The Last Mile: Managing Driver Helper Dispatching for Package Delivery Services

• Published date: 01 September 2020
• Author: Toyin Clottey,Yoshinori Suzuki,Shih‐Hao Lu
• Date: 01 September 2020
• DOI: http://doi.org/10.1111/jbl.12242

The Last Mile: Managing Driver Helper Dispatching for Package

Delivery Services

Shih-Hao Lu

1

, Yoshinori Suzuki

2

, and Toyin Clottey

2

1

National Taiwan University of Science and Technology

2

Iowa State University

Hiring seasonal driver helpers is one widely used approach by parcel delivery companies to deal with increased home-delivery volumes during

peak seasons. Nonetheless, driver helper-related issues have not received much attention in academic research. This study investigates how

driver helpers can be utilized in the most effective way by parcel delivery companies. We show that by separating a parcel delivery route into two

sub-routes, namely “no-helper”and “with-helper”routes, the utilization rates of driver helpers can be improved, and the carrier cost be reduced con-

siderably. Three main contributions of this study are as follows. First, based on costtrade-off insights, we develop a new mathematical model for the

last-mile distribution problem, which we call the Driver Helper Dispatching Problem. Second, using this mathematical model, we perform a series

of numerical experiments to identify the conditions under which the proposed “split-route solution”works most or least effectively. Finally, we per-

form sensitivity analyses to investigate the impact that changes in model parameters, such as fuel cost, would have on study results.

Keywords: last-mile distribution; driver helper; traveling salesman problem; transportation

INTRODUCTION

According to the National Retail Federation (2017), the winter

holiday shopping season from Black Friday to Christmas

accounts for 20%–25% of the annual retail sales in the United

States. E-commerce continues to grow, with many customers pre-

ferring the ease and convenience of home delivery (Statista

2016). Shipments during the holiday season have always been a

logistical challenge, but this problem has increased in importance

recently due to the growth of e-commerce. Major challenges

include last-minute shoppers, higher than expected e-commerce

volume, retailers with increasingly later delivery cutoff times,

and harsh winter weather (Soltes, 2014).

Shipment delays during peak shopping seasons have become

common in the past few years, especially among international

parcel delivery companies such as United Parcel Service (UPS),

Federal Express (FedEx), and Deutsche Post (DHL). For exam-

ple, during the Cyber Monday shopping period in 2017, only

89% of packages were delivered on time by UPS, and for pack-

ages that had been delayed, the average delay was two days

(Bhattarai, 2017). Given these facts, international parcel compa-

nies, who service millions of customers each day, are looking for

better (more efficient) ways to distribute parcels across their

logistics networks. This issue, that is, how to improve the effi-

ciency of parcel delivery operations, is an important problem not

only for parcel delivery companies, but also for many manufac-

turing companies. This is because last-mile logistics is known to

be one of the most expensive and least efficient segments of a

supply chain (Gevaers et al., 2011; Castillo et al., 2017), so that

improving the efficiency of this particular segment can eliminate

the bottlenecks and reduce costs in supply chains.

One approach that is increasingly used by these large parcel

delivery companies is to hire seasonal driver helpers during peak

seasons, who work with drivers to deliver packages, in order to

handle the increased home-delivery volumes (Soltes, 2014). This

approach seems to be working well for parcel delivery compa-

nies to minimize shipment delays. However, the approach that is

currently implemented in the field is not cost efficient. This paper

studies the driver helper dispatching problem of large parcel

delivery companies in an attempt to obtain insights into how dri-

ver helpers can be utilized in the most effective way to minimize

cost. The motivation of this study is the provision of an alterna-

tive approach for driver helper dispatching practices to improve

the cost and operational efficiency of parcel delivery companies,

so that they can better deal with the aforementioned logistical

challenges, especially during peak seasons.

STATE OF THE INDUSTRY AND STUDY SCOPE

During the holiday season, the need for drivers to work overtime is

an industry norm (Bhattarai, 2017). Working overtime, however,

has many drawbacks. First, based on the Fair Labor Standards Act

(FLSA), U.S. employees must receive overtime payment for hours

worked over 40 in a given workweek at a rate not <150% of their

regular rate. Thus, letting drivers work overtime increases direct

labor costs. Second, even with overtime payments, not every driver

is willing or able to work overtime. The unpleasant employment

condition, which requires overtime work, can not only put pressure

on the drivers’physical or mental well-being, but also increase the

turnover rate of drivers (e.g., Suzuki et al., 2009; Miller et al.,

2017). Third, there is a physical limitation, or upper bound, on the

drivers’daily work time (hours of service regulation) (Miller et al.,

2018; Miller et al., 2019).

Hiring seasonal employees is one widely used approach by

parcel delivery companies to deal with this challenge. For exam-

ple, in 2015, UPS and FedEx announced a plan to hire 95,000

and 55,000 seasonal employees, respectively, to support the

anticipated increase in package volume from November through

January (Schlangenstein 2015). In 2017, UPS again announced

plans to hire 95,000 temporary workers for November 2017

Corresponding author: Shih-Hao Lu, National Taiwan University of

Science and Technology, No.43, Keelung Rd., Sec.4, Da'an Dist.,

Taipei City 10607, Taiwan; E-mail: shlu@mail.ntust.edu.tw

Journal of Business Logistics, 2020, 41(3): 206–221 doi: 10.1111/jbl.12242

© 2020 Council of Supply Chain Management Professionals

through January 2018 (Bhattarai, 2017). A significant number of

these seasonal employees work as driver helpers, who assist dri-

vers in the delivery of packages. Driver helpers are not required

to drive vehicles, and they usually meet drivers at a mutually

agreed time and location (e.g., 7 am at the depot). Self-reported

salary data on Glassdoor.Com (2018) and our discussions with

industry experts (managers from UPS) revealed that the average

drivers’pay rate in the United States is about two to three times

higher than the average helpers’pay rate. Therefore, the strategy

of using driver helpers provides cost savings to parcel delivery

companies by replacing the drivers’work time (e.g., overtime)

with that of the helpers’.

