Guillermo Trejo and Sandra Ley tackle the deep-seated roots of Mexican violence in their new paper, "Why Did Drug Cartels Go to War in Mexico? Subnational Party Alternation, the Breakdown of Criminal Protection, and the Onset of Large-Scale Violence."

As Trejo and Ley point out, Mexican criminal groups enjoyed a longstanding period of peaceful coexistence prior to the 1990s. Violent conflicts disrupt business and attract increased attention from security agencies, making them essentially irrational. So, if it was not a prior mode of business and all participants suffer from criminal warfare, why indeed did Mexico descend from a stable, peaceful equilibrium into the current state of constant battle?

The story told by Trejo and Ley dates to the era of democratic turnover at the state level. This began in 1989 with the election of Ernesto Ruffo Appel of the opposition National Action Party (Partido Acción Nacional - PAN) in Baja California, the first time that the governing Institutional Revolutionary Party (Partido Revolucionario Insistucional - PRI) had ever failed to win a statehouse. Over the next decade, the PAN and the Democratic Revolution Party (Partido de la Revolución Democrática - PRD), the other major opposition force, scored victories in Jalisco, Mexico City, Chihuahua and various other areas.

As a result, for years prior to the PRI's loss of the presidency in 2000, growing swaths of the country were under the control of new political leadership. These new administrations changed much in their approach to governance, and as Trejo and Ley document, their first steps on security were to clear out "top- and midlevel personnel in the state attorney's office and the state judicial police." The authors base this contention on interviews with former opposition governors and other officials, giving their conclusions an atypical degree of certainty.

The incoming administrations upended the relationships that managed the country's drug trade. Whereas criminal groups used to be able to count on a small cohort of political partners -- and their vast police forces -- for protection in perpetuity, the statehouse turnover and expectations of future uncertainty left them adrift. Kingpins responded by building private militias, essentially as a hedge against changing political winds.

The existence of militarized wings in criminal groups across the country, combined with a steady flow of arms and the weakening of political authorities capable of decisively adjudicating underworld conflicts, fostered sustained bloodshed.

According to Trejo and Ley, Mexican states that saw the election of opposition governors in the 1990s experienced substantially greater rates of violence. Applying four different models to control for various outside factors, the authors found that having an opposition governor was responsible for between 55 percent and 79 percent more violence from 1995 to 2006.

Several examples fit this pattern: Ruffo Appel's election in 1989 preceded the militarization of the Tijuana Cartel, which then spent most of the 1990s engaged in bloody conflicts with rivals from Sinaloa and Juárez. The PRI's loss in Jalisco in 1995 encouraged the Sinaloa Cartel to build militarized wings under the Beltrán Leyva brothers and Ignacio Coronel. In Michoacán, the end of PRI rule brought about the entry of the Zetas and the emergence of the Familia Michoacana.

These armed groups not only went to war with one another, but they also often became the centers of gravity of their respective cartels -- or worse still, they became independent organizations. As the traditional capos succumbed to pressure from rivals or from the government, the militarized groups were poised to assume leading roles within the Mexican criminal landscape.

InSight Crime Analysis

One of the virtues of this study is that it takes a longer view of Mexican violence, unlike the more common approaches that begin in 2000 with the PRI's loss of the presidency, or in 2006 with the election of Felipe Calderón. To be sure, Calderón's heavily publicized militarization of Mexican security policy exacerbated problems that were already in evidence for years prior. But the bloodshed that Calderón sought to stem was part of a decade-long trend, and Trejo and Ley appear to have identified a key turning point.

Their work suggests that while Mexico's democratic opening was certainly laudable and probably inevitable, much more work was needed to build political and criminal justice systems capable of guaranteeing security and prosperity. The authors are careful to make clear that they are not calling for a return to a one-party system, but an effective political liberalization required more than just removing the PRI stranglehold on power.

It is unclear if the link between gubernatorial turnover and rising violence remains as strong today as it was from 1995 to 2006, but there is little question that politics and public security continue to exert substantial influence over one another.

One of the paper's more obvious implications for future security policy is that officials should principally target not kingpins, but rather their armed wings. While this sounds obvious enough, Mexican officials (and their allies within the US government) have prioritized kingpins for decades. However, taking down kingpins creates instability and increases the relative power within armed groups. While arresting or killing famous capos makes for good politics, doing so does not pave the way for greater security.

The authors' conclusion that governors contributed to the spiral of violence by removing high-level judicial and police officials also speaks to the need for a professional and non-partisan security bureaucracy. Removing the criminal justice system from the realm of politics would theoretically make public security less susceptible to ill effects stemming from political handovers.

At the same time, it is also clear that Mexico needs to establish a security equilibrium that does not rely on collusion between government officials and criminal groups. One way to look at the past 20 years in Mexico is a transition away from such a dynamic, and it remains to be seen how long it will take for the process to be completed.