In business with Wal-Mart;

(1)

In business with Wal-Mart;

How does dependence on Wal-Mart affect firms’ stock prices?

Student Name: T. Xi

Student Number: s1553062

Master Program: Operations and Supply Chains Faculty: Economics and Business

University: Rijksuniversiteit Groningen (RUG) First supervisor: Boyana Petkova

Second supervisor: Dr. Lammertjan Dam

(2)

1

Abstract

There are some academic researches on the case of Wal-Mart such as the dramatic growth of Wal-Mart and also Wal-Mart’s market power in relation to its suppliers, but these previous study findings have yielded mixed results whether it is good or bad to be in business with Wal-Mart. On one hand firms who are in business with Wal-Mart can benefit from increased sales volume and increased efficiency. On the other hand Wal- Mart has a good reputation and strong power of being very demanding to its suppliers and squeezing value out of them. This paper deploys a dynamic perspective to investigate how doing business with Wal-Mart affects performance. We conduct an event study using the REVERE Wal-Mart Top 30 supplier index. This index was constructed in 2006 and tracks the performance of the 30 firms that are most dependent on Wal-Mart (in terms of percentage sales volume).

We investigate how increased dependence on Wal-Mart (i.e. an entry in the index), potentially followed by decreased dependence (i.e. an exit from the index), affects the stock prices of suppliers of Wal-Mart. We consider the 87 entries and exits from the creation of the index (6 December, 2006) until July 13th 2011.

We furthermore explores if the stock prices of suppliers react differently based on differences in their resource dependence setting. We hypothesize that as Wal-Mart becomes more powerful, the stock price reaction to an index entry and exit will be increasingly negative.

Keywords: Wal-Mart, supply chain, event study, stock price, power, dependence,

performance

(4)

3

1. Introduction

Although the relations between Wal-Mart and its suppliers have been studied in literature, previous studies have yielded mixed results whether it is beneficial for suppliers to be in business with Wal-Mart. In this paper we conduct an event study and investigate how doing business with Wal-Mart affects the stock price of suppliers. We propose that announcements regarding increased respectively decreased dependence on Wal-Mart increase respectively decrease the stock price of suppliers of Wal-Mart. We attempt to explain variances in stock price reactions from two factors. First, we test if the stock price reaction to such announcements has become larger over time. Second, we test if stock prices of suppliers react differently based on differences in their resource dependence setting. We expect that the higher total dependence (i.e. the sum of resource dependences of Wal-Mart and a supplier) is, the more positive respectively less negative stock market reactions are to announcements regarding increased respectively decreased dependence.

The higher power imbalance (i.e. the difference in resource dependences) is, the less positive respectively more negative stock market reactions are to announcements regarding increased respectively decreased dependence.

Literature has been unclear if it is beneficial to be dependent on Wal-Mart. On one hand, firms who are increasingly dependent on Wal-Mart for their sales usually have higher total sales than non-Wal-Mart suppliers, better efficiency and higher process innovation (Vickery et al., 2003; Kulp et al., 2004; Sahin and Robinson, 2005; Christiaanse, 2005;

S. Mitra, V. Singhal, 2008). On the other hand, being in business with Wal-Mart also has its risks. As the dependence of suppliers on Wal-Mart grows, Wal-Mart is increasingly demanding towards its suppliers and squeezes value out of them (Bloomand, 2001;

Useem, 2003). Furthermore, because suppliers have to adjust their production to

accommodate the capacity needed by Wal-Mart, decreased dependence on Wal-Mart can

leave suppliers with excess capacity and thus harm their performance. These oppositional

views on how being dependent on Wal-Mart affects supplier performance leave a

theoretical gap. Our paper addresses this gap on if being in business with Wal-Mart is

beneficial or if it harms supplier performance. This research will increase the

understanding of firms who may want to be in business with Wal-Mart on how to set up

their relationship with Wal-Mart.

(5)

4

Our paper also attempts to explain if stock market reactions to increased respectively decreased dependence announcements vary due to two factors. First, we investigate if the effects of becoming more respectively less dependent on Wal-Mart have changed over time. Such effects could be expected since the interest of investors in Wal-Mart has increased over time. Second, we investigate if stock market reactions differ due to the resource dependence setting in which firms already operate. Based on resource dependence theory, we pose that the higher total dependence is, the more positive respectively less negative stock market reactions are to announcements regarding increased respectively decreased dependence. As total dependence is higher, stock prices react more positively, because higher total dependence promotes relationships between buyers and suppliers that are more likely to be mutually beneficial (Gulati and Sytch, 2007). We also posit that the higher power imbalance is, the less positive respectively more negative stock market reactions are to announcements regarding increased respectively decreased dependence. As power imbalance is higher, stock prices react less positively, because higher power imbalance brings about more conflicts, the supplier faces increasingly undesirable exchange conditions and Wal-Mart cares less about the pursuit of mutual benefit.

This paper makes two contributions. First, this paper complements prior studies that have only compared the performance of Wal-Mart suppliers with the performance of non-Wal- Mart suppliers and have not investigated how doing more respectively less business with Wal-Mart will influence the performance of current Wal-Mart suppliers. Second, our paper links dependence to stock market reactions. Although stock prices are the ultimate measure of firm performance, the effects of increased/decreased dependence on stock prices have not yet been studied.

We will adopt an event study in this paper to investigate the relationship between

announcements of entries and exits from the REVERE Wal-Mart Top 30 supplier index

and stock price reactions of suppliers. This index was constructed in 2006 and tracks the

performance of the 30 firms that are most dependent on Wal-Mart (in terms of percentage

sales volume). An entry respectively an exit signals increased respectively decreased

(6)

5

dependence. We consider 87 entries and exits of American firms that occurred from the creation of the index until the last time the index was changed (July 13th 2011).

The rest of our paper is organized as follows. In section 2 we develop our theoretical framework and several hypotheses. Section 3 reports the data set and methodology.

