Do suppliers benefit from supply chain sustainability programs? -the case of Wal-Mart

Master thesis

Business administration –Operations & Supply Chains

University of Groningen

Do suppliers benefit from supply

chain sustainability programs?

-the case of Wal-Mart

By Xiaowei Wang

Email: x.wang.8@student.rug.nl

Student number: 1553070

1

Abstract1

Key words: supply chain, sustainability, event study, Wal-Mart.

A large number of studies have shown that sustainability initiatives within firms increase their financial performance. However, when it comes to supply chain sustainability it is unclear how the benefits and costs are distributed among supply chain partners. In this paper we investigate if sustainability initiatives initiated by buyers are beneficial to their suppliers. Often, a large part of implementation costs of sustainability initiatives is bestowed upon suppliers while the added benefits to them are either unclear, or clearly in favour of the buyer.

We conduct an event study in order to determine whether the performance of suppliers is affected positively, negatively, or insignificantly by supply chain sustainability initiatives (SCSIs) proposed by the buyer. We test how announcements of SCSIs impact the stock price of the main suppliers of the announcing buyer. Furthermore, we test if the reaction of the stock price of the various suppliers differs due to three factors.

First, we consider the firm size. Bigger firms have more resources and cost advantages in the implementation of the sustainability activities over others.

Second. We take into account existing sustainability risks and strengths of the suppliers. Suppliers who have taken care of sustainability within their firm prior to external pressure by the buyer will face less implementation costs and less risks of abandonment by the buyer than suppliers who have not been pursuing sustainability.

Third, we apply a resource dependence perspective to evaluate the distribution of benefits between buyers and suppliers. We test how the two central tenets of resource dependence theory (power imbalance and total dependence) affect the value that a supplier derives from the initiative. We expect that suppliers’ stock prices will react less positively as buyers grow increasingly powerful (reflected by increased power imbalance), and supplier stock prices should react more positively, as the value created in the relationship increases (reflected by increased total dependence).

In this study we focus on Wal-Mart and its main suppliers. The case of Wal-Mart is well suited for several reasons. First, Wal-Mart has implemented several SCSIs in the past years. Furthermore, Wal-Mart is widely known for its ability to organize its supplier base and optimize the supply chain. Due to the quick reaction of suppliers to buyer requests, we expect that the market response to sustainability announcements should be clearly observable for the suppliers of Wal-Mart. We investigate how 12 sustainability

1

2

3

C

1. Introduction ... 4

2. Theoretical Framework ... 6

2.1 Supply Chain Sustainability Initiatives ... 7

2.2 Effects of SCSI on Supplier stock prices ... 8

2.3 Differences in Stock Price Reactions ... 9

2.3.1 Firm Size ... 9

2.3.2 Resources Dependence: Power Imbalance and Total Dependence ... 10

2.3.3 Self-enforced Sustainability ... 11

3. Methodology ... 12

3.1 Sample ... 12

3.2 Event Study ... 14

3.2.1 Size of Abnormal Returns (AR) ... 14

3.2.2 Significance of Abnormal Returns ... 16

3.2.3 Size of Cumulative Abnormal Returns (CAR) ... 17

3.2.4 Significance of Cumulative Abnormal Returns ... 18

3.3 Cross-section Regression... 19

3.3.1 Factor Measurement ... 19

3.3.2 Regression Models ... 20

4. Results ... 22

4.1 Abnormal Return and CARs ... 22

4.2 Cross-Sectional Regression ... 27

5. Discussion and conclusion ... 29

5.1 Summary ... 30

5.2 Implications ... 30

5.3 Limitations and Future Research... 31

References: ... 33

4

1.

I

Although many studies have shown that sustainability initiatives within individual firms can improve their financial performance, few studies have investigated the financial performance effects of supply chain sustainability initiatives (SCSI) between buying and supplying firms. The SCSI means the strategic, transparent integration and achievement of an organization’s social, environmental, and economic goals in the systemic coordination of key interorganizational business processes for improving the long-term economic performance of the individual company and its supply chains (Carter and Rogers, 2008). In this paper, we investigate the impact of supply chain sustainability initiatives launched by buyers on the stock price of the suppliers. We expect that when such programs are announced by buyers, suppliers’ stock prices increase. Moreover, we predict that stock price reactions vary due to firm size, resource dependence setting (total dependence and power imbalance) and self-enforced sustainability. In specific, the larger firm size, total dependence and self-enforced sustainability, the more positive the stock price reaction of suppliers to such announcements, and the larger power imbalance, the more negative the stock price reaction of suppliers to such announcements.

5

SCSIs by buyers influence supplier performance? This research gap needs to be addressed to develop a fuller understanding of the SCSIs in supply chain dyads. The confusion also leaves supply chain managers a dilemma. While suppliers’ managers hesitate to implement the SCSIs as the benefits or costs are unknown, Buyers’ managers find hard to convince their suppliers as the evidence of benefits is lacking. This paper analyses both the theoretical gap and the practitioner dilemma to provide contribution to future theoretical development and business practice.

