The impact of disclosing inventory‐scarcity messages on sales in online retailing

• Author: Rui Yin,Christopher S. Tang,Sungho Park,Elliot Rabinovich
• DOI: http://doi.org/10.1002/joom.1082
• Published date: 01 July 2020
• Date: 01 July 2020

RESEARCH ARTICLE

The impact of disclosing inventory-scarcity messages

on sales in online retailing

Sungho Park

1,2

| Elliot Rabinovich

3

| Christopher S. Tang

4

| Rui Yin

3

1

Marketing Department, W. P. Carey

School of Business, Arizona State

University, Tempe, Arizona

2

SNU Business School, Seoul National

University 1 Gwanak-ro, Gwanak-gu,

Seoul, 08826, South Korea

3

Supply Chain Management Department,

W. P. Carey School of Business, Arizona

State University, Tempe, Arizona

4

Business Administration, Anderson

School of Management, University of

California, Los Angeles, California

Correspondence

Elliot Rabinovich, Supply Chain

Management Department, W. P. Carey

School of Business, Arizona State

University, Tempe, AZ 85287-4706.

Email: elliot.rabinovich@asu.edu

Handling Editor: Michael Galbreth

Abstract

To influence demand, some online retailers post messages (e.g., “5 units or less

left in stock”) on their product pages to signal impending stockouts. These

“scarcity”messages provide consumers “partial”inventory information, reveal-

ing only an upper bound on the number of units available for sale. To examine

the impact of these messages, we obtained price and sales data from an online-

retailer website across a sample of durable goods before and after the retailer

posted the messages over multiple inventory-replenishment cycles. We then

used these data to assess empirically the effect of these messages on these

products' daily sales. We find that disclosing these messages can decrease daily

sales by an average of 17.60%. This finding suggests that scarcity messages such

as these can have a negative influence on the sales prospects of durable goods.

We also observe, on the other hand, that price discounts are quite effective in

increasing sales and offsetting the losses induced by scarcity messages. On

average, a reduction of 1% in stock keeping unit price increases daily sales by

approximately 3%. Therefore, relative to disclosing scarcity messages, price dis-

counts are a much more effective tool at increasing inventory turns.

KEYWORDS

econometrics, inventory management, online retailing, operations management-marketing

interface

1|INTRODUCTION

Low search costs on the Internet enable consumers to

compare many retail offers freely which, in turn, exacer-

bates certain purchasing behaviors, including the active

pursuit of bargains (Brynjolfsson & Smith, 2000; Zhang,

Fang, & Sheng, 2006) and the postponement of purchases

in anticipation of future price drops (Aviv, Tang, & Yin,

2009; Netessine & Tang, 2009). To counteract these

behaviors, many online retailers conduct “flash sales”

events of deeply discounted stock keeping units (SKUs)

for limited times (usually 24 hr) during which they make

available to consumers real-time information on every

SKU's inventory level available for purchase. The

retailers' goal is to induce consumers not to miss out on

these events by showing them how fast inventory is

decreasing for the SKUs. According to empirical research

by Cui, Zhang, and Bassamboo (2018) and Wagner,

Calvo, and Cui (2018), exposing consumers to informa-

tion on these inventory reductions significantly increases

the deals' frequency of purchase.

These findings, however, are limited to a small frac-

tion of online retail sales, since the vast majority of e-

commerce purchases involve durable goods available for

sale over indefinite time horizons and multiple inventory

replenishment cycles. The selling strategies in these

mainstream environments differ notably from those in

“flash sales”settings, where retailers typically promote

Received: 22 October 2018 Revised: 16 December 2019 Accepted: 30 December 2019

DOI: 10.1002/joom.1082

534 © 2020 Association for Supply Chain Management, Inc. J Oper Manag. 2020;66:534–552.wileyonlinelibrary.com/journal/joom

heavily discounted items in limited quantities (i.e., with

no inventory replenishments) and for extremely short

durations to attract bargain hunters who are inherently

interested in taking advantage of these temporary promo-

tions. Consequently, these consumers are naturally more

likely to purchase when they observe that inventory is

running low (Ferreira, Lee, & Simchi-Levi, 2016;

Sodero & Rabinovich, 2017).

This is not necessarily the case in mainstream settings

involving durable goods, where consumers often make

purchasing decisions based on their personal needs, as

opposed to opportunistically, upon encountering

discounted offers available during short time windows.

