Sustainable urban freight transport adopting public transport-based crowdshipping for B2C deliveries

Cities are experiencing a fast-rising demand for mobility linked both to urban and logistics sprawl. Furthermore, e-commerce, and especially B2C, generates a strong demand for home delivery services provoking, in turn, both social and environmental costs increase [48]. Proposing solutions that might contribute to improving the environmental sustainability of “last mile” delivery is fundamental especially due to the e-commerce-related trips, which, together with urbanization, is increasing substantially [1].

Crowdshipping, an innovative delivery model could, at least in principle, stimulate a better use of currently unexploited transport capacity thus reducing transport costs and emissions [9]. Crowdshipping is a sharing mobility service and implies delivering goods using the crowd [35]. Paloheimo et al. [40] question its ability to reduce congestion and polluting emissions since most of these initiatives rely on dedicated trips performed using private motorized vehicles. Paradoxically the greater the success the higher its social and environmental impact due to a rebound effect that might produce an increase in travel times and fuel consumption. Therefore, when proposing crowdshipping as a possible solution one should test its environmental and social friendliness rather than assume it.

This paper investigates a crowdshipping service implementation that circumvents all these considerations on environmental issues, together with other relevant externalities (e.g. fewer traffic accidents, less land use, etc.), by relying on the use of non-dedicated public transport trips. In fact, the least polluting trip is the one that does not take place. In our case, the service environmental friendliness is ensured ex-ante since it does not imply performing any additional trip and those that are taking place make use one of the least polluting modes. The paper focuses on commuters using the metro rather than the bus since the former is typically more frequent and reliable than the second thus making an additional stop/detour more easily acceptable also for low compensations which are typical for last mile deliveries.

The paper aims at testing under which conditions crowdshippers will produce the service and the customers buy it. The configuration of the delivery service, that can be produced/bought assuming crowdshippers use the metro during their regular home-to-work trips, is crucial to detect the feasibility of this form of crowdshipping. The paper sheds light on the necessary conditions to involve commuters, already using public transport and thus not producing additional trips, in delivering freight within the city in a timely and efficient manner. It is important to note that the type of goods to be transported is a critical issue, especially for green crowdshipping. In fact, size and weight can reduce people availability of acting as crowdshippers, while it is plausible that consumers would be less inclined to receive valuables or personal goods through the crowd using public transportation. However, this is not necessarily an overwhelming obstacle. In fact, one can still expect a large number of deliveries where parcels are compatible with green crowdshipping and, moreover, it is not intended to be the solution, but it should help the logistics system to deal with the requirements of the ‘on demand economy’.

European urban freight transport initiatives and actions have been promoted in the last years to promote research to identify innovative and effective solutions [25]. Policy interventions may produce unintended results also depending on the decision-making process adopted [33]. Participatory planning based on a living lab approach [16] or collaborative governance models [31] can help obtaining the desired results. As a first step, it is fundamental to perform ex-ante behavioral analyses focusing on policy/solution acceptability, taking into account stakeholders’ preferences, in order to provide policy-makers with an efficient and effective decision-support system [17, 23, 24].

The innovative crowdshipping concept is analyzed using the city of Rome as a case study where: i) almost 3 million people live, ii) citizens perform around 700,000 thousand trips during the morning peak, iii) congestion causes about 135 million hours lost, iv) operators employ in the city center approximately 32,700 vehicles daily to perform more than 35,000 loading and unloading operations [11].

Crowdshipping, also referred to as crowd logistics, crowdsourced delivery, cargo hitching or collaborative logistics, uses free capacity available in various transport modes to perform deliveries [10]. Most of the literature on this subject has focused on the various business models developed [43] while little is known about the behavior and perception of crowdshipping users and buyers.

