Modelling carrier type and vehicle type choice of small and medium size firms

Highlights

• We develop carrier and vehicle type choice models using discrete choice methods.

• A Commercial Travel Survey of shippers is used to develop the models.

• Firm size/industry/shipment type affect the choice of carrier and vehicle type.

• Firms are more likely to substitute between vehicle types before carrier types.

Abstract

The objective of this paper is to study freight transportation outsourcing and vehicle type choice decisions made by small and medium size firms. Carrier type and vehicle type choices are modelled independently using multinomial logit models and jointly using nested logit models. Mixed logit models are also developed for the defined structures to study the choice correlation among the choices made by the same firm, resulting in development of six different models. A commercial travel survey for the Greater Toronto and Hamilton Area is used to estimate the models. The models consider characteristics of the firm (employment, industry classification) and shipment (location, type). The results show significant improvement for the carrier type choice model using a mixing distribution while the vehicle type model performs slightly better using mixed logit model than using multinomial logit model structure. The nested logit model results show that firms are more likely to substitute vehicle type choices before substituting carrier type choices. The correlation of vehicle type choices is found to be highest (96%) among own-account carriers and driver-own carriers (90%) and least for the vehicles owned by a for-hire company (82%). Similar to the mixed vehicle type choice model, the mixed nested logit model also shows a slight improvement in model performance. Based on the strong correlation indicated by the nested logit model, a model reflecting the joint choice of carrier type and vehicle type is suggested.

Introduction

The outsourcing decision of freight transportation and the choice of vehicle type for the delivery of shipments are among the key decisions made by a firm. These decisions are part of logistics management which also includes storage, consolidation, and deconsolidation of goods. A firm can manage its own logistics functions, own another subsidiary for logistics management or outsource logistics to another firm (Razzaque and Sheng, 1998). The decision to outsource logistics is driven by efforts to reduce cost and maximize profit, and to allow the firm to focus on its core competency. However, a disadvantage of outsourcing is a loss of control of the delivery process, including contact with customers (Deepen, 2007).

Carriers are companies that are responsible for the transportation of goods. Firm-own/own-account carriers are firms that transport their own goods while the firms that provide transportation services are known as for-hire carriers (Holguín-Veras and Jaller, 2014). A firm’s decision of whether to use their own vehicles or a for-hire carrier could affect operations such as customer service and operating expenses, and depends upon factors such as geographic coverage, shipment size, and time-definite pickup and delivery services (Regan and Garrido, 2002). Firms can also adapt a hybrid approach where some logistics services are outsourced and others are performed internally (Reeves et al., 2010). Reeves et al. (2010) studies the role of internal versus external decision-makers in logistics decisions, making for potential sub-optimal decisions, within the automotive industry. However, the results are inconclusive, and the authors find little evidence of differences in decision-making by the two types of actors. Wallenburg, (2009) analyses logistics outsourcing relationships and finds that proactive cost and performance improvement affects customer loyalty. However, the degree of importance of the two measures depends upon the complexity of the service and the contracting period.

Distribution channels could also affect this choice as for-hire carriers are usually involved in direct shipments from manufacturer to end-consumer (Nuzzolo et al., 2013). Rowell et al. (2014) analyse factors that affect the choice of for-hire and private carrier types based on the survey results conducted in Oregon and Washington State. The results show that carrier type depends upon the truck size and is independent of the commodity type (Rowell et al., 2014). A survey of manufacturing companies in the metropolitan area of Turin reveals that outsourcing of transportation is a common practice especially for firms that supply a finished product to clients. Own-account transport has a minor share and is used mainly for intra-firm transport, local distribution, and short-distance transport services. Vehicle types also influence this decision as the cost of vehicle ownership is one of the reasons for outsourcing transport (Bertalero et al., 2020). The outsourcing decision also depends upon the type of shipper. Patterson et al. (2010) estimate a random-effects mixed-logit model to study differences in behaviour of third-party logistics companies (3PLs) and other shippers in the choice of carriers using a stated preference survey of shippers. The type of vehicle is one of the attributes of the carrier assessed by Patterson et al. (2010). The study finds that both types of shippers have different behaviours which should be considered while modelling mode and carrier choice (Patterson et al., 2010).