In this paper, we investigate how supply chain sustainability initiatives by buyers affect supplier stock prices through an event study. We expect that suppliers will benefit from supply chain sustainability initiatives proposed by buyers; i.e. that their stock price will rise when buyers announce such initiatives. Additionally we explore three factors which may impact the extent of the stock price reactions. First, we expect that supplier’ firm size positively influences their stock price because larger firms can enjoy economies of scale and scope and sustainability investments are more likely to weigh against potential benefits. Second, we expect that as suppliers have engaged more in self-enforced sustainability, their stock price will react more positively to supplier sustainability announcements by buyers because they are less likely to make additional investments in sustainability. Third, we expect that the resource dependence setting is of influence to the stock price reaction. We distinguish the effects of the two central tenets of resource dependence theory (Pfeffer and Salancik, 1978): total dependence and power imbalance. We expect that as total dependence (i.e. the sum of resource dependence of buyer and supplier) 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 predict that as power imbalance (i.e. the difference in resource dependence) is higher, stock prices react more negatively, because more power imbalance stimulates the more conflict and less interest in their relationship to perform the supply chain sustainability (Casciaro & Piskorski, 2005).

6

sustainability initiatives within firms, this article investigates supply chain sustainability. Second this paper tries to identify factors which may influence the extent to which suppliers benefit from supply chain sustainability initiatives. We provide insights to which extent the effect of SCSI is generalizable for all firms and to which extent it differs for firms.

In order to test our hypotheses, we acquire data from Wal-Mart and its main suppliers. The case of Wal-Mart is well suited to study the reaction of suppliers to supply chain sustainability initiatives for two reasons. First Wal-Mart has significantly increased its supplier sustainability efforts and has announced several supply chain sustainability initiatives over the last years. Abundant data is thus available which can be used for our study purpose. Second, suppliers are known to be responsive to requests by Wal-Mart and the stock market is known to follow such developments. Therefore, we expect that stock price reactions of suppliers to supply chain sustainability announcements by Wal-Mart should be detected, if the effect exists. Our sample includes 69 main suppliers and 12 sustainability announcements of Wal-Mart.

The paper is organized as follows: In section 2 hypotheses are presented. In section 3, the data collection and methodology are displayed. Results are stated in section 4. Section 5 contains the conclusion and discussion.

2.

T

F

self-7

enforced sustainability which can vary the stock reaction of suppliers to the SCSIs.

2.1

S

C

S

I

We define the SCSI as the strategic, transparent integration and achievement of an organization’s social, environmental, and economic goals in the systemic coordination of key interorganizational business processes for improving the long-term economic performance of the individual company and its supply chains (Carter and Rogers, 2008). Sustainability has three core aspects nature environment, society and economic performance. Environmental stewardship includes the preservation of natural resources, waste minimization and reduced emissions. Societal equity is concerned with poverty, injustice and human rights — from an operations management perspective employees’ health and safety is considered within this dimension of sustainability. The economic performance dimension assures that the economic needs of the company, workers and other stakeholders are met (Krause, Vachon and Klassen, 2009). The current sustainability literature suggests that at the intersection of social, environmental and economic performance, there are activities that organizations can engage in which not only positively affect that natural environment and society, but which also result in long-term economic benefits and competitive advantage for the firm.

8

2.2

E

SCSI

S

To investigate if SCSIs of buyers benefits suppliers, we study the effect of SCSIs on suppliers’ stock prices, as the most efficient indicator of changes in performance is stock market reaction (Brown and Warner, 1985). The efficient market hypothesis asserts that stock prices will respond rapidly to the information contained in public announcements, and that the market’s response will include the capitalization of future costs and benefits associated with this event. There is significant evidence in the finance literature that supports the efficient market hypothesis (Basu, 1997; Malkiel 2003). Moreover, in supply chain literature, many researchers have demonstrated a link between supply chain related events and stock market performance. Hendricks and Singhal (2003) use stock market reaction to study the effect of supply chain glitches, and Papadakis (2006) also found significant results in the study on financial performance after supply chain disruption. Mitral and Singhal (2007) document that the stock market does value supply chain related activities and managers must be proactive in communicating such activities in the market. Therefore, in this paper, we study the stock market reaction of suppliers on the announcements of Wal-Mart SCSIs to find if suppliers really benefit.

9

prevent image and reputation damage resulting from product, environmental waste, and worker and public safety (Shrivastava 1995). Finally, buyers should be willing to pay more for sustainable products and at a certain point they may only want to buy sustainable products, so the premium prices are able to set by these suppliers to obtain the higher profits (Bagnoli and Watts, 2002).

Overall, because there are significant benefits associated with SCSI, we expect that costs of SCSI on suppliers are outweighed by the benefits. The SCSI has a positive effect on suppliers’ performance, which will be directly reflected in the suppliers’ stock prices. We believe the first hypothesis can be as below:

H1. The announcements of Wal-Mart supply chain sustainability are positively related to the stock market reaction of suppliers.