Section 4 describes our results. In section 5 we end with the discussion and conclusion.

2. Theoretical framework and hypotheses

In this paper we study announcements regarding increased respectively decreased dependence on Wal-Mart increase respectively decrease the stock price of suppliers of Wal-Mart. We furthermore explore if the stock price reaction to such announcements has become larger over time and if the stock prices of suppliers react differently based on differences in their resource dependence setting. We describe the theories central to our study and the discussion is followed by the development of theoretical model and related hypotheses.

2.1. Changes in dependence: Addition and deletion to the Wal-Mart Top 30 supplier index

To find suppliers who become more relatively less dependent on Wal-Mart, we track the Wal-Mart top 30 supplier index created by ISE-REVERE. The ISE-REVERE Wal-Mart Supplier Index (ISE-REVERE-WMX) provides a benchmark for investors interested in tracking public companies deriving a significant percentage of sales from Wal-Mart (ISE- REVERE, 2006). Semi-annually firms are added to respectively deleted from the index.

Being added to the index signals increased dependence, while being deleted signals decreased dependence. Changes are announced on the publicly available website and are picked up by investors (ISE- REVERE, 2006).

Being added to the Wal-Mart Top 30 index has several advantages. First, suppliers in the

index have a stronger connection to Wal-Mart than others. They can cooperate with each

other much better, because they will know more about each other’s internal information

(7)

6

flow clearly. Second, suppliers in the index share more resources with Wal-Mart than others. Suppliers sometimes could share Wal-Mart’s third parties that provide needed ancillary services such as financial institutions, transportation and delivery services, law firms and system integrators (Mitra and Singhal, 2008). Third, suppliers’ revenue probably also has risen, because it is probably most feasible when business with Wal- Mart is relatively large, in other words it is beneficial for suppliers to become more dependent on Wal-Mart. We expect that stock prices of suppliers will increase as they become more dependent on Wal-Mart:

Hypothesis 1(a). Addition to the Wal-Mart Top 30 supplier index is associated with a positive stock market reaction.

Being deleted from the Wal-Mart Top 30 index has serious disadvantages. First, being deleted from Wal-Mart index means suppliers failed to establish a strong connection to the world’s largest retailer. Wal-Mart will reduce buying products from the supplier the revenue of the retailer is likely to drop. Second, when the business with Wal-Mart now is terminated, the supplier may be stuck with a lot of facilities and previous investments made to accommodate the large business of Wal-Mart. Consequently, we expect that stock prices of suppliers will decrease as they become less dependent on Wal-Mart:

Hypothesis 1(b). Deletion from the Wal-Mart Top 30 supplier index is associated with a negative stock market reaction.

2.2. Time effects

We expect that the impact of announcements of addition and deletion from Top 30 supplier index on stock market varies over time. Because recently Wal-Mart has grown and become a more important player in the market, the interest of investors in Wal- Mart’s announcements has increased over time. Financial markets should therefore be increasingly sensitive to Wal-Mart’s announcements of addition and deletion from the Top 30 supplier index. Accordingly, we posit the following hypothesis.

Hypothesis 2. The impact of announcements of addition and deletion from the Wal-Mart

Top 30 supplier index has risen over time.

(8)

7

2.3. Resource dependence theory: power imbalance and total dependence

In this section we explore how the two central tenets of resource dependence theory (RDT) affect the stock price reaction of firms to an announcement of increased respectively decreased dependence. RDT describes that firms lacking in essential resources intend to set up relationships with (i.e. be dependent on) others so as to obtain their needed resources (Pfeffer and Salancik, 1978). From this bi-lateral dependence between a buying and supplying firm, the two tenets of RDT can be derived total dependence (the sum of dependences) and power imbalance (the difference in dependences).

Total dependence is defined as the sum of the dependencies between firms, represented as d1 + d2. Total dependence creates for both parties the incentive and ability to achieve successful performance outcomes. When the total dependence between Wal-Mart and its supplier is low, the necessity for both parties to maintain the relationship is low (Casciaro and Piskorski, 2005). When total dependence is high, both parties do not want to jeopardize the relationship, and the costs of coordinating the flow of goods in the relationship are more likely to weigh against the value generated in the relationship. High total dependence can create substantial incentives for Wal-Mart and its suppliers to exchange with each other. In the stock market, investors will observe the fact that higher total dependence results in higher mutual beneficial behaviors. Nevertheless, when total dependence is low, no very strong coordination and long-term exchange are between Wal-Mart and its suppliers. The supplier is of little concern to the obstacles to negotiated exchange, because the supplier who does not depend on Wal-Mart so much for critical resources that can be easily procured from others. Consequently in the stock market, low total dependence will result in a less negative effect for the deleted firms. Hence, based on the view, we pose:

Hypothesis 3(a). As total dependence is higher, the stock market reaction to announcements regarding added firms is more positive.

Hypothesis 3(b). As total dependence is lower, the stock market reaction to

announcements regarding deleted firms is less negative.

(9)

8

Power imbalance is defined as the difference of the dependencies between firms. If the power imbalance was high and in favor of the buyer (i.e. buyer dependence << supplier dependence), when the supplier was added to the Wal-Mart Top 30 supplier index, the supplier would be constrained by Wal-Mart under the condition of unequal power even though adding to the index would post a positive impact on the supplier. Reflected in stock market, the financial performance of added firms would be less positive. However, if the power imbalance was high and in favor of the supplier (i.e. supplier dependence <<

buyer dependence), when the supplier was added to the Wal-Mart Top 30 supplier index, it would be more advantageous for the supplier in exchange with Wal-Mart more closely and more frequently. Reflected in stock market, the financial performance of added firms would be more positive.

