1 Market Response to Firm-Specific Shocks: The Arab Boycott of

Market Response to Firm-Specific Shocks:
The Arab Boycott of Danish Dairy Products
Alexis Antoniades*
Georgetown University
Sofronis Clerides†
University of Cyprus and CEPR
January 2016
PRELIMINARY AND INCOMPLETE
Abstract:
The 2006 boycott of Danish products in several Arab countries provides a natural
experiment setting to study how markets respond to firm-specific shocks. The boycott
was devastating for Danish firms in Saudi Arabia: their market share collapsed from
16.5% in January to <1% in March and never fully recovered: it was 6.3% in 2009.
Retailer participation was a key determinant of boycott effectiveness. The shock led to a
scramble for market share with increased promotional activity by large firms and entry
by fringe firms. The analysis highlights how a temporary shock can have a permanent
impact on market structure.
Keywords: boycotts, multi-product firms, demand shock, Saudi Arabia
Corresponding author. Address: School of Foreign Service in Qatar, Georgetown University, Education City, Doha,
23689, Qatar. E-mail: [email protected]. We would like to thank Orley Ashenfelter, Cheng Chen, Daniel
Westbrook and seminar participants at Princeton University for helpful comments made and Ganesh Seshan for
bringing to our attention the boycott against Danish products. Salma Mousa provided excellent research assistance.
This research was made possible by support of an NPRP grant from the Qatar National Research Fund. The
statements made herein are solely the responsibility of the authors.
† Address: Department of Economics, University of Cyprus, P.O.Box 20537, 1678 Nicosia, Cyprus. E-mail:
[email protected].
*
1
I. Introduction
On January 26th 2006, the imams in Saudi Arabia called for a boycott on Danish products during
the Friday prayers. The reason was the publication of 12 cartoons depicting the prophet
Mohammad by Jyllands-Posten, Denmark’s largest newspaper, which caused outrage amongst
Muslims. Within a week, the call for boycott had spread to more than ten Arab countries. Dairy
products became a focal point of the boycott campaign because they Denmark’s main export to
the Gulf countries. 1 Most large retailers responded to public pressure and joined the boycott,
removing Danish products from the shelves. Images of empty supermarket shelves appeared on
TVs around the world. The boycott was called off after four months, in late May 2006.
We exploit unique scanner data to study what impact, if any, this boycott had on the market for
cheese products in Saudi Arabia. We find that the boycott had an immediate and dramatic effect
on Danish brands. In February 2006, sales of Danish cheese dropped 52% relative to the previous
month, even though Danish products were still on grocery store shelves. When they were
removed from the shelves in March, the total sales decline reached 93%. The market share of
Danish brands collapsed from 16.5% in January to below 1% during March and April 2006.
Danish sales recovered somewhat after the boycott was officially called off in May 2006, but
never returned to their previous levels. In 2009, three years after the event, Danish brands
accounted for only 6.3% of all cheese sales, just over a third of their 18.1% share in 2005.
Because the change in consumer preferences was sudden, unexpected, and dramatic, the boycott
provides a great setting for evaluating how multiproduct brands respond to changes in market
Cheese and curd accounted for almost half of the Danish exports to the Gulf Cooperation Council
countries in 2005 (UN COMTRADE). The Gulf Cooperation Council (GCC) countries are Bahrain,
Kuwait, Oman, Qatar, Saudi Arabia, and the United Arab Emirates. Of these, Saudi Arabia is by far
the biggest economy. It accounts for about 84%, 70%, and 50% of the total land, population, and
GDP of the GCC.
2
1
conditions.2,3 Studying how the Danish and non-Danish brands responded in the months during
and after the boycott constitutes the main focus of this paper.
A few interesting facts come out of that analysis. First, Danish brands’ losses were driven
primarily by changes at the intensive margin, whereas gains of non-Danish brands were driven
by changes at the extensive margin. That is, the number of Danish products available for
consumption remained constant, but sales of each product suffered as a result of the boycott. In
contrast, non-Danish brands responded to the boycott by introducing new products, most of
which matched the Danish products in characteristics such as description, package type, and
weight.4
Second, the impact of the boycott on non-Danish brands was very heterogeneous. While
aggregate sales as a whole rose sharply, half these brands either shrunk or disappeared. In fact,
the biggest winners and biggest losers were all small brands, as measured by sales in the months
before the boycott.
Since brand size cannot explain performance what can? We show that a difference in strategies
in response to the boycott provides an explanation for the performance of the non-Danish brands.
The three dominant non-Danish brands introduced new products, raised prices of existing
products, and experienced a moderate increase in sales. However, for small brands, introducing
new products and raising prices of existing ones hurt sales. This occurred because the void left
by the boycott on Danish products encouraged a substantial introduction of new products into
Bernard, Redding, and Schott (2011) use US data from 1972 to 1997 to show that multi-product
firms (MPFs) are very common and important: they comprise 41% of all firms in the
manufacturing sector, they account for 91% of output, and product switching within these firms
explains a third of the total increase in real US manufacturing shipments during that period.
Similarly, using Indian data, Goldberg et al. (2011) show that MPFs tend to be larger, more
productive, and more likely to export.