2.3

D

S

P

R

We expect the stock price reactions of suppliers to announcements of SCSI to differ based on several factors: firm size, resources dependence setting and self-enforced sustainability. In next subsections, we will describe these factors and their expected influence.

2.3.1FIRM SIZE

10

sufficient funds and manpower to respond to stakeholders and to react to the sustainability-related pressures, larger suppliers can devote time and attention to sustainability-related details and to researching and using more exhaustive practices. Second, larger suppliers have cost advantages. Large firms generally produce large volumes (economies of scale) of many products (economies of scope). Moreover, large firms are able to produce more outputs with equal amount of inputs (technical efficiency) and higher profits (price efficiency) (Lau and Yotopoulos, 1971, 1972). Hence, they are able to absorb the increasing operation costs in short-run caused by SCSIs. In addition, larger firms have been in business long and thereby have huge capacity and capability for acquisition of knowledge (Nooteboom, 1993). In the financial market, these points may influence investors’ view on the effect of sustainability announcements on suppliers. They may think that the advantages can contribute to suppliers’ success in SCSIs. Reflected on the stock market, supplier’ stock values may positively rise. Therefore, our second hypothesis is:

H2. Stock prices of suppliers will react more positively to SCSI announcements as the firm size of suppliers is larger.

2.3.2RESOURCES DEPENDENCE:POWER IMBALANCE AND TOTAL DEPENDENCE

11

weigh against the costs of coordination (Paulraj and Chen, 2007). When SCSIs are initiated by buyers under higher total dependence, people can expect it mutual beneficial behaviour for suppliers and buyers, and it improves suppliers’ performance. Power imbalance indicates the difference in dependences in the relationship of two firms. The effect of power imbalance seems opposite to that of total dependence (Kumar, 1995). A large power imbalance indicates that one firm is much less dependent on the other firm than vice versa and interests in the relationship increasingly diverge if power imbalance grows. As the power imbalance in favour of the buyer increases, the supplier faces increasingly undesirable exchange conditions and higher levels of uncertainty (Casciaro and Piskorski, 2005). Hence, under higher power imbalance in favour of buyers, people will predict SCSIs less beneficial to suppliers. It lowers suppliers’ performance. In the financial market, investors may insight suppliers’ settings in resources dependence and suppliers’ stock reactions to the SCSI announcements can differ. Based on the literature, the following hypotheses can thus be formulated:

H3. Stock prices of suppliers will react more positively to SCSI announcements as the total dependence is larger.

H4. Stock prices of suppliers will react less positively to SCSI announcements as the power imbalance is larger.

2.3.3SELF-ENFORCED SUSTAINABILITY

12

have less risk to lose the business of buyers who value sustainability compared to suppliers who are not yet engaged in sustainability. Sustainable suppliers have the necessary technical and organizational experience and knowledge and fulfil the sustainability requirements of the buyer. Consequently they are much favoured by buyers. In the stock market, investors will view these suppliers more positively than those which have not produced sustainably. We have the following hypothesis:

H5. Stock price of suppliers will react more positively to SCSI announcements as the self-enforced sustainability is higher.

3.

M

In this section, we discuss the methods of sample and data collection and how our hypotheses are tested. First, we present how our announcements and suppliers samples are identified and collected. Second, we introduce the event study and our test of hypothesis 1 by calculating the abnormal return (AR) and cumulative abnormal returns (CAR). Final, we lay out how the factors varying ARs are measured and tested, which are our hypotheses 2, 3, 4 and 5.

3.1

S

13

of the environment. Wal-Mart has established 14 Sustainability Value Networks to ―green‖ it supply chain. Nowadays, sustainability is built into its business. It is completely aligned with its model, its mission and its culture (Heying and Sanzero, 2009). We expect that stock price reaction on suppliers of Wal-Mart to the SCSI announcements should de be clearly observed.

To identify sustainability announcements of Wal-Mart, we searched the Press Room on the website of Wal-Mart Corporate by using the string ―topic: sustainability‖ and years from January 2008 to September 2011, as this time period gives us sufficient number of announcements and includes some of most important SCSI announcements in recent years. We found a total of 89 announcements concerning the sustainability made by Wal-Mart. In order to distinguish the announcements really impacting the supply chain from the others, we scored each of announcements from 1 to 5 in which 5 indicates announcements which we expected to have the largest impact on the most suppliers. We used two scoring criteria: first, to what extent does the announcement influence all suppliers. Second, to what extent are plans concrete and does Wal-Mart plan to enforce sustainability towards suppliers. According to the criteria, we identified 13 announcements whose scores are higher than 4. Since there are two 5-score announcements on date July 16th 2009, we finally have 12 dates as our announcement data.

14

23 firms found either dead or data was unavailable in Datastream. Only 69 firms with complete data were found. For these firms, we downloaded the price index which shows the daily stock prices. Furthermore, we also collected NASDAQ stock composite data from Datastream as the market portfolio prices.

In sum, our study thus investigates how the 12 announcements mentioned above impacted the stock price of these 69 suppliers.