If the power imbalance was high and in favor of the buyer (i.e. buyer dependence <<

supplier dependence), when their exchange failed and the supplier was deleted from the Wal-Mart Top 30 supplier index, the supplier would be in worse exchange conditions and face greater uncertainty because it could hardly find critical alternative resources or services immediately. Reflected in stock market, the financial performance of deleted firms would be worse. However, if the power imbalance was high and in favor of the supplier (i.e. supplier dependence << buyer dependence), the supplier would be less negatively influenced by being deleted from the Wal-Mart Top 30 supplier index because the supplier is not so dependent on Wal-Mart compared with Wal-Mart’ dependence on the supplier, and deletion from the index would not bring about a substantial disadvantage to the supplier. Reflected in stock market, the financial performance of deleted firms would be less negative.

As mentioned above, in our event study, now we propose that:

Hypothesis 4(a). As power imbalance is more in favor of the supplier, the stock market reaction to addition announcements is more positive.

Hypothesis 4(b). As power imbalance is more in favor of the supplier, the stock market

reaction to deletion announcements is less negative.

(10)

9

3. Methodology

3.1. Data sample

To identify Wal-Mart and its top 30 suppliers we exhaustively searched ISE-REVERE Wal-Mart Supplier Index (WMX) for announcements established between 6 December 2006 (the founding date) and July 2011. This search resulted in 87 entries and exits totally (59 entries are firms added to Wal-Mart Top 30 Index and 28 exits are firms deleted from Wal-Mart Top 30 Index). However, we excluded 7 entries and 2 exits from our sample. In the addition group, CNS, Inc. and Spectrum Brands were removed because they did not exist anymore as they were acquired by GlaxoSmithKline Consumer Healthcare and Spectrum Brands Holdings Inc. respectively in 2006. Activision, Inc., Lance, Inc., K2 Inc. and Alberto-Culver Co. were deleted. These firms did not exist anymore, so we could not find stock prices on and around their announcement days and do further calculations. For Jarden Corp, there was no percentage of net sales to Wal-Mart for us to calculate total dependence and power imbalance. In the group with deletions, CNS, Inc. and Spectrum Brands were deleted for the same reason. The remaining 78 entries and exits formed our final sample in Table 1.

Table 1 describes our final sample. Panel A shows the addition firms with their announcement dates. Panel B shows the deletion firms with their announcement dates.

Table 1. (Panel A)Addition firms in our final sample

(11)

10 (Panel B)Deletion firms in our final sample

No. Addition firms Announcement date No. Addition firms Announcement date

1 Handleman Company December 6, 2006 27 Kellogg Co. March 17, 2007

2 Cott Corporation ( USA ) December 6, 2006 28 Smucker (J.M.) (New) March 17, 2007

3 Chattem Inc. December 6, 2006 29 THQ Incorporated March 17, 2007

4 Cal-Maine Foods, Inc. December 6, 2006 30 Dean Foods January 14, 2008

5 Del Monte Foods Company December 6, 2006 31 Lions Gate Entertainment Cor January 14, 2008 6 Playtex Products, Inc. December 6, 2006 32 Revlon Inc A January 14, 2008 7 LeapFrog Enterprises, Inc. December 6, 2006 33 CSS Industries Inc January 14, 2008 8 Iconix Brand Group Inc. December 6, 2006 34 FGX International Holdings Ltd July 14, 2008

9 The Clorox Company December 6, 2006 35 Pactiv Corp. July 14, 2008

10 Perrigo Company December 6, 2006 36 Smart Balance Inc July 14, 2008

11 Helen of Troy Limited December 6, 2006 37 Elizabeth Arden Inc July 14, 2008 12 Lifetime Brands Inc. December 6, 2006 38 Revlon Inc A January 12, 2009 13 Prestige Brands Holdings, Inc. December 6, 2006 39 American Italian Pasta Co January 12, 2009

14 Hasbro, Inc. December 6, 2006 40 Dean Foods Co July 13, 2009

15 Coinstar, Inc. December 6, 2006 41 Campbell Soup Co January 11, 2010

16 Tootsie Roll Industries, Inc. December 6, 2006 42 Mattel Inc January 11, 2010 17 Navarre Corporation December 6, 2006 43 Ralcorp Hldgs Inc January 11, 2010 18 Mattel, Inc. December 6, 2006 44 Lancaster Colony Corp January 11, 2010 19 Central Garden & Pet Co. December 6, 2006 45 Clorox Co July 14, 2010 20 Church & Dwight Co., Inc. December 6, 2006 46 Iconix Brand Group July 14, 2010 21 Energizer Holdings, Inc. December 6, 2006 47 Kellogg Co July 14, 2010 22 The Scotts Miracle-Gro Co. December 6, 2006 48 Green Dot Corp July 13, 2011 23 American Italian Pasta Company December 6, 2006 49 MoneyGram Intl Inc July 13, 2011 24 Diamond Foods, Inc. December 6, 2006 50 Spectrum Brands Holdings Inc July 13, 2011 25 General Mills, Inc. December 6, 2006 51 Majesco Entertainment July 13, 2011

26 Flowers Foods March 17, 2007 52 TreeHouse Foods Inc July 13, 2011

No. Deletion firms Announcement date No. Deletion firms Announcement date

1 Iconix Brand Group Inc. March 17, 2007 14 CSS Industries Inc July 13, 2009

2 Navarre Corporation March 17, 2007 15 Clorox Co January 11, 2010

3 American Italian Pasta Company March 17, 2007 16 Helen of Troy Ltd January 11, 2010

4 Central Garden & Pet January 14, 2008 17 Kellogg Co January 11, 2010

5 Handleman Company January 14, 2008 18 Lions Gate Entertainment January 11, 2010

6 Lifetime Brands Inc January 14, 2008 19 Campbell Soup Co July 14, 2010

7 The Scotts Miracle-Gro Co. January 14, 2008 20 Kraft Foods Inc A July 14, 2010

8 THQ Incorporated January 14, 2008 21 Lions Gate Entertainment July 14, 2010

9 Dean Foods July 14, 2008 22 Mattel Inc July 14, 2010

10 Mattel, Inc. July 14, 2008 23 Diamond Foods July 13, 2011

11 Revlon Inc A July 14, 2008 24 Central Garden & Pet Co A July 13, 2011

12 Cliffs Natural Resources Inc January 12, 2009 25 B&G Foods Inc July 13, 2011

13 Conexant Systems January 12, 2009 26 SMART BALANCE July 13, 2011

(12)

11

3.2. Calculating abnormal returns

The event study methodology is applied in this paper to estimate the changes in stock price (the abnormal return) attributable to the announcements of addition respectively deletion. The event is defined as the date when Wal-Mart announces added/deleted firms on its Top 30 Index.