Savelsbergh and Van Woensel [44] discuss the case of Walmart which contemplate the involvement of its in-store customers (occasional drivers) to deliver items that online customers order. It aims at guaranteeing same-day delivery to online customers and uses company drivers only in addition to perform occasional unforeseen deliveries. From a city logistics perspective, offering such a service is likely to increase the number of freight movements further, making coordination and consolidation of direct-to-consumer deliveries even more challenging. Furthermore, Mak [26] studies a novel crowdshipping service strategy foreseeing the use of in-store customers to deliver orders to online customers, looking at its potential impact on retailers’ operational and marketing strategies. However, this study focuses on economic aspects without delving on its environmental consequences.

Slabinac [46] illustrates the DHL case, where the German logistics service provider developed a crowdshipping service in Sweden. DHL involved, through monetary incentives, individuals going home to provide last mile delivery services from a retailer to a shopper’s home. The participation was organized via the use of an ad hoc-developed mobile application. Amazon also explored similar settings to those used by Walmart and DHL for the provision of crowdshipping services [6].

The recent increase in the number crowdshipping initiatives around the world has attracted the attention also of academic researchers [32] that started to investigate this subject focusing on both the service characteristics as well as on the related optimization problems. For example, Archetti et al. [2] formulated the Vehicle Routing Problem with Occasional Drivers representing Walmart’s vision of the vehicle routing problem. In fact, Walmart uses a fleet of capacitated vehicles and drivers to perform deliveries, as well as occasional drivers. The company seeks to minimize the costs of satisfying its customers’ demand (i.e., perform all deliveries). The paper addresses the problem using a multi-start heuristic procedure. The results indicate that employing occasional drivers may produce significant benefits especially if coupled with an appropriate compensation scheme.

Behrend and Meisel [4] analyze a platform combining shipping requests with community members’ planned trips through the development of mathematical and heuristic models with the aim of maximizing profits. The results quantify the benefit of integrating item-sharing (i.e. sharing of tools or leisure equipment between community’s members) and crowdshipping as a function of crowdshippers’ detour flexibility and compensations. The opportunity to combine limited flexibilities in a neighborhood delivery turns out to be additionally beneficial, as it provides consumers with access to items that are not reachable through the other modes. Due to these findings, the authors conclude that an integration of item-sharing and crowdshipping has the potential to push collaborative consumption by delivering through the crowd.

Yildiz and Savelsbergh [50] deal with crowd-sourced transportation for on-demand meal delivery. They investigate the interaction between courier satisfaction (i.e., couriers’ willingness-to-wait), courier compliance (i.e., couriers’ offer acceptance-probability) and profit.

Marcucci et al. [30] investigate the necessary pre-requisites a crowdshipping service needs to satisfy so to be attractive within the highly competitive urban freight delivery market. The exploratory investigation focuses on 200 university students in the city of Rome. While 87% of the students stated their willingness to operate as crowdshippers, the percentage decreases with the increase of the delivery box dimensions and with the decrease of the remuneration. As it is for the demand side, 93% would accept to receive goods via a crowdshipping service. However, the percentage drastically falls if: i) the customers cannot contact the crowdshipping company, ii) there is no direct contact with the crowdshipper or no package tracking is possible/available. The investigation also discovered that the maximum possible deviation from the usual path followed ranges from 1.5 km, in case the crowdshipper is using a non-motorized mode of transport, to 3.1 km for private transport.

Punel and Stathopoulos [42], using stated choice experiments, investigate the factors influencing the acceptability and preferences for crowdshipping. They provide some insights into the attributes affecting preferences for goods delivery performed via occasional drivers. The paper investigates delivery scenarios performed by non-professional shippers, compares them to traditional shipping options and evaluates them assuming a service user’s point of view (i.e. demand side). In a recent study Punel et al. [41] analyze how and to what extent attitudes, preferences, and characteristics of crowdshipping users differ from non-users. The main results show that crowdshipping is more prevalent among young people, men, and full-time employed individuals and that urban areas are preferential for the development of crowdshipping service. The green attitude is also relevant in the choice process, in fact, the individuals who have a strong sense of community and environmental concern are more likely to use crowdshipping services.

