Political corruption in Mexico is one of the most relevant problems of the last decades.
Widespread corrupt behavior in national politics has a negative effect on economic growth, reduces the availability of public funds, and increases social and economic inequality in modern democracies. The problem of corruption in Mexico is further exacerbated, partly due to systemic biases that limit prosecution, and the risks associated to the journalistic coverage of such cases. Major corruption scandals have been uncovered in the last decades by the actions and bravery of impartial journalists. “La Estafa Maestra,” where Mexican federal Government diverted billions of Mexican Pesos to finance electoral campaigns, is a recent example of how journalistic efforts are necessary to uncover these cases. Systematic analysis of journalistic coverage of corruption from a complex systems perspective may provide new insights on the nature of this phenomenon at the national level.
Here we used a corpus of more than 7,000 articles associated to corruption from the independent newspaper site Animal Político. This Corruption-Corpus (CC) was identified using a semi-automatic, expert-reviewed classification system, from a larger corpus of 65,000 articles published between 2010 and 2018. Using this database, we constructed a multi-layer network with layers representing political figures, places, institutions and “corruption keywords” co-occurring in the CC. We analyzed the topological features of the network. We identified the most relevant characters, places, institutions, and “corruption keywords,” in terms of their centrality degrees. We identified community structures in the network that can be associated to actors involved in particular scandals. We also explored clustering structures emerging from neighborhood similarities between geographical divisions such as states, which provides information on thematic closeness of corruption cases between geographical regions, regardless of physical proximity. We further explored the evolution of these networks in time windows, observing the appearance and disappearance of nodes and links, and changes in clustering and community structures. Finally, we scrutinized the terms associated to political figures in order to describe the main topics that associate different political actors to distinct facets of corruption.
Our results show a complex network that connects different layers of political actors, institutions, and geographical regions in a context of corruption. In this network, the most influential politicians have a clearly different subset of associated terms, indicating a “thematic independence” in terms of corruption topics. Political figure communities and geographical clustering structures exhibit abrupt changes through time, that can be associated to important events such as State and Federal elections, but are not dependent on physical (geographic) proximity. We argued that this approach may uncover relationships between political actors and acts of corruption that deserve further investigation, and may provide a clearer picture of the corruption phenomenon which may be useful for its eventual eradication.