RFID in Supply Chain Management
Vijay K. Vemuri
C.W. Post Campus, Long Island University, 720 Northern Blvd., Brookville, NY 11548, USA
Keywords: Radio Frequency Identification, Inventory Management, Supply Chain Management, Simulation.
Abstract: The usefulness of information systems critically depends on the accuracy of the data contained within it.
Errors in capturing data into the information systems are particularly vexing since these errors permeate the
entire information system(s), affecting every aspect of information use. The direct and indirect
consequences of unreliable data did not attract much attention as there were few alternatives to reduce them.
Newer technologies, especially Radio Frequency Identification (RFID), are enabling virtual elimination of
data entry errors in inventory management. We investigate the effect of accurate data on the performance of
supply chains utilizing lean production systems. Our simulation results indicate that time to fulfil a purchase
order (cycle time) is significantly reduced by improving the quality of the inventory data. The simulation
model we developed will enable us to examine other performance characteristics of a supply chain. We will
also investigate the sensitivity of supply chain performance due to changes in the parameters of the model.
Advances in information technology have led us to
expect the availability of accurate, reliable, and
timely information at a reasonable cost. Indeed,
synergies between microprocessor, fiber-optics,
satellite, and related technologies have increased the
promise of information systems further. However,
data capture still remains the weakest link in the
chain of information technology tasks needed to
provide accurate and timely information. A 1980’s
study of data capture error rates by the U.S.
Department of Defence (DoD) shows that in
entering 30 million characters; 250,000 errors in
handwritten, 100,000 errors in keyboard, 1,000
errors in OCR, and 10 errors in barcode entries were
made (Taylor, 2004). Despite the astonishing
advances in information technology since the DoD
study, inaccurate data still remains an elusive
problem. The level of data inaccuracies in patient
health care and the resulting dire consequences are
startling (up to 98,000 patients die in the U.S. die
every year due to medical errors (Kohn, et al,
2000)). Medical data errors are a leading cause of
these preventable errors. Information technology can
offer solutions in reducing the data errors - for
example, in one study, direct data entry by
physicians reduced errors by 55 percent (Wendel,
Although not to the extent those patients do,
businesses also bear the consequences of inaccurate
data. Due to handling of their records by many
functional areas within an organization, the high
volume of items, large number of transactions and
exchanges of information and goods with value
chain partners, inventories are especially vulnerable
to data errors. Kang and Gershwin (2005) report that
for a large global retailer, for an average of 49% of
its Stock Keeping Units (SKUs) its inventory
records did not exactly match with the physical
inventories. The error rate remains high at 24% if an
error of ± 5 units is tolerated.
1.1 Causes and Effects of Inventory
There are many causes of inventory inaccuracies. 1)
Stock loss due to obsolete inventory and theft by
employees, customers and others, 2) transaction
errors, 3) inaccessible inventory, and 4) incorrect
product identification are often cited as the most
common causes of incorrect inventory records. In
retail sector, customer, employee, and customer theft
is a major source of stock loss accounting for about
K. Vemuri V. (2006).
In Proceedings of the Eighth International Conference on Enterprise Information Systems - DISI, pages 221-225
DOI: 10.5220/0002466402210225
1.7% of sales (Dabney, et al, 2004). Although stock
loss, and especially loss due to theft, varies among
product categories and the types of business, the
inventory inaccuracies due to the causes mentioned
above remains high for most companies. Inventory
inaccuracies may have ripple effects in the entire
organization. Lee, et al, 2004, cite research
estimating the indirect loss due to inventory
shrinkage can be more than 30 times that of direct
loss. What are the probable causes of these high
indirect costs? Inventories are important assets that
affect many decisions of an organization. Inventory
levels form a basis for decisions pertaining to when
to produce, how much to produce, where and how
much to transport, and many other operating
decisions. The inaccuracies in inventory records will
propagate through out the organization setting-off a
chain of suboptimal decisions. The situation cause
by inventory errors can be much worse than
“garbage in garbage out,” contaminating the entire
processes that connect inputs and outputs of a
system. If not corrected, inventory inaccuracies can
lead to incorrect shelf replenishment policies, stock
outs, and loss of sales.
The consequences of inventory record inaccuracies
extend beyond cost of reconciling inventory system
counts and physical counts. Raman, et al, (2001)
report that in two stores they studied, inventory
inaccuracies reduced profits by 10% and 25%.
