Inventory record inaccuracy dynamics and the role of employees within multi‐channel distribution center inventory systems

• Author: Anníbal Camara Sodero,Thomas J. Kull,Mark Barratt
• DOI: http://doi.org/10.1016/j.jom.2018.09.003
• Published date: 01 November 2018
• Date: 01 November 2018

Contents lists available at ScienceDirect

Journal of Operations Management

journal homepage: www.elsevier.com/locate/jom

Inventory record inaccuracy dynamics and the role of employees within

multi-channel distribution center inventory systems

Mark Barratt

a,∗

, Thomas J. Kull

b

, Anníbal Camara Sodero

c

a

Department of Management, College of Business Administration, Marquette University, Milwaukee, WI, 53233, USA

b

Department of Supply Chain Management, W.P. Carey School of Business, Arizona State University, Tempe, AZ, 82801, USA

c

Department of Supply Chain Management, Sam M. Walton College of Business, University of Arkansas, Fayetteville, AR, 72701, USA

ARTICLE INFO

Accepted by: S. de Treville

Keywords:

Inventory systems

Inventory record inaccuracy

Retail distribution

Case study

System dynamics

Employees

ABSTRACT

The dynamics of inventory record inaccuracy (IRI) remain under-explored in a multi-channel distribution center

(MCDC) setting. Additionally, how employees interact with the rest of an MCDC's inventory system to impact IRI

is yet to be explored. To address these issues, our research adopts a multi-method exploratory approach to

understand the nature and existence of IRI between cycle counts in a multi-channel environment based on

continuous, daily IRI-related observations. These data are then used to develop a system dynamics model to

further explore the interaction between employees and the rest of the inventory system. Our case study finds

substantial levels of IRI on a daily basis, with varying levels of consistency and a bias toward negative errors,

raising concerns as to the suitability and stability of MCDC inventory systems. We also find varying IRI effects

depending on the channel, with inventory density, in particular, having mixed influences. Further, we provide

evidence for substantial volatilities of daily IRI likely generated by employee-system interactions in the MCDC.

Without the daily perspective, such evidence would otherwise have remained unobserved. Our system dynamics

model also provides a novel framework to show how employees interact with MCDC inventory characteristics to

affect the dynamic level and stability of IRI. This research also provides suggestions for managing and further

examining these phenomena.

1. Introduction

Inventory record inaccuracy (IRI), even in low levels, damages a

retailer's operational performance and its reputation with consumers

for product availability

1

(Gallino and Moreno, 2014;Kök and Shang,

2014). Following DeHoratius and Raman (2008), we define IRI as the

degree of system inventory record (SIR) inaccuracy relative to the

physical inventory of a specific item. IRI manifests itself in the form of

positive or negative errors, where there is more or less actual inventory

than recorded (Hardgrave et al., 2013;Kök and Shang, 2014). Cur-

rently, because retailers predominantly have employees perform in-

frequent cycle counting to track IRI, the dynamic (e.g., daily) visibility

of actual inventory on-hand is difficult and IRI dynamics remain un-

known (Kull et al., 2013). In this research, we explore the dynamics of

IRI produced by an inventory system within a retail multi-channel

distribution center (MCDC) setting to identify factors that influence the

stability of the SIR in accurately reflecting actual inventory.

Over the last twenty years, research has mostly considered IRI in the

manufacturing or retail store context (e.g., Brown et al., 2001;

DeHoratius and Raman, 2008), while overlooking the distribution

center context for the existence and impact of inventory record in-

accuracies. With the advent of multiple, Internet-enabled channels to

market, the role of the distribution center has transformed from one of

supporting retail brick and mortar (B&M) stores to one of directly im-

pacting consumer demand and service experience (Hobkirk, 2015;

Ishfac et al., 2016). This new multi-channel-role

2

still requires em-

ployees to ship regular orders to retail stores but also includes deli-

vering online orders either to retail stores for consumer pick-up or di-

rectly to consumers, which increases operational complexity for

employees and managers (Gallino and Moreno, 2014;Hübner et al.,

https://doi.org/10.1016/j.jom.2018.09.003

Received 15 March 2017; Received in revised form 17 September 2018; Accepted 21 September 2018

∗

Corresponding author.