3.2

E

S

Through an event study (MacKinlay, 1997); we can investigate how the stock price of the 69 suppliers reacted to the 12 SCSIs by Wal-Mart. We study changes in stock prices for the event window 10 days before and 10 days after the SCSI announcements. For this event window, we capture the stock price changes on the announcement days and watch for changes happening before and after in order to not omit effects caused by information leakage and by information delay. We refer to the announcement day as day 0 or t0, and the 10 days before and 10 day after the announcement day are t₋₁₀ and t₁₀ respectively. In following paragraphs of this part, we will explain the event study precisely. First, we will lay out how we calculated the stock price reaction to the event that cannot be explained from normal stock price fluctuations (i.e. the Abnormal Return, AR, at t=0). Thereafter we will show how we tested if this AR is significant. Finally, we will demonstrate how we calculated the cumulative abnormal return (CAR) which combines the ARs from several days around the event date, and its significance. The results of AR and CAR combined will test hypothesis 1.

3.2.1SIZE OF ABNORMAL RETURNS (AR)

In this section, we present the calculation of abnormal returns in general term and in two specific models: mean model and market model.

15

the firm, and the normal return is defined as the expected return without conditioning on the event taking place. The general formula can be shown as below:

ARit = Rit− E(Rit|Xt)

Where𝐴𝑅_𝑖𝑡, 𝑅_𝑖𝑡, 𝐸(𝑅_𝑖𝑡|𝑋𝑡) are the abnormal, actual, and normal returns respectively for firm i during time period t. the special term in this formula is 𝑋_𝑡. It is benchmark of the stock returns providing certain base for the normal returns. In the different specific models, it presents differently.

In order to calculate the accurate normal returns, an estimation period must be defined. We use day -100 to day -20. This estimation period is large enough to confirm that the standard normal return can be achieved, and it has 20 days away from the announcement days to prevent any abnormal returns biasing the normal returns.

Under the thought of the general formula, many specific models are developed to calculate the abnormal returns. In this paper, two most common models are applied. They are mean model and market model.

The formula for mean model is:

𝐴𝑅_𝑖𝑡 = 𝑅_𝑖𝑡− 𝑅̅̅̅ _𝑡

Where 𝑅̅̅̅ is the average normal return over the estimation period. _𝑡

Compared this formula to the general one, the only difference is on the last term. The mean model mainly tries to find abnormal return by using actual returns minus the average returns over the estimation period. Here, 𝑋𝑡 presents the mean values of normal returns as benchmark.

16

The market model posits a linear relationship between the return on a stock and return on the market portfolio over a given estimation period (Mitra, Singhal, 2007). In this model, different from the mean model, the benchmark is the market portfolio. To have the benchmark, we use formula as below:

𝑅_𝑖𝑡 = 𝛼_𝑖+ 𝛽_𝑖𝑅_𝑚𝑡+ 𝜀_𝑖𝑡

Where 𝑅_𝑖𝑡 is the return of stock, 𝑅_𝑚𝑡 is the market portfolio on day t and 𝛼_𝑖, 𝛽_𝑖, 𝜀_𝑖𝑡 are intercept, slope and error term for each firm. By running the OLS regression, we are able to have the 𝛼̂ , 𝛽_𝑖 ̂ for each firm respectively. Using the final market model formula below, _𝑖 the abnormal returns can be calculated.

The formula for market model is:

𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− 𝛼̂ − 𝛽𝑖 ̂ 𝑅𝑖 𝑚𝑡

The last two terms indicate the market portfolio as benchmark, by using the actual returns minus the benchmark, abnormal returns can be found.

3.2.2SIGNIFICANCE OF ABNORMAL RETURNS

In order to test our hypothesis 1 in ARs, we delineate two ways of significance test in this section.

Under the null hypothesis that announcements have no impact on market values, we use simple Z- test formula to find the significance of the abnormal returns for the mean model.

𝑍 𝑣𝑎𝑙𝑢𝑒 = 𝐴𝑅̅̅̅̅̅𝑡 √𝜎2

17 𝐴𝑅̅̅̅̅̅ =𝑡 _𝑁1∑𝑁𝑖=1𝐴𝑅𝑖𝑡 𝜎2 ₌ 1 𝑁2∑𝑁𝑖=1𝜎𝑖2 .

However, a disadvantage of the formula is that since it directly uses average daily abnormal returns, if variances of individual daily abnormal returns are substantially different from one another, the test results could be biased. In the market model, owing the calculation procedures, this bias could be big. To avoid this bias, we use an alternative test statistic in the market model: aggregates standardized abnormal returns which means each observation is weighted in inverse proportion of the standard deviation of the estimated abnormal return (Khotari and Warner, 2006)2

Standardized abnormal returns are defined as:

𝐴𝑅_𝒊𝒕′ _{= 𝐴𝑅}

𝒊𝒕/𝑆(𝐴𝑅𝒊)

Where S(ARi) is the standard deviation of the abnormal return and calculated as: S(ARi) = √σi2

And significance can be tested as:

𝑧 𝑣𝑎𝑙𝑢𝑒 = 𝐴𝑅₁ 𝑖𝑡′ 𝑁 ⁄

The main difference of standardized abnormal returns from the previous one is the every individual abnormal return is divided by own standard deviations (or ―standardized‖) before calculating the average daily abnormal returns. Hence, the accuracy is improved.

