Docstoc

sdarticle

Document Sample
sdarticle Powered By Docstoc
					Omega 27 (1999) 595±604

www.elsevier.com/locate/orms

Revisiting the stock price impact of quality awards
Greg Adams a, Grant McQueen b, Kristie Seawright a,*
a

Department of Business Management, Marriott School of Management, Brigham Young University, Provo, UT 84602, USA b Finance Department, College of Business, Arizona State University, Tempe, AZ 85287, USA Received 1 August 1997; accepted 1 March 1999

Abstract In an event study, Hendricks and Singhal [Hendricks KB, Singhal VR. Quality awards and the market value of the ®rm: an empirical investigation. Management Sci 1996;42:415±36.] ®nd evidence that ®rms that win quality awards are further rewarded with a stock price increase on the day of the award announcement. We revisit Hendricks and Singhal (1996), extend their research and ®nd four reasons why management, owners and analysts should be cautious about expecting an abnormal return when a ®rm wins a quality award. First, in our sample of Baldrige Award winners, the evidence of a stock price response on the announcement day is only marginally signi®cant. Second, in our sample of State quality award winners, the announcement day relationship between stock returns and winning awards is not signi®cant. Third, in the most recent subperiod, 1992±1997, we ®nd no evidence of positive abnormal returns. Fourth, the marginally signi®cant Baldrige results are actually driven by just four companies. A company-by-company microanalysis reveals that only 50% of the award winners experienced positive abnormal returns. The diminishing stock price response on event day does not necessarily imply a lack of stockholder rewards. Evidence from other studies suggests that the stockholders are rewarded for successful total quality management (TQM) implementation, but the rewards can come long before and after the formal award is presented. From a shareholder value perspective, TQM still matters but the award ceremonies may not. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Baldrige Award; Event study; Securities; Total quality management

1. Introduction Total quality management (TQM) is a philosophy intended to improve stakeholders' welfare. Theoretically, the needs of all stakeholders, including customers, employees and stockholders, should be better met through TQM program implementation. Rewards to customers are so integral to TQM that

* Corresponding author. Tel.: +1-801-378-3017; fax: +1801-378-5984 E-mail address: kristie_seawright@byu.edu (K. Seawright)

Juran [16] de®nes quality as ®tness for customer use. Ishikawa and Lu [15] extend this view to encompass total customer value, encouraging excellence at an acceptable price. Employee needs are also met by TQM. Deming [5] notes that job security increases as improved quality helps producers maintain and increase market share. Furthermore, Ishikawa [14] ties employee job satisfaction to production variation reduction and recommends that management practices which encourage employee commitment be included in TQM programs. Theory suggests that, along with customers and employees, stockholders too should be rewarded by

0305-0483/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 3 0 5 - 0 4 8 3 ( 9 9 ) 0 0 0 2 5 - 0

596

G. Adams et al. / Omega 27 (1999) 595±604

successful TQM implementation. Phillips et al. [19] show that improved product and/or service quality contributes to ®rm competitiveness which in turn improves ®rm performance. Crosby [4] notes that successful TQM implementation should increase cash ¯ows through increased sales and decreased costs and/ or lower discount rates through reduced risk, thus leading to higher stock prices. Hendricks and Singhal's [11] event study ®nds that quality-award-winning companies are further rewarded with a stock price jump. Easton and Jarrell [8] ®nd rewards in the form of stock returns and ratio improvements. Garvin [10] and Phillips et al. [19] associate quality programs with improvements in various ratios that measure pro®tability and operational eciency1. Hendricks and Singhal [11] compare the stock prices of 91 publicly traded ®rms before and after they won a quality award and ®nd that, as a group, these winners experienced a statistically signi®cant positive abnormal return on the day the award was announced. We extended Hendricks and Singhal's [11] analysis of a quality award's impact on stock prices in three ways. First, we add data from 1992 through 1997. Second, these added data allow us to test for temporal stability in the quality award/stock price relationship. A diminishing announcement day return is consistent with market participants becoming better informed about and more intuned to information about quality. Third, we performed a microanalysis of Malcolm Baldrige Award2 winners. The Hendricks and Singhal [11] study is a macroanalysis in the sense that the event study methodology simultaneously tests for positive abnormal returns for the group of 91 ®rms. We developed and performed a company-by-company microanalysis of the stock price response to winning the award. Both our micro- and macroanalyses suggest that managers, stockholders and traders should be cautious in their interpretation of Hendricks and Singhal [11]. Our macroanalysis extension ®nds that, as a group, the Baldrige winners experienced a positive abnormal return, but we ®nd that this return is only marginally signi®cant and primarily limited to the pre-1992 sample. Furthermore, state winners show no signi®cant stock price returns on announcement date. Our microanaly1 See [9] for a review of the academic research linking TQM with corporate ®nancial performance. 2 The Malcolm Baldrige National Quality Award was established by the Congress of the USA in 1987 to recognize US companies for their achievements in quality and business performance. The award was also created to raise national awareness about the importance of quality and performance excellence as a competitive edge. Several states have instituted similar quality awards; many of these awards are based upon the Baldrige criteria.