E-mail addresses: mark.barratt@marquette.edu (M. Barratt), Thomas.Kull@asu.edu (T.J. Kull), ASodero@walton.uark.edu (A.C. Sodero).

1

The National Retail Security Survey (2017), reports shrinkage in U.S. retailing increased to $48.8Bn in 2016, with the average shrinkage rate per retailer

increasing to 1.44% of sales.

2

APeerless Research Group report (2014), “Multi-Channel Distribution in the Apparel Industry” suggests that 47% of apparel retailers are utilizing a single MCDC

to support multiple channels. We note that these are industry-specific estimates and that, while the percentages will vary between industries, we have no reason to

believe these are not indicative of the growing prevalence of MCDCs.

Journal of Operations Management 63 (2018) 6–24

Available online 26 October 2018

0272-6963/ © 2018 Elsevier B.V. All rights reserved.

T

2016). In light of this increased complexity, MCDCs are highly sensitive

to IRI, because MCDC personnel need accurate, daily inventory on-hand

information across the range of products and locations to optimize al-

locations to retail or consumer orders (Gallino and Moreno, 2014;Kim

and Lennon, 2011). Despite a history of research (Delen et al., 2007;

Morey and Dittman, 1986), the problem of IRI continues, suggesting

unexplored opportunities remain.

Despite recent consideration of the impact of managerial incentives

and employee staffing levels on the occurrence of IRI in retail stores

(Chuang et al., 2016;DeHoratius and Raman, 2007), the literature on

employee-related aspects of inventory systems is relatively sparse

(Bruccoleri et al., 2014;Grosse et al., 2015). This sparsity is critical,

given the dominant levels of daily interaction that MCDC employees

have with inventory, i.e., product stowing and order picking, compared

to those in a retail store (Delen et al., 2007). The presence of consumers

interacting with products complicates understanding employee aspects

of IRI in a retail store. By contrast, the MCDC is a cleaner environment

without consumers; however, its employees need to cope with different

levels of channel-induced visual complexity.

3

Because the employee-re-

lated aspects of MCDC inventory systems are under-examined, this re-

search explores how employee interactions with the rest of the MCDC

inventory system (referred to as “employee-system interaction”) influ-

ence daily IRI variation. Recognizing the human element in managing

inventory (Fransoo and Wiers, 2006), we define an MCDC inventory

system as the interactions among employees, the Warehouse Management

System (WMS) software and hardware, the physical products and their

layout, and the inventory policies deployed.

We seek to answer two research questions: Considering that the use

of MCDCs is growing (Ishfac et al., 2016), and that channel differences

can elicit useful IRI insights (Kull et al., 2013), A) what inferences about

the emergence of IRI can be drawn from differences in dynamic IRI patterns

across channels? And B) How do employee interactions with an MCDC

inventory system influence the dynamics of IRI?

Considering the general lack of theory present in existing IRI re-

search, we adopt an exploratory, multi-method approach. We use an in-

depth empirical case study to ground a system dynamics simulation

model (Bruccoleri et al., 2014;Chandrasekaran et al., 2016;Chuang

and Oliva, 2016;Forrester, 1961;Repenning and Sterman, 2002;

Sterman et al., 1997), which guides us in following a more holistic and

continuous approach to explore IRI and the inventory system in an

MCDC setting. The case study provides us with daily inventory counts,

observations of the inventory system processes and daily operations,

but also IRI patterns that suggest important employee-system interac-

tions. The simulation then allows us to formalize and explore these

interactions within the dynamics of the holistic inventory system.

Without a simulation, case-based conceptualizations are limited; a si-

mulation helps reveal underlying structures that generate inferences

about the ongoing dynamics of an inventory system, including the

employee-system interactions. Without the case study, we could only

consider the employee-system interactions conceptually. Utilizing both

these approaches enables us to develop a more holistic and continuous

approach for exploring IRI and employee-system interaction in an

MCDC setting. Fig. 1 gives an overview of the research design and

process.