Mostafa and Roorda (2017) estimate discrete choice models for outsourcing of freight logistics and goods production for Canadian manufacturers. Binary logit models are developed for the independent choice while multinomial logit models are developed for joint simultaneous choice of outsourcing logistics and goods production. Factors related to firm characteristics and economic environment such as market competition and GDP growth are tested. The results suggest that freight activities are more likely to be outsourced by firms with focus on core activities (Mostafa and Roorda, 2017). Khakdaman et al. (2020) estimate discrete choice models to investigate the shift of mode choice decisions from shippers to logistics service providers. Data from a stated preference survey of global shippers are used to estimate the models. The study does not model the choice of mode but rather models the choice between the two synchromodal services in addition to the current service. The attribute of control over transportation mode and route is one of the key attributes of the synchromodal services. The study finds that the provision of better services and lower costs could result in the willingness of the shippers to give control over transportation modes and routes to logistics service providers.

The choice of mode or vehicle type is another critical component of the freight supply chain. Freight mode/vehicle type choice is a key part of freight demand models. De Jong provides a review of freight mode choice models (de Jong, 2014). Freight mode/vehicle type choices have been modelled independently (Keya et al., 2018, Samimi et al., 2011) and jointly with shipment size (Cavalcante and Roorda, 2010, Irannezhad et al., 2017, Keya et al., 2019, Pourabdollahi et al., 2013) because these two choices are correlated to each other. Shippers and carriers interact with each other to determine the mode choice (Holguín-Veras et al., 2011).

While freight mode choice is a crucial part of typical freight demand models, the choice of carrier type is less commonly modelled. Carrier type choice is modelled explicitly in some models while others consider the choice of carrier type but do not model it explicitly. Wisetjindawat and Sano (2003) estimate choice of carrier type and vehicle type as part of the micro-simulation freight demand model for Tokyo Metropolitan Area. The choice of carrier type between private truck and business truck, and choice of vehicle type between large truck and small truck is modelled jointly using a nested logit modelling framework. Carrier type choice is nested within vehicle type choice (Wisetjindawat and Sano, 2003). The commercial transport module of the second generation StateWide Integrated Model (SWIM2) of Oregon State also models carrier and vehicle choices. Carrier type is defined as private and for-hire and vehicle type is defined as single and multi-unit truck. Carrier type and vehicle type are assigned using a Monte Carlo process (Donnelly et al., 2017). Nuzzolo and Comi (2014) model carrier choice in combination with freight type as retailer on own account, retailer by third party, and same categories for wholesaler. Vehicle type is modelled in the subsequent modules with categories such as car, light and medium goods vehicle. Discrete choice methods are used for model estimation (Nuzzolo and Comi, 2014). The modelling framework SimMobility Freight assigns carriers for shipment transport. Carrier type choice is not modelled explicitly but rather carriers are assigned based on the vehicle availability of shipper and receiver firm. In case vehicles are not available at either facility, a multinomial logit model selects the carrier for each supplier (Sakai et al., 2020).

This study focuses on small and medium size shipper firms in the Greater Toronto and Hamilton Area (GTHA) and models their choice of carrier type and vehicle type. Data of shipper firms obtained from the 2012 Commercial Travel Survey are used to estimate a series of discrete choice models. We hypothesize that the shipping firm influences the choice of vehicle even when outsourcing freight transportation. While this hypothesis is obvious if the shipment is delivered by the firm using its own vehicles, it is also reasonable if outsourcing transportation to a for-hire carrier. This is because one of the primary reasons of the outsourcing decision could be that the transportation of a shipment requires a vehicle type not owned by the firm. This leads to another hypothesis that carrier type and vehicle type choices are correlated. To test whether shipper firms are more likely to substitute vehicle types or carrier types, the authors also estimate nested logit models for the joint choice of carrier and vehicle type. This contributes to the body of literature where vehicle type and shipments size choices are modelled jointly. Mixed logit models are also estimated to study the correlation among choices made by same firm. The estimated models have application in agent-based microsimulation models such as the framework presented by Roorda et al., (2010) with application in the GTHA. Moreover, the estimated models could be used to analyse policies related to the choice of vehicles used for shipment transportation. The models could also be used to understand the behaviour of carrier type selection of firms within GTHA.