Buldeo Rai et al. [10] systematically analyze a set of 42 papers and interview 11 logistics practitioners in order to reconstruct the state of practice of crowdshipping. The study suggests that three characteristics substantially affect the sustainability of such an initiative: third-party involvement (i.e. the involvement of professional third parties in case of insufficient or underperforming crowd), crowd motivation (i.e. financial stimulus or sustainability and community matters) and its modal choice. Specifically, the modal choice is important from an environmental perspective, especially when favoring soft modes, public transportation, and clean vehicles.

The present paper innovates with respect to the literature since it jointly investigates crowdshipping demand and supply, focusing on people preferences, while also considering a “green” service based on the use of public transportation.

The paper uses stated preference (SP) exercises to identify the most important features associated with the choice of acting as a crowdshipper (i.e. service supply side) and with the willingness to use the crowdshipping service (i.e. service demand side). The study investigates agent’s behavior using discrete choice models.

The SP scenarios are tailored to the city of Rome, used as a case study, and its metro network for B2C deliveries. The paper studies a hypothetical scenario where small packages can be picked-up/dropped-off in Automated Parcel Lockers (APLs) located either inside metro stations or in their surroundings.

The following section describes the data acquired via the survey.

3.1 Survey and data description

The administration of two specific surveys during October 2017 – one to 240 inhabitants of the city of Rome (demand-side survey) and the other one to 240 of its metro users (supply-side survey) – produced the data used in this study (for more detail, see also [45]). A convenient sampling strategy was adopted and approximately two-thirds of the interviews were performed via social media and e-mail, while one third “face-to-face”.¹

Four sections compose the questionnaire:

1. 1.

   Socio-demographic: collects gender, age, level of education, employment and behavioral variables useful to detect a green attitude (GA).

    
2. 2.

   Travel reconstruction: investigates the characteristics of the main home-based trip. Origin/destination points and travel times are common information collected in both the surveys. The supply-side survey collects data about the metro stations used, access/egress mode to/from metro stations, as well as public transport subscription.

    
3. 3.

   Hypothetical crowdshipping configurations evaluated through stated preferences scenarios:

    

• The supply-side survey investigates the role APLs location, remuneration, delivery booking, and alternative bank crediting modes (Table 1) play in stimulating people to act as crowdshippers. Remuneration was set considering current shipping costs in the B2C market and the rates applied by existing national crowdshipping companies; the “delivery booking” feature is representative of the high or low crowdshipper’s flexibility to react to an online delivery request.

• The demand-side survey explores the role of service time and cost, parcel tracking availability, delivery schedule date/time flexibility (Table 1) in stimulating potential e-commerce users to choose a crowdshipping service for delivering the goods they bought. Shipping fees and time refer to current national shipping companies operating in Italy.

1. 4.

   Additional behavioral statements: this section aims at discovering further service characteristics/perceptions as well as specific issues (e.g. willingness to deviate from the usual path) for the potential crowdshipper (leave the usual metro stations or moving outside them), or the most preferred time slot to pick-up the parcel for the potential crowdshipping buyer.

    

Demand-side survey		Supply-side survey
Features	Levels	Features	Levels
Shipping fee (with respect to current national shipping companies)	● Lower (+ 1)	Location of APL	● Inside metro stations (+ 1)
	● Typical (− 1)		● Outside metro stations/adjacent buildings (− 1)
Shipping time (with respect to current national shipping companies)	● Lower (+ 1)	Remuneration	● 3 €/delivery (+ 1)
	● Typical (− 1)		● 1 €/delivery (− 1)
Parcel tracking	● Available (+ 1)	Delivery booking	● Real-time booking (+ 1)
	● Not available (−1)		● Off-line booking (− 1)
Delivery date and Time schedule flexibility	● Yes (+ 1)	Bank crediting modes	● Single delivery (+ 1)
	● No (−1)		● Every 5 deliveries (− 1)

Following Gatta and Marcucci [15], who suggest using advanced experimental design techniques when performing SP exercises in the urban freight transport field, we use a Bayesian D-Optimality efficient design to define the choice tasks in both surveys. The results of a pilot survey (i.e. utility-neutral design, see [51]) were used as an input to develop the final choice tasks. D-optimality aims at minimizing the determinant of the covariance matrix of the model coefficients and maximize the expected value of the chosen alternatives [21].