sis extension ®nds that only half the winners had positive abnormal returns on the day they won the award. Furthermore, only four had statistically signi®cant positive abnormal returns: General Motors and Federal Express Corporation in 1990, Armstrong World Industries in 1995 and ADAC in 1996. We suggest several explanations for the weak and diminishing announcement-day stock price response to the Baldrige Award announcement. In light of the ®ndings of Easton and Jarrell [8] and two additional papers by Hendricks and Singhal [12,13], we suggest that the most likely explanation is that stockholders do bene®t from successful TQM implementation, but that most of the bene®ts do not occur on the announcement day. Rather, the stockholders are rewarded whenever tangible evidence of successful and ®nancially responsible TQM progress is recognized. In the extreme case, an announcement of a quality award becomes completely redundant with respect to the stock price as the stock price increase migrates back in time to the date or dates when the TQM-related progress was publicly known. Similarly, stock prices can respond forward in time from the announcement as new details about the TQM program's performance (better or worse than was expected on the announcement date) become public. In Section 2, we introduce a ®nancial model, common in the ®nance literature, that relates new information about a ®rm's quality to its stock return. In Section 3, we perform Hendricks and Singhal's [11] macroanalysis using the event study methodology for a group of 20 publicly traded Baldrige Award winners and 48 state award winners. In Section 4, we perform our microanalysis and individually examine these 20 ®rms for abnormal returns on the day they won the award. In Section 5, we venture possible explanations for the limited stock price response. We present our conclusions in Section 6.

2. Financial model A common model that links stock prices to information posits that stock prices equal the present discounted value of rationally forecasted future dividends. This model is represented as Pt ˆ
I ˆ E‰Dt‡t jOt Š tˆ1

1 ‡ rt‡t

,

…1†

where Pt is the price of the stock at time t, E[Á|Ot ] denotes the mathematical expectation conditional on information available at time t, Dt+t is the dividend paid at time t+t and rt+t is the stochastic discount factor for cash ¯ows that occur at time t+t. Economic announcements a€ect daily stock price

G. Adams et al. / Omega 27 (1999) 595±604

597

movement if the new information revealed by the announcements a€ects either expectations of future dividends (numerator of Eq. (1)), or discount rates (denominator of Eq. (1)) or both. Regarding quality, the new information is represented by the di€erence in the perceived quality on day t after the Baldrige announcement and the perceived quality as of day t À 1. The information set, Ot, includes all publicly available information investors have about the future prospects of the economy, industry and the company, including its product or service quality. If stock market participants consider the winning of the Baldrige Award as news (that is, the award tells them something about quality they did not already know) and if news of higher quality yields higher expected cash ¯ows and/or lower discount rates, then the award-winning company should be further rewarded with a higher stock price.

to customer satisfaction. The uniformity of the evaluation is exceptionally high. And, in most cases, a large part of the company is evaluated. Given the small sample of publicly traded Baldrige Award winners, we also analyze a larger sample of state award winners. Table 1 reports the award announcement day for each of the 20 companies (or the division of a company) and their stock returns on that day. Only seven of the 20 companies had positive raw returns on their announcement day. Looking only at the individual company raw returns can be misleading. For example, suppose that on the November 14, 1988, announcement day, not only did Motorola's stock price drop, but also new CPI ®gures were announced which indicated that unexpectedly high in¯ation was imminent. Such news should cause all stocks to fall, including Motorola's. Thus, our second step is to calculate abnormal returns which adjust the raw returns for event-day movements in the stock market. We calculate abnormal returns using three di€erent models. Inputs into these models require the market return on the announcement day, t = 0, as well as individual company and market returns for a 200-day estimation period, from 209 days before the announcement, t=À209, until 10 days before the announcement, t=À10. Following Hendricks and Singhal [11], our proxy for the market return is the return on an equally-weighted portfolio consisting of all stocks on the NYSE and AMEX exchanges. The announcement day market return and the individual companies' mean, standard deviation, high and low returns for the 200-day estimation period are reported in Table 14. From these data, we calculate abnormal returns using the following three models from Hendricks and Singhal [11]: 1. The mean adjusted model: " Ait ˆ Rit À R i , where Ait and Rit are the abnormal and total return " for stock i on day t and R i is the arithmetic average of stock i's daily return for the estimation period. 2. The market adjusted model: Ait ˆ Rit À Rmt ,

3. Macroevidence Of the Baldrige Award winners from 1988±1997, 20 were publicly traded on a major stock exchange at the time they won the award. In examining the stock price response to winning the Baldrige Award, we followed the three event study steps and the notation used by Hendricks and Singhal [11]3. In step 1, the stock returns for each of the 20 companies in the sample were found using data from the Center for Research in Securities Prices (CRSP), which reports daily returns for all stocks traded on the New York (NYSE), the American (AMEX) and the Nasdaq stock exchanges. The daily returns are the percentage change in the stock price from the previous day's closing price to the current day's closing price (plus an adjustment for any dividends). We also examine returns on the business day prior to and just after the announcement to test for leaks and delayed responses, respectively. Our primary focus is on Baldrige Award winners. Easton and Jarrell [9] develop six criteria for using third-party assessments in event studies on quality. They note that Overall, the Malcolm Baldrige Award meets the criteria the best. Speci®cally, the Malcolm Baldrige Award evaluates a broad set of issues, spanning seven major categories that range from leadership
3 See [2,3,6,18] for more detailed descriptions of the event study methodology. 4 In a sensitive analysis, we also examine results using only NYSE and only AMEX stocks as proxies for the market. We also examine results using a 200-day estimation period after the announcement (i.e. day t = 10 to day t = 209), as well as 100- and 300-day estimation periods.

