Tuesday, September 6, 2011

Kleiman's Scoring System

In addition to a drastic reduction in American drug demand, Mark Kleiman's security proposal for Mexico is based on a sophisticated scoring system that, coupled with attempts to put the worst offenders out of business, would change the incentives guiding the drug trade. He writes:
The Mexican government could craft and announce a set of violence-related metrics to be applied to each organization over a period of weeks or months. Such a “scoring system” could consider a group’s total number of killings, the distribution of targets (other dealers, enforcement agents, ordinary citizens, journalists, community leaders, and elected officials), the use or threat of terrorism, and non-fatal shootings and kidnappings. Mexican officials have no difficulty attributing each killing to a specific trafficking organization, in part because the organizations boast of their violence rather than trying to hide it. At the end of the scoring period, or once it became clear that one organization ranked first, the police would designate the most violent organization for destruction. In this case, “destruction” might not mean the arrest of the organizational leadership, as long as the targeted organization came under sufficiently heavy enforcement pressure to make it uncompetitive.
Even overlooking his inattention to Mexican security agencies' inability to implement any strategy competently, which is a far more important obstacle than the difficulty in properly identifying the most violent gang, I see a number of problems with the scoring scheme.

The first is that Kleiman seems to be operating with the idea that the gangs in Mexico are well defined entities, and that the 500,000 or so Mexicans who live off the drug trade are full-time employees of one or another of the big six. To a large degree, this is not the case; that half-million people is organized more into a series of loose federations and patterns of alliances, which are constantly in flux, especially the further down the food chain you go. He refers to the six big organizations as though there were some formal distinction bestowed upon these groups. I'm not sure which six he has in mind, but I think you could easily make a case for eight dominant organizations (Sinaloa, Gulf, Zetas, Beltrán Leyvas, Juárez, Tijuana, Familia Michoacana, Caballeros Templario, though I imagine he is ignoring Tijuana and considering the latter two as a single group) that have significant production or smuggling or territorial-control networks (or all three), and there are scores of upstarts (CIDA, South Pacific, La Mano con Ojos, et cetera), local gangs (these groups aren't as famous, but I remember Los de la Casa del Cerro in Torreón, for example), and enforcer groups that have varying degrees of autonomy from their ostensible bosses (La Línea, Los M, many, many others).

This murky reality makes it a lot harder to positively identify one of the big six with the crimes. Are the killings in Durango, a product of the Sinaloa Cartel, or are they a product of different sects of Gente Nueva targeting each other? What about two rival narcomenudista bosses in a Zeta-controlled town, who pay their quota but aren't part of the organizational structure, killing each others' hoppers? Are those attributable to the Zetas, even though they have nothing to do with Lazca's ambitions and they could be occurring regardless of who was in control of the town? Related to this, how much confidence should we have in the Mexican government to positively determine who is behind the killings? Kleiman writes that "Mexican officials have no difficulty attributing each killing to a specific trafficking organization", but that's quite a sweeping assertion that he doesn't back up in the slightest. Sometimes there is a note attached, but my impression is that far more often there is not. Sometimes there is a logical suspect; often, however, the killings remain shrouded in mystery. I don't know the proportion of the killings in which the federal government would be capable of assigning blame, but it's worth noting that this is a country where 98 percent of all crimes remain unpunished, and where the federal government, which is the most competent level, is unable to secure even a trial for 70 percent of the people it arrests, to say nothing of a conviction. That certainly cuts against the notion that the government always knows whose behind a murder.

A second problem is that such a method of determining policy would leave itself open to manipulation by the very gangs that the government seeks to combat. For instance, if Chapo wanted his rivals to come out with the largest number of killings, he could just send 100 poor, ill-armed saps to Veracruz, and also let it be known who are the wardens who are taking money from Sinaloa so as to make life hard on La Línea in Chihuahua jails. All of a sudden, his enemies would be the authors of a handful of crimes that would make their score higher. I think the idea is that the scoring system would impose a race toward better behavior by the gangs, in order to avoid government attention. But if you think about it from the perspective of the clever gangster, it could just as easily spark an ever-escalating series of provocations.

Or how about the inevitable result that Chapo et al would spend their millions buying off the people who determine who is responsible for a given murder? A trustworthy score of technocrats working from Los Pinos would not alone be able to make much sense of 15,000 murders; a dependable scoring system would require lots of local knowledge, which would, in turn, create many juicy targets for corruption, and make the scoring system far less reliable even on its own terms.

Another problem relates to the policy-makers' understandable desire to remain flexible. Such a system could certainly be a guide to government policy if used internally, but I can't imagine a president wanting to tie himself to the results of a publicly available scoring system ahead of time. What if the metrics seemed logical beforehand, but then they told you that the Cártel de Jalisco Nueva Generación was Mexico's biggest threat? In other words, what if it became obvious halfway through the process that the system was crap? (Is anyone else reminded of the BCS?) The government is then tied to something that it no longer believes in. Perhaps there real-world examples that refute my conclusions, but it's hard to believe that a politically competent president would voluntarily subject himself to the possibility.

And this is not just a matter of insufficient commitment; a president should want to remain flexible. A scoring system devised in 2007 would almost certainly have placed little emphasis on extortion, but today, I'd say that's among the most important crimes plaguing the nation. Organized crime groups in Mexico are quite flexible, and there's no way a president could know now what will be the most dangerous element of their behavior years or even months ahead of time. I guess you could just constantly tinker with the system, but then this would undercut this sense of objectivity and empirical certainty, which is presumably what appeals to Kleiman to start with. Although it remains unclear to me why Kleiman thinks that the reasoned judgment of the professionals who have dedicated their careers to analyzing and combating organized crime is insufficient, which is probably the most fundamental flaw behind the proposal.

Kleiman certainly is right that the goal needs to be a less violent pattern of behavior from the Mexican groups, and that a reversal of the incentives driving the violence today is necessary. But I don't think the program he lays out is capable of engineering that switch.

No comments: