Coordination of demand and supply in the Sharing Economy

• Applicants: Professor Dr. Nils Boysen; Professor Dr. Dirk Briskorn
• Subject Area: Accounting and Finance
• Term: from 2019 to 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 425058257

In the sharing economy, the central operative planning decision is the assignment of resources to demand over time. Resources as well as demand may be characterized in multiple dimensions, which renders demand highly individual. We highlight the renting of parking lots as an example. Even if we restrict the characterization to location and time interval two requests will most likely differ in one of both dimensions. Nevertheless, multiple requests are not independent.Most sharing platforms offer (or even require) requests to be made online. Therefore, data for an automated decision-making (or decision support) is available. Common decision mechanism are list-based and decide one by one in order of arrival about whether to accept requests or not and which resource to assign to them. The inherent combinatorics of multiple requests are ignored, since the decision is made for one request at a time. This results in non-optimum assignments in terms of number of requests satisfied or total revenue. In the course of this project, we investigate the use of deterministic optimization approaches embedded in rolling horizon frameworks in order to take acceptance and assignment decision for requests. In a rolling horizon framework, we collect requests over a certain time interval and, then, take the decision for acceptance and assignment of these requests simultaneously. This approach brings the advantage that we can take the combinatorics of multiple requests into account. The challenge is to design algorithms that are able to find assignment fast enough to be employed in real world settings. Hence, in a first step we focus on optimization problems concerned with the acceptance of requests and assignment to resources. We will analyze the computational complexity of these optimization problems and develop suitable solution approaches. Once these approaches are available, we will embed them in a rolling horizon framework and compare its performance with the list-based approaches common in practice. In a second step, we will consider further influences that might interfere with the rolling horizon approaches. One direction of research will be to aim for acceptance and assignment decisions that are robust with respect to stochasticity and selfish behavior. Foreseeably, the approaches as developed so far have to be generalized to account for robustness. We expect to see a trade-off between the efficiency of decisions and their robustness. This trade-off will be investigated, as well.Finally, the generic approaches developed so far will be reified for several specific applications (which will be supported by our partner from the industry). It is to be expected that we need to once again adapt the generic concepts to fit specific applications and we will do so. The performance of the resulting mechanism as compared to the approaches used in the real world will be evaluated.

• DFG Programme: Research Grants