where Rmt is the market return on day t. 3. The market model: ” ” Ait ˆ Rit À ai À bi Rmt , ” ” where the ai and bi are the intercept and slope coef®cients obtained from ordinary least squares (OLS)

598

G. Adams et al. / Omega 27 (1999) 595±604

Table 1 Publicly traded Baldrige Award winners stock return data (1988±1994) Baldrige Award winner Announcement day returns (t = 0) date 11/14/88 11/14/88 11/02/89 10/10/90 10/10/90 10/10/90 10/09/91 10/14/92 10/14/92 10/18/93 10/18/94 10/18/94 10/16/95 10/16/95 10/16/96 10/16/96 10/15/97 10/15/97 10/15/97 10/15/97 company return % À1.97 À0.50 À1.07 1.75 À1.78 2.40 4.17 À0.89 0.00 0.21 À0.47 À0.81 4.11 À1.45 8.24 0.00 À1.01 À0.58 À0.14 1.19 Estimation period return statistics (t=À209 to t=À10) high %; 5.11; 4.67; 4.04; 5.33; 4.12; 8.99; 11.76; 3.70; 9.69; 7.78; 2.96; 4.98; 6.71; 6.18; 20.31; 6.22; 5.26; 9.66; 7.53; 3.91; low % À9.92 À4.80 À4.94 À4.69 À2.81 À9.83 À11.81 À2.92 À4.76 À9.79 À2.12 À3.41 À3.83 À8.84 À8.13 À4.33 À3.43 À4.98 À6.32 À3.83

Company Name Motorola Westinghouse Electric Corp., Commercial Nuclear Fuel Division Xerox Corporation, Business Products and Systems General Motors, Cadillac Motor Car Division IBM, Rochester Federal Express Corporation Solectron Corporation AT&T, Network Systems Group, Transmission Systems Business Unit and AT&T, Universal Card Services Texas Instrument, Defense Systems and Electronics Group Eastman Kodak, Eastman Chemical AT&T, Consumer Communications Services GTE, Directories Corporation Armstrong World Industries, Building Products Operations Division Corning, Telecommunications Products Division ADAC Laboratories Dana Commercial Credit Corporation 3M, Dental Products Division Merrill Lynch and Company, Credit Corporation Solectron Corporation Xerox Corporation, Business Services

market mean % return % (S.D. %) À0.53 À0.53 À0.31 À1.17 À1.17 À1.17 À0.52 À0.08 À0.08 0.08 À0.23 À0.23 À0.18 À0.18 À0.03 À0.03 À0.16 À0.16 À0.16 À0.16 À0.07 (2.14) 0.05 (1.51) 0.06 (1.22) À0.06 (1.56) 0.07 (1.20) À0.1 (1.99) 0.60 (3.37) 0.09 (1.12) 0.26 (2.25) 0.21 0.00 À0.06 0.21 0.00 0.35 0.03 0.10 0.35 0.24 0.26 (1.75) (0.98) (1.16) (1.52) (1.60) (3.98) (1.56) (1.46) (2.20) (2.38) (1.60)

regressions of company i's daily return on the market's daily return over the estimation period. In step 3, the event-day abnormal returns of all 20 " stocks are averaged and this average, A t , is tested for statistical signi®cance against a null hypothesis that the " event day mean abnormal return is zero, H0 :A t ˆ 0. The test statistic is " At TSt ˆ , ”" S
A

We report mean abnormal returns with their t-statistics from Eq. (2) in Tables 2±4 for all three models of abnormal returns. The macroanalysis mean abnormal returns on the announcement day are 0.44% for the mean adjusted
Table 2 Mean abnormal returns and t-statistics for macrotests of a stock price response to winning the Baldrige Award. t-statistics are in parentheses Panel A: all years, N = 20

…2†

where: 4 ”" SA ˆ and
À10 1 ˆ " " Aˆ A tX 200 tˆÀ209 À10 ˆ À Á 1 " 2 " At À A 200 À 1 tˆÀ209

51a2 ,

Abnormal return model Mean adjusted Market adjusted Market model

t=À1 À0.44 (0.97) À0.06 (0.14) À0.09 (0.21)

t=0 0.44 (0.96) 0.92a (2.27) 1.07a (2.63)

t=1 À0.46 (1.01) 0.02 (0.06) 0.21 (0.52)

a Indicates signi®cance at the two-tailed, 10 and 1% critical values, respectively.

