Operations Management of On-Demand and Crowdsourced Systems

Abstract

Over the last few years, crowdsourced and on-demand service platforms have provided unconventional accessibility to services and products, and hence, studying their operations is of great importance. Research in this thesis focuses on studying operations and pricing of these service platforms and providing policies and guidelines on how such systems should be optimally implemented in practice. Some crowdsourced platforms do not provide on-demand services and they merely match crowd providers with customers. An example of these platforms is a crowdfunding platform which is the focus of the second chapter in this thesis. Additionally, there are some crowdsourced platforms that provide on-demand services, such as crowdsourced last-mile delivery platforms that provide fast 1-hour and 2-hour delivery services to customers. Studying these platforms is the focus of the third chapter in this thesis. Finally, some on-demand service platforms do not base their operations on the crowd, but they revolve around matching independent and third-party logistics service providers with retailers; an example is an on-demand warehousing platform, which is the focus of my research in the fourth chapter. In this thesis, we first study the application of revenue-sharing contracts in crowdfunding. In particular, we study an emergent model of crowdfunding that allows firms to raise capital from a pool of investors and repay them a multiple of their investment by sharing a percentage of its future revenue, under a revenue-sharing contract. This means that investors will receive M>1 dollars for every dollar that they have invested over an investment horizon of uncertain duration. This contract, as a flexible repayment agreement, is linked with the financial performance of the firm, allowing variable payments and investment horizons and, thus, reducing financial stress on the firm. If the firm’s business does well, the firm is obligated to increase the payments, thus reducing the investment horizon, which results in a higher effective interest rate for the investors. Therefore, revenue-sharing contracts intuitively align firms’ and investors’ incentives in a way that was not possible with traditional fixed-rate loans. We also compare revenue-sharing contracts with equity crowdfunding and fixed-rate loans and observe that revenue-sharing contracts result in higher net present values and lower bankruptcy probabilities due to their flexible nature. We next study labor planning and pricing for crowdsourced last-mile delivery systems that are utilized to deliver on-demand orders. We develop our optimization model by combining crowdsourcing, robust queuing, and robust routing theory, which allows for capturing uncertainties, trend and seasonality in: customer demands, crowd availability, service times, and traffic patterns. For a given delivery time window and a guarantee level to deliver on-time, we analytically derive the optimal delivery assignments to available independent crowd drivers and compute their optimal hourly wages. We evaluate the performance of our robust model against a stochastic counterpart and a modification (to accommodate crowdsourcing) of the well-known Savings Algorithm via a realistic simulation study, based on the Seattle transportation network, that allows for nonstationary customer purchasing patterns, heterogeneous crowd drivers, as well as time-varying traffic patterns. We show that our robust solution performs significantly better than the two benchmarks, both in terms of the percentage of on-time deliveries and cost savings. We show that crowdsourcing last-mile deliveries can significantly reduce logistical costs for retailers. Finally, we study on-demand warehousing platforms that match independent warehouse providers who possess excess capacity with retailers who seek on-demand warehouse capacity without imposing large fixed costs or long-term leases. Although traditional warehousing may be cheaper, it must be acquired based only on demand forecasts and not the realized demand. However, on-demand warehousing can be acquired during the selling season after observing the demand. We study a firm’s capacity decision in the presence of these on-demand platforms. Our results indicate that companies can rely on on-demand warehousing to supplement their traditional warehousing, with hybrid strategies constituting the optimal outcome in many cases. We find that, as firms' demand variability increases, firms benefit more from on-demand warehousing. Our results further show that as there is less uncertainty in the capacity provided by independent warehouse providers, firms' benefit from on-demand warehousing increases as they can rely on on-demand capacity to cope with their demand uncertainty.