The effect of supply chain glitches on shareholder wealth

Journal of Operations Management 21 (2003) 501–522

The effect of supply chain glitches on shareholder wealth

Kevin B. Hendricksa,1, Vinod R. Singhalb,∗a Richard Ivey School of Business, The University of Western Ontario, London, Ont., Canada N6A-3K7

b DuPree College of Management, Georgia Institute of Technology, Atlanta, GA 30332, USA

Received 23 April 2002; received in revised form 6 January 2003; accepted 10 February 2003

Abstract

This paper estimates the shareholder wealth affects of supply chain glitches that resulted in production or shipment delays.The results are based on a sample of 519 glitches announcements made during 1989–2000. Shareholder wealth affects areestimated by computing the abnormal stock returns (actual returns adjusted for industry and market-wide influences) aroundthe date when information about glitches is publicly announced. Supply chain glitch announcements are associated withan abnormal decrease in shareholder value of 10.28%. Regression analysis is used to identify factors that influence thedirection and magnitude of the change in the stock market’s reaction to glitches. We find that larger firms experience a lessnegative market reaction, and firms with higher growth prospects experience a more negative reaction. There is no differencebetween the stock market’s reaction to pre-1995 and post-1995 glitches, suggesting that the market has always viewed glitchesunfavorably. Capital structure (debt–equity ratio) has little impact on the stock market’s reaction to glitches. We also providedescriptive results on how sources of responsibility and reasons for glitches affect shareholder wealth.© 2003 Elsevier B.V. All rights reserved.

Keywords:Supply chain management; Stock price performance; Financial and economic analysis

1. Introduction

In recent years, supply chain management (SCM)has been heralded as the next source of building, sus-taining, and winning competitive advantage. Manyhave alluded to the compelling bottom-line benefitsand tremendous payoff that accrue to firms from de-veloping effective supply chains while others havetalked about the strong correlation between excellencein SCM and shareholder value (Edward et al., 1996;Raman, 1998; Tyndall et al., 1998; Quinn, 1999;

∗ Corresponding author. Tel.:+1-404-894-4908;fax: +1-404-894-6030.E-mail addresses:[email protected] (K.B. Hendricks),[email protected] (V.R. Singhal).

1 Tel.: +1-519-661-3874; fax:+1-519-661-3959.

Chopra and Meindl, 2001; Mentzer, 2001).2 Yet, hardevidence to support these claims seems to be limited.Much of the evidence that we have come across isanecdotal and case study oriented, and often basedon non-financial metrics. Little evidence exists that

2 There is no dearth of references on the link between SCM,profitability, and shareholder value. The practitioner literature isfull of such references (see, for example, articles in publicationssuch as Supply Chain Management Review, Purchasing, LogisticsManagement and Distribution Report, and Inside Supply Manage-ment). Other sources of reference are the websites of solutionproviders such as SAP, Inc., Oracle, i2 Technologies, Manugistics,and numerous other small and large providers. Of interest hereare the various white papers that are available from these websitesas well as the literature that describe their solutions. The link be-tween SCM, profitability, and shareholder value is also stressed inacademic journals such as Management Science, Journal of Op-erations Management, and Decision Sciences.

0272-6963/$ – see front matter © 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.jom.2003.02.003

502 K.B. Hendricks, V.R. Singhal / Journal of Operations Management 21 (2003) 501–522

systematically links effective SCM to shareholdervalue creation. This assertion is based on our review ofrecent SCM books; websites of supply chain solutionproviders such as SAP, Inc., Oracle, i2 Technologies,and Manugistics; popular practitioner publicationssuch as Supply Chain Management Review and In-side Supply Management; as well as articles on SCMin academic journals.3

This paper is concerned with empirically estimatingthe shareholder value creation potential of effectiveSCM. Supply chains create value by being reliableand responsive in matching demand and supply. Reli-ability is delivering the right product in right quantityat the right time to the right place at the lowest cost.Responsiveness is the ability to respond quickly tochanging market conditions. One approach to esti-mating the value creation potential of supply chainsis to identify a set of firms that have improved thereliability and responsiveness of their supply chains,and compare their stock price performance against aset of firms that have not improved the reliability andresponsiveness of their supply chains. The differencein stock price performance, appropriately adjusted forindustry and market performance, provides an esti-mate of the value creation potential of reliable andresponsiveness supply chains. Although intuitivelyappealing, this approach is hard to implement becauseof the difficulty in measuring the reliability and re-sponsiveness of supply chains from publicly availabledata. One could resort to samples based on anecdotesor case studies that appear in press. However, this isnot the most objective method.

An alternative approach is to estimate the share-holder value lost, if any, when supply chains are unre-

3 See, for example, recent textbooks on SCM byGattorna (1998),Tyndall et al. (1998), Handfield and Nichols (1999), Simchi-Leviet al. (2000), Chopra and Meindl (2001), and Shapiro (2001).Recent academic research has focused on developing analyti-cal models of supply chain issues to understand how alternateways of managing supply chains affects performance (Cachon andFisher, 2000; Lee et al., 2000; Aviv, 2001; Barnes-Schuster et al.,2002; Milner and Kouvelis, 2002; Taylor, 2002). Others have at-tempted to empirically establish the relationship between supplychain practices and performance (Narasimhan and Jayaram, 1998;Narasimhan and Das, 1999; Krause et al., 2000; Shin et al., 2000;Frohlich and Westbrook, 2001). However, most of the currentlyavailable evidence is based on hypothetical or self-reported data.Hence, it is not clear how well the evidence correlates to actualperformance, and in particular to shareholder value.

liable and unresponsive. Unreliable and unresponsivesupply chains are more likely to suffer from glitches inmatching supply and demand. Glitches could be dueto many reasons including inaccurate forecast, poorplanning, part shortages, quality problems, productionproblems, equipment breakdowns, capacity shortfall,and operational constraints (Fisher and Raman, 1996;Fisher, 1997; Raman, 1997). Glitches could be due tosuppliers, customers, or internal sources. Glitches doaffect a firm’s short- and long-term profitability, whichin turn affects shareholder value. By calculating howmuch shareholder value is lost due to glitches, one canestimate the value creation potential of more reliableand responsive supply chains. The rationale is that ifsupply chains were more reliable and responsive theywould not have experienced the glitches, and hence,would not have experienced the loss in shareholdervalue.

This paper measures the effect of supply chainglitches on shareholder wealth. Our focus is onglitches that resulted in production or shipment de-lays. Examples of such glitches include Sony’s in-ability to deliver Playstation 2 s for the 2000 HolidaySeason due to part shortages; Nike’s inability in 2001to match demand with supply due to complications inimplementing a supply chain management systems;and disruption in Ericsson’s ability to meet the de-mand for mobile phones in 2000 due to internal andsupplier production problems. Our results are basedon a sample of 519 such supply chain glitches thatwere publicly announced during 1989–2000.

To estimate the shareholder wealth affects, eventstudy methodology is used to compute abnormal re-turns around the date when information about glitchesis publicly announced. Abnormal returns are the dif-ference between the actual change in stock pricesand a benchmark to adjust for the overall industryand market-wide influences. Thus, abnormal returnsare normal stock returns purged for industry andmarket-wide influences. We examine how factors suchas size, growth prospects, capital structure (debt–equity ratio), and the timing of the glitches influencethe magnitude of shareholder wealth affects. We alsoprovide descriptive results on how sources of respon-sibility and reasons for glitches affect shareholderwealth.

Section 2discusses the hypothesis examined in thispaper.Section 3describes the collection of the sample.

K.B. Hendricks, V.R. Singhal / Journal of Operations Management 21 (2003) 501–522 503

The basics of the event study methodology are de-scribed inSection 4. Section 5presents and interpretsthe abnormal stock market reaction and the associ-ated shareholder value loss.Section 6analyzes howthe shareholder value loss associated with glitches ismoderated by various factors.Section 7discusses theimplications of our results on dealing with glitches.The final section summarizes the paper.

2. Issues and hypotheses examined

To develop our hypothesis that supply chain glitchesadversely affect shareholder value, we present a frame-work (seeFig. 1) on the link between supply perfor-mance and shareholder value that is similar to whathas been proposed in the literature (Evans and Danks,1998; Tyndall et al., 1998; Chopra and Meindl, 2001).The framework first links supply chain strategy tooperational metrics. Supply chain strategy could in-clude elements such as the design of the supply net-work; integration strategies (Frohlich and Westbrook,2001); supplier development and sourcing strategies(Narasimhan and Das, 1999), and investments in in-formation technology (Davenport, 1998). The choiceof strategies and how they are executed affects key op-erational metrics.Handfield and Nichols (1999)and

Fig. 1. Linking supply chain performance to shareholder value.

Simchi-Levi et al. (2000)present a set of operationalmeasures that relate supply chain performance to areassuch as forecasting and planning accuracy, supplierperformance, delivery performance, lead time, inven-tory, capacity, and quality. Although the choice andimportance of operational measures will vary acrossfirms, the performance of a firm on its chosen opera-tional metrics will determine the efficiency, reliability,and responsiveness of its supply chains (Tyndall et al.,1998; Simchi-Levi et al., 2000; Chopra and Meindl,2001). Efficiency, reliability, and responsiveness affectthe cash flows and earnings of a firm, and thereforeshareholder value. Furthermore, the performance ofa firm’s supply chain on these dimensions of perfor-mance affects its reputation and credibility. Investorsassign a premium valuation to firms that have a rep-utation and credibility of superior supply chain man-agement and execution capabilities (Francis, 2002).