G. Adams et al. / Omega 27 (1999) 595±604 Table 3 Mean abnormal returns and t-statistics for macrotests of a stock price response to winning the Baldrige Award. t-statistics are in parentheses Panel B: state awards, t = 0 Abnormal return model Mean adjusted Market adjusted Market model N = 48, 1991±1997 0.23 (0.61) 0.14 (0.39) 0.11 (0.30)

599

Table 4 Mean abnormal returns and t-statistics for macrotests of a stock price response to winning the Baldrige Award. t-statistics are in parentheses Panel C: subsamples, t = 0 Abnormal return model Mean adjusted Market adjusted Market model N = 7, 1988±1991 0.35 (0.41) 1.20 (1.70) 1.54a (2.25) N = 13, 1992±1997 0.49 (0.86) 0.77 (1.45) 0.82 (1.56)

a Indicates signi®cance at the two-tailed, 10 and 1% critical values, respectively.

model, 0.92% for the market adjusted model and 1.07% for the market model. Two of the three (market adjusted and market model) models' results are signi®cantly greater than zero at the 10% critical value. These three means correspond to mean abnormal event-day returns of 0.67, 0.65 and 0.59% found by Hendricks and Singhal [11] using their sample of 91 ®rms. All three of Hendricks and Singhal's models of abnormal returns were signi®cantly positive. On the day prior to the announcement, t=À1, estimates of abnormal returns are negative and insigni®cant consistent with no news leaks. Similarly, on the day after the announcement the abnormal returns are insigni®cantly di€erent from zero suggesting an e5 Following Hendricks and Singhal [11] a Wilcoxon signedrank test and a paired-di€erence test both failed to reject a null hypothesis that betas prior to the announcement equaled betas after the announcement. Thus, we ®nd some evidence that stock prices increased after the announcement, but this price change is apparently driven by higher expected cash ¯ows not by decreases in systematic risk. 6 We identi®ed 60 winners of state quality awards who were publicly traded on the NYSE, AMEX or Nasdaq exchanges. However, 12 of these companies were excluded for a lack of complete data in the 200-day estimation period.

cient response to the announcements. In contrast, Hendricks and Singhal [11] not only found more signi®cant positive announcement day returns than we did, but also found more evidence of negative and occasionally signi®cant post-announcement day returns. That is, much of the abnormal return found by Hendricks and Singhal on the announcement day was eliminated on the following day. Although our point estimates con®rm Hendricks and Singhal's [11] ®nding that a typical Baldrige Award winner is further rewarded with over a 0.5% abnormal return, our test results yield less evidence that this reward is statistically signi®cant. We ®nd the event-day return is only signi®cantly di€erent from zero at the 10% level and then for only two of the three models of abnormal return5. We explore two explanations for our less signi®cant results: our smaller sample size and our extended sample period. First, with 91 events, rather than 20, Hendrick and Singhal's test has greater power. However, their subsample of 12 large ®rms receiving awards from independent organizations corresponds closely with our 20 Baldrige Award winners, yet they report signi®cant results (see their Table 9) despite having only 12 events. Furthermore, our pre-1992 sample yields statistically signi®cant results with a sample of seven ®rms. Thus sample size alone does not explain why the Hendricks and Singhal results are more signi®cant than ours. Nevertheless, in Panel B we report the results of 48 publicly traded winners of state quality awards6. This larger sample yields positive abnormal returns ranging from 0.11 to 0.23%, but these returns are statistically indistinguishable from zero. Thus we ®nd weak evidence of a stock response in our sample of Baldrige winners and almost no evidence in our state winners. The earliest state awards in our sample were given in 1991 (by Maine and North Carolina), thus the insigni®cance of the state awards may be driven by our second explanation: the extended sample period. Second, our data run through the 1997 winners; Hendricks and Singhal's sample ends in 1991. If in the 1980's investors learned that TQM award winners had positive abnormal returns, then in the 90's these investors would have clear ®nancial incentives to gather data on quality and then trade on these data. As the data on quality are incorporated into stock prices, the news or signal of the actual award becomes more and more redundant. In Panel C we test for a diminishing stock price response predicted by investor learning and increasingly ecient (with respect to quality) markets. Panel C reports results for a subsample ending in 1991 (n = 7) and for a second subsample from 1992 to 1997 (n = 13). Interestingly, the earlier subperiod corresponding to Hendricks and Singhal's [11] yields one signi®cant rejection (market model) of the null hypoth-

600

G. Adams et al. / Omega 27 (1999) 595±604

esis in favor of positive abnormal returns. For the 13 winners since 1991, however, insigni®cant and generally lower mean abnormal returns are found. Thus we con®rm Hendricks and Singhal's ®nding of a stock price response using their sample period, but our subsequent subsample and our state results suggest that the link between quality awards and stock price increases has diminished. Before performing our microanalysis, we report the results of a sensitivity analysis in Table 5. The base case results, from Tables 2±4 at t = 0, are repeated in the ®rst column for comparison purposes. In cases 1 (2), we rede®ned the market return as an equallyweighted portfolio of NYSE (AMEX) stocks only, rather than the portfolio of NYSE and AMEX stocks combined. These two perturbations yielded results similar to those in the base case. In case 3, the estimation period is 200 days after the announcement, t = 10 to t = 209. We also considered 300- and 100day estimation periods in cases 4 and 5, respectively. These perturbations yield results similar to the base case, although the market model results are not stable with t-statistics varying between 1.05 and 3.13. One possible explanation for the weak evidence of a stock price response is that other, non-quality related, negative news o€set any price measure due to the news about winning the Baldrige Award. We control for such compounding news in two ways. In case 6, we exclude from the sample Motorola, IBM and AT&T in 1992 because they had Wall Street Journal articles about them on the same day as the Baldrige Award article. In case 7, we exclude IBM, AT&T and Corning because they had Wall Street Journal articles that were negative in tone sometime between the