Substantial inventory errors were present despite the
use of modern information systems for inventory
Technology provides a solution to mitigate the
problems due to inventory inaccuracies. RFID
provides solutions to track and manage inventory
items in the entire transformation process from raw
materials to finished products. It enables wireless
exchange between tagged items and a reader. Unlike
barcodes, RFID tracking does not require line-of-
sight, and accommodates greater distances between
reader and the tag. The array of RFID applications
within businesses include manufacturing, asset
management, production tracking, inventory control
and logistics. RFID enables improvements of
internal efficiencies of many business processes. For
example, the “group select” feature of RFID utilizes
information contained in the RFID tags for locating
containers with a particular product in a large
shipment consisting of many different types of
2.1 RFID and Inventory Accuracy
RFID may not be able to eliminate all stock losses,
but it can identify discrepancies in system and
physical inventories and recalibrate these accounts
much more efficiently than periodic inventory
audits. By establishing connectivity with the
products real-time inventory information can be
obtained. The accuracy of the data extends beyond
physical presence of the inventory item within the
premises. By embedding expiration data of
perishable products and other information in the
RFID tags other causes of inventory inaccuracies
such as obsolescence, damage, and expiry can be
detected and inventory records can be updated in
real-time. Incidence of data entry errors mentioned
above are much less prevalent in RFID data entry.
The DoD study reports that the use of transponders
(RFID tags), on the average, results in one error in
entering 30 million characters. The RFID technology
has advanced since the DoD study and the error rates
are expected to be even less – resulting in virtual
elimination of inventory record inaccuracies.
Due to their ability to preserve information in their
tags across organizational boundaries, RFID can be
useful in tracking and managing inventories over the
entire supply chain. Retailers can achieve lower
stockouts, cost savings and increased
responsiveness. Distribution centres and warehouses
can improve accuracy and reduce costs due to
automated routing, and cross-docking.
Manufacturers in the supply chain can also benefit
from the RFID tags by using them in their receiving,
shipping and inventory management. Suppliers can
set the automated identification in motion by
sending their shipments with tags. These tags can
also be useful for suppliers in their own shipping
operations and inventory management. That is,
every echelon in the supply chain can achieve better
planning, control, and coordination with automated
identification. The RFID technology can nearly
eliminate inventory record inaccuracies and improve
visibility of inventory data across the entire supply
chain. In addition sharing of information each
member of the supply chain can get timely, accurate
information at a low cost.
3.1 Prior Research
Two streams of prior research is relevant for this
study. The first deals with inventory errors and their
effect on inventory management. Iglehart and Morey
(1972) consider the combined effect of inventory
errors, safety stock and frequency of inventory
counts, service levels and inventory costs. They
derive the level of safety stock and frequency of
inventory count to minimise total costs of holding
and inventory count while maintaining a desired
service level. The present paper and the others
studies dealing with RFID in supply chains attempt
to answer what happens in the limit if inventory
errors tend to zero, or cost of inventory count tends
to zero and frequency of inventory count tends to
infinity. However, these are very difficult issues to
tackle analytically and simulations are used to
answer the above questions. Ganeshan, et al, 2001
deals with forecasting error (similar to inventory
errors) on supply chain performance. Their
simulation results confirm that forecasting errors
have a significant impact on service levels and cycle
time of a supply chain.
Three other recent papers have a direct
connection to this study. All of them deal with
inventory errors on the supply chain and how RFID
technology by nearly eliminating inventory affects
different supply chain performance measures. Kang
and Gershwin, 2005 compare different methods to
cope with inventory inaccuracies. They compare
stockout percentage and average inventory using 1)
additional safety stock, 2) manual inventory
verification, 3) manual reset of the inventory count,
4) periodic write-down to reflect stock loss, and 5)
automatic inventory identification. Not surprisingly,
the last choice achieves the best trade-off between
low stockout and low levels of inventory cost.
Lee, et al, 2004 study a supply chain consisting of a
retailer, distribution centre, and a manufacturer.
They mainly study the impact of inventory accuracy
and inventory replenishment policies on inventory
levels at each echelon. They find that inventory
levels are more stable and lower with RFID
inventory tracking.