With the framework ofFig. 1, one can argue thatsupply chain glitches adversely affect the short- andlong-term net cash flows of the firm. On the revenueside, glitches can lead to loss in sales and market share,lower salffes price due to markdowns of excess inven-tories, and can prevent the firm from capitalizing onstrong market demand due to unavailability of prod-ucts. On the cost side, glitches can increase the costsassociated with expediting, premium freight, obsolete


inventory, additional transactions, overtime, storageand moving, selling, and penalties paid to customer.Glitches can also negatively impact the productivityand utilization of assets. For example, the firm may endup with excess inventory for some products and ex-perience stockouts and backorders for others. Glitchescan negatively impact customer service if customersare unable to get the products they want at the timethey asked for. Poor customer service leads to highercustomer dissatisfaction and lower loyalty and com-fort levels among customers, and poor word-of-mouthpublicity.

To succeed in the marketplace, the fundamental pro-cesses required to match supply with demand must beexecuted well. Supply chain glitches are an indicationthat these basic processes are not working as smoothlyas desired. This can adversely impact the reputationand credibility of the firm in the mind of investorsas it raises concerns about the ability of managementto execute fundamental business processes. The lossof reputation and credibility can have negative conse-quences. Investors may view a firm’s future prospectswith skepticism and may value it at a discount whencompared to similar firms. The loss of reputation andcredibility may also make it more expensive to raisecapital. Finally, to restore credibility, top managementmay have to spend more time meeting and talking toinvestors, a costly activity that takes top managementtime away from other activities. Our first hypotheses,stated in alternate form (as are all hypothesis in thispaper) is:

H1. The announcement of supply chain glitches willhave a negative stock market reaction.

Hypothesis H1 is about the overall stock marketreaction to glitch announcements. However, it is ofinterest to identify factors that could influence thedirection and magnitude of the market reaction. Thispaper provides evidence on these issues by studyinghow the market reaction is moderated by factors suchas firm size, growth prospects, debt–equity ratio, andthe time of glitch announcements.

Our second hypothesis is that supply chain glitchannouncements by smaller firms will experience amore negative stock price reaction than glitch an-nouncements by larger firms. There are a numberof reasons for proposing this hypothesis. First, the

economic impact of a glitch can be more severe forsmaller firms as such firms are more likely to be highlyfocused, and their profitability is critically dependenton the flawless execution of the supply chains for theirlimited set of products (Kuper, 2002). Second, smallerfirms may take longer to recover from glitches sincethey may not have the capital to invest in technologiesor supply chain-management software that could helpspeed recovery. Third, their small size reduces theirpower and clout to influence and change the behaviorof other supply chain partners to help recover fromglitches (Kuper, 2002). Finally, small firms are lesslikely to be tracked closely by analysts and investorswhen compared to large firms.Bhushan (1989)findsthat the aggregate supply of and demand for analystservices is an increasing function of firm size.Brownet al. (1987)report that the market reaction to quar-terly earnings announcements is stronger for smallerfirms, suggesting that information about smaller firmsis not as well anticipated when compared to largerfirms. Thus, the announcement of a glitch may bemore of a surprise for smaller firms when comparedto larger firms.4 Thus, our hypothesis on firm size is:

H2. The stock market’s reaction to supply chainglitches will be more negative for smaller firms thanlarger firms.

Our third hypothesis is that the stock market’s re-action will be more negative for supply chain glitchesannounced by firms with high growth prospects thanfor firms with low growth prospects. Growth prospectsdepend on the nature of the product, product-market,and industry. Products with high growth potentialgenerally have shorter product life cycles, higher con-tribution margin, and require shorter delivery timerelative to low growth products.Fisher (1997)arguesthat reliable and responsive supply chains are criticalfor success in products with high growth potential.The direct economic impact of supply chain sup-ply chain glitches on high growth products is likelyto be more negative when compared to low growthproducts.

4 Firm size has often been used as an explanatory or moderatingvariable in empirical studies in finance and accounting, and morerecently in operations management (Hendricks et al., 1995; Klassenand McLaughlin, 1996; Hendricks and Singhal, 1997, 2001).


The indirect negative impact of glitches on highgrowth products can also be severe. In the case of highgrowth products, new customers are entering the mar-ket. Since these customers typically do not have estab-lished loyalty to existing products, they are less likelyto wait for the firm’s product if the firm experiencesdelays in delivering the product. This could result inlost sales for both the current and future periods. Fur-thermore, high growth product markets are likely tobe characterized by more competition. Thus, unreli-able and unresponsiveness supply chains could causeexisting customers to migrate to competitors, leadingto perhaps permanent loss in market share. These is-sues may be less of a concern in low growth prod-ucts as the products are standard, margins are low,and the basis of competition is more on cost. Ourhypothesis is:

H3. Supply chain glitches by high growth prospectsfirms will have a more negative stock market reactionthan low growth prospects firms.

Our next hypothesis is that the stock market’s re-action to supply chain glitches is moderated by thedebt–equity ratio (or capital structure). Conditional onthe fact that glitches decrease the market value of thefirm and increase the risk of the firm, we hypothe-size that the higher the debt–equity ratio of the firm,the less negative will be the abnormal returns experi-enced by its shareholders. We have argued that glitchesdecrease the current and future revenues of the firmwhile increasing the current and future operating costs(both fixed and variable), which decrease the valueof the firm. Furthermore, a decrease in revenues andan increase in costs increase the operating leverageof the firm, which increases the risk of the firm (Lev,1974; Gahlon and Gentry, 1982; Lederer and Singhal,1988).

The existing literature has theoretically and empiri-cally documented how changes in the market value andrisk of the firm affect the market values of debt and eq-uity (Jensen and Meckling, 1976; Galai and Masulis,1976; Smith and Warner, 1979; Masulis, 1980). Tworesults from this literature are relevant for developingour hypothesis. The first is that the consequences ofany action that causes a change in the market value ofthe firm are borne by both the debtholders and share-holders. An increase (decrease) in the market value of

the firm will increase (decrease) the market values ofdebt and equity. Furthermore, the extent of change inthe market values of debt and equity is a function ofthe firm’s debt–equity ratio. Specifically, the higherthe debt–equity ratio, less (more) of the change in themarket value will be borne by shareholders (debthold-ers). The second result is that any change in the riskof the firm will change the market values of debt andequity (Galai and Masulis, 1976; Smith and Warner,1979). Specifically, if the market value of the firm re-mains the same but the risk increases, then the valueof debt will decrease and the value of equity will in-crease. Furthermore, the higher the debt–equity ratio,the more will be the increase (decrease) in the valueof equity (debt). Since supply chain glitches are likelyto decrease the market value of the firm and increasethe risk of the firm, our hypothesis is:

H4. The higher the debt–equity ratio, the less nega-tive will be stock market’s reaction to supply chainglitches.

We expect that the stock market’s reaction will beless negative for the earlier (in terms of calendar date)supply chain glitches than the more recent glitches.The justification for this hypothesis is from the recentliterature on SCM that mention the changing compet-itive environment as one of the key motivating factorsfor paying attention to improving the effectiveness ofSCM (Lee, 2001; Selen and Soliman, 2002; Heikkila,2002). Intense global competition, short product lifecycles, rapid changes in technologies, better informedand more demanding customers, the need to providehigher levels of customer service, and the constantpressure of reducing costs and improving asset utiliza-tion are among the many factors that are motivatingfirms to pay attention to supply chains (Narasimhanand Das, 1999; Cachon and Fisher, 2000; Frohlichand Westbrook, 2001; Milner and Kouvelis, 2002;Swafford et al., 2003). Handfield and Nichols (1999)andSimchi-Levi et al. (2000)state that these factorshave become more intense in recent years with mostindustries facing a tougher competitive environmenttoday than a few years ago. An implication of thesechanges in the competitive environment is that thenegative economic consequences of ineffective supplychains are likely to be much higher today than in thepast. Thus, our hypothesis is:


H5. More recent supply chain glitches will be penal-ized more by the market than earlier glitches.

In addition to these hypotheses, and to better ex-plore the nature of supply chain glitches we also pro-vide descriptive results on the stock market reaction bythe source of responsibility for glitches. The responsi-bility for the glitches could be internal or external. Byinternal we mean that the firm that is announcing theglitch is responsible for the glitch. Similarly, externalsources could also cause a glitch in the supply chain.Examples include suppliers who fail to deliver partson time or customers who change their requirements atthe last moment. We also provide descriptive results onthe stock market reaction by the reasons for glitches.

3. Sample selection procedure and datadescription

We searched the full text articles in Wall Street Jour-nal (WSJ) and the Dow Jones News Service (DJNS)to collect a sample of supply chain glitch announce-ments. The search covered the time period from 1989to 2000, and looked for announcements that dealt withproduction delays or shipping delays. Key words usedin the search include combinations of words such asdelay, shortfall, shortage, manufacturing, production,shipment, delivery, parts, components, and other rel-evant phrases. We read the full text of articles thatcontained combinations of these keywords, and elim-inated the following types of articles/announcements:

• Articles in which firms did not actually announcesupply chain glitches. Some articles discussed ingeneral why certain industries were facing supplychain problems.

• Announcements relating to firms with insufficientdaily stock price information available from CRSP(Center for Research in Security Prices). This ex-cludes firms that are not publicly traded on the NewYork, American, or Nasdaq stock exchanges.

• We excluded earnings announcements that men-tioned supply chain glitches as the main or one ofthe main reasons for the level of reported earningsbecause such announcements are likely to provideinformation about other factors that are affectingfirm performance. Furthermore, in some cases weobserved that supply chain glitch information was

made public days before the release of earnings.In these cases, earnings announcements are notthe first indication of glitches, and the valuationimpact of glitches is likely to be captured beforethe earnings announcements. There were 593 suchannouncements.

• We excluded earnings pre-announcements wheresupply chain glitches were mentioned as one ofthe many factors affecting earnings expectationsbecause these announcements are contaminatedand confounded by other factors that have affectedearnings expectations. There were 283 such an-nouncements.