3 To study how MPFs respond to economic shocks, some recent work examines intra-firm
adjustments of MPFs after trade liberalizations. Hallak (2000) and Eckel and Neary (2010)
provide a theoretical treatment and Baldwin and Gu (2006) and Bernard et al (forthcoming) an
empirical. But since trade liberalizations are associated with a market-size effect (Eckel and
Neary 2010), a competition effect (Eckel and Neary 2010), and a production outsourcing effect
(Hallak 2000), disentangling the effects of the various shocks on the intra-firms adjustments is
challenging.
4 Like Bernard et al. (2011), Goldberg et al. (2011), and Broda and Weinstein (2010), we find that
product switching within the non-Danish brands is a far more important channel for adjustments
than the entry and exit of brands.
3
2
the market, which brought about tougher competition. And because small brands responded
with very different strategies, the impact of the boycott varied by brand.
To summarize, this study provides perhaps the most striking example of a successful boycott
ever documented in the literature. Most importantly, because the boycott had such a sharp of an
effect on sales of Danish products and because the data are available at the scanner level and
across hundreds of outlets, we are able to analyze the response of Danish and non-Danish firms
and complement a growing body of the literature that examines how multi-product firms
respond to economic shocks, while highlighting the differences in behavior between top brands
and fringe brands.
The paper contributes to the recent revival of the literature on consumer boycotts. Early work in
this area is summarized by Friedman (1991, 1999). For theoretical treatments of consumer
boycotts see Innes (2006), Barron (2001, 2002, 2003) and references therein. John & Klein
(2003) and Sen, Gürhan-Canli, and Morwitz (2001) offer psychological explanations as to why
consumers choose to participate or not. The recent literature was kicked off by the work of Chavis
and Leslie (2009), who study the impact of the US boycott on French wine during the 2003 Iraqi
war. They find that French wine sales dropped by about 13% over the six months they estimated
the boycott to have lasted. Using more aggregate data, Ashenfelter et al. (2007) fail to find that
the boycott on French wine has an impact on sales. Michaels and Zhi (2010) find that strained
relations between US and France led to a reduction in bilateral trade by a remarkable 10-12%.
Hong, Hu, Prieger and Zhu (2011) find that Chinese boycotts of French products in 2008 led to a
25-33% drop in sales of French automobiles. Clerides, Davis and Michis (2015) find a significant
but economically small impact on sales of US products in Arab countries after the Iraq war.
Pandya and Venkatesan (forthcoming) show that during the 2003 US-France dispute over the
Iraq War the market share of French-sounding, US supermarket brands declined. Hendel, Lach
and Spiegel (2015) study the boycott of cottage cheese in Israel that was motivated by consumer
perceptions of excessive pricing.
The paper also contributes to a more general line of research that examines the market response
to different types of shocks. Some papers look at the stock market response to negative news
such as aircraft crashes (Borenstein and Zimmerman 1988, Bosch, Eckard and Singal 1998) and
labor strikes (de Fusco and Fuess 1991). The focus of this literature is on whether a negative
4
shock affects a firm’s value. We differ in that we do not ask whether an effect exits (it clearly does
and it is massive) but we focus instead on examining in detail the market response to the negative
news in terms of pricing, sales and product introduction. In that sense our work is closer to
research looking at the demand effects of negative events. Grafton, Hoffer and Reilly (1981) and
Reilly and Hoffer (1983) examined the effect of automobile recalls on automobile demand. More
recently, Freedman, Kearney, and Lederman (2012) studied the effects of toy recalls. Also related
is work by Cawley and Rizzo (2008) that studies the spillover effects of drug withdrawals.
Our work is also related to industrial organization research on how prices respond to demand
shocks. Chevalier, Kashyap and Rossi (2003) examine why prices do not rise during periods of
(predictable) peak demand. Hatzitaskos and Nevo (2006) study the same question and propose
an alternative explanation. Copeland and Hall (2011) study the impact of temporary changes in
the demand for cars; they find that most of the impact is absorbed by sales and there is little price
adjustment. Gagnon and López-Salido (2014) study the pricing response of U.S. supermarkets to
large demand shocks and find that responses are small. The main takeaway from this literature
is that prices are not the main margin of adjustment. Our findings are consistent with this but we
go a step further in analyzing medium-term firm response in terms of entry and exit.
The remainder of the paper is organized as follows. In Section 2 we present some background
information on the boycott against Danish products. We describe the dataset used to study the
impact of the boycott in Section 3 and present the analysis in Section 4. Section 5 concludes.
II. The Boycott
On September 30th 2005, Jyllands-Posten, Denmark’s largest newspaper printed 12 drawings
depicting the Prophet Mohammad. Danish Muslim groups called the depiction of the Prophet in
cartoons blasphemous and protested against the Jyllands-Posten publications. The debate
between those who supported freedom of speech and those who supported respect for religion
went global when the cartoons were reprinted in various publications in more than fifty
countries in early 2006.
On January 26th 2006 Saudi Arabia recalled its ambassador from Denmark and initiated a boycott
on Danish products. The call for the boycott was spearheaded by the imams during the Friday
5
prayers. Within a week the boycott spread to Iraq, Yemen, Syria, Palestine, Bahrain, Qatar, the
United Arab Emirates, Jordan, Algeria, Morocco, Tunisia, and Oman. On January 28th, the
Denmark-based Arla Dairy Group placed advertisements in Middle Eastern newspapers in an
effort to stop the boycott of its products. Two days later, the EU Trade Commissioner threatened
to take the issue to the World Trade Organization if the boycott persisted. The next day Saudi
hospitals refused to buy Danish insulin. In the following days violence broke out in many regions,
including Syria, Lebanon, Pakistan, Iran and Gaza, where the Danish embassies were set on fire.
