Using a novel database of 189,000+ Colombian firms and 500,000+ firm executives' names, I study the effect of financial factors, CEOs' centrality (corporate power), and political connections on access to a government bailouts program launched to subsidy wages in the first stages of COVID 19 crisis. Natural Language Processing algorithms and complex networks metrics are used to unveil ownership and control links of politic/economic elites and gauge their closeness to the Colombian President. I find that firm size factors and firm age, instead of political-connections or being run by prominent CEOs/shareholders, explain access to the program. In addition, I find that impacts of the program are positive in terms of salaries and liquidity, but they do not increase with firm size and age. The differential effect on firms run by powerful economic elites was on liquidity instead of employment. These findings thus suggest a preference for protecting systemically-important firms (without ex-post economic efficiency) rather than special interests of elites.