The final sample consists of 519 announcements.We read the text of each to collect information on sup-ply chain partners that are responsible for the glitchesand the reasons for the glitches. Some examples ofglitch announcements included in our sample are:

• “Sony Sees Shortage of Playstation 2s for HolidaySeason”, The Wall Street Journal, 28 September2000. The article indicated that because of compo-nent shortages, Sony has cut in half the number ofPlaystation 2 machines that it can manufacture fordelivery.

• “Motorola 4th Quarter Wireless Sales GrowthLower Than Order Growth”, The Dow Jones NewsService, 18 November 1999. In this case Motorolaannounced that its inability to meet demand wasdue to shortages of certain type of components andthat the supply of these components is not expectedto match demand sometime till 2000.

• “Apple Computer, Inc. Cuts 4th-period ForecastCiting Parts Shortages, Product Delays”, The WallStreet Journal, 15 September 1995. In this case Ap-ple announced that earnings would drop because ofchronic and persistent part shortages of key com-ponents and delays in increasing production of newproducts.

Panel A ofTable 1presents statistics on the sam-ple based on the most recent fiscal year completed be-fore the date of the supply chain glitch announcement.The median observation represents a firm with marketvalue of equity of US$ 225.9 million, total assets ofUS$ 143.7 million, and sales of US$ 152.2 million.The sample has firms from nearly 50 distinct two-digitSIC Codes. Panel B presents the number of announce-


Table 1Description of the sample of 519 announcements of supply chain glitches

Measure Mean Median S.D. Maximum Minimum

Panel A: Descriptive statistics for the sample of 519 announcements of supply chain glitchesMarket value (million US$) 6482.9 225.9 24626.2 274428.2 1.9Total assets (million US$) 7823.7 143.7 33153.0 279097.0 2.2Sales (millions US$) 6189.9 152.2 21329.5 162558.0 0.05Net income (million US$) 226.7 5.9 1540.4 7314.0 −2348.0Employed (thousands) 26.1 1.0 77.8 750.0 0.01Debt–equity ratio 0.16 0.09 0.19 0.87 0.0

Year Number of announcements Announcements (%)

Panel B: Distribution of the announcement year for the sample of 519 announcements of supply chain glitches1989 27 5.201990 19 3.661991 16 3.081992 20 3.851993 28 5.391994 38 7.321995 51 9.831996 46 8.861997 68 13.101998 87 16.761999 64 12.332000 55 10.591989–2000 519 100.00

ments by year. Nearly 60% of the announcements inour sample were made during 1996–2000 with the re-mainder during 1989–1995.

Out of the 519 announcements, 81 announcementsdid not give any information on who is responsiblefor the glitches. Of the remaining 438 announcements381 announcements gave a single source and 57 gavemultiple sources of responsibilities for the glitches.Panel A of Table 2 indicates that the responsibilityfor the glitches is solely attributed to internal sourcesin 189 cases, to customers in 99 cases, and to sup-pliers in 85 cases, and to other sources (government,regulatory agencies, and nature) in 25 cases. Manyannouncements also gave reasons for the glitches.Panel B of Table 2 indicates that parts shortages,last minute changes requested by customers on theirorders, production problems, ramping and rollout,quality and development problems are the primaryreasons cited for glitches.5

5 Some announcements simply mentioned that glitches weredue to production problems whereas others were more specificand included reasons such as yield, technical issues, changes toprocesses, scheduling, maintenance, and equipment installation.

4. Methodology

We use the event study methodology to estimate theshareholder value loss associated with supply chainglitch announcements. This methodology provides arigorous approach to estimate the stock market’s re-action to events, while adjusting for both industry andmarket-wide influences on stock prices (seeBrownand Warner, 1980, 1985, andMacKinlay, 1997, for areview of this methodology). These adjusted returns(commonly referred to as abnormal returns) providean estimate of the percent change in stock price asso-ciated with the event. The underpinning of event studymethodology is that in an efficient market, the wealthimpact of an event will be immediately reflected instock prices. Thus, a measure of the wealth impact canbe obtained by observing stock price behavior overrelatively short time periods.

We briefly describe the key features of the eventstudy methodology and how to estimate the abnormalreturns (seeBrown and Warner, 1985, for details). Dif-ferent models have been developed to estimate abnor-mal returns. We use the market model on daily stock


Table 2More descriptive results on the sample of 519 announcements ofsupply chain glitches

Responsibility Number ofannouncements

Panel A: Distribution of the primary sources ofresponsibility supply chain glitches

Internal 189Customers 99Suppliers 85Othersa 25Internal and supplier 20Internal and customer 5Internal and others 3Customer and supplier 10Supplier and other 2None provided 81

Reasons Number ofannouncements

Panel B: Distribution of the primary reasons forsupply chain glitches

Parts shortages 117Order changes by customers 70Production problems 53Ramping up and rollout problems 48Quality problems 36Development and engineering changes 23Weather related problems 15Capacity and equipment problems 11Information technology problems 11Regulatory approval delays 10Project and program delays 5Construction problems 5Reorganization delays 5Cash and financial problems 3None provided 81

a The other category includes government, regulatory and na-ture as sources of responsibility.

price returns as this is the best specified model andcontrols for the systematic risk of the stock, a key fac-tor in explaining stock returns. This model posits alinear relation between the return on a stock and thereturn on the market portfolio (the market return) overa given time period as:6

rit = αi + βirmt + εit (1)

whererit is the return of stocki on Dayt, rmt the mar-ket return on Dayt, αi the intercept of the relationship

6 The market model is derived from the empirically verifiedCapital Asset Pricing Model (CAPM), which postulates a linearrelationship between a firm’s stock return and the market return.

for stock i, βi the slope of the relationship for stocki with the market return, andεit is the error term forstocki on Dayt. αi is an estimate of the constant dailyreturn for stocki, βirmt is the portion of the returnfor stocki that is due to market-wide movements, andthe error term,εit , is the part of the return of stocki that cannot be explained by market movements andcaptures the effect of firm-specific information.

For each sample firm, we estimateαi, βi, and S2εi

(the variance of the error term,εit ,) using ordinary leastsquare regression (seeEq. (1)) on data over an estima-tion period of 200 trading days. A minimum of 40 re-turn observations in the estimation period is requiredor that announcement is removed from the analysis.7

For this study, the proxy for the market portfolio is theequally weighted index of all securities traded on theNew York, American, and Nasdaq stock exchanges.

Ait , the abnormal return for stocki on Day t is thedifference betweenrit , the actual return of stocki onDay t, and (αi+Birmt) the normal (or expected) returnon stocki on Dayt, and is expressed as

Ait = rit − (αi + βirmt) = rit − αi − βirmt (2)

Averaging the abnormal returns across the samplefirms on any Dayt, the daily mean abnormal,At , canbe expressed as:

At =N∑

i=1

Ait

N(3)

whereN is the number of sample observations on Dayt. The cumulative abnormal return over a given timeperiod (t1, . . . , t2), is the sum of the daily mean ab-normal returns,At , and is expressed as:

CAR (t1, t2) =t=t2∑t=t1

At (4)

7 To prevent any potential bias, the data used to estimate theparameters of the market model must be isolated from the impact ofthe event itself (the estimation period should not include the eventperiod) while attempting to overcome any potential non-stationarityin estimating the parameters (it should be close in time since theparameters may change). Thus, a short time period of 10 tradingdays (two weeks in calendar time) is typically used to separatethe estimation and event periods. Each firm’s estimation periodended 10 trading days prior to the announcement date. Generally,estimating two parameters with less than 40 or so points is notstatistically sound and the presence of missing data can have issuesfor non-stationarity of the underlying estimates.


To test the statistical significance of the daily meanabnormal return ofEq. (3), each abnormal return,Ait ,is first divided by its estimated standard deviation,Sεi ,to yield a standardized abnormal return,AS

it :

ASit = Ait

Sεi

(5)

The test statistic, TSt , for any Dayt is given by

TSt =N∑

i=1

ASit√N

(6)

The rationale behindEq. (6) is that under the nullhypothesis abnormal returns are assumed to be inde-pendent across firms with mean 0 and varianceS2

εi.

From the central limit theorem, the sum ofN stan-dardized abnormal returns is therefore normal withmean 0 and varianceN, which leads toEq. (6). Thetest over multiple days (t1, . . . , t2) is derived simi-larly with the additional assumption that abnormalreturns are independent and identically distributedacross time. The multiple day test statistics, TSc, isgiven by

TSc =N∑

i=1

(∑t=t2t=t1

Ait)/

√∑t=t2t=t1

S2εi√

N(7)

Consistent with the approach used in most eventstudies, we use a 2-day event period. If the glitch an-nouncement is made in WSJ, the event period includesthe day of the announcement and the trading day be-fore the announcement date to account for the possi-bility that information about the event could have beenpublicly released the day before the publication of theWSJ article. If the supply chain glitch announcementis made in DJNS, the event period includes the dayof the announcement and the trading day after the an-nouncement date to account for the possibility that theDJNS announcement was made after the market closedfor trading, in which case the market would react tothe announcement on the next trading day. Calendarday is translated to event time as follows. The WSJannouncement calendar day is Day 0 in event time,the next trading day is Day 1, and trading day preced-ing the announcement day is Day−1, and so on. TheDJNS announcement calendar day is Day−1 in eventtime, the next trading day is Day 0, and trading daypreceding the announcement is Day−2, and so on.

We use different conventions for translating calendartime into event time for announcements made in WSJand DJNS to align the 2-day event period of WSJ andDJNS announcements. In both cases event Days−1and 0 give the 2-day event period.8

5. Empirical results: event study results

Panel A of Table 3 presents the daily abnormalreturns based on 507 out of the 519 supply chainglitches. Abnormal returns for 12 announcementscould not be estimated either because firms didnot have a minimum of 40 returns observations inthe estimation period (eight announcements) or hadmissing returns during the event period (four an-nouncements). The evidence clearly shows that onaverage, announcements of supply chain glitches areassociated with severe stock price decreases. Thestock market reaction is negative on both days of theevent period. The mean abnormal return on Day−1is −7.65% (t-statistic of −49.21) and on Day 0 itis −2.63% (t-statistic of−14.60). For days−1 and0 combined (the event period), the mean abnormalreturn is−10.28% (t-statistic of−45.12).

