a bitter, snark-infested debate, convulsing the blogosphere– in Brazil.

I missed it, but was caught up thanks to Drunkeynesian, who published a kind of online guide to the debate. It all erupted last month, when the Veja newsweekly’s Reinaldo Azevedo, who writes one of the country’s most-visited blogs, came across an academic study that said government welfare payments to poor families had contributed to lower crime rates in metropolitan São Paulo.

Bolsa Família, as the program is known, is considered by supporters of the governing Worker’s Party, and even by some detractors, as one of Brazil’s main policy achievements of the last dozen years. It channels cash payments to families, with a few strings attached– children must attend school and receive basic pediatric care.

The link between crime reduction and Bolsa Família was too much for Azevedo, known as an acerbic critic of the Worker’s Party, and of starry-eyed praises of Bolsa Família as a poverty and inequality cure-all. He begins his post dismissing the researchers’ work as “silliness,” or bobajada, on its own a rather strong term in usually cordial Brazil. Azevedo then sprinkles in more such remarks, saying “it’s been a long time since I read so much nonsense,” and calling the study “a bunch of claptrap.”

Azevedo’s more substantive point is that crime has actually dropped precipitously over the last decade in relatively well-off São Paulo, where the academic study focused, but has climbed alarmingly in northeast Brazil, where poverty is more prevalent and Bolsa Família reaches a larger portion of the population. In other words, if Bolsa Família reduces crime, then why has crime gone up so dramatically precisely in the region of Brazil where the cash transfers to families reach the most households? And finally, Azevedo accuses the researchers of a kind of politically correct bias, which conflates poverty and crime.

Someone needs to tell the so-called researchers … that,  just as we shouldn’t come down with police on someone who needs Bolsa Família, we shouldn’t try to contain, with Bolsa Família, those who need to be better-policed. And a final recommendation: stop it with that disgusting prejudice against the poor, under the pretext of social benevolence. If poverty really induced people to violence, nobody would be able to stick their nose out the door.

Azevedo’s post, only the very first parry in this war of words, triggered a three-page open letter from one of the study’s authors, microeconomist João M. P. de Mello. Mello’s very public reply undoubtedly escalated a discussion that otherwise might have quietly gone away.

Mello, unable to contain his disdain for Azevedo’s quick-fire dismissal of his empirical and statistical labors, fires off a few epithets of his own. He characterizes Azevedo’s post as “primitive,” “arrogant,” “ignorant,” and “aggressive.” He accuses Azevedo of not bothering to look at the study and of a knee-jerk ideological reaction, due to his anti-Worker’s Party biases. Mello attempts to explain the subtleties of empirical research and statistically significant correlations. But meanwhile he cannot help condescending:

The majority of intelligent and educated people, without scientific training, have difficulty in distinguishing correlation and cause. Don’t feel diminished by that, blogger. I’m going to refer to you as ‘blogger,’ because in the end, that’s your title, isn’t it?

In the end, Mello buckles down and tries to explain why Azevedo’s specific criticisms are statistically wrong-headed (the fact that crime increased in northeastern cities like Maceio and Salvador does not necessarily mean that Bolsa Família does not decrease crime– crime might have increased more without the program).

At this point, before wading further into the debate, it might be helpful to pause and review Bolsa Família, its history, what it actually does, and what Mello’s research sought to show.

Early versions of Bolsa Família began under previous governments but the programs were consolidated, significantly expanded, and championed federally by former Worker’s Party President Luiz Inácio “Lula” da Silva (2003-2011). His hand-picked successor, President Dilma Rousseff, has also pushed Bolsa Família, expanding it by increasing payments for very young children, and covering day-care costs.

For opponents, Bolsa Família is a thinly disguised Worker’s Party vote-buying program that is woefully managed, and hence shot through by waste and inefficiency. The critics also fret about the long-run consequences of dependency on state largesse, especially since Bolsa Família is not the only government assistance program out there (São Paulo, for example, has its own state-level cash transfer program). So, the Azevedo vs. Mello debate has to be understood as part of a wider discussion about the Worker’s Party social welfare policies.

