Timing Matters: Crowd-sourcing Workers in On-demand Freight Matching Platforms

Jingxuan Geng, Ziqi Dong, Guangwen Kong, Qiuping Yu

The movement of goods over long distances is essential for economic growth, as it facilitates the efficient flow of production, trade, and consumption activities. However, traditional long-haul transportation firms face significant expenses in maintaining their vehicles and employing full-time drivers. In addition, these firms operate with fixed service capacities (e.g., drivers and vehicles), which often results in over-capacity or under-capacity risks when demand fluctuates. The emergence of freight-matching platforms provides an alternative business model for transportation firms seeking to improve the cost effectiveness of their operations. These platforms connect shippers with a crowdsourced workforce, offering lower set-up costs and flexible capacities to serve shippers compared to traditional transportation models. However, freight-matching platforms also face challenges. Because these platforms allow crowdsourced carriers to self-schedule their work, they have no control over the carriers’ participation in serving shippers, making it difficult for the platform to balance the supply and demand sides of the market, especially when carrier supply is low. Additionally, shippers have varying request lead times (i.e., the time between when they send their freight-matching requests and the pickup dates of their freights), further complicating matters. In this paper, we examine how request lead time is associated with two performance metrics of freight matching: freight-matching probability and sourcing costs. We collected a large and detailed dataset from an online freight-matching platform that operates one of the largest transportation networks in the United States. We focus on request-level data (38,575 freight-matching requests) from December 2017 through February 2018, including basic request-level information (e.g., freight information, scheduling information, and financial information), as well as detailed information about carriers’ offers quoting sourcing costs. Based on this data and a comprehensive estimation strategy, we examine the impact of shippers’ request lead time on freight-matching performance. Specifically, we focus on two important performance measures of freight matching in the industry: freight-matching probability and sourcing costs. Freight-matching probability reflects whether the platform can successfully match freight requests with carriers who provide freight transportation services, while sourcing costs represent the payments to carriers in exchange for their services. Our study reveals that shippers’ request lead time significantly impacts freight-matching performance in a nuanced way. Specifically, increasing shippers’ request lead time can enhance the probability of matching freight requests when the request lead time is short, but it can reduce the probability when the request lead time is long. Furthermore, we find a U-shaped association between shippers’ request lead time and the platform’s sourcing costs. When the request lead time is shorter than a threshold, the sourcing cost decreases with the request lead time; but when the request lead time is higher than the threshold, it increases with request lead time. Therefore, our results suggest that by motivating shippers to shift their timing of requesting freight-matching services, the platform can improve the overall performance of freight-matching services.