3.2.3SIZE OF CUMULATIVE ABNORMAL RETURNS (CAR)

In this section we introduce the method of measuring and testing Cumulative abnormal

18

returns (CAR) which indicates that the sum of daily abnormal returns over the time period under observations. CAR can be used to prevent the bias caused by information leakage or delay before or after the announcements, if abnormal returns on other days than on day 0 are detected.

The formula for the CAR is as below:

CAR(t_x, t_y) = ∑ty AR_t t=tx

3.2.4SIGNIFICANCE OF CUMULATIVE ABNORMAL RETURNS

To test null hypothesis that the mean CAR is equal to zero, we define the variance of CAR as following:

var (CAR(t1,t2)) = _N12 ∑Ni=1σ2i (t1, t2)

Where, σ_i2(t1, t2) indicates the variance of abnormal returns for individual firms over period t₁ and t₂.

To test the hypothesis 1 in CARs, we set up first the value of two-tails 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.

upper boundary = 1.65 ∗ √var(CAR)

lower boundary = −1.65 ∗ √var(CAR)

19

windows (-3, 1) (-1, 1) (--1, 3) are defined. Because we do not have any information stressing that if announcements are leaked or sunk before or after the publishing dates and if markets absorb the announcements quickly or slowly, these three windows which covers stock market effects before , in and after the announcements respectively ensure us not to miss any impact.

3.3

C

-

R

We expect that the abnormal returns vary due to several factors which are firm size, resource dependence settings and self-enforced sustainability. In following parts, we will show how these factors are measured and how the regressions are performed.

3.3.1FACTOR MEASUREMENT

Firm Size

Prior work on Corporate Sustainability Reporting (CSR) or environmental management has found total asset, Fortune 500 ranking and revenue to be significant (Gallo and Christensen, 2011). Because of these empirical precedents, we chose to represent firm size using revenue/net sales. We derived revenue/net sales data from the Datastream.

Self-enforced Sustainability

20

Total Dependence and Power Imbalance

Power imbalance is dependence differential between two organizations, and total dependence is the sum of their dependencies (Casciaro and Piskorski, 2005). Thereof, we define their formulas as:

Power imbalance can be calculated as:

𝑃_𝑖 = 𝐷_𝑖− 𝐷_𝑤𝑖 Total dependence can be calculated as:

𝑇_𝑖 = 𝐷_𝑖 + 𝐷_𝑤𝑖 Where: 𝐷𝑤𝑖 = 𝑃𝑆𝑤𝑗∗_𝑇𝑆𝑆𝑖

𝐽

Where 𝑃𝑖 is the power imbalance value for supplier i. 𝑇𝑖 is the total dependence value for supplier i. 𝐷_𝑖 is the dependence of supplier i on Wal-Mart, and 𝐷_𝑤𝑖 is dependence of Wal-Mart on supplier i. 𝑃𝑆_𝑤𝑗 is the percentage of shares of industry j in Wal-Mart, 𝑆_𝑖 is the net sales for supplier i and 𝑇𝑆_𝐽 is the total sales of industry j. the second term in the last equation presents the market share of supplier i in its own industry.

We used yearly percentage of net sales of firms to Wal-Mart as dependence of firms on Wal-Mart. The data are collected from the 10-k forms of our sample firms. Meanwhile, in the 10-k form of Wal-Mart, we found the percentage of shares of industries in terms of Wal-Mart net sales. For the total sales of industries, we went to the database ―EU KLEMS Growth and Productivity Accounts‖ and found US total sales data per industry. Owing to the availability, we barely had data in 2007, but we expected it would not substantially bias the results, as the growth on the data in each year is very slight, and in terms of percentages, the differences are very small.

3.3.2REGRESSION MODELS

21

Model 1: 𝐴𝑅_{0𝑚𝑒𝑎𝑛} = 𝛽₀+ 𝛽₁𝑥₁+ 𝛽₂𝑥₂+ 𝛽₃𝑥₃ Model 2: 𝐴𝑅_{0𝑚𝑎𝑟𝑘𝑒𝑡} = 𝛽₀+ 𝛽₁𝑥₁+ 𝛽₂𝑥₂+ 𝛽₃𝑥₃

Where 𝐴𝑅0𝑚𝑒𝑎𝑛, 𝐴𝑅0𝑚𝑎𝑟𝑘𝑒𝑡 are dependent variables showing the abnormal returns at day 0 in mean and market models respectively. 𝑥₁ is the firm size. 𝑥₂ is the total dependence. 𝑥₃ is power imbalance. In both model 1 and 2, we use 828 entries (12 events and 69 firms per event).