With the announcements categorized, we first selected the event window that is a period over which stock prices react to an event. We defined the announcement day is t

₀

, the trading day preceding t

₀

is t

₋₁

, and the trading day after t

₀

is t

₁

, etc. Sometimes informational spillovers may occur before the official publication, and sometimes informational delay may also occur, so we chose and examined a long-term effect by using a 20-day event window from t

₋₁₀

to t

₁₀

. The next step was to specify an estimation period. We chose -100 to -20 which is large enough to ensure the standard normal return.

We selected 20 days prior to the announcement day, because it could produce the least bias.

For all the events, we first collected the price indexes (PI) for the firms through DATASTREAM for a time period before and after the event (i.e. the event window). The stock return is calculated by the following formula:

₋₁

−1

where

is the return of stock i on date t, where

is the price of stock i on day t, and where

₋₁

is the price of stock i on day t-1.

The abnormal return is the actual ex post return of the security over the event window minus the normal return of the firm over the event window. The normal return is defined as the expected return without conditioning on the event taking place. For firm i and event date t the abnormal return is below:

𝐴

=

- E(

|𝑋 )

where 𝐴

and E(

|𝑋 ) are the abnormal, actual and normal returns respectively

for time period t. 𝑋 is the conditioning information for the normal return model

(Mackinlay, 1997). Abnormal returns are normally calculated by both the market model

(13)

12

and mean adjusted model according to Brown and Warner (1985). In the market model, 𝑋 is the market return. In the mean model, 𝑋 is a constant (Mackinlay, 1997). Although it is common to use the market model in estimating abnormal returns, we use the mean adjusted model returns to account for any potential bias as well. The potential bias is that stock returns may be normally larger for firms with better operating performance. The mean adjusted model accounts for such biases by subtracting the mean daily return of this stock. Thus, in analysis of Hypothesis 1, we will present both the market model and the mean model.

To estimate abnormal returns around event dates, we start by setting up the following market model regression which is a linear relationship between the return on a stock and the return on the market portfolio over a given time period:

𝑟

𝛼 + 𝛽 𝑟

_𝑚

+ 𝜀

where 𝑟

is the return of stock i on day t

,

𝑟

_𝑚

is the return of market portfolio (NASDAQ EXCHANGE) on day t, 𝛼 is the intercept of the relationship for stock i

,

𝛽

is the slope of the relationship for stock i

,

𝛽 𝑟

_𝑚

is the return to stock i on day t which can be attributed to market wide movements and 𝜀

is the unexplained part of the return that captures the effect of firm specific events on day t (Mitra and Singhal, 2008). By estimating the above equation for each firm, we obtain the estimated parameters 𝛼 ̂ and 𝛽

_𝑖

̂

_𝑖

by using ordinary least squares (OLS) regression over the estimation period. To calculate the abnormal return for the stock of each firm on day t, we use the market model defined as:

𝐴

𝑟

- 𝛼 ̂ - 𝛽

_𝑖

̂𝑟

_𝑖 _𝑚

In contrast to the market model, abnormal return also can be predicted by the mean adjusted model by subtracting the mean return of the stock over the estimation period (- 100 to -20) from the actual return on stock i on day t. The formula is as following:

AR

_it

𝑟

- 𝑟 ̅

_𝑖

In order to find for leakage of information or market reaction to announcements delay, the daily abnormal return is calculated during our event window from -10 to +10 as:

𝐴 ̃ ∑

^𝑁₌₁^𝑡

𝐴

/

𝑁

(14)

13

where 𝑁 is the

number

of firms in the sample for which 𝐴

values are available on day t (Mitra and Singhal, 2008).

We use both two methods to test abnormal returns of firms added to respectively deleted from Wal-Mart Top 30 index. After getting daily abnormal returns in market model and mean model, daily mean of abnormal return will be tested through z test in order to know the effects of Wal-Mart’s announcements on the stock price of firms added on respectively kicked out from that index.

For Hypothesis 1, under the null hypothesis that announcements have no impacts on market value, t t calculate Z test to find the signifiance of the abnormal returns for both the mean and the market model.

Z value =

^̅̅̅̅̅̅

^/

,

Var (𝐴 ̅̅̅̅̅) =

_𝑁¹

∑

^𝑁₌₁

𝑟 .

3.3. Calculating cumulative abnormal returns (CAR)

We also investigate the cumulative abnormal returns (CAR) in our paper to see the aggregate returns over time and across firms. The formula for the CAR is as below:

𝐴 ̅̅̅̅̅̅

₁

) ∑ 𝐴 ̅̅̅̅

=

This formula represents that CAR is calculated based on the abnormal returns of firms over a period. To test null hypothesis that if mean abnormal performance is equal to zero, we define the variance of CAR as following:

𝑟 𝐴

₁

))

𝑁

∑

𝑁

=1

₁

)

Where,

₁

) indicates the variance of abnormal returns for individual firms over period t

₁

and t .