Outliers are not driving the negative mean abnor-mal returns on Days−1 and 0. The median abnor-mal return is negative and highly significant on Day−1, Day 0, and the event period (Days−1 and 0).The median abnormal return for the event period is−8.06% (Z-statistic of the Wilcoxon signed rank testis −10.28). The proportion of negative abnormal re-turns also indicates that outliers are not driving thenegative mean abnormal returns. 75.94% of the ab-normal returns are negative on Day−1, and 60.75%are negative on Day 0. Over the event period, 82.65%of the abnormal returns are negative. If for a givenfirm the probability of observing a negative event pe-riod abnormal return equals 0.5, then the probabilityof observing 82.65% or more negative returns out of asample of 507 is less than 1%. The distribution of theevent period abnormal returns (Panel B ofTable 3)also indicates that outliers do not drive the overall re-sults. The distribution is negatively skewed with nearly223 out of 507 announcements (44% of the announce-

8 The 200-day estimation period spans Days−210 through−11.


Table 3Event study results

Day –1 Day 0 Event period(Days−1 and 0)

Panel A: Event period (Days−1 and 0) abnormal returns for 507 announcements of supply chain glitchesMean abnormal return −7.65% −2.63% −10.28%t-Statistics −49.21 −14.60 −45.12Median abnormal return −4.84% −1.00% −8.06%Wilcoxon signed-rank testZ-statistic −14.30 −5.11 −10.28Abnormal returns negative (%) 75.94% 60.75% 82.65%Binomial sign testZ-statistic −11.81 −4.90 −14.87

Range of abnormal returns (%) Number of observations Observations (%)

Panel B: Distribution of 2-day event period’s (Days−1 and 0) abnormal returns (R) for 507announcements of supply chain glitchesR ≤ −40.0 19 3.75−40.0 < R ≤ −30.0 18 3.55−30.0 < R ≤ −25.0 23 4.53−25.0 < R ≤ −20.0 34 6.71−20.0 < R ≤ −15.0 55 10.85−15.0 < R ≤ −10.0 74 14.60−10.0 < R ≤ −5.0 88 17.36−5.0 < R ≤ 0.0 108 21.300.0 < R ≤ 5.0 58 11.445.0 < R ≤ 10.0 17 3.25R > 10.0 13 2.56

ments) experiencing abnormal returns lower than−10%.

We also computed the mean dollar change in theshareholder value for the sample firms. For each firm,the Day −1 (Day 0) dollar change in value is theproduct of the market value of its equity on Day−2(Day −1) and its Day−1 (Day 0) abnormal return.The equity value on any trading day is the number ofcommon shares outstanding times the share price atthe end of that trading day. For ease of comparison,the dollar changes in value are converted to 2000dollars using the S&P 500 index. In using the S&P500 index as a basis of comparison, we are assumingthat the dollar change due to the glitch announce-ment is invested in the S&P 500 at the time of theannouncement and held till the end of 2000. Themean (median) dollar change in the shareholder valueis US$ −251.47 million (US$−26.29 million) in2000 dollars. The mean (median) unadjusted changein shareholder value is US$−152.18 million (US$−11.53) over the two days. Clearly, supply chainglitches have significant negative shareholder wealthimpacts.

5.1. Sensitivity analysis of the event study results

Our estimate of the stock market reaction to sup-ply chain glitches is based on using the marketmodel to estimate the abnormal returns. To makesure that the results are not driven by the choice ofthe model, we estimate abnormal returns using themarket-adjusted model, mean-adjusted model, andsize and industry adjusted (Brown and Warner, 1985).The market-adjusted model uses the market returnas the benchmark. The abnormal return,Ait , in thismodel is computed as

Ait = rit − rmt (8)

Compared to the market model (Eq. (2)), this modelassumes that each firm in the sample has characteris-tics similar to the overall market (αi = 0, andβi = 1).

The mean-adjusted model uses as benchmark thestock’s daily average return over the estimation period.The abnormal return,Ait , in this model is computedas:

Ait = rit − rt (9)


rt = 1

Dest

∑t∈EstP

rit (10)

wherert is the simple average of stocki’s daily returnover EstP, the estimation period (Days−210 through−11 in our case), andDest the number of trading daysin the estimation period (200 days in our case). Thestatistics to test for the statistical significance of themean abnormal returns from the market-adjusted andmean-adjusted models are the same as that of the mar-ket model (seeEqs. (6) and (7)).

To strengthen our sensitivity analysis, we repeatthe test of hypothesis H1 using an independent ap-proach which does not rely on an estimation periodsor choice of underlying abnormal return generatingmodels. This approach uses a matched-control groupand buy-and-hold returns to generate the abnormal re-turns. We refer to this approach as the size and indus-try adjusted model. Under this approach for each firmin the sample, we identify a control firm that is similarin size and has at least the same three-digit SIC codeas that of the glitch announcing firm. The size and in-dustry adjusted model then estimates the buy-and-holdreturns of both the sample and matched control firmsand calculates the abnormal return,Ait , as:

Ait = rit − rcit (11)

wherercit is the return on Dayt for the control firm

chosen for sample firmi.Summary results for all of these three models are:

Model Mean(%)

Abnormal returns

Median(%)

Negative(%)

Market-adjusted −10.27 −7.97 83.87Mean-adjusted −10.15 −7.81 82.85Size and industry

adjusted−10.08 −8.32 80.88

The results shows that the magnitude of abnormalreturns associated with glitches are very similar acrossthe different models. We also conducted additionalanalysis to test the robustness of our results using dif-ferent lengths of estimation periods and data from bothpre and post-announcement periods. The results arevery similar to those presented inTable 3and the threemodels above. Since the results across different mod-

els are very similar, the rest of the results reported inthis paper are those obtained from the market model.

5.2. Significance of the stock market reaction tosupply chain glitch announcements relative to othertypes of announcements

To provide a perspective on the significance of themagnitude of the stock market reaction to glitches,Table 4 summarizes the market reaction during theevent period from a sample of previous event studies.The event period can vary slightly from study to studybut generally includes the announcement day and theday before the announcement. The idea in these stud-ies is the same as ours—estimate the shareholder valueeffects of a specific type of events using the eventstudy methodology. Most of the mean abnormal re-turns reported inTable 4are statistically significant atcommonly accepted levels of significance.

The magnitudes of the stock market reaction to op-erational event such as changes in the level of capitalexpenditure and research and development expenses,effective quality management performance, and effec-tive environmental management performance are inthe range of 1–2%. Most of the marketing relatedevents highlighted inTable 4show a small positivemarket reaction of around 0.5% except in the case ofdelaying the introduction of new products, which isassociated with a strong negative reaction of−5.3%.Although the market does not react much to informa-tion technology (IT) investments, announcements,DosSantos et al. (1993)find that the mean abnormal returnfor innovative IT investments are about 1%. The meanabnormal return from IT failures is−1.7%. The meanabnormal return of 7.5% associated with e-commerceannouncements is surprising given the current state ofthe internet industry.Subramani and Walden (2001)suggests that these results may have limited general-izability as they are based on announcements madeduring the last quarter of 1998, which was the timewhen there was a lot of hype and speculation about thepotential of e-commerce initiatives. Financial eventssuch as stock repurchase, capital structure changes,equity offerings, and corporate control events such asproxy contests draw a strong market reaction of about3–8% in absolute terms. Give the results highlightedin Table 4, the mean abnormal return of nearly−10%due to supply chain glitches is on the high end in terms


Table 4Summary of mean abnormal returns during the event period in select event studies

Mean abnormal return (%)

Operational eventsIncrease in capital expenditure (McConnell and Muscarella, 1985) 1.01Decrease in capital expenditure (McConnell and Muscarella, 1985) −1.78Increase in R&D expenditures (Chan et al., 1990) 1.38Plant closings (Blackwell et al., 1990) −0.72Effective TQM implementation (Hendricks and Singhal, 1996) 0.59Effective environmental performance (Klassen and McLaughlin, 1996) 0.63

Marketing eventsChange in firm name (Horsky and Swyngedouw, 1987) 0.64Brand leveraging (Lane and Jacobson, 1995) 0.32Celebrity endorsement (Agrawal and Kamakura, 1995) 0.20New product introduction (Chaney et al., 1991) 0.25Delays in introduction of new products (Hendricks and Singhal, 1997) −5.25

Information technology (IT) eventsInnovative IT investments (Dos Santos et al., 1993) 1.01IT operational problems (Bharadwaj and Keil, 2001) −1.79e-Commerce announcements (Subramani and Walden, 2001) 7.50

Financial eventsOpen market share repurchase (Ikenberry et al., 1995) 3.5Seasoned equity offerings (Speiss and Affleck-Grave, 1995) −3.0Increasing financial leverage (Masulis, 1980) 7.63Decreasing financial leverage (Masulis, 1980) −5.27Proxy contests (Ikenberry and Lakonishok, 1993) 4.21

of the effect on shareholder value when compared withthe effects reported in previous studies.9

5.3. Post-announcement stock price performance

Since the stock market reaction is estimated over a2-day interval, it might give the impression that we aremainly capturing the short-term impact of supply chainglitches. At a theoretical level one can argue that this is

9 Given the negative outcomes of glitch announcements a naturalquestions is what prevents management from not releasing badnews. Although management may have an incentive to delay badnews as long as possible, they have to balance this against the riskof personal litigation if material bad news is not released to themarket in a timely manner, loss of reputation among customers,analysts, and the investment community as well as the loss of theirpersonal human capital (Skinner, 1994). These factors mitigatethe incentive to hold on to bad news for too long. Furthermore,unless the firm can recover very quickly from a bad outcome suchas a supply chain glitch, the bad news will come out in the nextearnings announcement. To avoid personal litigation and/or lossof reputation, managers may choose to disclose bad news soonerrather than later.

not the case because an underpinning of event study isthe fact that in an efficient market, the effect of an eventwill be rapidly reflected in stock prices. Therefore,estimating the change in stock prices over relativelyshort time period yields an unbiased estimate of thestock market’s assessment of the event’s total share-holder value implications. Nonetheless, some studieshave documented statistically significant abnormal re-turns during the post-announcement period for certainevents. To test whether this behavior is observed withsupply chain glitch announcements, we estimate theabnormal stock price performance over a 60 tradingday period subsequent to the glitch announcement.10

Fig. 2 depicts the cumulative behavior of the60-day post-announcement period together with thestock price reaction during the event period (Days−1

10 Sixty trading days is roughly equivalent to a quarter in calendartime. There are really no guidelines in the literature on what isthe optimal length of the post-announcement period that should beused to judge whether the market has overreacted or underreactedto the announcement.