Because of the difference on available data, the independent variable ―self-enforced sustainability‖ can only be tested on 624 entries (12 events, 52 firms per event). Therefore, we create the model 3 and model 4.

Model 3: 𝐴𝑅_{0𝑚𝑒𝑎𝑛} = 𝛽₀+ 𝛽₁𝑥₁+ 𝛽₂𝑥₂+ 𝛽₃𝑥₃+ 𝛽₄𝑥₄ Model 4: 𝐴𝑅0𝑚𝑎𝑟𝑘𝑒𝑡 = 𝛽0+ 𝛽1𝑥1+ 𝛽2𝑥2+ 𝛽3𝑥3+ 𝛽4𝑥4

Where 𝑥4 is the self-enforced sustainability.

All these 4 models help test how the benefits of SCSIs vary among all the suppliers in terms of abnormal returns on day 0. Moreover, after testing the hypothesis 1, we suspect that markets absorb the information from announcements of Wal-Mart SCSIs slowly and also after the announcements. Specifically, stock market reaction of suppliers to the SCSIs exists both on day 0 and 1, as we also detect some significant abnormal returns on the day 1. Hence, in order to prevent bias, we also use CAR (0, 1) as the dependent variable to retest our hypotheses 2 to 5. We develop the following models:

22

In Model 5 to 8, only dependent variable is changed to CAR (0, 1), and all independent variable are the same as those in model 1 to 4.

4.

R

In this section, we first present results of whether the Wal-Mart SCSI announcements are significantly related to a positive abnormal return for suppliers. Then, we discuss the results that if the stock price reactions to such announcements vary based on firms size, resources dependence settings and firms’ self-enforced sustainability.

4.1

A

R

CAR

Hypothesis 1 states that the announcements of Wal-Mart supply chain sustainability are positively related to the stock market reaction of suppliers. To test this hypothesis, for our sample of 828 firms, we calculated the abnormal returns with the mean model and the market model from the 10 days (day -10) before the announcement day (day 0) to the 10 days (day 10) after the announcement day (Table 1).

TABLE 1 ABNORMAL RETURNS

DAY AR MEAN Z VALUE AR MARKET Z VALUE

23 4 0.00569 3.84931 0.134659 3.87481 5 0.00119 0.80768 -0.04794 -1.37952 6 0.00644 4.35511 -0.01086 -0.31254 7 0.00297 2.00684 0.063025 1.81353 8 0.00053 0.36392 -0.05969 -1.71759 9 -0.00134 -0.91007 -0.03273 -0.94185 10 0.00722 4.87893 0.02319 0.66729

The results show that abnormal returns on day 0 in both models are positive and significant. In the mean model, the mean abnormal return on day 0 is 0.54 % and is significant at the 1% level based on a two-tailed test (P-value is 0.0022). In the market model, the mean abnormal return on day 0 is 6.7% and is significant at the 10% level based on a two tailed-test (P-value is 0.0562). The stock market reaction on the day before the announcement (day -1) and after the announcement (day1) are not statically significant on all reported measures, except on day 1 in market model. Overall results in table 1 indicate marginally significant positive stock market reactions of the suppliers to announcements of SCSIs by Wal-Mart.

24

FIGURE 1 RESULTS COMPARISON BETWEEN TWO MODELS

Since we do not have any knowledge if the announcements’ information is leaked or delayed in the stock market, in order to not create any bias in following calculations, we also perform calculation of cumulative abnormal returns (CAR) in three event windows: (-1, 1) (-1, 3) (-3, 1). All graphs are as below:

FIGURE 2 CUMULATIVE ABNORMAL RETURNS (CAR) (-3, 1) IN THE MARKET MODEL

25

FIGURE 3 CUMULATIVE ABNORMAL RETURNS (CAR) (-3, 1) IN THE MEAN MODEL

FIGURE 4 CUMULATIVE ABNORMAL RETURNS (CAR) (-1, 1) IN THE MARKET MODEL

26

FIGURE 5 CUMULATIVE ABNORMAL RETURNS (CAR) (-1, 1) IN THE MEAN MODEL

FIGURE 6 CUMULATIVE ABNORMAL RETURNS (CAR) (-1, 3) IN THE MARKET MODEL

FIGURE 7 CUMULATIVE ABNORMAL RETURNS (CAR) (-1, 3) IN THE MEAN MODEL

27

In all results, the CARs are positively significant on day 0, which proves our results in the abnormal returns. Moreover, we also find significant CARs on day 1 in all graphs in the market model. This probably suggests the stock markets react to the information of announcements on SCSIs not only on day 0 but also on day 1. In other words, the markets absorb the information slowly and stock reaction to the information also exists after the announcements. In the following tests, we should take this fact into account. In addition, we observe that in the event window (-3, 1) in the mean model the CAR becomes positively significant at day -2, but the strongest positive significance of the CAR is still at day 0. It could imply that the stock market might be positively shocked by prior information leakage to the event or by other events before the SCSIs, but the stock price mainly reacted on day 0. However, in rest of graphs, we cannot find the similar results. Overall, this suggests that there is no leakage of information prior to the announcement day, but information is absorbed slowly in the market after the announcement. In order to prevent errors, we use both abnormal returns on day 0 and CAR (0, 1) to calculate our cross-sectional models.