announcement and ®ve days before the announcement, inclusive. The dropping of events with potentially compounding in¯uences yields slightly more signi®cant evidence of a stock price increase on the announcement day. Overall, we found that the link between winning the Baldrige Award and a stock price increase is marginally signi®cant in our full sample of Baldrige winners and insigni®cant in our larger sample of state award winners. We ®nd stronger evidence using the pre-1992 data and ®nd that the stock price response is weaker in the 1992 to 1997 period. 4. Microevidence In the preceding macroanalysis, the mean of all 20 Baldridge award winners' abnormal return on the " announcement day, A 0 , was simultaneously tested for signi®cance. With only 20 observations, we do not need to rely solely on two summary statistics (mean and standard deviation); rather, we can examine the entire distribution to see which ®rms are driving the joint results. In our microanalysis, each individual company's abnormal return on the announcement day, Ai0, was tested for signi®cance using: TSi0 ˆ where: 4 ” Si ˆ
À10 Á 1 ˆ À " 2 Ait À A i 199 tˆÀ209

" Ai0 À A i , ”i S 51a2 ,

…3†

Table 5 Sensitivity analysis of the macrotest of a stock price response to winning the Baldrige Award (mean abnormal returns on day 0, with t-statistics in parentheses). Base case: market return includes NYSE and AMEX stocks and the estimation period is 200 days before the announcement, t=À209 to t=À10. Case 1: market return includes only NYSE stocks. Case 2: market return includes only AMEX stocks. Case 3: estimation period is for 200 days after the announcement, t = 10 to t = 209. In case 3, the 1997 award winners are dropped from the sample since 1998 return data are needed (N = 16). Case 4: estimation period is for 300 days prior to the announcement, t=À309 to t=À10. Case 5: estimation period is for 100 days prior to the announcement, t=À109 to t=À10. Case 6: excludes Motorola, IBM and AT&T in 1992 because they had compounding articles in the Wall Street Journal on the same day as the Baldrige Award announcement (N = 17). Case 7: excludes IBM and AT&T in 1992 and Corning in 1995 because they had an article with a negative tone in the Wall Street Journal sometime between the announcement and ®ve days before the announcement, inclusive (N = 17). t-statistics are in parenthesis. NC=no change from base case Abnormal return model Base case Case number 1 Mean adjusted Market adjusted Market model
a b

2

3

4

5

6

7

0.44 (0.98) NC NC 0.63 (1.19) -0.02 (0.04) 0.51 (1.24) 0.94a (1.76) 0.65 (1.33) 0.92a (2.27) 0.95a (2.39) 0.83a (1.95) 1.14a (2.39) 0.92a (1.91) 0.92a (2.53) 1.41a (2.89) 1.10a (2.49) 1.07a (2.63) 1.07a (2.65) 0.92a (2.18) 1.46b (3.13) 0.55 (1.05) 1.15b (3.04) 1.63b (3.17) 1.28b (2.89)

Indicates signi®cance at the one-tailed, 10 and 1% critical values, respectively. Indicates signi®cance at the one-tailed, 10 and 1% critical values, respectively.

G. Adams et al. / Omega 27 (1999) 595±604

601

and
À10 1 ˆ " Ai ˆ Ait X 200 tˆÀ209

Thus, each company's announcement day abnormal return, Ai0, was tested to see if it is di€erent from its " typical daily abnormal return, A i . Before presenting the microanalysis, a caveat is in order. The 20 individual tests are anecdotal in nature because they cannot be combined into one statistic with a con®dence interval or p-value. Thus, confounding news unrelated to quality is not averaged out over many events. Table 6 reports the company-speci®c event-day abnormal returns (using all three models of abnormal return) with t-statistics from Eq. 3 reported in parentheses. For example, using Motorola, the market adjusted abnormal return was À1.44, which was not signi®cantly di€erent from the abnormal returns experienced in the 200-day estimation period (t-statistic of 0.68). When the market adjusted and market model measures are used, 11 of the 20 individual stocks'

abnormal returns are positive but only four of them are signi®cantly higher than their typical abnormal return. Speci®cally, General Motors, Federal Express, Armstrong World Industries and ADAC Laboratories received unusually high and statistically positive abnormal returns on the day their companies won the quality award. (Interestingly, General Motors was so aggressive in using the Baldrige Award in its advertisements that new rules governing advertising were instituted). When the mean adjusted model is used, only seven of the 20 ®rms received positive abnormal returns and only two of those are statistically signi®cant. The preceding microanalysis results depend on three models of abnormal returns and assume individual stock returns are normally distributed. To avoid the reliance on assumptions and models, the last column of Table 6 reports the results of a crude, but assumption free, nonparametric measure of the individual announcement day return's uniqueness. For each company, the returns from the 200-day estimation period are ranked from high to low and the percentile ranking of the announcement day is reported. For example,