-16

-14

-12

-10

-8

-6

-4

-1 11 23 35 47 59

Trading day relative to announcement date

Ave

rage

sha

reho

lder

ret

urns

(%

)

Fig. 2. The average market-adjusted shareholder loss when the loss is cumulated on a daily basis starting on the day before the announcementof the supply chain glitch to 60 trading days after the announcement of the glitch.

and 0). The sharp drop around Day−1 is the−10%drop during the event period that we have alreadydiscussed inTable 3. Fig. 2 shows that after the eventperiod, the abnormal stock price performance is rela-tively flat. The mean (median) CAR over the 60-daypost-announcement period is−1.59% (0.11%). Boththe mean and median CARs are insignificantly dif-ferent from zero. 50.21% of the announcements hadnegative cumulative abnormal returns over the 60-dayperiod, which is insignificantly different from 50%.

We also computed the 60-day abnormal buy-and-holdreturn using the matched-control group approachwhere for each firm in the sample we identify a con-trol firm that is similar in size and has at least the samethree-digit SIC code as that of the glitch announcingfirm. The abnormal mean (median) buy-and-hold re-turn is −2.87% (−2.71%), insignificantly differentfrom zero. 53% of the 60-day buy-and-hold abnor-mal returns are negative, which is insignificantlydifferent from 50%. Overall there is no evidence ofstatistically significant abnormal performance (neg-ative or positive) after the glitch announcement. It

appears that the market captures most of the wealthaffects of glitches very close to the announcementitself.

5.4. Some descriptive results

To provide additional insights into the stock mar-ket reaction associated with supply chain glitches, weestimate the abnormal returns by categorizing the an-nouncements by the sources that are mainly responsi-bility for causing the glitches.Table 5presents theseresults. The stock market’s reaction is negative for allcategories. For the 80 announcements where no infor-mation is given to determine who is responsible forthe glitches, the mean abnormal return is−10.91%.When the glitch is caused by internal problems, themean abnormal return is−9.20%. Glitches caused bycustomers have a mean abnormal return of−14.31%,whereas glitches attributed to suppliers have a meanabnormal return of−8.70%. Other sources and com-binations of responsibilities also experience negativestock market reaction.


Table 5Summary of the 2-day event period (Days−1 and 0) abnormal returns by primary sources of responsibility for supply chain glitches

Sources of supplychain glitches

Number ofobservations

Mean abnormalreturns (%)

Median abnormalreturns (%)

Negative abnormalreturns (%)

No reason given 80 −10.91 (−16.23) −9.06 (−6.56) 86.25 (−6.48)Internal only 185 −9.20 (−26.45) −6.16 (−8.88) 79.46 (−8.01)Customer only 95 −14.31 (−25.33) −11.68 (−7.49) 84.22 (−6.67)Supplier only 83 −8.70 (−15.41) −6.65 (−6.74) 83.14 (−6.04)Others onlya 24 −6.47 (−7.79) −4.52 (−3.64) 91.66 (−4.08)Internal and supplier 20 −7.88 (−7.13) −9.25 (−3.11) 80.00 (−2.68)Supplier and customer 10 −10.86 (−6.53) −6.69 (−2.03) 80.00 (−1.90)

For each subsample, thet-statistic for the mean abnormal returns, the Wilcoxon signed-rank testZ-statistic for the median abnormal returns,and the binomial sign testZ-statistic for the % negative abnormal returns are reported in parentheses.

a The other category includes government, regulatory, and nature as sources of responsibility.

The basic conclusion fromTable 5 is that thestock market severely penalizes firms that experienceglitches irrespective of which link in the supply chainis responsible for the glitch. The results show theheavy price one link in the supply chain pays for thepoor performance by other links in the supply chains.It also suggests that firms should be cautious in re-moving buffers from the supply chain. Also note thatfirms that experience glitches caused by customers orsuppliers still lose a significant portion of their share-holder value. Such significant losses should providean incentive for various links in the supply chain tocollaborate and co-operate to minimize disruptions insupply chains. For example, many have argued thatgetting information from point-of-sale data and usinga single point of forecasting can significantly reducethe demand-supply mismatch problems (see, for ex-ample, Lee et al., 1997). While such solutions areintuitively appealing, implementing these solutionsrequire investments and changes in the relationships.Our results provide an economic rationale why suchinvestments and relationship building efforts could bevery beneficial.

Table 6presents results on the stock market’s reac-tion to the seven most cited reasons for supply chainglitches.11 Parts shortages are associated with a meanabnormal return of−8.16%. Parts shortages are of-ten caused by poor forecasting, poor planning, depen-

11 The results are based on only those announcements that givea single reason for the glitches. The results are similar whenannouncements with multiple reasons for the glitches are alsoincluded.

dency on a single supplier, long lead times, low inven-tory levels, and poor communication of information,and complex supply chains, among other things. Thebenefits of single sourcing and low inventory levelshave been heavily touted in the academic and prac-titioner literature. While such initiatives have led tobenefits, it is not clear if there is much awareness ofhow significant the cost can be if low inventory levelsand single sourcing cause severe parts shortages thatdisrupts the supply chain.

Order changes by customer have a mean abnor-mal return of −13.38%. Last minute changes incustomer needs are normal occurrences in today’senvironment where competition is intense, productlife cycles are getting shorter, and product and pro-cess technologies are changing rapidly. Firms canrespond to changes in customer needs by developingmore flexible and responsive supply chains. The re-sults for glitches due to changes in customer ordersshow that the cost of a non-responsive and inflexiblesupply chain can be very high. This suggests thatfirms might want to spend more time enhancing theirforecasting, sales and operations planning, and mas-ter scheduling efforts to deal with customer drivenchanges.

The importance of rapid ramp-up of new technolo-gies and rapid rollout of products is underscored by thesignificant penalties incurred by firms when ramp-upsand rollouts are delayed. The market takes a very dimview of such problems and penalizes such mishapsseverely. Production problems, development and engi-neering changes, and quality and testing problems thatcause significant disruptions in the efficient operation


Table 6Summary of the 2-day event period (Days−1 and 0) abnormal returns by primary reasons for supply chain glitches

Reasons for supply chain glitches Number ofobservations

Mean abnormalreturns (%)

Median abnormalreturns (%)

Negative abnormalreturns (%)

Parts shortages 86 −8.16 (−14.79) −7.17 (−5.98) 79.07 (−5.39)Order changes by customer 65 −13.38 (−19.13) −11.08 (−6.06) 81.54 (−5.09)Production problems 41 −12.41% (−15.39) −10.17 (−4.52) 78.05 (−3.59)Ramp-up and roll-out problems 39 −12.75 (−17.49) −9.89 (−5.14) 89.75 (−4.96)Quality and testing problems 30 −8.10 (−7.88) −4.21 (−3.05) 76.66 (−2.92)Development and engineering changes 21 −11.08 (−10.62) −8.45 (−3.30) 85.72 (−3.27)

For each subsample, thet-statistic for the mean abnormal returns, the Wilcoxon signed-rank testZ-statistic for the median abnormal returns,and the binomial sign testZ-statistic for the % negative abnormal returns are reported in parentheses. The results are reported for only thoseevents where only one reason for the supply chain glitches is given. The results including multiple reason announcements are very similar.

of supply chains are associated with a 8–12% loss invalue.12

6. Results from regression analysis

This section discusses results that test our hypothe-ses on the effect of size, growth prospects, debt–equityratio, and earlier or later announcements on the di-rection and magnitude of the abnormal returns dur-ing the event period. We use the following regressionmodel:

Abreti = β0 + β1 Sizei + β2 Market-to-booki

+ β3 Debt−−equityi + β4 Timei + εi,

where Abreti is the event period abnormal return forfirm I. Sizei is measured as the natural logarithm ofsales in the most recent fiscal year ending prior to theannouncement date.13 Predicted sign of the coefficientis positive. Market-to-booki is the proxy for growthpotential, measured as the ratio of the market value of

12 We would suggest caution in interpreting the descriptive resultsof Tables 5 and 6, as some might be tempted to look at thedifferences in abnormal returns across sources of responsibility orreasons for glitches and conclude that some are more critical thanothers. We have purposely stayed away from doing this, as thedescriptive analyses are not backed by hypotheses. We have alsobeen careful in making assertions because we have not controlledfor other factors that may affect the results of our descriptiveanalysis.13 The logarithm transform is commonly used to remove the skew

in the distribution of firm size. Other commonly used measuressuch as total assets and market value of the firm are generallyhighly correlated with sales.

equity to the book value of equity.14 We compute thisratio using the market value of equity 10 trading daysbefore the glitch announcement date, and the bookvalue of the equity reported in the most recent fiscalyear ending prior to the announcement date. Predictedsign of the coefficient is negative. Debt–equityi is mea-sured by the ratio of the book value of debt to the sumof the book value of debt and the market value of eq-uity. We use book value of the debt as reported in themost recent fiscal year ending prior to the announce-ment date and the market value of equity 10 tradingdays before the glitch announcement date. Predictedsign of the coefficient is positive. Timei is an indicatorvariable that measures when the announcement wasmade in calendar time. It takes a value of zero if theannouncement was made on or before 31 December1995, and a value of 1 otherwise. Predicted sign of thecoefficient is negative.εi is the random error term.