To test Hypothesis 1, now we have no knowledge about information spillover or delay,

but we should know that whether we use AR(0) is precise or not, so we will use CAR

(15)

14

curves to take a look at. The value of two-tailed 5% confidence interval of Z test is 1.65 and -1.65. According to the transformation of Z test formula below, we can have the series values of upper and lower boundaries of 5% confidence interval.

√ AR) √ AR)

If the CAR exceeds the upper or lower boundary which means there is an information leakage or delay. If not, it means using AR(0) to do all our calculations is correct.

Eventually, in order to show the results clearly, we plot the line graphs which combine the upper boundary, lower boundary and CAR curves together. We choose three windows for CARs, which are (-1, 2), (-3, 3) and (-2, 6). The window (-1, 2) is close to day 0, and not extended largely, so it best reflects effects of events around day 0. However, because the index announcements may not immediately impact the stock market, due to low information spreading speed, or time needed by investors to analyze, the market reactions to the events can delay to the later days. To prevent this effect, we also study the CARs in window between day -2 and day 6. In this window, as it includes wider finishing boundary, the significant results at later days can be detected. Moreover, information may be leaked earlier than the announcement days, and its impact on markets may happen more rapidly than that on day 0. Hence, it is necessary to develop the window between day -3 and day 3 to see if earlier effects before day 0 occur.

For Hypothesis 2 we test whether the impact of announcements of addition and deletion from Wal-Mart Top 30 index has risen over time. To estimate both of the market model and mean model we use a five month window (-20 to -170) but we perform a robustness check to control whether our results are confirmed with a different (2 month) window (- 20 to -80). The estimated model is:

| 𝐴 |

=

𝛼

₀

+ 𝛼

₁

Trendyear + 𝜀

where the dependent variable is the absolute cumulative abnormal return of stock i for

which an event of entry or exit from index occurred and trend year is an independent

variable. Since informational spillovers may occur before the announcement day we

calculate AR (0) and CAR (-1, 0).

(16)

15

3.4. Calculating total dependence and power imbalance

To test Hypothesis 3 and Hypothesis 4, we need to calculate total dependence and power imbalance. Total dependence is noted as ds + dw; power imbalance is noted as ds - dw.

We used the percentage of net sales to Wal-Mart from 10-K for each firm in announcement year as the dependence of the supplier on Wal-Mart. US based firms that get more than 10% of their revenue from one buyer have to report this on the 10-K form, so we can establish which suppliers are very dependent on Wal-Mart. However, for calculating the dependence of Wal-Mart on the supplier, first every firm should be categorized by each industry. Second, in EU KLEMS, industry total sales are listed from 1977 to 2007, so we used the newest available 2007. Third, we collected net sales for firms which are obtained in DATASTREAM according to their entry or exit year divided by industry total sales in 2007 based on each firm’s industry to get market share per firm.

And then we used market share per firm multiplies shares of per industry in Wal-Mart based on Wal-Mart annual report to obtain our final result for dependence of Wal-Mart on supplier.

3.5. The cross-sectional regression model

We make the following regression models to test Hypothesis 3 and Hypothesis 4:

Model 1: AR-Mean at day 0 = 𝛽

₀ +

𝛽

₁

(total dependence) + 𝛽 (power imbalance) Model 2: AR-Market at day 0 = 𝛽

₀ +

𝛽

₁

(total dependence) + 𝛽 (power imbalance) Where AR-Mean is the mean model on day 0 abnormal return (%) and AR-Market is the market model on day 0 abnormal return (%), since now no any evidence shows information leakage or information delay on the announcements. AR(0) is the dependent variable, and total dependence and power imbalance are independent variables in both the mean and market models.

4. Results

In this section, we first test whether addition or deletion to the Wal-Mart Top 30 supplier

index is associated to a significant abnormal return. Second, we test if the stock price

reaction to such announcements has become larger over time. In the end, we test if stock

(17)

16

prices of suppliers react differently based on differences in their resource dependence setting.

4.1. Results of addition firms

Hypothesis 1(a) states that addition to the Wal-Mart Top 30 supplier index is associated with a positive stock market reaction. To test this hypothesis, for 52 entries between 2006 and 2011, we calculated the abnormal returns for the period of 10 days prior to the event until 10 days after the event. We show the data in Table 2.

Table 2. Abnormal returns for addition firms after the effect of announcement daily average AR_mean Z test daily average AR_market Z test

-10 0.0003 0.07 0.0014 0.38

-9 -0.0026 -0.68 -0.0049 -1.37

-8 -0.0044 -1.15 -0.0021 -0.60

-7 -0.0094 -2.45 -0.0007 -0.19

-6 -0.0016 -0.41 -0.0007 -0.19

-5 0.0080 2.08 0.0037 1.03

-4 0.0046 1.20 0.0044 1.23

-3 -0.0085 -2.23 -0.0039 -1.11

-2 0.0043 1.11 0.0010 0.27

-1 -0.0035 -0.90 -0.0012 -0.34

0 -0.0006 -0.15 0.0000 -0.01

1 -0.0060 -1.57 -0.0001 -0.02

2 0.0013 0.35 -0.0019 -0.53

3 0.0070 1.83 0.0075 2.09

4 -0.0007 -0.18 0.0000 0.00

5 -0.0020 -0.53 -0.0005 -0.14

6 0.0061 1.60 0.0034 0.95

7 0.0029 0.77 0.0006 0.16

8 -0.0076 -1.98 -0.0029 -0.81

9 0.0019 0.50 0.0045 1.26

10 -0.0022 -0.58 0.0027 0.74

In this table, on day 0, we have no significant results through the mean model and the market model. The results from both two models are very similar as shown in Figure 1.

Figure 1: Trends in the mean model and market model for all addition firms

(18)

17

The first 25 events in our sample all enter on 6 December, 2006 (when the index was first created). It is possible that at the beginning the influence of the index was not very important, so the announcement of those 25 firms might not trigger a substantial market reaction. In order to avoid this bias, now we only take a look at the addition of firms from 2007 to Year 2011. Our results in Table 3 are below.