Table 6 Individual abnormal returns and t-statistics for micro-tests of a stock price response to winning the Baldrige Award. t-statistics are in parentheses Company name Abnormal returns % (t-statistics) mean adjusted Motorola Westinghouse Electric Corp., Commercial Nuclear Fuel Division Xerox Corporation, Business Products and Systems General Motors, Cadillac Motor Car Division IBM, Rochester Federal Express Corporation Solectron Corporation AT&T, Network Systems Group, Transmission Systems Business Unit and AT&T, Universal Card Services Texas Instrument, Defense Systems and Electronics Group Eastman Kodak, Eastman Chemical AT&T, Consumer Communications Services GTE, Directories Corporation Armstrong World Industries, Building Products Operations Division Corning, Telecommunications Products Division ADAC Laboratories Dana Commercial Credit Corporation 3M, Dental Products Division Merrill Lynch and Company, Credit Corporation Solectron Corporation Xerox Corporation, Business Services
a b

Percentile market model À0.34 (0.21) 0.70 (0.66) À0.61 (0.57) 3.31b (2.63) À0.73 (0.75) 3.77a (2.07) 4.72 (1.46) À0.85 (0.80) 0.17 (0.09) 0.05 (0.03) À0.27 (0.29) À0.66 (0.57) 4.30b (2.96) À1.13 (0.72) 7.93a (1.99) 0.12 (0.09) À0.73 (0.53) À0.04 (0.02) 0.23 (0.10) 1.48 (1.03) 15 35 13 90 6 92 86 14 49 49 30 24 99 17 96 55 23 38 44 75

market adjusted À1.45 (0.70) 0.02 (0.06) À0.76 (0.66) 2.93a (2.27) À0.61 (0.77) 3.57b (1.97) 4.68 (1.32) À0.80 (0.73) 0.08 (0.03) 0.13 (0.02) À0.24 (0.23) À0.59 (0.42) 4.29b (2.88) À1.27 (0.73) 8.26b (2.01) 0.03 (0.07) À0.85 (0.58) À0.41 (0.31) 0.02 (0.03) 1.35 (0.84)

À1.91 À0.55 À1.13 1.82 À1.85 2.50 3.56 À0.98

(0.90) (0.37) (0.92) (1.16) (1.56) (1.26) (1.06) (0.88)

À0.26 (0.11) 0.00 (0.00) À0.50 (0.47) À0.75 (0.65) 3.90a (2.57) À1.45 (0.91) 7.88a (1.98) À0.03 (0.02) À1.11 (0.76) À0.93 (0.42) À0.39 (0.16) 0.92 (0.57)

Indicates signi®cance at the two-tailed, 10 and 1% critical values, respectively. Indicates signi®cance at the two-tailed, 10 and 1% critical values, respectively.

602

G. Adams et al. / Omega 27 (1999) 595±604

Motorola's À1.97% return (see Table 1) was in the 15th percentile of all 200 returns suggesting, if anything, the award hurt Motorola. Our nonparametric measure yields similar results to the parametric ®ndings, namely that GM, Federal Express, Armstrong and ADAC, (percentiles from 90 to 99) had atypical announcement day returns, but that the other 16 ®rms' announcement day returns were not particularly noteworthy.

5.2. The award does not matter The Baldrige Award may have limited in¯uence on stock prices because it just does not matter, meaning that it has no implications for either future cash ¯ows (numerator in Eq. (1)) or for risk-adjusted discount rates (denominator in Eq. (1)) and consequently is unrelated to stock prices. Even some of the Award's Board of Overseers argue that the award may not matter ®nancially. For example, Garvin [10, p. 83] states, To fault the Baldrige Award for not rewarding ®nancial success is meaningless - it was never meant to. There are no pro®t guarantees accompanying a high score. Indeed, winning is neither a necessary nor sucient condition for ®nancial success leading to what one examiner has called ``the problem of the Baldrige-winning buggy-whip manufacturer''. The award also may not matter if it is given to companies with products of either marginal quality or excessively high quality. On the one hand, some critics argue that the award is given not to the ®rms with real quality products, but rather to the ®rms that satisfy the requirements of the award, many of which may not be integral to the making of quality products. (See, for example, the criticisms, with replies, in the Harvard Business Review, January±February 1992 and in Quality Progress, May 1991). On the other hand, an excessive focus on quality can actually hurt stockholder value if it leads a ®rm to spend money on quality beyond the point where the marginal cost of the quality improvement equals its marginal bene®t. The potential for expenditures on quality to either add or subtract value may explain why some ®rms (50% in our study and about 55% in the Hendricks and Singhal study) have positive abnormal returns while others have negative abnormal returns. Another possible explanation for the unimportance of award announcements could be the decreasing importance of quality as a competitive element; this lack of in¯uence would be re¯ected in stock price performance. Dobyns and Crawford-Mason [7] suggest that quality no longer di€erentiates, but is only the required ticket into the current international marketplace. Product and service quality have been emphasized for over a decade. Because of increased understanding of the importance of quality in many industries, managers in most organizations have worked to improve their systems of quality assurance and production. If strategic parity is increasing in the area of quality, future cash ¯ows of award winners may not be seen as very di€erent from nonrecipients7. Our ®nding of marginal and diminishing stock price responses on the announcement date is consistent with