It is believed that more than 140 people died during the violent protests. On April 24 th an AlQaeda video of Osama Bin Laden emerged urging Islamic nations to continue the boycott.
In response to the public outcry against Denmark, and after they observed that consumers had
stopped buying Danish goods, most retail outlets in Saudi Arabia joined the boycott in mid-March
by removing all Danish products from their shelves. In May, Yusuf Al-Qaradawi, the most
prominent Sunni religious cleric called off the boycott. Despite the end of the boycott, a local
supermarket chain decided to keep Danish products off its shelves indefinitely. This particular
chain accounted for about 40% of all supermarkets in Saudi Arabia. A timeline of the boycott is
presented in Appendix A.
III. The Data
We study the impact of the boycott on Danish and non-Danish brands selling processed cheese
products in Saudi Arabia. We chose Saudi Arabia because it is a large country with a conservative
Muslim population and for the fact that monthly scanner data at the store level during and after
the boycott are available. We chose processed cheese because it is the product with the strongest
presence by Danish firms in Saudi grocery stores.
The data come from Nielsen and cover the years 2005-2009. The frequency is monthly and each
observation describes the total quantity of a product (barcode) sold at a retail outlet in a given
month and the price paid during the day of the audit. Nielsen also provides information on the
brand, on the distributor or manufacturer, on the weight, on the package, on the variant, and on
whether the product is under a promotion or not. “Kraft White 240 Glass Jar Blue” and
“Lavachequirit White 240 Glass Jar Blue” are descriptions of two such barcodes.
6
We also have information on outlets. Each outlet is classified as supermarket, large grocery,
small grocery, or mini-market/self-service, based on size. We can identify outlets belonging to
the same chain and the region where each operates. The majority of the chains are local, but
international retailers such as Carrefour also operate in Saudi Arabia.
We eliminated extreme price movements by dropping observations where the monthly price
changes is above +300% or below -75%.5 Most of our analysis will utilize a balanced set of outlets
(outlets that are present in the sample in all months between January of 2005 and December
2009) in order to ensure that our results are not driven by entry and exit of outlets. 6 In cases
where the entire sample is used, it is stated explicitly.
Table 1 presents descriptive statistics for the data. In 2005 550 cheese products (EAN codes)
were sold in Saudi Arabia. These products belonged to 68 brands and 28 distributors or
manufacturers. Over time the number of products increased, the number of brands dropped
slightly, and the number of distributors/manufacturers remained fairly constant.
Table 1 – Descriptive Statistics for the Market of Cheese Products in Saudi Arabia: 2005-2009
Year
2005
2006
2007
2008
2009
Products
550
539
670
718
699
Brands
68
60
71
73
63
Supplier Outlets Chains Channels Regions
28
334
6
4
4
27
334
6
4
4
28
334
6
4
4
28
334
6
4
4
27
334
6
4
4
In 2005, Danish brands sold 68 different cheese products in Saudi Arabia. These products
accounted for 12% of all products sold, and 17% of all revenue. 7 At the same time, the market for
These thresholds are recommended by Statistics Netherlands on their work on inflation
measures using monthly scanner data (de Hann and van der Grient 2009).
6 A new outlet entry in the dataset does not imply that the outlet is newly established. While this
may be one reason, another reason may be that Nielsen decided to include it as part of their
sampling strategy, or because the company just received permission to audit the outlet. Similarly,
an exit could imply that the outlet went out of business, it dropped from the sample that Nielsen
chooses to follow, or the management decided to stop sharing information with Nielsen. The
dataset covers sales of cheese products across 1952 retail outlets. The number of outlets is
reduced to 322 when we restrict the sample to these outlets that existed in all periods. Most of
the outlets dropped represent mini-markets and groceries, and account for a very small fraction
of total sales.
7 Using a balanced panel of outlets, the Saudi market for cheese products in 2005 was about US
$30 million. Based on data from all the outlets in our sample, and discussions we had with Nielsen,
7
5
cheese products was very concentrated. During 2005-2009, the top 5 brands had a 50% share of
all products sold and generated 87% of all revenue. In 2005, a Danish brand was among the top
5.8 Moreover, half the brands were multi-product, in the sense that they sold multiple cheese
varieties.
IV. Boycott impact
Figure 1 plots the market share of Danish brands in the processed cheese category from January
2005 to December 2009. The call for boycott had a devastating impact on Danish brands, both in
the short-run and long-run. The share of Danish products collapsed from 19% in 2005 to about
8% in February 2006 (the first full month of the boycott) and to less than 1% in March 2006 (the
month retailers joined in). 9 Practically overnight, Danish brands went from dominating to
vanishing.
When the boycott ended in May, there was a partial but short-lived recovery. For a couple of
months after the boycott was called off, sales of cheese products rebounded to as much as 14%
of the market share, but then started to fall again. From the end of 2007 until the end of our
sample two years, later, the market share of Danish brands fluctuated between 5-10% and
averaged 7. 5%.
We next look at revenue. Figure 2 plots the time series of category revenue broken down by
Danish versus non-Danish brands (note that the bars are not stacked). Danish revenue goes to
almost zero during the boycott while non-Danish revenue shows a substantial increase. Total
category revenue does not seem to deviate from its trend levels. Regression analysis that we
conducted (not shown here) also failed to pick up any significant impact of the boycott on total
category sales. This suggests that the loss in sales of Danish products was essentially completely
offset by a corresponding increase in sales of other brands.
we estimate the total Saudi market for cheese products in 2005 to be closer to US $100 million
and to have doubled in value by 2009.
