This paper examines the effects of limited liability, one of the most common insurance instruments entrepreneurs use worldwide, on output, firm size, and technology adoption. I show that despite higher financing costs, limited liability firms are larger than their unlimited counterparts, and their productivity is higher and more volatile. I propose a theory where limited liability is endogenously chosen by relatively rich entrepreneurs to insure themselves against losses, leading to the adoption of more profitable and riskier technologies, and larger firms. Computing general equilibrium effects, I show that limited liability produces output gains of 3.0%.

Firms have inherent and permanent characteristics that drive their performance until several years after establishment, constituting projects with different expected returns. If lenders cannot distinguish between types of firms (different potential returns), how do they allocate credit among them? Does this allocation translate into misallocation and output losses? In an optimal contract setting, I show that if investors cannot observe firms' productivity, the optimal financing contract will generate inefficient capital allocation: high (low) productivity firms will receive more (less) funding than optimal, leading to inefficient firm size and output losses. Then, I evaluate the model quantitatively and find that asymmetric information on firms' productivity can generate output losses of almost 9.7% compared to a frictionless benchmark.

We use matched administrative data from Brazil to study entrepreneurial selection and firm upgrading through the lens of legal form choice. We compare entrepreneurs who enter under the MEI regime—introduced in 2009 to formalize small firms—with those who start limited liability firms. MEI entrepreneurs are negatively selected on education, labor earnings, and prior work experience, consistent with necessity-driven entrepreneurship. In contrast, limited liability founders are positively selected, suggesting stronger opportunity-based motives. We track individuals over time to examine legal status transitions of the firms they create and find that MEI-to-limited liability transitions account for nearly 20 percent of new limited liability firms in 2018, revealing an overlooked path to formal upgrading. Exploiting industry-level MEI eligibility in a difference-in-differences design, we show the program increased firm creation but also displaced some limited liability entries. However, this substitution is more than offset by upgrading among MEI firms that later grow and hire, resulting in a small net positive effect on the creation of successful, job-generating firms.

We study whether subsidized innovation credit creates new innovation or primarily reallocates it across firms. Using linked employer-employee records, a national patent census, and administrative data from Brazil’s FINEP Crédito Direto program (2005-2019), we document a stark disconnect: recipients file significantly more patents andexpand their research workforce, yet the share of firms actively patenting does not expand, and patenting among incumbent innovators is flat or declining. Using propensity score matching combined with a two-way fixed effects model and a Bartik shift-share instrument, we find that recipient firms absorb researchers disproportionately from other innovating firms in the same sector, while exposed non-recipients experience declines in patent output and employment. Subsidized credit reorganizes the distribution of innovation activity rather than expanding its frontier, and standard firm-level evaluations systematically overstate aggregate returns by ignoring the negative externalities imposed on competing innovators