To date, parcel delivery companies have primarily utilized two

types of driver helpers (Rhodes et al., 2007). Dependent driver

helpers, who travel with delivery trucks, are the first type. Such

helpers work together with (or sometimes separately from) the

drivers to deliver packages. Independent driver helpers are the

second type. These helpers do not travel with delivery trucks,

but rather work at a predetermined point that typically covers a

high customer density area (e.g., a small area with a large num-

ber of apartments or business offices). Typically, an independent

helper will deliver all of the packages to customers by using

bicycles or other tools such as a hand truck.

There is no question that the use of driver helpers is a valu-

able cost-saving technique for parcel delivery companies. How-

ever, based on our interviews with several practitioners

(including two UPS senior managers, namely, the Vice President

of Engineering and the Director of Engineering), we believe that

there is room for improvement in the way the parcel delivery

companies are currently using driver helpers. In this paper, we

develop a new approach for parcel delivery companies that can

better manage the cost of dependent driver helpers.

There are three reasons why we study dependent (rather than

independent) helpers. First, while dependent helpers travel with

vehicles, independent helpers do not, which suggests that the

problem that deals with dependent helpers is more complex, but

may provide higher cost-saving opportunities, than the problem

which deals with independent helpers. Second, independent help-

ers may require special equipment such as trikes (a specialized

tricycle with a basket in the rear) or additional trailers (for on-

site storage), which must be provided by parcel delivery firms

(Rhodes et al., 2007). This means that the use of independent

helpers may be more costly, and thereby less attractive, to parcel

delivery firms than the use of dependent helpers. Third, a driver

has less oversight of an independent helper than a dependent

helper who moves with the driver.

According to our interviews with practitioners, the current

practice for using dependent helpers in industry follows a simple

solution procedure; that is, the routing problem is solved by min-

imizing the travel distance, assuming that in each route the driver

helper will accompany the driver from the start to the end. This

shortest-route approach, however, has limitations because it

forces driver helpers to visit locations (nodes) that have only a

single customer, which is inefficient (helpers are most useful at

nodes with multiple customers). To obtain better solutions, we

propose an approach that (1) relaxes the shortest-route assump-

tion and (2) reduces, in addition to the truck’s travel cost, the

cost of servicing customers (driver and helper costs combined) at

each node.

LITERATURE ON LAST-MILE PROBLEMS

The problem this paper considers, which we call the Driver

Helper Dispatching Problem (DHDP), is a type of last-mile logis-

tics problem. Last-mile logistics refers to the last portion of tran-

sit in supply chains, where goods are delivered from the last

transit point to the final drop point (Lee and Whang 2001; Esper

et al., 2003; Boyer et al., 2009; Wang and Odoni 2016). This

part of logistics often involves routing a fleet of vehicles for

physical distribution and plays a crucial role in ensuring that the

products are delivered to customers in correct quantities and

within time limits.

Although the literature on last-mile logistics can be split into

two main bodies, namely freight and passenger logistics, our

research focuses on the former. Readers that are interested in the

latter are referred to Wang and Odoni (2016). Issues that have

been investigated in last-mile freight delivery include examining

fulfillment strategies (e.g., Lee and Whang 2001; Punakivi et al.,

2001); delivery options (Esper et al., 2003); effects of customer

demand characteristics (e.g., Boyer et al., 2009; Song et al.,

2009); crowd-sourced deliveries (e.g., Castillo et al., 2017);

effects on the environment (e.g., Brown and Guiffrida 2014; Liu

et al., 2019); and humanitarian aid (e.g., Balcik et al., 2008). Our

study is geared towards optimizing the process by which parcel

delivery companies use driver helpers, an issue that has received

scant attention.

To the best of our knowledge, Rhodes et al. (2007) is the only

previous study that has focused on driver helper decisions. They

proposed a dispatch tool for UPS’s delivery system that finds the

optimal number of helpers based on cost–benefit analyses. In

Rhodes et al. (2007), the authors focused on answering the fol-

lowing three questions: (1) how many helpers should be

deployed, (2) which routes always need dependent helpers, and

(3) which locations always need independent helpers during the

holiday season. Notice that these are all strategic-level questions.

In contrast, this study extends the scope by developing an opera-

tional level problem of finding the optimal vehicle routing and

scheduling decisions based on the cost of operating both drivers

and helpers, as well as that of operating a vehicle, which can fur-

ther improve cost savings.

Though the literature on driver helpers is limited, there are

several types of studies that are relevant to this study. The first is

those that studied the vehicle routing problem with time windows

(VRPTW). This type is relevant to our study because parcel

delivery service is often considered as a time-constrained routing

problem. Since VRPTW is difficult to solve (NP-hard), many

studies have focused on developing efficient solution techniques.

They include J€

ornsten et al. (1986) (Lagrangian relaxation),

Kolen et al. (1987) (dynamic programming), Desrochers et al.

(1992) (column generation), Fisher (1994) (K-tree), and Errico

et al. (2016) (a-priori optimization). These studies, however,

mostly used “hard”time window constraints. In practice, time

windows may take the form of “soft”constraints (which can

sometimes be violated), as Dondo and Cerd

a (2007) point out.

This may be particularly true for package delivery services.

Specifically, an article written by UPS executives (Holland et al.

2017), as well as our interviews with UPS managers, confirmed

that UPS is using soft, rather than hard, constraints to specify

time windows. In fact, our interviews suggest that, for many

Driver Helper Dispatching Problem 207