8 A confidentiality agreement with Nielsen prevents us from disclosing the brand names.
9 This is the only instance where data from all 1852 outlets were used. Since entry and exit of
outlets does not bias the computation of market shares, we take advantage of the large, and more
representative, sample of outlets. The results do not change if we restrict the data to a balanced
panel of the outlets that appear in all periods.
8
Figure 1 –Market Share of Danish Cheese Products in Saudi Arabia between Jan 05 and Dec 09
Figure 2 – Category revenue by Danish/non-Danish
9
Two factors contributed to the drop in Danish sales after the boycott: changes in market
conditions and changes in the distribution of Danish products. The former comes from a shift in
preferences against Danish products and in favor of non-Danish (the demand shock) and changes
in the product mix (the supply shock). The latter captures the decision of a major supermarket
chain not to carry Danish products, even after the boycott ended. This particular chain accounted
for about 40% of all supermarkets at that time.
Figure 3 – Number of Outlets Selling Danish Products
1-Jan-09
1-Oct-08
1-Jul-08
1-Apr-08
1-Jan-08
1-Oct-07
1-Jul-07
1-Apr-07
1-Jan-07
1-Oct-06
1-Jul-06
1-Apr-06
1-Jan-06
1-Oct-05
1-Jul-05
1-Apr-05
1-Jan-05
The Boycott
Because the number of outlets carrying Danish products changed over time, we want to know
what part of the drop in Danish sales came from changes in the distribution channel and in
market conditions. We start by counting each month the number of outlets carrying Danish
products between 2005 and 2009. We report the number in Figure 3 where we also provide a
breakdown by outlet type. To be included in the count, the outlet must report positive sales for
at least one Danish product within a month.
10
A third of the outlets that carried Danish cheese products before the boycott dropped them
during the boycott. Most importantly, almost all the supermarkets, which accounted for the
majority of sales, dropped them. However, the outlets did not join the boycott instantaneously.
According to interviews we conducted, customer accounts, and press releases from that period,
it took about six weeks for retailers to move Danish products off their shelves. This is reflected
in the diagram from the collapse in the number of outlets, supermarkets in particular, that carried
Danish products between March and April 2006. Interestingly, once the boycott ended, about
40% of the supermarkets decided to keep Danish products off the shelves indefinitely. Most of
these outlets belonged to a local chain.
Table 2: Market share of Danish products in supermarkets
2005
2006
2007
2008
2009
Droppers
Keepers
1.4%
13.3%
12.8%
0.0%
0.0%
20.9%
17.5%
14.3%
8.3%
11.2%
0.0%
All
11.6%
8.9%
6.4%
6.8%
A back-of-the-envelope calculation helps us estimate how much of the drop was due to changes
in market conditions or changes in distribution. In each year we break down the market share of
Danish brands across the two types of supermarkets; those that re-stocked their selves with
Danish products after the boycott ended and those that did not. The shares are reported in Table
2. In 2005, Danish brands accounted for 17% of all cheese sales in supermarkets across Saudi
Arabia. By 2009, they accounted for only 7%. Sales in stores that dropped Danish goods for good
went from 5% in 2005 to 0% each subsequent year (with the exception of positive sales in
January 2006). Similarly, sales of Danish products in other supermarkets went from 12% to 7%.
We conclude that half of the lost market share was due to the choice of some supermarkets to
drop Danish goods forever, and half was lost due to changes in consumer preferences and
changes in the product mix of non-Danish brands. 10
10
In principle, consumers that could not find Danish products in certain stores could visit other
stores. In such case, the analysis above potentially overstates the importance of sales lost to
11
Which outlets permanently removed Danish brands? Table 2 shows the share of Danish products
by outlet type: “keepers” are outlets that re-stocked Danish products after the boycott was called
off; “droppers” are outlets that did not. In 2005, Danish products had a 20.9% market share in
keepers but only 12.8% in droppers. Outlets that opted to permanently drop Danish products
had lower Danish sales to begin with.
Figure 4 – Price index of Danish and non-Danish products
Figure 4 depicts price indices for Danish and non-Danish products, each normalized to be 1 in
January 2005. There is no noticeable shock to prices during the boycott. In the few months after
the boycott prices of Danish products registered a drop of the order of 5%. This came mostly as
a result of increased promotional activity (we will revisit this issue in the next section). This price
stores that dropped Danish products indefinitely and understates the importance of changes in
preferences and product mix. However, because most consumers decide where to go shopping
based on store location, product variety and prices, produce quality, and ambience, we do not
expect the availability or not of Danish products to be a major factor in their decision making
process. So even if such bias exists, we expect it to be small.
12
drop is consistent with and explains the large spike in sales of Danish products in July, August
and September of 2006. The large increase in cheese prices in 2008 is associated with a period
of high overall inflation in Saudi Arabia.
What products did consumers substitute to? Did they switch to local or other international
brands? Figure 5 depicts market shares by country of origin for each of the five years of our
sample. The “regional” category includes products from Saudi Arabia and neighboring countries,
such as Egypt. There was a large gain for regional and US brands in 2006 and 2007. US brands
lost some market share thereafter, returning close to their 2005 level. Regional brands on the
other hand continue to increase their market share. This cannot of course be solely attributed to
the boycott.
