Subsidizing startups under imperfect information

We study the early stages of firm creation under imperfect information. Because startups make error-prone decisions due to rational inattention, the model generates both inefficient entry and labor misallocation. We show that information frictions alter the effects of lump-sum transfers to startups: the total employment gain is amplified due to an unintended increase in inefficient entry, most entrants hire fewer workers, and misallocation goes up. The transfer makes low-size, previously dominated actions profitable, affecting the entire endogenous learning problem and making even productive startups lean toward more conservative hiring. We show that this novel information channel works against well-known mechanisms (for example, financial frictions) and also dampens the effects of alternative policies such as wage subsidies.