The optimal design produced four different questionnaire blocks in both surveys, each including three choice’s exercises with three alternatives: two unlabeled linked to a crowdshipping service (i.e. option A and B) and a “no choice” option. The “no choice” alternative represents the status quo implying not using the crowdshipping service (for the demand-side) or acting as a crowdshipper (for the supply-side). Notwithstanding the use of the “no choice” option could also lead to a serious reduction in the information gathered [12], we prefer to include it so to increase both the realism of choice tasks and the robustness of respondents’ estimated preferences [3]. However, less than 10% of the sample selected the “no choice” option in both surveys.

Table 2 reports a summary of respondents’ socio-demographic characteristics. Although the sample does not perfectly reflect census data distribution for the entire population of Rome, the high number of young respondents moving for study/work represents an added value given the specific purpose of the study. In fact, young people are typically keener to use crowdshipping platforms (Briffaz and Darvey, [8]). Indeed, some studies suggest that web surveys have high rates of technologically advanced respondents [22].

	Demand-side survey		Supply-side survey		Entire Pop [%]
	Sample size	Sample [%]	Sample size	Sample [%]
Gendera
Man	112	48.9	105	46.7	46.7
Woman	117	51.1	120	53.3	53.3
Agea
16–30	102	44.5	95	42.2	16.5
31–44	49	21.4	67	29.8	23.0
45–58	50	21.8	40	17.8	26.5
> 58	28	12.2	23	10.2	34.0
Level of educationb
Primary school diploma	4	1.7	2	0.9	16.9
Middle School diploma	15	6.6	12	5.3	28.0
High school diploma	113	49.3	146	64.9	36.5
University degree (any level)	82	35.8	55	24.4	15.1
Post university degree	15	6.6	10	4.4	3.6
Professional statusb
Employed	88	38.4	104	46.2	47.9
Student	66	28.8	70	31.1	7.7
Retired	14	6.1	12	5.3	22.4
Working student	14	6.1	5	2.2	–
Housewife (no male)	24	10.5	21	9.3	10.5
jobless	23	10.0	13	5.8	11.5
Income
< 5.000 €	66	32.7	73	35.3
€ 5000 - € 10,000	36	17.8	27	13.0
€ 10,000 - € 20,000	53	26.2	66	31.9
€ 20,000 - € 30,000	29	14.4	28	13.5
€ 30,000 - € 40,000	11	5.4	11	5.3
€ 50,000 - € 80,000	6	3.0	1	0.5
> 80.000 €	1	0.5	1	0.5

^aResident population 1st January 2017, age group 16–99 (ISTAT data – http://demo.istat.it/)

^b2011 Census, age group 16–99 (ISTAT data - http://datiopen.istat.it/)

The supply-side respondents state that the home-based trips involving the use of the metro are mainly taking place in the morning peak (79%). A similar frequency for the work-to-home trips in the afternoon (84%) is linked to the survey taking place during regular weekdays. Access/egress to/from metro stations usually takes place by walking (52.9%) and 50 min is the average travel time associated to the entire origin-destination trip. Interestingly, 75% of the demand-side respondents pass close to, at least, one metro station during his/her home-based trip.

Most of the potential users of the crowdshipping service (Fig. 1) declared to prefer to pick-up the parcel during the afternoon (38%) or evening (33%) and to have the withdrawal option available at least for 24 h (44%). Only 9% of the respondents declared to prefer having a short withdrawal time (less than 3 h), mainly for safety reasons.