We dropped 21 announcements from our regressionanalysis because Compustat did not have any informa-tion about these firms (3 announcements) or had miss-ing data (18 announcements) that prevented us fromcomputing one or more of the independent variables.Therefore, our regression results are based on 486 an-nouncements. To control for outliers, all results arereported after symmetrically trimming the dependentvariable (abnormal returns) at the 2.5% level in eachtail. The conclusions are similar using capping at the

14 Many other variables have been used in the literature to mea-sure growth including the ratio of the market value of the firm tothe book value of assets (Smith and Watts, 1992), the level of re-search intensity (Skinner, 1993), and revenue or return variability(Smith and Watts (1992)). The market-to-book ratio of equity isthe most commonly used measure.


Table 7Estimated coefficients (t-statistics in parentheses) from regressions of event period (Days –1 and 0) abnormal returns on size, market-to-bookratio, debt–equity ratio, early vs. late announcements without and with separate intercepts for various industry groupings

Variables Predicted sign Model 1 (no industry intercepts) Model 2 (with industry intercepts)

Intercept ? −0.159 (−12.34) aSize + 0.012 (6.76) a 0.011 (5.66) aMarket-to-book ratio − −0.001 (−2.32) b −0.001 (−2.41) bDebt–equity ratio + 0.014 (0.59) 0.0001 (0.01)Time − −0.004 (−0.49) −0.003 (−0.41)Industry1 ? −0.123 (−4.98) aIndustry2 ? −0.144 (−7.55) aIndustry3 ? −0.139 (−9.17) aIndustry4 ? −0.168 (−11.49) aIndustry5 ? −0.136 (−5.87) aIndustry6 ? −0.178 (−5.67) aIndustry7 ? −0.175 (−8.35) aIndustry8 ? −0.170 (−8.98) a

Number of observations 486 486Model F value 17.11 47.92 aR2 13.03% 56.10%AdjustedR2 12.27% 54.93%

Significance levels (two-tailed tests): (a) 1% level, (b) 2.5% level.

2.5% level in each tail, and without any trimming orcapping.

Model 1 in Table 7presents these results. The re-gression results support two of the four hypotheses.As predicted the estimated coefficient of size is posi-tive and highly significant. Therefore, larger firms ex-periencing supply chain glitches have lower negativeabnormal returns when compared to smaller firms. Wehad predicted a negative coefficient for market-to-bookratio, our proxy for growth prospects. The estimatedcoefficient for market-to-book ratio is negative and sta-tistically significant indicating that the negative con-sequences of supply chain glitches are more severe forfirms with high growth prospects compared to firmswith low growth prospects. We had predicted a posi-tive relation between debt–equity ratio and abnormalreturns. Although the estimated coefficient is positive,it is statistically insignificant. There is no support forthe relation between debt–equity ratio and abnormalreturns associated with glitches.

The estimated coefficient of time, which segmentsthe sample into pre and post 31 December 1995 an-nouncements, is insignificantly different from zero.The evidence does not support our hypothesis thatearly glitches are penalized less by the market thanmore recent glitches. Glitches have always been

viewed unfavorable by the market, and have beenaccompanied by significant shareholder value loss.While the current focus on improving the reliabil-ity and responsiveness of supply chains is timelyand relevant, it is important to note that the mar-ket has always placed a high value on the ability ofwell-functioning supply chains to create value.

Overall the model is highly significant with anFvalue of 17.11.R2 and adjustedR2 values are 13 and12%, respectively, which are quite strong given thatour regressions are based on cross-sectional data, andare also high relative to those observed in previousstudies on cross-sectional regression models that at-tempt to explain abnormal return behavior.

The intercept in Model 1 ofTable 7captures theeffect of other factors that could affect the abnormalreturns. Some of these could be industry specificfactors such as the level of competition, the basis ofcompetition, innovation rate, and cost of switchingfor customers, etc. While an in-depth study of thesefactors is beyond the scope of this paper, we did someexploratory analysis of this issue by allowing forseparate intercepts for announcements that fall intoeight broad industry groups. Due to the distributionof sample points and the clustering in batch/repetitivemanufacturing (SIC codes ranging from 3000 to


3999), industry groups outside of these ranges weredefined based on single digit SIC codes to generategroupings that included natural resources, continu-ous processing, logistics and supply, wholesale andretailing, and services. Because more sample pointsin our sample are in batch/repetitive manufacturing,industry groups were formed based on two and threedigit SIC codes. The specific industry groupings andSIC ranges used are as follows:

Industry1= 1 if the SIC code is between 0001 and1999 (agriculture, natural resources), 0 otherwise.

Industry2= 1 if the SIC code is between 2000 and2999 (food, tobacco, textiles, lumber, wood, furniture,paper, and chemicals), 0 otherwise.

Industry3= 1 if the SIC code is between 3000 and3569 or 3580 and 3659 or 3800 and 3999 (rubber,leather, stone, metals, machinery, equipment, other),0 otherwise.

Industry4= 1 if the SIC code is between 3570 and3579, 3660 and 3699 or 3760 and 3789 (computers,electronics, communications, defense), 0 otherwise.

Industry5= 1 if the SIC code is between 3700 and3759, or 3790 and 3799 (automobile, airlines, trans-portation), 0 otherwise.

Industry6= 1 if the SIC code is between 4000 and4999 (logistics, supply), 0 otherwise.

Industry7= 1 if the SIC code is between 5000 and5999 (wholesaling, retailing), 0 otherwise.

Industry8= 1 if the SIC code is between 6000 and9999 (services, financial services, government), 0 oth-erwise.

Using these industry variables, we estimate the fol-lowing regression:

Abreti = α1 Industry1+ α2 Industry2

+ α3 Industry3+ α4 Industry4

+ α5 Industry5+ α6 Industry6

+ α7 Industry7+ α8 Indystry8+ β1 Sizei

+ β2 Market-to-booki + β3 Debt−−equityi+ β4 Timei + εi

Model 2 in Table 7gives the regressions results withthe eight industry variables and four predictor vari-ables. The basic conclusions regarding the coefficientsfor size, market-to-book, debt–equity ratio, and timeare the same as that of Model 1 ofTable 7. Further-more, all of the industry variables are statistically

significant. The intercepts range from−12 to −18%depending on the industry, suggesting that supplychain glitches have statistically significant large neg-ative stock market reaction in all industry groupings.The models with separate industry intercepts havevery high explanatory power—R2 and adjustedR2

values are 56 and 54%, respectively.

6.1. Sensitivity analysis of the regression results

To test for the robustness of the regression resultsof Table 7, the following additional control variablesare considered in our analyses:15

• Capital intensity: It is measured as the ratio of prop-erty plant and equipment to the number of employ-ees in the year prior to the year of the glitch. Thisratio is not available for 22 sample firms becauseof missing or non-reported data on number of em-ployees.

• Research and development intensity: It is measuredas the ratio of research and development expenses tosales in the year prior to the year of the glitch. Thisratio is not available for 115 sample firms becauseof missing or non-reported data on research anddevelopment expenses.

• Industry competitiveness: We use the Herfindahl in-dex as a proxy for the degree of competition. Whilethis index is traditionally a measure of concentra-tion, it has been widely used as a proxy for compet-itiveness because the degree of concentration andthe degree of competition are generally inverselyrelated (Zeghal, 1983; Lang and Stulz, 1992). Foreach firm in our sample, we computed the Herfind-ahl index using sales of all firms in the Compustatdatabase with the same primary three-digit SIC codegrouping as that of the firm announcing the glitch.The Herfindahl index for an industry is defined asthe sum of the squared fraction of industry sales byfirm, based on reported sales in the most recent fis-cal year completed before the glitch announcement.

• Multiple glitch indicator: It is defined as a binaryvariable with a value 1 if the firm announcing theglitch had a previous glitch announcement, 0 other-wise.

15 We thank two reviewers for suggesting the variables for thesensitivity analysis.


Table 8Estimated coefficients (t-statistics in parentheses) from regressions of event period (days –1 and 0) abnormal returns on size, market-to-bookratio, debt–equity ratio, early versus late announcements with various combinations of control variables including capital intensity, researchand development intensity, industry competitiveness, and multiple glitch indicator

Variables Predictedsign

Model 1 Model 2 Model 3

Intercept ? −0.164 (−10.86) a −0.155 (−11.36) a −0.151 (−10.90) aSize + 0.014 (6.66) a −0.013 (6.93) a 0.011 (5.29) aMarket-to-book ratio − −0.001 (−1.98) c −0.001 (−2.55) a −0.001 (−2.60) aDebt–equity ratio − 0.0001 (0.01) 0.01 (0.44) 0.015 (0.61)Time − 0.0004 (0.05) −0.0009 (−0.11) −0.001 (−0.18)Capital intensity ? 0.0077 (0.40) 0.01 (0.97) 0.01 (0.98)Research and development intensity ? 0.0085 (0.79)Industry competitiveness ? −0.029 (−0.59) −0.055 (−1.40) −0.050 (−1.26)Multiple Glitch indicator ? 0.012 (1.04)

Number of observations 357 447 447Model F value 9.36 a 12.01 a 10.50 aR2 15.78% 14.04% 14.25%AdjustedR2 14.09% 12.87% 12.89%

Significance levels (two-tailed tests): (a) 1% level, (b) –2.5% level and (c) 5% level.