Table 3. Abnormal returns for partial addition firms after the effect of announcement daily average AR_mean Z test daily average AR_market Z test

-10 -0.0031 -0.50 0.0007 0.13

-9 -0.0027 -0.44 -0.0083 -1.45

-8 -0.0043 -0.69 -0.0026 -0.46

-7 -0.0028 -0.45 -0.0012 -0.22

-6 0.0008 0.13 0.0030 0.53

-5 0.0133 2.17 0.0089 1.57

-4 0.0054 0.88 0.0037 0.65

-3 -0.0056 -0.90 -0.0028 -0.49

-2 0.0021 0.34 0.0037 0.65

-1 -0.0085 -1.39 -0.0044 -0.77

0 0.0033 0.54 0.0014 0.24

1 -0.0104 -1.69 -0.0048 -0.85

2 0.0049 0.79 -0.0001 -0.01

3 0.0054 0.88 0.0064 1.13

4 0.0007 0.11 -0.0022 -0.38

5 -0.0032 -0.52 -0.0013 -0.22

6 0.0060 0.98 0.0050 0.88

7 0.0115 1.88 0.0066 1.15

8 -0.0052 -0.85 -0.0030 -0.52

9 0.0040 0.64 0.0061 1.08

10 -0.0050 -0.82 0.0026 0.46

-3 -2 -1 0 1 2 3

-10 -8 -6 -4 -2 0 2 4 6 8 10

Z test_mean model Z test_market model

(19)

18

It is can be seen that in both the mean model and the market model the results are very similar (Figure 2). We still cannot get significant results on day 0 for the partial addition group between 2007 and 2011. Hence, whether eliminating addition firms in 2006 or not will not bias our analysis.

Figure 2: Trends in the mean model and market model for partial addition firms

Since we have no any knowledge about information leakage or information delay, we use AR(0) in our calculations. However, in order to reduce any bias in the research, we will apply cumulative abnormal returns (CARs) to see and check whether any information spillover or delay exits. We plot CARs, lower boundaries and upper boundaries in three different windows (-1, 2), (-3, 3) and (-2, 6) respectively. We do all these three windows.

If there is no information spillover or delay, it shows that it is correct for us to use abnormal returns on day 0 to confirm or reject our hypotheses.

Since the trends of our market model and mean model are always almost the same, we will only show the figures of the mean model below. Figures 3 - 5 show full addition data (including firms added in 2006) in the mean model and Figures 6-8 show the partial addition data between 2007 and 2011 in the mean model. In the mean model with all three windows, all CAR curves are relatively flat between both 5% confidence intervals.

We are not able to obtain any significant result before/after day 0 as no any point before/after day 0 is outside the upper or lower boundary. Hence, there is no information leakage or delay and it is correct for us to do calculations with AR(0).

Figure 3: CAR and significance boundaries for window (-1, 2) _Mean model for All Additions

-2 -1 0 1 2 3

-10 -8 -6 -4 -2 0 2 4 6 8 10

Z test_mean model Z test_market model

(20)

19

Figure 4: CAR and significance boundaries for window (-3, 3) _Mean model for All Additions

Figure 5: CAR and significance boundaries for window (-2, 6) _Mean model for All Additions

Figure 6: CAR and significance boundaries for window (-1, 2) _Mean model for Additions from 2007

Figure 7: CAR and significance boundaries for window (-3, 3) _Mean model for Additions from 2007

-0.02 -0.01 0 0.01 0.02

-1 0 1 2

CAR

significance level significance level(-)

-0.04 -0.02 0 0.02

-3 -2 -1 0 1 2 3

CAR

significance level siginificance level(-)

-0.04 -0.02 0 0.02 0.04

-2 -1 0 1 2 3 4 5 6

CAR

-0.04 -0.02 0 0.02 0.04

-1 0 1 2

CAR

(21)

20

Figure 8: CAR and significance boundaries for window (-2, 6) _Mean model for Additions from 2007

Finally from our results in Table 2 and Table 3, it becomes clear that addition to the Wal- Mart Top 30 supplier index does not result in significant abnormal returns. Consequently, we cannot prove Hypothesis 1(a) that addition to the Wal-Mart Top 30 supplier index is associated with a positive stock market reaction.

4.2. Results of deletion firms

Hypothesis 1(b) states that deletion from the Wal-Mart Top 30 supplier index is associated with a negative stock market reaction. To test this hypothesis, for 26 entries between year 2006 and 2011, we calculated the abnormal returns for the period of 10 days prior to the event until 10 days after the event. For 26 exits in our final sample, our test results are in Table 4.

Table 4. Abnormal returns for deletion firms after the effect of announcement daily average AR_mean Z test daily average AR_market Z test

-10 -0.0058 -0.73 -0.0014 -0.23

-9 -0.0029 -0.36 -0.0096 -1.51

-0.04 -0.02 0 0.02 0.04

-3 -2 -1 0 1 2 3

CAR

-0.04 -0.02 0 0.02 0.04

-2 -1 0 1 2 3 4 5 6

CAR

(22)

21

-8 0.0027 0.34 0.0021 0.32

-7 -0.0033 -0.42 -0.0002 -0.04

-6 0.0050 0.63 0.0032 0.50

-5 0.0046 0.58 0.0000 0.00

-4 -0.0058 -0.72 -0.0059 -0.94

-3 -0.0063 -0.78 -0.0020 -0.31

-2 0.0017 0.21 0.0005 0.07

-1 -0.0013 -0.16 0.0035 0.55

0 0.0005 0.07 -0.0007 -0.11

1 -0.0051 -0.64 -0.0017 -0.26

2 -0.0114 -1.43 -0.0128 -2.02

3 0.0020 0.25 -0.0003 -0.05

4 0.0145 1.82 0.0118 1.86

5 0.0002 0.03 0.0008 0.13

6 0.0076 0.95 0.0121 1.90

7 0.0169 2.12 0.0083 1.31

8 -0.0086 -1.08 -0.0064 -1.01

9 0.0048 0.60 0.0072 1.14

10 -0.0134 -1.69 -0.0102 -1.61

No significant results can be obtained on day 0 in Table 4. Hence, for deleted firms, no significant result can be detected. The trends of the mean model and mean model are similar (see Figure 9).