With respect to the maximum deviation from the usual path, it is interesting to note that about half of the potential crowdshippers (43.1%) is not willing to modify the path if the APL is outside the metro stations while 39.2% would accept to deviate the path for a maximum of 300 m. Only 15.3% is willing to travel an additional distance of 600 m, while the percentage of those willing to travel more than 600 m is negligible.

The measure of GA relies on four qualitative levels (no GA, low, medium, high), eliciting a self-stated weight of individual-specific pro-environmental behavioral attitudes. These attitudes are considered a proxy for the interest in/concern about environmental problems. This, in fact, should also correlate to the adoption of sustainable transport modes or the consumption of organic products. Each attitude has been assessed through a Likert scale during the survey: 68% of the respondents self-stated a medium level of green attitude (Fig. 2, a).

Finally, respondents were asked to express a judgment on the chances of success of the crowdshipping service both in urban and suburban areas, under the hypothesis of expanding it using APLs located in the main metro and urban rail network city stations. In the urban case, a substantial part of the sample (46%) foresees a successful service while 41% are not sure. As for the suburban area, the percentage of respondents confident in the success of the service decreases to 25% (Fig. 2, b).

Crowdshipping provides a good delivery service in the last mile. It is particularly suitable for e-commerce and it is progressively developed and adopted worldwide. Crowdshipping is both innovative and with a great growth potential, both economic and environmental. Its success much depends on organizational and performance issues. Standard crowdshipping, using dedicated motorized trips, might squander most of the potential economic/environmental benefits, also considering the strong rebound effects that might arise from the success of the initiative.

The paper proposes two SP surveys, in the city of Rome, and studies “green” crowdshipping potential demand and supply. In particular, it explores under which conditions commuters, moving from home to work and vice-versa, would possibly act as crowdshippers, investigating specific population segments that are using the metro for their systematic trips. The underlying belief is that the most efficient crowdshipping service is the one based on non-dedicated trips.

Model results show that APLs location is the most relevant feature even more important than remuneration. Additionally, the paper analyses the demand side by investigating how potential e-commerce users can be stimulated in choosing a crowdshipping service for delivering the goods they bought. Results show that the possibility to plan the delivery date and its time schedule has the highest impact on consumers’ utility.

Moreover, comparing demand and supply results, a scenario analysis is performed given different service configurations. Preliminary results are comforting since the estimated number of potential crowdshippers in Rome is higher than those that would satisfy the potential demand.

The main strength is the identification of the most important instruments needed to develop a crowdshipping service capable of relying on a sufficiently large base of crowdshippers so to provide a reliable and low-cost solution for last-mile deliveries. The results will be made available to Rome Mobility Agency who is interested in this topic and willing to explore/test alternative service configurations in a real-life pilot study should financial resources be procured.

The likely increase of small-volume/ low-weight/ parcel market in the near future, due to the diffusion of e-commerce, and the recent opening of the C-metro line in Rome suggest investigating this crowdshipping configuration further.

Future research will focus on: i) increasing the sample size, ii) acquiring additional preference information (e.g. frequency with which agents are available to work as crowdshippers, freight type impact on agents’ choice, etc.), iii) adopting more advanced estimation models (e.g. latent classes) and calculating willingness to pay/accept measures to possibly develop a pre-business pilot study; iv) quantifying the environmental effects obtainable from the development of crowdshipping services based on public transport use (e.g. environmental pollutants, noise abatement), v) testing financial viability, vi) linking regular logistics chains with the crowdshipping service envisioned, vii) defining necessary pre-requisites for crowdshipping service integration, viii) evaluating the potential effects of associating green-label recognition to public transport crowdshipping, ix) designing logistics networks to optimize network structure and nodes location, x) investigating further social implications on wages and labor force employed in the logistics sector.