Table 8 presents the results with these controlvariables. Model 1 ofTable 8includes capital inten-sity, research and development intensity, and industrycompetitiveness. The estimated coefficients of sizeand market-to-book ratio are in the predicted directionand highly significant, whereas the coefficients fordebt–equity ratio and time are statistical insignificant(same as those from Model 1 ofTable 7). The esti-mated coefficients for capital intensity and researchand development intensity are positive, and for indus-try competitiveness it is negative. However, none ofthese coefficients are significant different from zero.

Model 1 of Table 8is based on 357 observations,about 25% less observations than the base model(Model 1 of Table 7). The reason for losing so manyobservations is that data on research and developmentexpenses is missing or non-reported for many of oursample firms. To test whether or not these missing ob-servations drive the results of Model 1 inTable 8, weran the model without research and development in-tensity (Model 2 ofTable 8). The results are basicallythe same, and whether we include or exclude researchand development intensity makes little difference.Model 3 of Table 8 is the same as Model 2 exceptthat we include an indicator variable to capture multi-ple glitches. The coefficient of this indicator variableis positive but statistically insignificant suggestingthat the market treats each glitch independently andthere is no additional penalty for multiple-glitch an-

nouncing firms. We also ran Model 3 ofTable 8withseparate intercepts for announcements that fall intothe eight broad industry groups that we defined ear-lier. The results are very similar to those reported forModel 2 in Table 7. Overall, the results ofTable 8suggest that abnormal returns are not influenced bythe capital intensity, research and development inten-sity, industry competitiveness, and multiple glitches.

7. Implication for managers in dealing withglitches

The analysis of the shareholder value loss due tosupply chain glitches is valuable because it providesfirms with a sense of the economic impact of poor sup-ply chain performance. The evidence clearly indicatesthat ignoring the possibility of supply chain glitchescan have severe negative economic consequences. Anobvious question for managers is what are the strate-gies for avoiding glitches and/or mitigating the nega-tive effect of supply chain glitches. In our view someof these strategies could include:

7.1. Reducing the frequency (probability) of glitches

Better forecasting and planning can go a longway in reducing the frequency of demand and sup-ply mismatches. Forecasting and planning tools and


techniques are available in Enterprise Resource Plan-ning (ERP), SCM, and Customer Relationship Man-agement (CRM) systems. Investments in such systemstogether with appropriate mechanisms and incentivesto motivate supply chain partners to communicate,collaborate and share information with each other,as well as synchronize plans should improve theaccuracy of forecasts and plans.

7.2. Developing the ability to predict glitches

Firms often come to know about problems too late toavoid or minimize the adverse consequences. A desir-able capability would be the ability to predict potentialglitches. Developing predictive capabilities involvesselecting and tracking leading indicators of futurebusiness performance; extracting, integrating andtransforming data from different systems to generatethe leading indicators; delivering information on theseindicators on a real-time basis; and providing visibil-ity into the extended supply chain, including internaloperations, suppliers, and customers. Business perfor-mance analytic packages can help serve as the foun-dation for developing the ability to predict glitches.

7.3. Reducing the elapsed time between theoccurrence and detection of glitches

The negative consequences of glitches are amplifiedwhen glitches go undetected. Firms need to developthe capability to learn about glitches sooner rather thanlater, and aim for an elapsed time of zero betweenoccurrence and detection. A new breed of softwareapplications called Supply Chain Event Management(SCEM) offers the potential to learn about and respondto glitches on a timely basis. This application allowsthe tracking of critical supply chain events and param-eters in real time, setting threshold levels and rulesfor detecting abnormal performance, generating auto-matic alerts when performance is outside the pre-setthreshold limits, and automatically notifying appropri-ate people when exceptions occurs.

7.4. Reducing the time it takes to resolve glitches

The longer it takes to resolve the glitches, the morenegative is its impact. Thus, firms need to develop

the ability to quickly resolve the problem and pre-vent escalation and worsening of the situation. Thisrequires developing a systematic process for dealingand responding to glitches with clear identification ofresponsibilities and allocation of resources (true sce-nario analysis for likely supply/demand mismatches).This also involves learning from past glitches, iden-tifying the root causes of the problem, correcting thevarious processes to deal with the root causes, and en-suring that these problems do not surface again.

The evidence presented in this paper will be im-portant for making an economic case for the majororganizational changes that are needed to improvereliability and responsiveness of supply chains. Suchorganizational changes include integrated planningacross various functions; collaboration with otherpartners in the supply chain; sharing of information,plans, and data with important supply chain partners;and change in mindset, behavior, and performancemetrics (Gavirneni et al., 1999; Cachon and Fisher,2000; Lee et al., 2000). While such organizationalchanges are being talked about extensively, imple-menting these changes is difficult since it requirestrust, cooperation, and alignment of incentives amongthe various partners. Evidence on the extent of valuelost due to supply chain glitches can provide strongincentives for implementing such organizationalchanges for more effective SCM.

Advances in technologies, particularly informa-tion, software, and communication, are widely be-lieved to be key enablers of effective SCM. Thebenefits from successfully implementing these tech-nologies could be more efficient, responsive, andreliable supply chains (Lewis and Talalayevsky, 1997;Greis and Kasarda, 1997; Flynn and Flynn, 1999).Brynjolfsson and Hitt (1996)and Hitt et al. (2002)document the positive impact of investments in infor-mation technology on firm-level performance mea-sures, whereasCotteleer (2002), McAfee (2002), andRabinovich et al. (2003)document the impact on op-erational measures. On the other hand, implementingthese technologies require major investments, andthe implementation itself can be difficult and risky(Davenport, 1998; Ettlie, 1998). To the extent thatthese new technologies could reduce the probabilityof supply chain glitches occurring, the reduced prob-ability with the evidence on the expected value lossfrom supply chain glitches can be used to evaluate


the return on investment from these new technologiesand also balance the implementation risk.

8. Summary

This paper has examined the abnormal stock mar-ket reaction around the dates of supply chain glitchannouncements. Based on a sample of 519 announce-ments made during 1989–2000, we find that glitch an-nouncements decrease shareholder value by 10.82%.We also find that the larger the firm, the less negativeis the stock market reaction, and the higher the growthprospects, the more negative is the stock market reac-tion. Furthermore, the negative stock market reactionis not a recent phenomenon as the market has alwaysviewed glitches negatively. Finally, irrespective ofwho is responsible for the glitch or what caused theglitch, shareholders of firms that experience glitchespay dearly.

There are a number of directions in which futureresearch could prove useful. One is to estimate howa glitch in one link affects the value throughout thesupply chain. Our focus has been on estimating theshareholder value implications of the firm that has an-nounced the supply chain glitches. Glitches could havea negative impact on the upstream and downstreamlinks. Thus, identifying the key customers and sup-pliers of the firm that has experienced the glitch, andestimating the shareholder value implications to cus-tomers and suppliers would provide a more completeassessment of the impact of supply chain glitches. An-other direction would be to estimate the impact of afirm’s supply chain glitches on its competitors. Argu-ments can be made that predict an increase as well adecrease in the competitors’ stock prices. It would beof interest to study the impact of supply chain glitcheson accounting based performance measures, the mag-nitude of revisions in earnings forecast by analysts, aswell as the long-term stock price performance due tosupply chain glitches. This could shed some light onhow long it takes to recover from supply chain glitchesand how permanent are the negative consequences ofpoor supply chain performance.

The results for the source of responsibility for theglitch and the reasons for the glitch require furtherexamination. Future research should explore the de-velopment and testing of hypotheses to explain how

the responsibility of glitches and reasons for glitchesinfluences the stock market reaction. Of particular in-terest would be to see why customer caused glitchesare penalized more than other glitches caused by othersources and the conditions under which this resultholds.

Acknowledgements

We are very grateful to three referees and the as-sociate editor whose constructive comments have sig-nificantly improved the paper.

References

Aviv, Y., 2001. The effect of collaborative forecasting on supplychain performance. Management Science 47, 1326–1343.

Agrawal, J.W., Kamakura, A., 1995. The economic worth ofcelebrity endorsement: an event study analysis. Journal ofMarketing 59, 56–62.

Barnes-Schuster, D., Bassok, Y., Anupindi, R., 2002. Coordinationand flexibility in supply contracts with options. Manufacturing& Service Operations Management 4, 171–207.

Blackwell, D.W., Marr, M.W., Spivey, M.F., 1990. Plant-closingdecisions and the market value of the firm. Journal of FinancialEconomics 26, 277–288.

Bharadwaj, A., Keil, M., 2001. The Effects of InformationTechnology Failures on the Market Value of Firms: AnEmpirical analysis. Working Paper. Goizeuta School ofBusiness, Emory University, Atlanta, GA.

Bhushan, R., 1989. Firm characteristics and analyst following.Journal of Accounting and Economics 11, 255–274.

Brown, S., Warner, J., 1980. Measuring security price performance.Journal of Financial Economics 8, 205–258.

Brown, S., Warner, J., 1985. Using daily stock returns: the caseof event studies. Journal of Financial Economics 14, 3–31.

Brown, L.D., Hagerman, R.L., Griffin, P.A., Zmijewski, M.E.,1987. An evaluation of alternate proxies for the market’sassessment of unexpected earnings. Journal of Accounting andEconomics 9, 159–193.

Brynjolfsson, E., Hitt, L.M., 1996. Paradox lost: firm-levelevidence on the returns to information systems spending.Management Science 42, 541–558.

Cachon, G., Fisher, M., 2000. Supply chain inventory managementand the value of shared information. Management Science 46,1032–1048.

Chan, S.U., Martin, J.D., Kensinger, J.W., 1990. Corporate researchand development expenditure and share value. Journal ofFinancial Economics 26, 255–276.

Chaney, P.K., Devinney, T.M., Winer, R.S., 1991. The impact ofnew product introductions on the market value of firms. Journalof Business 64, 573–610.


Chopra, S., Meindl, P., 2001. Supply Chain Management: Strategy,Planning and Operation. Prentice-Hall, Upper Saddle River,New Jersey.