Figure 9: Trends in the mean model and market model for deletion firms

Again, we calculate CARs for the deleted firms in the mean model. Figures 10-12 show that the CARs in the mean model are between the two significance boundaries.

Apparently, there is no any information spillover or delay, as no any point before/after

-3 -2 -1 0 1 2 3

-10 -8 -6 -4 -2 0 2 4 6 8 10

Z test_mean Z test_market

(23)

22

day 0 is beyond the upper or lower boundary. Our calculations for deletion group with AR(0) are accurate.

Figure 10: CAR and significance boundaries for window (-1, 2) _Mean model for Deletions

Figure 11: CAR and significance boundaries for window (-3, 3) _Mean model for Deletions

Figure 12: CAR and significance boundaries for window (-2, 6) _Mean model for Deletions

Finally from our results in Table 4 above, it is obvious that deletion to the Wal-Mart Top 30 supplier index does not result in significant abnormal returns. Consequently, we cannot prove Hypothesis 1(b) that deletion to the Wal-Mart Top 30 supplier index is

-0.06 -0.04 -0.02 0 0.02 0.04 0.06

-1 0 1 2

CAR

-0.04 -0.02 0 0.02 0.04

-3 -2 -1 0 1 2 3

CAR

significance level

significance level(-)

-0.06 -0.04 -0.02 0 0.02 0.04 0.06

-2 -1 0 1 2 3 4 5 6

CAR

(24)

23

associated with a negative stock market reaction.

In a word, clearly, all our addition and deletion data were detected and no any information leakage or delay. Hence, we will use AR at day 0 for the following calculations in our study.

4.3. Results of time effects

Hypothesis 2 states that the impact of announcements of addition and deletion from the Wal-Mart Top 30 supplier index has risen over time. To test this hypothesis, since the data from 2006 have not brought any bias, in Hypothesis 2 we use data from all the added and deleted firms. We show the results in Table 5 and Table 6. We see that our results are not significant. Therefore, we cannot prove the impact of announcements of addition and deletion from Wal-Mart Top 30 supplier index has risen over time. Our Hypothesis 2 is not supported.

Table 5. The impact of time trend on the absolute AR (CAR) for addition and deletion from Wal-Mart Top 30 supplier index (in the market model)

AR(0)_

market model

2 months

5 months

CAR(1,0)_

market model

2 months

5 months Trendyear -5.841E-12

(.909)

-5.453E-13 (.991)

Trendyear 5.622E-11 (.457)

6.691E-11 (.374) Constant 0.094

(.891)

0.023 (.973)

Constant -0.726 (.475)

-0.870 (.390)

Table 6. The impact of time trend on the absolute AR (CAR) for addition and deletion from Wal-Mart Top 30 supplier index (in the mean model)

AR(0)_

mean model

2 months

5 months

CAR(-1,0)_

mean model

2 months

5 months Trendyear 2.560E-12

(.962)

1.005E-11 (.854)

Trendyear 2.560E-12 (.962)

1.005E-11 (.854) Constant -0.017

(.982)

-0.118 (.873)

Constant -0.017 (.982)

-0.118

(.873)

(25)

24

4.4. Results of the cross-sectional regression

Hypotheses 3 and 4 state that as total dependence is higher respectively lower, the stock market reaction to announcements regarding added respectively deleted firms is more positive respectively less negative; as power imbalance is more in favor of the supplier, the stock market reaction to addition respectively deletion announcements is more positive respectively less negative. To test these hypotheses, we conduct a cross-sectional regression.

For all 52 added firms and 26 deleted firms in Table 7 below, in both of the mean model and market model, our results are not significant. Consequently, our hypotheses 3(a), 3(b), 4(a) and 4(b) are not supported.

Table 7. Regression results

Additions Deletions

AR on day 0 Mean Model

AR on day 0 Market Model

AR on day 0 Mean Model

AR on day 0 Market Model Total

dependence

.509 (.473)

.068 (.915)

.214 (.716)

.357 (.507) Power

imbalance

-.628 (.374)

-.198 (.755)

-.076 (.896)

-.209 (.692)

Constant .026

(.001)

.030 (.000)

-.026 (.076)

-.033 (.019)

5. Discussion

5.1. Summary of results

This paper analyzes if it is good or bad for firms to become heavily dependent on Wal-

Mart. It utilizes the event study methodology to investigate the relationship between

stock market reactions (abnormal returns) of suppliers and announcements of addition

and deletion from Wal-Mart Top 30 supplier index. For 52 respectively 27 entries (all

(26)

25

entries respectively entries without the first addition group in 2006) and 26 exits, the market reaction based on the mean model and market model is statistically insignificant.

Therefore, we are not able to conclude that addition and deletion from Wal-Mart Top 30 supplier index have any impact on stock market reaction. Moreover, we do not find indications that the impact of announcements of addition and deletion from Wal-Mart Top 30 supplier index has risen over time. Finally, we do not find that the resource dependence setting influences the abnormal returns due to an entry or exit.

5.2. Implications

According to Hendricks and Singhal, the stock market does value supply chain related activities and managers must be proactive in communicating such activities to the market. Yet, while we expected otherwise, our study indicates that news of firms added to or deleted from Wal-Mart Top 30 supplier index is “no news” to the stock market. A reason for our finding could be that Wal-Mart does not announce any substantial bonus or punishment on the addition and deletion firms. Investors might think that the net economic impact of addition or deletion to the index is likely to be very limited. For entry it could be that being added to the index is just a public act and does not indicate much higher reputation, more shared resources and higher economic value of firms.