5. Possible explanations In this section, several conjectures are made as to why we ®nd only limited evidence to support the hypothesis that stockholders are rewarded on the day their ®rms receive a quality award and why the evidence may be diminishing with time. These conjectures are included to give direction to further work on the subject. The conjectures can be categorized into two broad areas: the award is not news and the award does not matter. 5.1. The award is not new news For several reasons, the announcement of the Baldrige award may not actually contain any new information or news about publicly traded companies. First, stock analysts are typically already informed about a company's TQM programs. For example, analysts often go on plant tours, visit company headquarters and test or sample products. Likewise, management often visits Wall Street to inform the analysts of the company's status. Second, a reputation for high quality products and/or services, coupled with high customer satisfaction and loyalty, may have impacted ®nancial statements, which, in turn, impact stock price. No matter how the information was obtained, prior knowledge of the TQM program or the award may cause the stock price increases to be achieved prior to the award announcement. Finance theory would suggest that as stock analysts learned of the relationship between TQM and stock prices, they would have incentives to collect information on quality and incorporate it into the stock price as soon as possible. Such behavior would explain the diminishing stock price response over time.
7 A third explanation for limited stock price response is that the tests lack power. Daily abnormal stock returns are quite noisy. In fact, the market adjusted model's full sample standard deviation of abnormal returns is 0.41%; thus, the news needs to generate about a 1% daily stock return in order to be detected statistically (2 standard deviations equals 0.82%).

G. Adams et al. / Omega 27 (1999) 595±604

603

both the `not news' and `does not matter' explanations. However, several studies using complementary methodologies lead us to believe that stockholders are rewarded for successful TQM implementation but the rewards now come before and after the formal recognition of the award. 5.3. Other tests of quality's ®nancial impact Three studies, one by Easton and Jarrell [8] and two additional papers by Hendricks and Singhal [12,13], ®nd evidence relating superior ®rm performance to TQM programs8. In Easton and Jarrell [8], interviews of senior quality executives by a former senior examiner for the Baldrige Award were used to ascertain the extent and approximate time of TQM implementation. Easton and Jarrell [8] also used control groups and adjusted for analysts' expectations. They found that successful TQM implementation leads to better performance in a number of measures of pro®tability. Hendricks and Singhal [13] use winning a quality award as evidence of successful TQM implementation and ®nd evidence that these award-winning ®rms outperform a control group in various measures of pro®tability, growth and eciency. The evidence of superior performance is not only limited to the years after the award is won, but is also evident in the years prior to the award. In Hendricks and Singhal [12], the long-run stock returns of quality award winners are shown to be higher than non-award winners. However, the superior performance is observed four and ®ve years after the award. Thus the results con¯ict with the ecient markets hypothesis and are subject to the methodology criticisms found in Kothari and Warner [17] and Barber and Lyon [1]. The ®ndings of Easton and Jarrell [8] and Hendricks and Singhal [12,13] suggest an interpretation for our ®nding of a weak and diminishing stock price response on announcement day that di€erentiates between the competing explanations of `not news' and `does not matter'. They show that successful TQM implementation yields lasting rewards in many measures of eciency, pro®tability and returns. We found marginal, and in recent years no, signi®cant stock price increase on the day of the quality award announcement. Our ®ndings do not con¯ict with those of Easton and Jarrell [8] and Hendricks and Singhal [12,13] if the stock price increase is distributed over months and years as stockholders and analysts gain tangible evi8 Many other papers compare accounting measures for ®rm performance for companies with and without TQM programs (see Easton and Jarrell [9]). We focus on these three papers because they do not rely on self-reported data and use control groups to account for macroeconomic and industry e€ects.