Figure 5 – Market shares by country of origin
V. Firm response: pricing, entry and exit
The previous section established that the boycott of Danish products led to a dramatic decline in
their market share, from 18.1% in 2005 to practically zero during the boycott. The market share
13
recovered only partially after the boycott was called off, settling down at almost one third of the
level of the pre-boycott period. Total category revenue was unaffected, indicating that consumers
fully substituted to other products. There was no aggregate price adjustment during the boycott
but Danish prices registered a decline in the few months after the boycott. This decline could
reflect an effort by Danish firms to recover market share.
We now delve deeper into the data in order to understand the forces that shaped this market
outcome. How did firms respond to the large negative shock that hit one of the market’s dominant
players? Did consumers substitute to existing products or to new products? Was there entry of
new brands or of new products by existing brands? Did the boycott have a lasting effect on
market structure? We examine these questions next.
Multi-product brands and intra-brand adjustments
We start by looking at intra-brand adjustments. We examine whether the change in the fortunes
of Danish and non-Danish brands was a result of a change in sales of existing products, a change
in the product mix, or a combination of the two changes. Following Broda and Weinstein (2010),
we break down total sales growth between two periods into sales of products that survive and
products that are new or disappear. Specifically,
(1)
Vt - Vs
C - Cs
D
N
º t
- s + t
Vs
Vs
Vs
Vs
Total growth
Common Destruction Creation
products
growth
where Vt and Vs measure total sales in periods t and s, respectively. Ct and Cs measure total sales
in periods t and s of products that are common in both periods. Ds measures total sales of
products that existed in period s but disappeared in period t, and Nt measures sales of new
products that appear in period t but not in period s. We also define the intensive and extensive
margins of growth as:
(2)
Intensive = Common Products Growth
14
(3)
Extensive = Creation - Destruction
The intensive margin accounts for the share of the growth that is driven by growth in sales of
products that existed in both periods. The extensive margin accounts for the share of the growth
that is driven by the creation of new products, after taking into account losses in sales of products
that exit the market.
We first compute the growth in total sales of Danish and non-Danish brands between 2005 and
2006. Then, we decompose the growth into intensive and extensive margins based on the
formulas above. We also extend the analysis by looking at two-, three-, and four-year horizons.
The results are shown in Table 3.
Table 3 – Sales Growth Decomposition into Intensive and Extensive Margin by Brand Type
Year
Total
2005-2006
2005-2007
2005-2008
2005-2009
-0.43
-0.47
-0.54
-0.47
Growth Rates
Danish
Non-Danish
Intensive Extensive
Total Intensive Extensive
(i) All Outlets
-0.35
-0.08
0.21
0.05
0.15
-0.48
0.02
0.19
-0.06
0.25
-0.47
-0.06
0.37
-0.08
0.45
-0.41
-0.06
0.46
-0.06
0.52
2005-2006
2005-2007
2005-2008
2005-2009
(ii) Supermarkets that kept Danish products
-0.22
-0.12
-0.11
0.31
0.17
-0.26
-0.29
0.04
0.16
-0.05
-0.34
-0.28
-0.06
0.36
-0.07
-0.26
-0.23
-0.04
0.49
-0.05
0.14
0.22
0.42
0.54
2005-2006
2005-2007
2005-2008
2005-2009
(iii) Supermarkets that dropped Danish products
0.09
-0.05
0.21
-0.06
0.40
0.05
0.44
0.07
0.13
0.28
0.35
0.38
Note:
Intensive = Growth in sales of products that are common in both periods: (C t-Cs)/Cs
Extensive = Creation – Destruction: Nt/Vs-Ds/Vs
What we observe is that intra-brand adjustments for the brands that experienced a negative (the
Danish) and a positive (the non-Danish) demand shock took place at opposite margins. The drop
in sales of Danish brands in 2006 is explained almost entirely by the intensive margin, whereas
15
gains by non-Danish brands by the extensive margin. The decomposition suggests that Danish
brands continued to sell the same products, but revenue per product dropped as a result of the
boycott. In contrast, non-Danish brands expanded their product mix in order to fill the void left
by the absence of Danish products and attract new customers. Interestingly, for the non-Danish
brands revenue per existing product decreased in all but one year, suggesting that the
introduction of new products came at the expense of existing ones. We check robustness by
repeating the analysis separately for the group of supermarkets that re-introduced Danish
products and those that did not. As shown in Panels (ii) and (iii), the results do not change.
Figure 6 –Brand Revenue Growth and Brand Size
*
Based on brand revenue in January 2006 (in Saudi Riyals)
An interesting finding is that not all non-Danish brands benefited from the boycott against Danish
products. Half of them grew in revenue, and half of them either shrank or disappeared. We
illustrate this in Figure 6 (left panel), where we rank non-Danish brands according to sales
growth between January 2006 and January 2007. We also plot the size of each brand pre-boycott
16
(January 2006) on the right panel. Throughout the paper we measure size by brand revenue in
January 2006, the month before the boycott took effect.
As the figure reveals, the post-boycott experience of the brands varied substantially. Moreover,
their performances were not related to brand size. The three brands that accounted for most of
sales experienced moderate sales growth, but the biggest winners and losers were all small
brands by market size. What could account for this?