Furthermore, investor might even think that adding to the index indicates the more control of Wal-Mart over the suppliers, and suppliers’ profits could be more squeezed by Wal-Mart, because added suppliers in the index have higher dependence and less bargain power on Wal-Mart. For exit it may be that deletion from the index will not bring about consumer boycotts, negative publicity, employee actions, pressures from Wal-Mart.

Investors might even view deletion from the index as getting away from Wal-Mart’s power over suppliers and deletion does not mean a loss on sales or profits to suppliers.

For instance, some suppliers can have bigger sales revenues from other buyers, though it is deleted from the index because of the decreasing sales revenues to Wal-Mart.

Consequently, addition does not mean a bonus and deletion never signals a loss.

Second, we do not find the impact of announcements of addition and deletion

from the Wal-Mart Top 30 supplier index has risen over time. This probably indicates the

(27)

26

importance of the index in the stock markets is low. The reason might be that the index is created based on the sales revenues of suppliers to Wal-Mart, and shareholders do not actually value the suppliers’ short-term sales data to the Wal-Mart. Instead they may be willing to pay more attention to the long-term strategic relationships between suppliers and Wal-Mart. For instance, we find that some suppliers added to the index in this period were deleted in the next period or other way around, so investors might view this just as a short-term change and also realize addition and deletion will not cause a big bonus or loss.

Hence, they become less and less concerned to the index.

Third, in our cross-sectional study, we cannot find significant proof that the resource dependence setting affects the abnormal return due to index announcements. It could be that our sample size is too small. We have only 87 entries and exits from the creation of the index (6 December, 2006) until July 13th 2011, and semi-annually firms are added to respectively deleted from the index. Normally 3 to 5 firms are added or deleted from the index per half year. That might be the reason why our study does not have sufficient power to capture the effects of both tenets of resource dependence theory exactly. Alternatively, although we believe total dependence and power imbalance affect relationship between suppliers and Wal-Mart, the index is not importantly considered by investors and we are not able to derive the significant abnormal returns in addition or deletion in the index, which therefore leads to fail in finding the significant relationship between resources dependence settings and suppliers’ financial market performance.

5.3. Limitations and further research

Our study carries some limitations and opportunities for the further research.

First, our sample size is too small as mentioned above. Maybe in the future research, a well-defined index in terms of methodology and sample size should be established before studying the abnormal returns. In this paper, the limited data might bias some findings.

Second, the Top 30 supplier index is not published by Wal-Mart itself, but by

another organization. Therefore, it is unclear if this index is immediately acknowledged

by investors and reflected in the stock market, and it is hard to decide the real day on

(28)

27

which the most market reactions exert. Namely, the exact announcement days cannot be

well defined. In the future research, the acceptability and punctuality of the index should

be taken into account. The accuracy of abnormal returns must be ensured.

(29)

28

References

Becchetti, L., Ciciretti, R. and Hasan, I. (2007). “Corporate social responsibility and shareholder’s value: An event study analysis”. Working paper.

Bloom, P.N., Perry, V.G. (2001). “Retailer power and supplier welfare: the case of Wal- Mart”. J Retail 2001; 77(3):379–96.

Bowen, D.E., Jones, G.R. (1986). “Transaction cost analysis of service organization- customer exchange”, Acad. Management Rev. Vol. 11, No.2, pp4 28-441.

Bradach J., Eccles, R. (1989). “Price, authority, and trust: from ideal types to plural forms”, Annual Rev. of Sociology, Vol. 15, pp. 97-118.

Brown, S.J., Warner, J.B. (1985). “Using daily stock returns: the case of event studies”.

Journal of Financial ECONOMICS 14 (1), 3-31.

Casciaro, T., Piskorski, M. (2005). “Power imbalance, mutual dependence, and constraint absorption: A closer look at resource dependence theory”. Administrative Science Quarterly, 50 (2005): 167-199.

Chandran, C.P., Gupta, V. (2003). “Wal-Mart’s supply chain management practices”.

ICFAI Center for Management Research.

Christiaanse, E. (2005). “Performance benefits through integration hubs”.

Communications of the ACM 48 (4), 95–100.

Gulati, R., Sytch, M. (2007). “Dependence Asymmetry and Joint Dependence in Interorganizational Relationships”, Administrative Science Quarterly, Vol. 27, No.1, pp.

33-80.

Heide, J.B., John, G. (1990). “Alliances in industrial purchasing: The determinants of joint action in buyer-supplier relationships”, J.Market. Res, Wol.27, No. 1, pp. 24-36.

Hendricks, K.B., Singhal, V.R. (2003). “The effect of supply chain glitches on shareholder value”. Journal of Operations Management 21 (5), 501-522.

MacKinlay, A. C. (1997). “Event studies in Economics and Finance”. Journal of Economics Literature, Vol. XXXV (March 1997), pp. 13-39.

Mitra, S., Singhal, V. (2008). “Supply chain integration and shareholder value: Evidence

from consortium based industry exchanges”. Journal of Operations Management 26

(2008) 96-114.

(30)

29

Mottner, S., Smith, S. (2009). “Wal-Mart: Supplier performance and market power”.

Journal of Business Research 62 (2009) 535-541.

Mudambi, R., Helper, S. (1998). “The ‘close but adversarial’ model of supplier relations in the US auto Industry”, Strategic Management Journal, Vol. 19, pp. 775-792.

Pfeffer, J., Salancik, G.R. (1978). “The external control of organizations: a resource dependence perspective. New York: Harper and Row.

Scholtens, B., Dam, L. (2006). “Banking on the equator. Are Banks that Adopted the Equator Principles Different from Non-Adopters?”. World Development Vol. 35, No. 8, pp. 1307-1328.

In business with Wal-Mart;