Conventional wisdom depicts corruption as a tax on incumbent firms. This paper challenges this view in two ways. First, by arguing that corruption matters not so much because of the value of the bribe ("tax"), but because of another less studied feature of corruption, namely bribe unavoidability. Second, we argue that the social costs of corruption arise not because corruption hurts incumbent firms, but mostly because it acts as a powerful barrier to the entry of new firms. Corruption sands and greases in tandem: it helps incumbent firms (on balance) and it hurts potential entrants. We put forward a model in which a bureaucrat chooses entry barriers to optimize bribe revenues. When the capacity to collect bribes is high, it is optimal to allow high levels of oligopoly power to incumbents. Conversely, the more avoidable are the bribes, the more firms are allowed into the market. These ideas are tested using a unique, representative sample of Brazilian manufacturing firms. Consistently with our theoretical model, we show that corruption (a) is ranked as the most important barrier to entry (above finance, taxes and regulation) and (b) while bribes' unavoidability is positively related to firm performance, the size of the bribe is not.