Governments often cannot contract on the quality of what they procure. When quality is non-contractible, screening may improve performance by restricting which firms can compete. Many governments screen contractors indirectly, conditioning eligibility for larger contracts on the execution of smaller ones. I study this trade-off in Rwanda’s public procurement of infrastructure, focusing on one key measure of quality: project completion time. I link the near-universe of electronic procurement records for 2016–2024 to records of government payments that trace contract execution. I estimate a dynamic auction model in which low- and high-quality firms compete in first-price auctions and completed work determines access to larger contracts. High-quality firms complete work and accrue experience necessary for promotion faster than low-type firms. Therefore, high-quality firms value winning tenders more than low-quality firms, which induces them to bid more aggressively, partially offsetting their higher costs. I conduct two sets of counterfactuals. First, I modify the eligibility criteria and highlight the government’s trade-off between price and quality. Second, I change the government’s promotion policy to estimate the value of screening on past performance.

Connections between firms are critical for economic growth, providing new business, opportunities for learning by doing, and access to higher-quality inputs. What prevents firms from forming these connections? One possibility is information frictions. Industrial trade fairs are a long-standing solution to such frictions, coordinating a centralized meeting point for buyers and suppliers. In partnership with the largest operator of industrial fairs in Chile, we run a two-sided randomized controlled trial at the country’s largest trade fairs in aquaculture and mining. On the supplier side, we subsidize randomly selected firms to host a booth at the fair; on the buyer side, we introduce randomly selected buyers to suppliers in randomized business meetings. This two-sided randomization distinguishes whether new connections crowd out incumbent suppliers or represent new aggregate business. Combining the experimental variation with detailed administrative tax data and firm surveys, we trace the direct effects on participants and the indirect effects on their competitors and trading partners. To assess the aggregate value of these new linkages, we build a structural model of firm-to-firm matching.