Cheese category breakdown
In order to better understand the entry and exit process, we turn to a more detailed analysis of
the data at the item level. Nielsen divides up products in the processed cheese category into seven
segments on the basis of packaging: glass jar, plastic jar, slices, triangles, squares, tubes, and tin
(cans). Glass jars contain cheese spreads and are by far the largest segment with about 45% of
the product category (Table 4). Slices, triangles, tins and squares have 10-15% of the market each.
Plastic jars and tubes have a market share of the order of 2% each.
Table 4: Market shares of product segments
2005
2006
2007
2008
GLASS JAR
0.447
0.441
0.445
0.453
TRIANGLE
0.156
0.156
0.149
0.147
TIN
0.133
0.138
0.146
0.138
SQUARE
0.119
0.120
0.124
0.119
SLICES
0.103
0.108
0.106
0.108
TUB
0.021
0.017
0.016
0.015
PLASTIC JAR
0.020
0.020
0.014
0.020
2009
0.461
0.141
0.138
0.122
0.111
0.014
0.013
Table 5 shows the market share of Danish firms within each segment. Danish firms are very
strong in the glass jar segment, where they had 38.3% of the market in 2005. They also have a
substantial presence in tubes (18.3%) and to a lesser extent squares (9.3%) and triangles (4.7%).
For the first three of these segments the market share of Danish firms roughly halves in 2006
compared to 2005, and continues to decline thereafter. Triangles are a bit different; the share
drops to less than a quarter in 2006, partially recovers in 2007, but vanishes by 2009. Danish
firms were essentially non-active in the plastic jar, slices and tin segments. The existence of these
segments is important because they serve as a (quasi) control group.
17
Table 5: Share of Danish products within segments
2005
2006
2007
2008
2009
GLASS JAR
0.383 0.194 0.186 0.142 0.161
TUB
0.183 0.081 0.059 0.061 0.048
SQUARE
0.093 0.047 0.027 0.012 0.003
TRIANGLE
0.047 0.011 0.025 0.006 0.000
TIN
0.009 0.006 0.010 0.014 0.000
SLICES
- 0.000 0.000 0.000 0.000
PLASTIC JAR
Overall
0.282 0.136 0.123 0.095 0.121
Note: a dash indicates that no Danish products are available
in that segment. Segments are ordered by Danish market
share in 2005.
Looking back at Table 4, it is interesting to note that the glass jar segment did not shrink despite
the sharp loss of sales suffered by some of its largest participants. This indicates that consumers
substituted to other products in the same segment rather than switch to other segments. Did they
substitute to existing products or new products? How did firms react to the shock? How did
Danish firms try to control the damage and what strategies did competing firms employ to take
advantage of the opportunity?
Table 6: Number of brands/items by segment
Change
2005-07
Danish share
in 2005
2005
2006
2007
2008
2009
Brands
GLASS JAR
TUB
SQUARE
TRIANGLE
TIN
PLASTIC JAR
SLICES
Active brands
16
9
5
47
9
1
16
66
19
8
5
37
8
1
18
60
25
8
5
45
11
2
20
71
23
5
6
49
10
2
22
73
23
6
4
42
10
2
20
63
9
-1
0
-2
2
1
4
38%
18%
9%
5%
1%
0%
0%
Items
GLASS JAR
TUB
SQUARE
TRIANGLE
TIN
PLASTIC JAR
SLICES
183
27
32
111
60
2
105
205
23
32
99
65
6
109
265
22
41
128
80
10
124
261
19
45
151
73
20
149
273
15
36
146
64
25
140
82
-5
9
17
20
8
19
38%
18%
9%
5%
1%
0%
0%
18
Table 6 documents entry in each segment of the cheese category. Entry targeted the segments
where Danish firms were most active. The number of brands in the glass jar segment increased
from 16 in 2005 to 19 in 2006 and 25 in 2007. The number of active brands increased in other
segments also during this period, but the amount of entry was much smaller. A similar picture
emerges when we look at number of items offered, which increased by 45% in the glass jar
segment. It seems clear that there was a lot of activity in this segment. We now focus on this
particular segment to study the behavior of both incumbent firms and new entrants.
The glass jar segment
The glass jar segment is dominated by three firms with roughly 90% of the market. Entry could
have come from completely new firms or from existing firms that were already active in other
segments. In 2006 there was a net increase of three firms in the glass jar segment. This resulted
from the entry of five new firms and the exit of two firms. One of the five entrants was a
completely new firm, while the other four were already active in other segments. In 2007 there
were seven additional entrants and one exit, giving a net entry of six firms. One of the entrants
was a completely new firm, while another had entered in 2006. All entrants were local companies,
with the exception of one international company that entered the glass jar segment in 2007.
We now turn to the behavior of major incumbents. Did they introduce new products to exploit
the hit taken by the Danish firms? The answer is no, because the bit firms were already competing
head-to-head with Danish products. Most of the increase in items offered in 2006 and 2007 were
promotions (for example, a two-pack with price reduction). Table 7 shows the number of regular
priced and promotion items that were offered in 2005 and 2007 – before and after the boycott.
The offerings are reported separately for Danish incumbent firms (that were active in 2005),
non-Danish incumbents and entrants (firms that entered in 2006 or 2007). The number of
regular priced products remained the same between 2005 and 2007for both Danish and nonDanish incumbents. But the number of promotions by non-Danish incumbents increased very
substantially, while it decreased for Danish incumbents. This is a very striking difference. It
appears that non-Danish firms very aggressively went after Danish market share. The share of
their revenue that came from promotions went from 15.1% in 2005 to 23.9% in 2007. Danish
firms offered fewer promotions but they were big in terms of volume. As a result the share of
19
revenue from promotions also increased. Finally note that entrants have relatively few
promotions (this is generally true of small firms).