Fleisch and Tellkamp, 2005 use simulations to
study the impact of inventory inaccuracy on a retail
supply chain. They disaggregate the sources of
inventory inaccuracy into its component factors
(theft, unsaleables, and misplaced item and incorrect
deliveries) and study their impact on probability of
out-of-stock and cost of inventory inaccuracies.
Their results indicate that theft is the most important
factor impacting supply chain performance and the
level of unsaleable items do not affect supply chain
performance significantly.
Two additional studies are also relevant for this
research as they study the performance of supply
chains with imperfect information. Chen, et al, 2000
study the effect of sharing of customer demand
information at every stage of the supply chain and
conclude that information sharing will mitigate the
so called bullwhip effect (amplification of demand
variability along the supply chain away from the
customers). Their analytical model and simulation
results show that sharing of customer demand
forecasts reduces forecasting errors resulting in
softening of bullwhip effect; however, reduced
forecasting error will not completely eliminate the
bullwhip effect.
Joshi, 2000 in his masters’ thesis develops a
comprehensive framework to improve visibility of
information in supply chains by reducing the delays
in information flows.
The recent surge in RFID implementations raises the
questions regarding supply chain performance and
RFID. Supply chains are complex dynamic systems
with complex flows of products and information.
With uncertainties at, and complex interactions
between, various levels it is nearly impossible to
analytically solve research questions. Simulation
models facilitate tight control of research
environment and ability to manipulate extraneous
factors resulting in excellent internal validity.
However, genralizability, or external validity of
simulation studies are low. Most of the simulation
studies reviewed above do not model “pull”
production environments where customer demand
sets the production process in motion. Lean
production systems and just-in-time inventory
management are characterized by “pull-type”
production processes. Lee, et al, 2004 consider a
production system operating under “push” system
and the retailer using a “pull” system.
In this study will concentrate on “pull” system
where the inventory movement is triggered by
customer demand. In these systems stockout rate,
service level or fill rate is not an appropriate
performance measure since the customer generates
an order and waits for the fulfilment of the order.
Appropriate performance measure is these supply
chains is the cycle time for inventory. The cycle
time in the supply chain refers to the time it takes to
fill an order.
The research question of interest in this study is:
What is the effect of RFID deployment in a supply
chain characterized by pull type production
Supply Chain Management
environments on the customer order cycle time?
This question is relevant since the predominant trend
is to implement pull type production environments
with just-in-time inventory management. In these
environments inventory errors are expected to be
critical since level of inventories are kept to a
minimum. A typical strategy of carrying extra safety
stock to cope with inventory errors is contrary to the
motives for implementing lean environments.
The supply chain we model consists of two
customers calling at random for varying amounts of
goods. The marketing department receives the calls
and sends the orders to nearest distribution centre
(DC): DC1 for customer 1, and DC2 for customer 2.
If enough units are not available at the closest DC,
the sales order is sent to the other DC.
Replenishment at the DCs are based on (s, S) policy,
if the inventory level drops below s units, an order is
sent to bring the inventory level to S units. The
timing of a purchase order is triggered by incoming
purchase orders. The products are manufactured in
the Assembly department. Figure 1 shows overall
flow of goods and information and the location of
the decisions made. Each of the rectangles
represents a business process within the supply chain
and includes subprocesses. Due to space limitations,
the details contained within these subprocesses are
not shown, but implements the description of the
supply chain above.
We used SimProcess 4.2 program to conduct the
simulations. This modelling environment enabled us
to easily construct hierarchical business processes,
activity based costing for allocating overhead costs
and optimisation routines to determine optimal
strategies within the simulation environment.
Table 1 shows the model parameters and their
Table 1: Model Parameters.
Customer 1 Customer 2
Normal (5,
1) Hours
(2.5, 1) Hours
Order Size
(25, 100)
(50, 200)
s S
t at DC1
200 500
t at DC2
500 2,000
Distance to
Distance to
1 minute per unit
In order to answer the research question,
simulations are run under two different assumptions.
Under the first assumption, inventory shrinkage
takes place but the RFID system will detect the
shrinkage immediately and the inventory
replenishment decisions are made based on the
correct level of inventory in stock. This scenario is
compared with cycle times attained when the system
is not aware of the inventory loss.
