Using a database of 150,000+ Colombian firms in the real sector, I study the impact of CEOs and shareholders having economic power and political connections on access to PAEF (Programa de Apoyo al Empleo Formal), a government bailouts program during COVID-19 crisis in 2020, and the economic efficiency of this policy. Natural Language Processing algorithms and complex networks are used to identify ownership and control links of economic elites based on their complete names, gauge centrality of CEOs and their closeness to politicians. Following neighbor-pairs fixed effects and differences-in-differences as identification strategies, I find that firms exploited the network prominence of its CEOs and shareholders to obtain government support, and the impact of the program on earnings is positive for small-medium sized, not large or politically-connected firms. These results must not be interpreted as evidence of any illegal behavior, but of special preferences when it comes to shield society from collective risks.