Table 7: Product offerings by firm type
2005
Non-Danish
incumbents
Danish
Entrants
Regular Promotions
% revenue
from
promotions
-
-
73
26
-
41
43
2007
Regular Promotions
0.151
0.283
72
26
43
80
31
13
% revenue
from
promotions
0.239
0.355
0.146
We have already seen evidence that entry targeted the glass jar segment, where Danish firms had
their biggest presence. We now consider the precise mode of entry. Did new competitors match
Danish offerings (head-to-head competition) or did they try to fill in the blanks in the product
space? The table below shows that entry targeted the sizes dominant sizes. The sizes are ranked
by their market shares. Danish firms had 86.9% of their sales in 2005 in the first six sizes. Note
also that entry covered most of the major sizes: most entrants offered the .5, .24 and .14 sizes and
half of them offered the .91 size also.
VI. Demand analysis (sketch)
We want to look at demand before and after the boycott; for Danish and non-Danish products;
and for droppers vs keepers (outlets that permanently dropped Danish products versus those
that re-stocked them).
We specify demand for product in store at time as:
(4)
where
ln
(… ) means that
=
+ (… ) ln
+ controls +
can vary across products/stores/time. Control variables include
month dummies and origin-specific trend and effect of Ramadan. Estimates of elasticities are
summarized in Table 8. Overall, elasticities are higher for non-Danish product and for dropping
outlets. Post-boycott, elasticities for non-Danish products are lower than before, while for Danish
product elasticities in keeper stores are higher.
20
Table 8: Estimated elasticities from equation (4)
Before
Non-Danish
-2.273
**
-2.270
**
Equality test
p-val
0.455
Non-Danish
-0.756
**
-0.632
**
0.000
Danish
Droppers
-2.414
0.235
**
Danish
-0.132
†
-0.201
**
0.000
Droppers
After
0.038
Keepers
0.031
Keepers
0.078
0.038
0.031
0.078
Equation (4) is a demand equation that allows for demand rotations. We estimated a variant of
(4) that allows for demand to shift rather than to rotate:
(5)
ln
controls +
=
+ (… ) ln
We obtain estimates
+
= −.176 and
∗ (1 −
)+
(
∗
)+
= +.126. The estimates capture the magnitudes of the
demand decline and increase for Danish and non-Danish brands respectively.
The welfare impact of the boycott is substantial. The overall impact on producers is likely neutral
but there was a significant re-distribution, with large losses for targeted firms and big gains by
competitors. Retailer participation implies some consumer welfare loss as some consumers are
unable to purchase their preferred product. There is a seemingly permanent change in
preferences: decreased demand for some products, increased demand for others. The welfare
impact of this change depends on the shape of the demand curves, but likely to be small (zero in
some models) if total sales stay the same.
One way to assess the magnitude of the demand shock is to ask the following question: how much
would prices have to change to induce the same reduction in sales as produced by the boycott?
21
The answer can be obtained with our estimates of equation (5). The drop in demand for Danish
products is equivalent to price multiplying by a factor of 2.35; the rise in demand for non-Danish
is equivalent to what would be observed after an 18% price decline.
VII. Conclusion
The boycott against Danish products in Saudi Arabia in 2006 had a devastating impact on sales
of Danish cheese products in the country. The market share of Danish firms dropped from 16.5%
the month before to less than 1% the month after. Even after the boycott was called off, Danish
market share never fully recovered: in 2009 Danish firms only accounted for 6.3% of all cheese
sales in Saudi Arabia.
We use monthly scanner data on sales of cheese products across hundreds of retail outlets in
Saudi Arabia to document the impact of the boycott. In contrast to past work on boycotts, we
find that the boycott had a strong impact, both in the short run and the long run. Because we
have access to highly disaggregated data on sales of cheese products at each store, because the
boycott did matter, and because the start date and end date of the boycott are well defined, we
are able to then explore how Danish and non-Danish firms changed prices and their product mix
in response to the boycott. Our results bring to light several asymmetries. Firstly, Danish firms
kept the same number of products, whereas non-Danish firms responded by introducing new
products. Secondly, half of the non-Danish brands benefited but the other half did not. Thirdly,
the handful of well-known brands that accounted for the majority of sales benefited from raising
their prices but also from the increased competition that resulted from the introduction of new
products. In contrast, the increased competition hurt sales of fringe brands, especially those that
raised prices in response to the boycott.
22
Bibliography
Ashenfelter, Orley, Stephen Ciccarella, and Howard J. Shatz, 2007, "French Wine and the US
Boycott of 2003: Does Politics Really Affect Commerce?", Journal of Wine Economics 2(1): 55-74.
Borenstein, Severin and Martin B. Zimmerman, 1988, “Market Incentives for Safe Commercial
Airline Operation”, American Economic Review, 78(5), 913-935.
Baldwin, John, and Wulong Gu, 2009, "The impact of trade on plant scale, production-run length
and diversification," in Producer dynamics: New evidence from micro data. University of Chicago
Press, 557-592.
Baron, David P., 2001, "Private politics, corporate social responsibility, and integrated
strategy." Journal of Economics & Management Strategy 10(1): 7-45.