Many use Retail Security Survey (Dabney, et al,
2004) estimate of 1.7% of gross annual sales as the
annual loss due to inventory shrinkage. However,
mistakenly, they apply this percentage to inventory
Figure 1: Top Level Supply Chain Model.
levels (Lee et al, 2004, and Fleisch and Tellkamp,
2005). Based on average turnover ratio of about 5
and gross profit ratio of 70%, inventory shrinkage is
closer to 12% of inventory levels. Inventory
shrinkage cannot go undetected forever. We assume
that once every quarter (at the time of preparation of
quarterly financial statements) the shrinkage will be
detected and the information system is corrected to
reflect actual count. Many use quarterly reset of the
system inventory for the detection of inventory
shrinkage. We simplify this situation by running
simulation for a quarter at a time. To mitigate the
effect of initial conditions, and to obtain steady state
results, we use a warmup period of 10 days (run
length is 120 days). To obtain robust results we
repeated the simulation for 25 quarters.
Table 2 compares cycle times with incorrect
inventory records and perfect inventory counts
obtained by deploying RFID technologies.
Table 2: Comparison of Supply Chain Performance.
Mean 2.002 Hours 2.646 Hours
1.125 Hours 1.165 Hours
Sample Size 25 25
Hypothesis test resulted in a t-value of 1.988 and
a p-value of .026. Based on these statistics we
conclude that accurate inventories result in lower
cycle time and faster fulfilment of purchase orders.
The analysis of the simulation results is not yet
completed. Sensitivity of cycle times to
manufacturing times, capacities and other important
parameters are being established. Other important
supply chain performance criteria are being
examined. The framework and the model we have
established so far will enable us to extend the
analysis of supply chain performance measures with
accurate inventory counts now possible with RFID
Our research addresses important issues related to
the role of RFID in supply chain management.
Accurate inventory data is especially important in
lean manufacturing environments as large inventory
buffers and excess safety stock are not available to
forgive inventory inaccuracies. Our simulation
model consisting of a pull production environment
concludes that the cycle time (time it takes to fill a
purchase order) is reduced by accurate inventory
counts. The performance of a supply chains using
alternative performance criteria and the sensitivity of
performance to parameters of the production
environment are currently developed.
Chen, F., Drezner, Z., Ryan, J.K., and Simchi-Levi, D.,
2000. Quantifying the Bullwhip Effect in a Simple
Supply Chain: The Impact of Forecasting Lead Times,
and Information. Management Science, 46, 436-443.
Dabney, D., Hollinger, R., and Dugan, L., 2004. Who
Actually Steals? A Study Of Covertly Observed
Shoplifters, Justice Quarterly, 21, 693-728.
Fleisch, E., and Tellkamp, C., 2005)\. Inventory
Inaccuracy and Supply Chain Performance: A
Simulation Study of Retail Supply Chain. International
Journal of Production Economics, 95, 373-385.
Ganeshan, R., Boone, T., and Stenger, A., 2001. The
Impact of Inventory and Flow Planning Parameters on
Supply Chain Performance: An Exploratory Study.
International Journal of Production Economics, 71,
Iglehart, D.L., and Morey, R.C., 1972. Inventory Systems
with Imperfect Asset Information. Management
Science, 18, B388-B394.
Joshi, Y., 2000. Information Visibility and Its Effect on
Supply Chain Dynamics. Masters Degree Thesis, MIT.
Kang, Y., and Gershwin, S.B., 2005. Information
Inaccuracy in Inventory Systems: Stock Loss and
Stockout. IEEE Transactions, 37, 843-859.
Kohn, L., Corrigan, J. and Donaldson, M., (Eds.), 2000.
To Err Is Human: Building a Safer Health System,
National Academies Press, Washington, D.C.
Lee, Y.M., Cheng, F., and Leung, Y.T., 2004. Exploring
the Impact of RFID on Supply Chain Dynamics.
Proceedings of the 2004 Winter Simulation
Conference. December 5-8, 2004. Washington, D.C.
Raman, A., DeHoratius, N. and Ton, Z., 2001. Execution:
The Missing Link in Retail Operations. California
Management Review, 43, 136-152.
Taylor, D., 2004. Supply Chains: A Manager’s Guide,
Addison Wesley, Boston, MA.
Wendel S., 2000. Medical errors: Institute of Medicine
report raises questions about technology. Healthcare
Informatics, 17, 37-38.
Supply Chain Management