Cotteleer, M.J., 2002. Order Lead-Time Improvement FollowingEnterprise—IT Implementation: An Empirical Study. WorkingPaper. Harvard Business School, Boston.

Davenport, T., 1998. Putting the enterprise into the enterprisesystem? Harvard Business Review, July–August, 121–131.

Dos Santos, B.L., Peffers, K., Mauer, D.C., 1993. The Impactof information technology investment announcements on themarket value of the firm. Information Systems Research 4, 1–23.

Edward, M.A., Cornelia, L.M., Vickery, S.K., 1996. Strategicsupply chain capabilities for competitive advantage and firmsuccess. Journal of Business Logistics 17, 1–22.

Ettlie, J.E., 1998. The ERP challenge. Automotive Manufacturing& Production 6, 16–17.

Evans, R., Danks, A., 1998. Strategic supply chain management:creating shareholder value by aligning supply chain strategywith business strategy. In: Gattorna, J. (Ed.), Strategic SupplyChain Alignment. Gower, Aldershot, pp. 18–37.

Flynn, B.B., Flynn, E.J., 1999. Information-processing alternativesfor coping with manufacturing environment complexity.Decision Sciences 30, 1021–1052.

Fisher, M., 1997. What is the right supply chain for your product?Harvard Business Review, March–April, 105–116.

Fisher, M., Raman, A., 1996. Reducing the cost of uncertaintythrough accurate response to early sales. Operations Research44, 87–99.

Francis, S.J., 2002. The view of supply chain from Wall Street. In:Proceedings of the Presentation Made at the First Annual SupplyNetwork Conference. San Jose, California, 18 September.

Frohlich, M.T., Westbrook, R., 2001. Arcs of integration: aninternational study of supply chain strategies. Journal ofOperations Management 19, 185–200.

Gahlon, J., Gentry, J., 1982. On the relationship between systematicrisk and the degree of operating and financial leverage. FinancialManagement 11, 15–23.

Galai, D., Masulis, R., 1976. The option pricing model and therisk factor of stock. Journal of Financial Economics 3, 53–81.

Gattorna, J., 1998. Strategic Supply Chain Alignment. Gower,Aldershot, pp. 18–37.

Gavirneni, S., Kapuscinski, R., Tayur, S., 1999. Value of infor-mation in capacitated supply chains. Management Science 45,16–24.

Greis, N.P., Kasarda, J.D., 1997. Enterprise logistics in the infor-mation era. California Management Review 39, 55–78.

Handfield, R.B., Nichols, E.L., 1999. Introduction to Supply ChainManagement. Prentice-Hall, Upper Saddle River, New Jersey.

Heikkila, J., 2002. From supply to demand chain management:efficiency and customer satisfaction. Journal of OperationsManagement 20, 747–767.

Hendricks, K.B., Singhal, V.R., 1996. Quality awards andthe market value of the firm: an empirical investigation.Management Science 42, 415–436.

Hendricks, K.B., Singhal, V.R., 1997. Delays in new productintroductions and the market value of the firm: the consequencesof being late to the market. Management Science 43, 422–436.

Hendricks, K.B., Singhal, V.R., 2001. Firm characteristics, totalquality management, and financial performance. Journal ofOperations Management 19, 269–285.

Hendricks, K.B., Singhal, V.R., Weidman, C.I., 1995. The impactof capacity expansion on the market value of the firm. Journalof Operations Management 12, 259–272.

Hitt, L.M., Wu, D.J., Zhou, X., 2002. Investment in enterpriseresource planning. Journal of Management Information System19, 71–98.

Horsky, D., Swyngedouw, P., 1987. Does it pay to changeyour company’s name? A stock market perspective. MarketingScience 6, 320–335.

Ikenberry, D., Lakonishok, J., 1993. Corporate governance throughthe proxy contests: evidence and implications. Journal ofBusiness 66, 405–436.

Ikenberry, D., Lakonishok, J., Vermaelen, T., 1995. Marketunderreaction to open market repurchases. Journal of FinancialEconomics 39, 181–208.

Jensen, M., Meckling, W., 1976. Theory of the firm: managerialbehavior, agency costs, and ownership structure. Journal ofFinancial Economics 3, 305–360.

Klassen, R.D., McLaughlin, C.P., 1996. The impact of envi-ronmental management on firm performance. ManagementScience 42, 1199–1214.

Krause, D.R., Scannell, T.V., Calontone, R.J., 2000. A structuralanalysis of the effectiveness of buying firms’ strategies toimprove supplier performance. Decision Sciences 31, 33–55.

Kuper, A., 2002. Hardening vulnerable links in the supply chain.TOTALsupplychain, February 2002, 1–4.

Lane, V., Jacobson, R., 1995. Stock market reactions to brandextension announcements: the effects of brand attitude andfamiliarity. Journal of Marketing 59, 63–77.

Lang, L.H.P., Stulz, R.M., 1992. Contagion and competitiveintra-industry effects of bankruptcy announcements: anempirical analysis. Journal of Financial Economics 32, 45–60.

Lederer, P., Singhal, V.R., 1988. The effect of cost structure anddemand risk in the justification of new technologies. Journalof Manufacturing and Operations Management 1, 339–371.

Lev, B., 1974. On the association between operating leverage andrisk. Journal of Financial and Quantitative Analysis 9, 627–642.

Lee, H., 2001. Introduction to focused issue: the use ofinformation in managing supply chains. Manufacturing &Service Operations Management 3, 51–52.

Lee, H., Padamanabhan, P., Whang, S., 1997. Informationdistortion in supply chain: the bullwhip effect. ManagementScience 43, 546–558.

Lee, H., So, K., Tang, C., 2000. The value of information sharingin a two-level supply chain. Management Science 46, 626–643.

Lewis, I., Talalayevsky, A., 1997. Logistics and informationtechnology: a coordination perspective. Journal of BusinessLogistics 18, 141–157.

Masulis, R., 1980. The effects of capital structure changes onsecurity prices: a study in exchange offers. Journal of FinancialEconomics 8, 139–177.

McAfee, A., 2002. The impact of enterprise informationtechnology adoption on operational performance: an empiricalinvestigation. Production and Operations Management 11, 33–53.


MacKinlay, C.A., 1997. The event studies in economics andfinance. Journal of Economic Literature 35, 13–39.

McConnell, J.J., Muscarella, C.J., 1985. Corporate capitalexpenditure and market value of the firm. Journal of FinancialEconomics 14, 399–422.

Mentzer, J.T., 2001. Supply Chain Management. Sage, ThousandsOak, California.

Milner, J.M., Kouvelis, P., 2002. On the complementary valueof accurate demand information and production and supplierflexibility. Manufacturing & Service Operations Management4, 99–113.

Narasimhan, R., Jayaram, J., 1998. Casual linkages in sup-ply chain management: an exploratory study of NorthAmerican manufacturing firms. Decision Sciences 29, 579–605.

Narasimhan, R., Das, A., 1999. An empirical investigation of thecontribution of strategic sourcing to manufacturing flexibilitiesand performance. Decision Sciences 30, 683–718.

Quinn, F., 1999. The payoff potential in supply chain management.In: Achieving Supply Chain Excellence Through Technology,Montgomery Research, Inc., San Francisco, California.

Rabinovich, E., Dresner, M.E., Evers, P.T., 2003. Assessing theeffects of operational processes and information systems oninventory performance. Journal of Operations Management 21,63–80.

Raman, A., 1997. Module Overview: Coordinating and ManagingSupply Chains. Matching Supply and Demand. HarvardBusiness School, Boston, MA.

Raman, A. 1998. Toward finding the perfect match: matchingsupply and demand in supply chains. In: Gattorna, J. (Ed.),Strategic Supply Chain Alignment. Gower, Aldershot, pp. 171–187.

Selen, W., Soliman, F., 2002. Operations in today’s demand chainmanagement framework. Journal of Operations Management20, 667–673.

Shapiro, J.F., 2001. Modeling the Supply Chain. Duxbury, PacificGrove, California.

Shin, H., Collier, D.A., Wilson, D.D., 2000. Supply managementorientation and supplier/buyer performance. Journal ofOperations Management 18, 317–333.

Simchi-Levi, D., Kaminsky, P., Simchi-Levi, E., 2000. Designingand Managing the Supply Chain. McGraw-Hill, New York, NY.

Skinner, D., 1993. The investment opportunity set and accountingprocedure choice: preliminary evidence. Journal of AccountingResearch 19, 407–445.

Skinner, D., 1994. Why firms voluntarily disclose bad news.Journal of Accounting Research 32, 38–60.

Smith, C., Warner, J., 1979. On financial contracting: an analysisof bond covenants. Journal of Financial Economics 7, 117–161.

Smith, C., Watts, R., 1992. The investment opportunity setand corporate financing, dividend, and compensation policies.Journal of Financial Economics 32, 263–292.

Speiss, D.K., Affleck-Grave, J., 1995. Underperformance inlong-run stock returns following seasoned equity offerings.Journal of Financial Economics 38, 243–267.

Subramani, M., Walden, E., 2001. The impact of e-commerceannouncements on the market value of the firm. InformationSystems Research 12, 135–154.

Swafford, P.M., Ghosh, S., Murthy, N., 2003. The Antecedents ofSupply Chain Agility: Scale Development and Model Testing.Working Paper. DuPree College of Management, Atlanta.

Taylor, T.A., 2002. Supply chain coordination under channelrebates with sales effort effects. Management Science 48, 992–1007.

Tyndall, G., Gopal, C., Partsch, W., Kamauff, J., 1998. Super-charging Supply Chains: New Ways to Increase Value ThroughGlobal Operational Excellence. Wiley, New York, NY.

Zeghal, D., 1983. Industry, market structure, and the informationalcontent of financial statement. Journal of Accounting and PublicPolicy 2, 115–131.

Documents

The effect of supply chain glitches on shareholder wealth