Family, business and politics: mechanisms of tie formation in a multilayer network of Guatemalan elite

Previous research showed that Latin American countries are still to a large extent dominated by business-family elites, resulting in unequal distribution of opportunities and resources in these societies. Scholars argue that these elites are repeatedly substitutes for solid institutions and competitive markets. In Guatemala, elites have preserved a long-lasting network –largely by marriage and economic alliances– and are characterized by high levels of cohesion. Moreover, business relations have become internationalized in the 21st century, and the Guatemalan elite have established a (parallel) business network via offshore companies in Panama. The database includes 6902 elite members linked in several two-mode layers to 1537 offshore companies, 5178 Guatemala-based companies, 128 business associations, 236 non-profit organizations, 55 political parties and 431 public entities. The persons are also related in a one-mode layer by 1980 kinship relations. Current network-based elite studies focus mostly on interlocking directorates or kinship relations, but less attention is paid to examination of how elite cohesion emerges and reproduces in multiple layers of business, social, political and kinship relations. We integrate these approaches and assume that elite cohesion –defined as embeddedness of elites within business, offshore, social, political and kinship relations– is driven by network mechanisms in and between multiple relational contexts. Moreover, we expect that elite cohesion increases their political influence. We introduce an approach to modeling elite networks comprising multiple relations using the exponential random graph model (ERGM) and examining cohesion and influence in interdependent connectivity patterns among different layers. Specifically, we compare two models testing how both the economic connectedness and political influence are being shaped by node attributes and network homophily. The elite members' surnames and gender are taken into consideration as node attributes, and homophily across two-mode layers is based on two different types of ‘bipartite homophily terms’: kinship-interconnectedness and mutually corresponding connectivity. As the former, family-owned business groups are supposed to show similarity between the one-mode kinship ties and the two-mode business ties. As the latter, comemberships in firms, business associations, non-profit organizations, political parties and public entities are expected to predict not only similar business ties (in Guatemala and Panama), but also political influence (in Guatemala). The preliminary results show a well-defined gender inequality in business and political networks, and also that traditional elite-surnames still hold privileged positions in the network layers. The homophily terms indicate that business ties in Guatemala tend to predict business ties in Panama, and vice versa. The mechanisms of political influence are less clearly defined and seem to be based more on business representations in public entities than through the weak Guatemalan political party system. The research applies ERGM on a large dataset combining one-mode and two-mode networks, and contributes to Latin American elite studies with an empirical case.