Our investigation finds significant levels of IRI in the form of both

positive and negative errors on both a daily and continuing basis, giving

rise to concerns regarding the suitability and stability of inventory

systems in multi-channel settings. We also find that elements of em-

ployee-system interaction can have significant and varying IRI effects

depending on the channel in which they are working. Our study makes

three primary contributions to inventory management theory. First, we

provide initial evidence of different, significant and even volatile levels

of dynamic IRI across both channels, generated by possible employee-

system interactions that, without the daily perspective, would other-

wise have remained unseen. Second, by considering the direction of

errors, we identify the propensity of the inventory system to under-

estimate the actual inventory on-hand. Finally, our employee-system

interaction perspective offers a different viewpoint compared to the

extant literature, which takes more of a technology-process solution

perspective (e.g., DeHoratius et al., 2008), implying that employees are

more of a hazard than an asset for correcting IRI. Our perspective

provides insights into how employees in an MCDC affect variation in

IRI, resulting from their sensitivity to fulfillment disruption, their levels

of impact (error) awareness, and the visual complexity arising from the

storage of inventory for picking purposes.

In what follows, we first provide an overview of IRI-related em-

ployee-system interactions and consider the previous research ap-

proaches to understanding their impact on IRI and the contexts within

which IRI has been investigated, highlighting their applicability (or lack

of it thereof) to our research context. Next, we present our methodology

that comprises a unique and revelatory, single case study with multiple,

embedded sub-units of analysis. After discussing our empirical results,

we propose and explore a system dynamics model that incorporates

insights from the case study to simulate the employee-system interac-

tions within the MCDC. Finally, we develop a set of propositions re-

lating to the dynamic variability of IRI and the employee-system in-

teractions within an MCDC.

2. Literature review

2.1. IRI drivers: an employee-system interaction perspective

The extant inventory management literature only identifies drivers

of SIR inaccuracy that appear to explain the nature and causes of IRI via

discrete snapshots through cycle counting-based audits or single period

counts (DeHoratius and Raman, 2008). This approach of using single

period/cycle counting data results in the ongoing day-to-day nature of

IRI remaining unknown (Kull et al., 2013). For instance, auditing is a

significant employee-inventory system interaction, and audit frequencies

can contribute to lowering IRI levels (Chuang et al., 2016;Hardgrave

et al., 2013;Neeley, 1983). Indeed, DeHoratius and Raman (2008)

found support for these suggestions and argue the more frequently SKUs

are counted, the lower the IRI levels. Also, studies suggest the more

time between audits, the less likely that employees will correct the

increasing discrepancies (Morey and Dittman, 1986). Less apparent,

however, is the effect that an item's dollar value and dollar volume have

on IRI. Per DeHoratius and Raman (2008), the higher the dollar value

and the dollar volume of an SKU, the lower the IRI level. This reduction

occurs because managers and employees are aware of the impact of IRI

for high-value items; hence they pay more attention and react to errors

quicker when SKUs have this characteristic (Brooks and Wilson, 1993;

Hanna and Newman, 2001). That is, with high-value items IRI levels are

lower because impact awareness is higher.

Extant literature, predominantly based on the periodic approach of

cycle counting (Arnold and Chapman, 2004;Bernard, 1985) also sug-

gests that an SKU's transaction frequency allows more opportunities for

employee error, thus being another major driver of IRI (Bernard, 1985;

Rinehart, 1960). DeHoratius and Raman (2008) found that IRI levels

increase with sales velocity. More recently, research has expanded on

the role of transaction frequency in multi-channel retailing to find that

the volume and frequency of orders are likely to be considerably dif-

ferent across channels and could affect IRI levels in different ways

(Metters and Walton, 2007).

Additionally, increases in product variety, reflecting wider SKU of-

ferings, negatively affect operational performance indicators, including

IRI (DeHoratius and Raman, 2008). Increases in product variety

3

Visual complexity captures the concept of richness or lack thereof, relating

to the quantity, irregularity, detail, and dissimilarity of elements (e.g. SKUs),

the asymmetry of element (SKU) arrangement, and variation in colors and

contrasts (Orth and Crouch, 2014: 525).

M. Barratt et al. Journal of Operations Management 63 (2018) 6–24

7