dence of the program's success. Over time, information on quality is re¯ected in a stock's price at an earlier stage in the implementation. Additionally, even years after an award, stock analysts will adjust prices to re¯ect information on a TQM program's actual results relative to the expected results. Apparently, this migration of information and the associated stock price response has made the actual award process a ceremonial formality with little new informative content. We show that in recent years the stock price response on event day has diminished. Our evidence, together with other ®ndings, suggests that the stockholder rewards have not vanished altogether but have migrated in time. 6. Conclusions Prior research shows that many stockholders are rewarded through the successful implementation of TQM programs. Hendricks and Singhal [11] explored the impact of TQM on stockholder value by comparing the stock price of publicly traded ®rms before and after they receive a quality award. In their macroanalysis event study, they found that quality award winning companies, as a group, are further rewarded with a statistically signi®cant positive abnormal return on the announcement day. After revisiting and extending Hendricks and Singhal's [11] event study several ways, we found four reasons why stockholders should be cautious about expecting a stock price jump on the day of a quality award announcement. First, our event study of Baldrige Award winners found marginally signi®cant evidence of a stock price response in only two of the three measures of abnormal returns. Second, our state award winners showed no signi®cant stock price response. Third, Hendricks and Singhal's [11] event study ended in 1991; our subsample tests found signi®cance only in the pre-1992 subperiod. Our point estimates in the 1992±1997 subperiod are signi®cantly positive and are signi®cantly less than the pre-1992 subperiod. Fourth, in a company-by-company microanalysis, we found that only half of the Baldrige companies had positive abnormal returns, and only four (General Motors, Federal Express, Armstrong World Industries and ADAC Laboratories) had signi®cantly positive abnormal returns. The weak and diminishing stock price response on the award announcement day does not mean that stockholder rewards for TQM programs are ignored. Stockholders may still bene®t from successful TQM implementation, but the bene®ts may no longer be realized on the announcement day. Rather, forward-looking stock analysts increase a company's stock price as tangible evidence of successful and ®nancially respon-

604

G. Adams et al. / Omega 27 (1999) 595±604 [4] Crosby PB. Quality is free: the art of making quality certain. New York: McGraw-Hill Book Company, 1979. [5] Deming WE. Out of the crisis. Cambridge: Massachusetts Institute of Technology Center for Advanced Engineering Study, 1986. [6] Dyckman T, Philbrick D, Jens JS. A comparison of event study methodologies using daily stock returns: a simulation approach. J Account Res 1984;22:1±30. [7] Dobyns L, Crawford-Mason C. Quality or else: the revolution in world business. Boston: Houghton-Mi‚in Company, 1991. [8] Easton GS, Jarrell SL. The e€ects of total quality management on corporate performance: an empirical investigation. J Business 1998;71:253±307. [9] Easton GS, Jarrell SL. The emerging academic research on the link between total quality management and corporate ®nancial performance: a critical review. In: Stahl M, editor. Perspectives in total quality. Malden: Blackwell Publishers, 1998, in press. [10] Garvin DA. Managing quality: the strategic and competitive edge. New York: The Free Press, 1988. [11] Hendricks KB, Singhal VR. Quality awards and the market value of the ®rm: an empirical investigation. Manage Sci 1996;42:415±36. [12] Hendricks KB, Singhal VR. The long-run stock price performance of ®rms with e€ective TQM programs as proxied by quality award winners. Working Paper. School of Management, Georgia Institute of Technology, 1996. [13] Hendricks KB, Singhal VR. Does implementing an e€ective TQM program actually improve operating performance? Empirical evidence from ®rms that have won quality awards. Manage Sci 1998, in press. [14] Ishikawa K. Quality Control in Japan. In: Sasaki N, Hutchins D, editors. The Japanese approach to product quality: its applicability to the West. Oxford: Pergamon Press, 1984. [15] Ishikawa K, Lu DJ. What is total quality control? The Japanese way. Englewood Cli€s: Prentice-Hall, 1985. [16] Juran JM. Juran on planning for quality. New York: The Free Press, 1988. [17] Kothari SP, Warner JB. Measuring long-horizon security price performance. J Finan Econ 1997;43:301±40. [18] Malatesta PH, Thompson R. Partially anticipated events: a model of stock price reactions with an application to corporate acquisition. J Finan Econ 1985;14:237±50. [19] Phillips LW, Chang D, Buzzell RD. Product quality, cost position, and business performance. J Market 1983;47:26±43.

sible TQM progress is made. Such evidence is discovered in plant visits by Wall Street analysts and marketplace acceptance by customers. Complete or partial knowledge of a ®rm's TQM programs and their e€ects on future cash ¯ows make the actual announcement of a quality award completely or partially redundant in a ®nancially informative sense. Insigni®cant results in quality award announcement event studies alone cannot distinguish whether stockholders received no rewards from TQM or their rewards were received long before the announcement. Easton and Jarrell [8] and Hendricks and Singhal [12,13] ®nd evidence that stockholders do bene®t from successful TQM implementation. Their evidence of a reward, combined with our ®ndings of diminishing event-day rewards, leads us to believe that the TQM rewards to stockholders do exist. However, the stock response now occurs as relevant information about the program and its success becomes known by stock market participants, not only on the day of formal recognition. In a strictly ®nancial sense, TQM matters but awards may not. Acknowledgements The Harold R. Silver Fund and Brigham Young University provided ®nancial support for databases and research assistants. Grant McQueen received ®nancial support from the Goldman Sachs fellowship. Kristie Seawright received ®nancial support from the Reed Dame Fellowship. The authors thank Vinod Singhal, Steven Thorley and Stanley Fawcett for their comments. References
[1] Barber BM, Lyon JD. Detecting long-run abnormal stock returns: the empirical power and speci®cation of test statistics. J Finan Econ 1997;43:341±72. [2] Brown SJ, Warner JB. Measuring security price performance. J Finan Econ 1980;8:205±58. [3] Brown SJ, Warner JB. Using daily stock returns, the case of event studies. J Finan Econ 1985;14:3±31.