For these consumers, low inventory levels may actually

undermine their confidence in the retailer's quality of ser-

vice (Balakrishnan, Pangburn, & Stavrulaki, 2004). They

may also have an adverse influence in their inferences

regarding products' quality in relation to other items that

are available for sale currently or may become available

in the future. Specifically, a low inventory amount may

lead consumers to infer that the retailer has chosen to

forego replenishing its inventory because the products

are undesirable, of low quality, or likely to become obso-

lete (Koschat, 2008). To the extent that low search costs

on the Web make it easy for consumers to look for prod-

ucts elsewhere (Bakos, 1997), low stock quantities may

actually induce them to forego buying these items and

reduce sales (Koschat, 2008; Wolfe, 1968).

Another important consideration is that SKUs sold

through flash sales events are offered at prices that typi-

cally remain fixed regardless of how fast inventory is

depleted. This is generally not the case for products in

mainstream environments. Because these items are avail-

able for sale over indefinite time horizons and multiple

inventory replenishment cycles, their prices normally fluc-

tuate with inventory levels. Such conditions are not com-

monly present in flash sales settings, making it difficult to

evaluate whether lower prices or inventory availability are

more effective in promoting online sales in those settings.

Our goal is to address these deficiencies in the litera-

ture. A challenge in fulfilling this objective is that many

online retailers in mainstream settings have chosen not

to share with the public exact inventory level information

because disclosing such information may give their com-

petitors and suppliers sensitive insights into their inven-

tory management and pricing policies (Fisher, Gallino, &

Li, 2017). However, there are retailers (e.g., Walmart and

Overstock) that have chosen to disclose “partial”inven-

tory information contained in messages alerting con-

sumers about the maximum number of units available in

stock when actual inventory levels drop below this upper

bound. For instance, by posting a message alerting shop-

pers that an SKU has “5 units or less left in stock”when

inventory drops below this threshold, retailers can let

consumers know of the maximum amount of inventory

available for sale. However, since these “scarcity mes-

sages”contain only partial inventory information, con-

sumers (as well as competitors and suppliers) will not

know the exact inventory levels available for sale at any

point before purchase.

To our knowledge, no research has evaluated the

impact of these scarcity messages' disclosure on durable

consumer goods' sales in mainstream online retail envi-

ronments. We are only aware of a study by Sodero,

Rabinovich, Aydinliyim, and Pangburn (2017) that also

focused on these settings. However, this study focused on

the link between purchasing frequencies (not total sales)

and inventory depletion at low levels, in which

inventory-level information is displayed continuously to

consumers at all times. Therefore, Sodero et al. (2017)

were unable to estimate the impact of the initial disclo-

sure of scarcity information on sales because they focused

exclusively on consumers' purchasing rates and had no

way of making comparisons “before”versus “after”con-

sumers' exposure to scarcity information. Because we can

evaluate the differences in sales that exist before versus

after a retailer posts this information for its products, we

can expand on Sodero et al.'s (2017) analysis. Moreover,

because the disclosure of scarcity messages is potentially

applicable as a widespread mechanism to influence con-

sumers' purchasing behavior, our research not only has

academic relevance, but also carries significant implica-

tions for practitioners in the retail industry.

To evaluate the aforementioned issues, we collected

data (prices, sales, etc.) from a sample of SKUs sold by a

focal retailer that posted scarcity messages with partial

inventory scarcity information (“5 or less left in stock”)

every time the inventory level of an SKU dropped below

6 units. Because this retailer sells durable consumer

goods available for purchase with no time or inventory-

cycle restrictions, we can examine the impact that post-

ing these messages has on sales for each SKU over

extended periods. In so doing, we can partial out time

effects from the effects of inventory messages' disclosure

on sales, since it is possible that the time a product has

been available for sale will influence demand. Further-

more, having no limitations imposed by the retailer on

the amount of time products are available for sale means

that we can examine the effects of these messages' disclo-

sure on sales for each SKU longitudinally while isolating

significant differences in these effects across products

selling in large versus small daily amounts due to season-

ality. Finally, because prices in our data vary over time,

we can also isolate these effects from longitudinal price

fluctuation effects on sales. To our knowledge, no paper

has evaluated these effects in an online setting. This is

PARK ET AL.535