V. BARGAINING IN FLUX: LABOR AND MANAGEMENT RESPONDS TO A PERIOD OF UNCERTAINTY

Post-Strike Effects of Labor Conflict on Retail Consumers: Preliminary Evidence from the 1998 Northwest Airlines and General Motors Strikes

RICHARD N. BLOCK AND BRIAN D. SILVER
Michigan State University

Abstract

      This paper, reporting results from survey questions on the effect of the 1998 GM and Northwest strikes on potential retail customers in Michigan, suggests that a strike generates negative attitudes on the part of consumers toward purchasing the struck product/service, that consumers act accordingly, and that a product that is undifferentiated from its competitors’ products will suffer more than a differentiated product. The results also suggest the existence of an “anger effect” toward high–market share Northwest shortly after the strike from those who were dependent on Northwest. This “anger effect” may have started to dissipate 9–10 months after the strike.

Introduction

      Industries in flux often experience employee dissatisfaction and uncertainty. When employees in a changing industry are represented by a union, this dissatisfaction and uncertainty can take the form of labor conflict such as strikes. With production or service curtailed during a strike, customers whose demand for a product or service cannot be deferred must shift their demand to competitors. Strike effects on consumers, therefore, are likely to be predictable.

      Much less certain, however, are the post-strike effects on consumers. Does the disruption in production or the provision of the service caused by the strike have a negative reputation effect such that there is a possibility of a long-term or permanent loss of the firm’s customer base with a possible reduction in post-strike employment opportunities for the employees represented by the union? If so, then labor conflict associated with industry flux or change has the potential to increase rather than decrease uncertainty.

      The summer of 1998 saw two strikes associated with industry flux, both of which affected the state of Michigan. The auto industry is subject to continual pressure from foreign competitors, causing, among other things, an ongoing reallocation of work among facilities as the companies continue to attempt to reduce production costs. Strikes against General Motors by two UAW locals in Flint, Michigan, that started on June 5 and June 11 essentially halted all production at GM by mid-June. The strikes ended on July 29 (Christian 1998; “UAW Announces” 1998; “UAW Strikes”1998).

      Airlines continue to experience labor conflict associated with flux and change as the industry restructures and unions attempt to address concessions from the late 1980s and early 1990s. A strike against Northwest Airlines by the Airline Pilots Association, an attempt to recoup old concessions, affected Michigan because Detroit is a Northwest hub and Northwest controlled approximately 76 percent of the market at the Detroit Metropolitan Airport.¹ The strike began on August 28, 1998, and ended on September 13, 1998 (Zuckerman 1998a, 1998b). Northwest was flying a full schedule by September 21, 1998 (Kennedy 1998).

      How did these strikes affect potential customers of General Motors and Northwest? In this paper, we exploit the occurrence of these two large strikes that were well-publicized in Michigan and report the results from experimental survey questions on the effect of these strikes on the likelihood that potential retail customers of GM and Northwest in Michigan would purchase a GM vehicle or fly Northwest. Whether a strike exacerbates or ameliorates long-run changes due to potential shifts in consumer demand away from the struck product is a question that has never been addressed. It is hoped that the results reported here will begin to fill that gap by providing preliminary empirical evidence regarding whether there may be a post-strike shift in consumer demand away from the struck firm.

Literature Review

      Previous work on the post-strike effect of strikes has focused on firms and industries as the unit of analysis. In an event study of airline strikes during the period 1963–1986, DeFusco and Fuess (1991) found evidence consistent with post-strike effects. For intervals that included the 30-day period after the strike settlement, nonstruck carriers had positive abnormal returns. By contrast, struck carriers experienced negative abnormal returns for the mutual aid pact period (1963–1978) but not for the nonpact deregulated period (1978–1986).