Baron, David P., 2003, "Private politics." Journal of Economics & Management Strategy 12(1): 3166.
Bernard, Andrew B., Stephen J. Redding, and Peter K. Schott. 2010, Multi-product firms and
product switching. American Economic Review, 100(1): 70-97.
_______________ , 2011, “Multi-Product Firms and Trade Liberalization,” Quarterly Journal of
Economics, 126 (3): 1271-1318.
Bosch, Jean-Claude, E. Woodrow Eckard, and Vijay Singal, 1998, “The Competitive Impact of Air
Crashes: Stock Market Evidence”, Journal of Law and Economics, 41(2): 503-519.
Broda, Christian, and David Weinstein, (2010). “Product Creation and Destruction: Evidence and
Price Implications”, American Economic Review 100: 691-723.
Cawley, John and John A. Rizzo, 2008, “Spillover effects of prescription drug withdrawals,” in
Lorens Helmchen, Robert Kaestner, Anthony Lo Sasso (ed.) Beyond Health Insurance: Public
Policy to Improve Health (Advances in Health Economics and Health Services Research, Volume 19),
Emerald Group Publishing Limited, 119–143.
Chavis, Larry, and Phillip Leslie, 2009, "Consumer boycotts: the impact of the Iraq war on French
wine sales in the US." Quantitative Marketing and Economics 7(1): 37-67.
Chevalier, Judith A., Anil K. Kashyap and Peter Rossi, 2003, “Why Don’t Prices Rise During Periods
of Peak Demand? Evidence from Scanner Data,” American Economic Review, 93(1): 15-37.
Clerides, Sofronis, Peter Davis and Antonis Michis, 2015, “National Sentiment and Consumer
Choice: The Iraq War and Sales of US Products in Arab Countries,” Scandinavian Journal of
Economics, 117(3): 829:851.
Copeland, Adam and George Hall, 2011, “The Response of Prices, Sales, and Output to Temporary
Changes In Demand,” Journal of Applied Econometrics, 26: 232–269.
23
de Fusco, Richard A. and Scott M. Fuess, Jr., (1991), “The Effects of Airline Strikes on Struck and
Nonstruck Carriers”, Industrial and Labor Relations Review 44(2): 324-333
Eckel, Carsten, et al. (2011), "Multi-product firms at home and away: Cost-versus quality-based
competence."
Eckel, Carsten, and J. Peter Neary (2010), "Multi-product firms and flexible manufacturing in the
global economy", The Review of Economic Studies 77(1): 188-217.
Friedman, Monroe (1991), "Consumer boycotts: A conceptual framework and research
agenda." Journal of Social Issues 47(1): 149-168.
_______________ (1999), Consumer boycotts: effecting change through the marketplace and media.
Routledge.
Freedman, Seth, Melissa Kearney, and Mara Lederman (2012), “Product Recalls, Imperfect
Information, and Spillover Effects: Lessons from the Consumer Response to the 2007 Toy
Recalls,” Review of Economics and Statistics , 94(2): 499-516 .
Gagnon, Étienne and David López-Salido, (2014), “Small Price Responses to Large Demand
Shocks,” Federal Reserve Board Finance and Economics Discussion Series 2014-18.
Goldberg, Pinelopi K., et al. "Multiproduct firms and product turnover in the developing world:
Evidence from india." The Review of Economics and Statistics 92.4 (2010): 1042-1049.
Grafton, Steven M., George E. Hoffer and Robert J. Reilly, 1981, “Testing the Impact of Recalls on
the Demand for Automobiles,” Economic Inquiry, 19(4): 694–703.
Hallak, Juan Carlos. Domestic firms vs. Multinationals: the Effects of Integration. mimeo, 2000.
Hatzitaskos, Kostis and Aviv Nevo (2006), “Why Does the Average Price Paid Fall During High
Demand Periods?” NBER Working Paper 11572.
Hendel, Igal, Saul Lach and Yossi Spiegel, 2015, “Consumers’Activism: the Facebook boycott of
Cottage Cheese”, CEPR discussion paper 10460.
Hong, C., W.-M. Hu, J. E. Prieger, and D. Zhu, 2011, “French Automobiles and the Chinese Boycotts
of 2008: Politics Really Does Affect Commerce,” The B.E. Journal of Economic Analysis & Policy
(Topics), 11(1), Article 26.
Innes, Robert. “A Theory of Consumer Boycotts under Symmetric Information and Imperfect
Competition,” The Economic Journal 116 (2006): 355-381.
John, Andrew, and Jill Klein (2003), "The boycott puzzle: consumer motivations for purchase
sacrifice." Management Science 49(9): 1196-1209.
Michaels, G., and X. Zhi, 2010, “Freedom Fries,” American Economic Journal: Applied Economics,
2(3): 256-281.
24
Pandya, Sonal S. and Rajkumar Venkatesan (forthcoming), “French Roast: Consumer Response
to International Conflict—Evidence from Supermarket Scanner Data”, The Review of Economics
and Statistics.
Reilly, Robert J. and George E. Hoffer (1983) “Will Retarding the Information Flow on Automobile
Recalls Affect Consumer Demand? “ Economic Inquiry, 21(3): 444–447.
Sen, Sankar, Zeynep Gürhan-Canli, and Vicki Morwitz (2001), "Withholding consumption: a social
dilemma perspective on consumer boycotts." Journal of Consumer Research 28.3: 399-417.
25