In sum, our results using AR and CAR confirm that there is a positive and significant stock prices reaction on suppliers when the announcement of the SCSI is made by Wal-Mart. Therefore, our hypothesis 1 is supported.

4.2

C

-S

R

Hypothesis 2-5 state that different factors might influence the size of the abnormal return to a SCSI announcement (H2: Firm size, H3: Total dependence, H4: Power imbalance, H5: Self-imposed sustainability). To test these hypotheses, we conducted cross-sectional regression analyses.

28

4). Results for models 3 and 4 include the Self-enforced sustainability variable.

TABLE 2 CROSS-SECTIONAL REGRESSION ANALYSES FOR MODELS 1-4

(t-Values are in parenthesis)

Variable Proxy for Model 1 Model 2 Model 3 Model 4

Dependent AR (MEAN) AR (MARKET) AR (MEAN) AR (MARKET) Intercept (𝛽₀) Mean Return .006 (1.509) .003 (.732) .015 (1.400) .008 (.819) Net Sales (𝛽₁)

Firm Size -5.510E-11 (-.302) -5.757E-11 (-.339) -3.754E-11 (-.256) -4.177E-11 (-.333) TD (𝛽2) Total Dependence -.078 (-.378) -.070 (-.365) -.133 (-.804) -.103 (-.730) PI (𝛽₃) Power Imbalance .071 (.348) .068 (.363) .114 (.708) .090 (.658) S (𝛽₄) Self-enforced Sustainability .000 (-.569) -4.621E-5 (-.295) Sample Size 828 828 624 624 F .290 (.833) .322 (.809) .780 (.538) .613 (.653) 𝑅2 _{.001 .001 .005 .004} Adj𝑅2 -.003 -.002 -.001 -.002

Overall averagely in all four models, F-values are not significant at 10% level and R2 -values are very small for all the models, and less than 5%

Opposite to our predictions, all the estimated coefficients are statistically insignificant at even 10% level in all four models. All coefficients in mean model and market are similar which once again proves that results in the mean model are generally the same as these in market model.

29

and 8). Results for models 7 and 8 include the Self-enforced sustainability variable.

TABLE 3 CROSS-SECTIONAL REGRESSION ANALYSES FOR MODEL 5-8

(t-Values are in parenthesis )

Variable Proxy for Model 5 Model 6 Model 7 Model 8

Dependent CAR (MEAN) CAR (MARKET) CAR (MEAN) CAR (MARKET) Intercept (𝛽₀) Cumulative Mean Return .004 (.823) .217 (1.502) .017 (1.233) .636 (1.291) Net Sales (𝛽1)

Firm Size -5.701E-11 (-.252) -7.235E-10 (-.113) -2.026E-11 (-.106) -1.443E-10 (-.022) TD (𝛽₂) Total Dependence -.101 (-.394) -6.187 (-.861) -.171 (-.799) -9.688 (-1.298) PI (𝛽₃) Power Imbalance .106 (.422) 5.956 (.842) .148 (.709) 8.676 (1.194) S (𝛽4) Self-enforced Sustainability .000 (-.577) -.004 (-.461) Sample Size 828 828 624 624 F .374 (.771) .638 (.591) .640 (.634) 1.096 (.358) 𝑅2 _{.001 .002 .004 .007} Adj𝑅2 -.002 -.001 -.002 .001

Generally after changing the dependent variable to CAR (0, 1), F-values and R2_{-values are somewhat higher on average than those in model 1-4. Both highest values} are 1.09 and 0.007 in model 8. However, we still cannot derive significant results at 10% level for any coefficients.

Conclusively, since all results from model 1 to 8 are all insignificant in our tests, our hypotheses 2, 3, 4, and 5 cannot be supported.

30

5.1

S

Through an event study, this paper analyzes if suppliers benefit from the supply chain sustainability initiatives of the buyer. We investigate how SCSI announcements made by the buyer impact the stock market reactions (or abnormal returns) of the suppliers. In addition, by cross-sectional analysis, this paper tries to reveal if and how abnormal returns vary with firm size, resources dependence setting and the supplier’s own sustainability efforts.

In both the mean model and the market model in terms of AR and CAR, we find positive and significant results to support our hypothesis that the suppliers benefit from the buyer’s supply chain sustainability initiatives. Nevertheless, we are not able to determine any factors which indicate that abnormal returns fluctuate for different event settings. Size, resource dependence setting, and self-enforced sustainability could all not explain variations in abnormal returns.

5.2

I

There are several broad implications of our results.