      Other studies that have examined the post-strike impact of strikes generally, rather than solely in the airline industry, have found evidence consistent with post-strike effects. Neumann (1980) found that firms (shareholders) incur costs during and after a strike, and that a firm valued at $500 million dollars on the day of the settlement would see its value reduced by $750,000 less 14 days after the strike. Using an event study methodology, taking the event as the period from 30 days before the strike to 30 days after the strike, Becker and Olson (1986) found that the cumulative average return to struck firms (for strikes involving 1,000 workers or more) was approximately 4 percent less than the cumulative average return for nonstruck firms, estimating the average strike cost shareholders to be between $72 million and $87 million (in 1980 dollars). On the other hand, Kramer and Vasconcellos (1996) found that the cumulative average return for 21 struck firms between January 1982 and July 1990 increased approximately 1.8 percent in the 30-day post-strike period relative to the pre-strike period, suggesting that any losses of market share or customers were either fully recouped or minimal or that the strike resulted in cost reductions that more than offset any revenue reductions associated with consumer diversion to competitors.

      Although not directly examining the post-strike period, Neumann and Reder (1984) studied the strike-associated changes in annual industry-level output of 63 industries between 1958 and 1978. They found that reductions in annual industry output were small, less than 0.65 percent in all industries except ordinance and accessories, suggesting that, along with inventory draw-downs, output increases in nonstruck firms (competitors) were compensating for output losses in the struck firms. Such output increases in nonstruck firms would be consistent with post-strike gains for nonstruck firms and post-strike losses for struck firms.

Theoretical Considerations

      Two questions may be asked: (1) Is there post-strike customer diversion from struck firms?; and, if so, (2) are there differences in the level of diversion based on substitution possibilities? If the products of nonstruck competitors are perfect substitutes for the struck good, and if purchase of the goods cannot be deferred, then, in principle, 100 percent of the demand of the struck good or service could be diverted to nonstruck competitors. Diversion, both during the strike and post-strike, will be less than otherwise if the competitors’ goods are not perfect substitutes, if the transaction costs of switching are nonzero, and if there is an absence of substitutes available for the struck product or service.

      Assuming that the nominal price for comparable goods/services is market determined, imperfection in substitution may result from differences in specific attributes of the product or service, such as styling with respect to autos, quality, implicit price² and, in the case of airlines, scheduling. Transaction costs would include such costs as investment in time to learn about the attributes of substitutes. Availability of substitutes, at least in the short-run, would depend on the extent to which the struck firm dominates the market.

      Based on the foregoing, we hypothesize that post-strike diversion of demand from Northwest and GM would be nonzero, as there are substitutes for the products of both firms. Based on product differentiation, we hypothesize that post-strike diversion from Northwest would be greater than from GM. Other airlines provide flights that are highly substitutable, if not perfectly substitutable, for Northwest flights, and other modes of transportation would be available, at least for relatively short trips. On the other hand, while the products of other auto companies may be viewed as substitutes for GM products, there are sufficient differences in styling and reputation among the products of different automobile companies to create some product differentiation.

      The potential post-strike diversion from Northwest would be lower, however, the lower the availability of substitutes, e.g., non-Northwest flights. Thus, regarding the second question, we would hypothesize that the greater the market share held by Northwest at the airport that serves the residence of a customer, the lower the potential diversion.

Data and Method

      The basic source of data for this study is the State of the State Survey (SOSS) conducted by the Institute of Public Policy and Social Research at Michigan State University. This survey is administered four times per year to a stratified random sample of approximately 960 respondents in Michigan for the purpose of monitoring the views of citizens on public issues in Michigan.³

      We collected data on the perception of SOSS respondents (consumers) to these two strikes, asking respondents if the strike made it more or less likely that they would buy a car from GM or travel on Northwest, or had no effect on the likelihood. To determine the longevity of any post-strike effects, data were collected in two waves; in the fall of 1998, two and one-half months after the GM strike and about a month after the Northwest strike, and then in the summer of 1999, 1 year after the end of the GM strike and approximately 10 months after the end of the Northwest strike. While limiting the sample to Michigan respondents may make the sample less representative of all consumers than a national sample, all respondents were likely to be familiar with the strikes, as these strikes were prominent Michigan news stories between June and September 1998.

      The SOSS also contained data on the county of residence of each respondent. By combining the SOSS county data with data on the noncharter commercial flight market share of Northwest at each airport in Michigan, we were able to estimate the availability of Northwest substitutes for any respondent. It would be expected that the greater the market share of Northwest in the airport most proximate to that county, the lower the availability of substitutes for Northwest, and the smaller the percentage of respondents who should state that they are “less likely” to fly Northwest due to the strike.