The academic study that sparked all the fuss begins with the basic numbers on Bolsa Família. The program costs about .4% of Brazil’s GDP, but that figure disguises its massive coverage: over 11 million Brazilian households. The payments appear modest, a maximum of about $100 (200 Brazilian reais) monthly per family, depending on income level and number of children. But it is hardly a token amount considering that $70 is the maximum an enrolled household can earn, on a “per-head” basis.

Formerly restricted to low-income families with children 15 years of age and younger, in 2008 the program expanded to families with adolescent children aged 16 and 17. And that is the event the researchers– Mello and his co-authors Laura Chioda and Rodrio R. Soares– exploit to build their study (Chioda is with the World Bank, while Soares and Mello are with the Pontifical Catholic University in Rio).

They collect detailed data on high school and middle school students in São Paulo, and link it with geo-referenced crime records. Using regression analyses, and controlling for other variables, they show that on average neighborhoods around schools with a greater proportion of 16 and 17-year-olds experienced a significantly greater drop in crime after the expansion of Bolsa Família.

Going by these results, 59 additional students covered by Bolsa Família (that was the median increase in Bolsa recipients at each school between 2006 and 2009) translated to 21% fewer crimes, or 94 fewer crimes per year in each school-area neighborhood.

Here are the researchers explaining the results in their own words, using the acronym “CCT,” for conditional cash transfers, when referring to the program:

This instrument allows us to show that CCT coverage in a school has a negative impact on crime in the neighborhood. The evidence also indicates that the reduction in crime is not concentrated in school days and seems to be mostly driven by economically motivated crimes (robberies). So it is likely that … time spent in school is not a particularly relevant mechanism, while the income component of CCTs seems to play an important role.

In other words, it is not the school attendance requirement, but the extra family income, which accounts for the positive spillover effect on crime rates.

Returning to the online furor, Azevedo did not let the matter rest, and rebutted Mello’s open letter, point by point, in a new blog post, with a rather long title: “Economist confesses (unmissable!) how he constructed his research and thinks that if he attacks and insults me, it will prove he’s right. He explains everything except what’s essential.”

Despite all the verbiage, nothing was really added to the debate, except maybe for a slight escalation in rhetoric, with Azevedo referring to Mello “an immense dope.” This stage of the affaire did seem to attract a lot more eyeballs. Azevedo’s second post received 772 comments (the first did not break 200 comments).

One of his commenters, wrote on June 28, “for the good of our friend the Phd, I hope he didn’t read your reply to his reply. If he did read it, I don’t think his ego will recover very soon.”

And it was after Azevedo’s second post that the blogosphere erupted. Some took Azevedo’s side, some the professor’s, and many noted that neither looked too hot, and that the two hadn’t managed to truly communicate: the journalist and the statistician had spoken past each other.

A post from Selva Brasilis was characteristic of this last sort of lamentation:

“Azevedo has emerged with his reputation tarnished amongst those of his readers that know economics, and Mello has had his reputation roasted without having had the chance to build one.”

The Drunkeynesian, a well-known and prolific Brazilian economics blogger, also was late to the party (though far less late than I am, and had a vantage point from which to judge the conflagration, linking to at least a dozen takes on the issue.

Drunkeynesian ended by characterizing the Azevedo vs. Mello fracas as a kind of watershed, which showed the extent to which the Brazilian blogosphere could serve for the airing of an important issue. And unlike may bloggers, who lamented the vitriolic tone, he chose to look on the bright side: a patently wonky issue had incited passions, public input, and received a wide airing.

“I don’t remember any other issue having mobilized the economics blogosphere to the same extent, and it should welcome this type of dialogue, which in the end has enriched it.”