31

expands our understanding of sustainability within firms to sustainability within the context of supply chains. It proves that sustainability benefits not only individual companies but also related supply chain actors. Therefore, when it comes to sustainability issues, we should study the supply chain as a whole. Practically, our findings may change how suppliers perceive supply chain sustainability issues. Suppliers may be more willing to cooperate with buyers in the implementation of SCSIs. Using our results, buyers could be more able to convince suppliers to implement SCSIs.

Second, our paper provides a starting point for understanding fluctuations of benefits in SCSIs. We believed that firm size, total dependence and self-enforced sustainability might positively influence the financial benefits of SCSIs on suppliers, and power imbalance might negatively influence the financial benefits of SCSIs on suppliers. Nerveless, we could not find significant evidence to support our thoughts. It could be that sustainability is good, but evaluation or assessment on SCSIs is still lacking. Shareholders might not think these suppliers would seriously implement and be inspected on SCSIs by buyers, so own capability or resources dependence with buyers could not differ the share of benefits among suppliers. For instance, Wal-Mart own sustainability is not good (its sustainability score is 38, and rating is CCC which are both lowest in Global Socrates index). Investors might not think buyers like Wal-Mart would assess its suppliers rigidly on SCSIs. Therefore, it results in that suppliers would not implement seriously. Consequently, although suppliers with better resources and dependence with Wal-Mart have advantages in the SCSI over others, they are not valued by investors.

5.3

L

F

R

32

33

R

:

Anthony, B., Zellner, W., 2003. Is Wal-Mart too powerful?" http://www.businessweek.com/magazine/content/03_40/b3852001_mz001.htm (accessed Feb.2012)

Baron, D. P., 2001. Private politics, corporate social responsibility, and integrated strategy. Journal of Economics Management Strategy, 10(1), 7-45.

Becchetti, L., Ciciretti, R., Hasan, I., 2009.

Corporate Social Responsibility and Shareholder'sValue: An Event Study Analysis; Bank of Finland Research Discussion Paper No. 1/2009

Brown, Stephen J., Warner, Jerold B., 1985. Using daily stock returns: The case of event studies, Journal of Financial Economics, Elsevier, vol. 14(1), pages 3-31

BSR, 2011. Maximizing benefits from a sustainable supply chain‖

http://www.bsr.org/reports/BSR_Maximizing_Benefits_From_A_Sustainable_Supply_Chain.pdf

(accessed in Feb.2012)

Carter, C., Rogers, D., 2008. A framework of sustainable supply chain management: moving toward new theory. International Journal of Physical Distribution & Logistics Management Vol. 38 N. 5, 2008 360-387

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

Colicchia, C., Melacini, M., Perotti, S., 2011. Benchmarking supply chain sustainability: insights from a field study, Benchmarking: An International Journal, Vol. 18 Iss: 5, pp.705 - 732

Hart, S. L. 1995. A natural-resource-based view of the firm. Academy of Management Journal, 37: 986- 1014.

Hald, K. S., Albin Olsen, M., 2010. Supplier Incentives to Invest in Buyer Promoted Sustainability Activities in the Supply Chain. Paper presented at The 17th International Annual EurOMA Conference 2010, Porto, Portugal

34

MacKinlay, A. C., 1997. Event studies in economics and finance. Journal of Economic Literature, 35(1), 13-39

Mahmood H., Dennis R., 1979. Effects of farm size on economic efficiency: the case of Pakistan American, Journal of Agricultural Economics, Vol. 61, No. 1 pp. 64-69

Manning, S., Boons, F., Von Hagen, O., Reinecke, J., 2011. National contexts matter: The co-evolution of sustainability standards in global value chains. Ecological Economics, forthcoming.

Nooteboom, B., 1993, Adoption, firm size and risk of implementation, Economics of Innovation and New Technology, 2 p.203-216

Bauer, R., Koedijk, C. G., Otten, R., 2002.

International Evidence on Ethical Mutual Fund Performance and Investment Style . LIFE Working Paper

Paulraj, A., Chen, I.J. 2007. Environmental uncertainty and strategic supply management:A resource dependence perspective and performance implication. Journal of Supply Chain Management, 43(3), 29–42

Pfeffer, J., Salancik, G. R., 1978. The External control of organizations: A resource dependence perspective. , Harper and Row, New York.

Rezabakhsh, B., Hansen, S., 2000. Consumer Power: A comparison of the old Economy and the Internet Economy. Journal of Consumer Policy, Vol.29, pp 3-36

Mitra, S., Singhal, V., 2007. Supply chain integration and shareholder value: evidence from consortium based industry exchanges. Journal of operations Management, Vol. 26, 96-114.

Scholtens, B., & Dam, L. 2007. Banking on the Equator. Are banks that adopted the equator principles different from non-adopters? World Development, 35(8), 1307-1328.

Wal-Mart (2009), ―Supplier Sustainability Assessment‖

35

A

:

36

9 -0.002211414 -1.6273286 -0.032731597 -0.94185

10 2.59358E-05 0.01908554 0.023189973 0.667291

FIGURE 1 COMPARISON BETWEEN AR MEAN, NONSTANDARDIZED AR MARKET AND STANDARDIZED AR MARKET