Randomization Unit
We will randomize the intervention at two levels: tender and firm-level.
First, at week-level we will randomize at the level of active tenders, we will conduct an equal split with computer generated random selection of tenders subject to intervention, and others that serve as control.
Second, within those treated tenders, we will randomly allocate firms with the expertise to participate in upcoming public tenders, determining which of these firms will receive a) information about the active tender, and b) a detailed report about the active tender. Within the treated tenders, firms will be randomly assigned to three groups: i) treatment arm 1 (information about the active tender), ii) treatment arm 2 (a detailed report about the active tender), and iii) a control group.
This approach allows us to capture nuanced variations in competition and bidding behavior across different tenders, ensuring that the intervention is relevant to the specific context of each procurement opportunity. By selecting certain tenders to receive the treatment while others serve as a control, we can effectively evaluate the overall impact of the intervention on competition at the public tender level. The second level of randomization is implemented within the treated tenders, where firms with the expertise to participate in upcoming public tenders are randomly allocated to receive targeted information about those tenders. This individual-level randomization enables us to analyze changes in firm behavior, such as participation rates and bidding strategies, while providing insights into how access to information influences decision-making among various firms, particularly smaller firm and firms further away from the public buyer. By incorporating both tender and firm-level randomization, our study will yield comprehensive findings on the effectiveness of the intervention and its implications for enhancing competition in public procurement.