Results

      Table 1 presents the means for variables GMLESS and NWLESS, the percentage of respondents who stated they were “somewhat less likely” or “less likely” to purchase/lease a GM product or fly Northwest, respectively, as a result of the strike. As can be seen, in the fall 1998 administration, 26 percent of the respondents said they were less likely to purchase a GM product as a result of the strike, while 41 percent of the respondents stated they were less likely to fly Northwest as a result of the strike. Both percentages are significantly different from zero at p .01 (t = 25.22 for NWLESS; t = 17.6 for GMLESS), suggesting that these results did not occur by chance. These results are consistent with the existence of short-run negative reputation effects. These two percentages are also significantly different from each other at p
.01 (t = 7.55), suggesting that any negative reputation effects were greater for Northwest than for General Motors. This result is consistent with what would be expected based on substitution principles as Northwest produces a service that is less differentiated from the product of its competitors than is the GM product.⁴

      Table 1 also presents the comparable percentages for the summer 1999 SOSS administration, about 10 to 12 months after the strikes for all respondents. With a different set of respondents, negative reputation effects persist. Both percentages are significantly different from zero at p .01 (t = 24.98 for NWLESS; t = 17.2 for GMLESS), and the percentages are different from each other at the p .01 level (t = 8.22).

      Further insights can be obtained by examining responses regarding vehicle purchase or lease intention. Respondents in the fall 1998 survey were asked whether they were leaning toward or intended to purchase a GM vehicle when the strike began; 78 responded yes. Of these, 33.7 percent responded that they were less likely to purchase/lease a GM vehicle as a result of the strike. Of those 78 respondents, 54 actually purchased or leased a vehicle during or after the strike. Of those 54 who actually purchased or leased a vehicle, 12 (22.2 percent) purchased/leased from another manufacturer. This is close to the 25 percent of all fall 1998 respondents who stated they were less likely to purchase a GM vehicle. A majority of car purchasing “GM Leaners/Intenders” who stated they were less likely to purchase/lease a GM vehicle appear to have acted in accordance with their stated preference, which suggests the validity of the attitudinal questions.

      An additional perspective on post-strike consumer effects can be obtained by attempting to measure consumer substitution options. As the Northwest market share was approximately 76 percent at the Detroit airport, but only about 47 percent in Michigan outside of Detroit during 1998 and 1999,⁵ it was hypothesized that negative post-strike consumer effects on Northwest would be greater outside the Detroit metropolitan area than within the Detroit metropolitan area. This was because the smaller market share of Northwest outside of Detroit would mean that consumers outside of the Detroit metropolitan would have more Northwest substitutes available than consumers in the Detroit metropolitan area. Therefore, we analyzed the differences in responses toward Northwest by whether the respondent’s county of residence was within, or outside, the metropolitan Detroit area.

      Because these results could be affected by the definition of metro Detroit, we defined metro Detroit in two ways. Definition 1 considered metro Detroit as consisting of Lenawee, Livingston, Macomb, Monroe, Oakland, St. Clair, Washtenaw, and Wayne Counties.⁶ Definition 2 removed St. Clair and Livingston Counties from metro Detroit and placed them in the “outstate” category, as portions of both of these counties are close enough to the airport in Flint that it might be rational for some portion of the population of these counties to use the Flint airport.⁷

      The results, presented in the 2nd and 3rd columns of Table 2, are precisely opposite of those that would be predicted based on elasticities of substitution. Despite less choice of air carriers, respondents in metropolitan Detroit were significantly more likely than those from outstate to respond that they were less likely to fly Northwest. This suggests that, at least shortly after the strike, when the strike was still likely fresh in the minds of respondents, the existence of an “anger effect” toward Northwest outweighed economic rationality for this population with a low elasticity of substitution away from Northwest.

      These results from the summer of 1999 are presented in the 4th and 5th columns of Table 2. They demonstrate that the gap between “outstate” respondents and the metropolitan Detroit respondents closed. Using Definition 1, the percentage of metropolitan Detroit respondents who stated they were “less likely” to fly Northwest declined by 5 percentage points (10.4 percent), while the percentage of outstate residents who stated they were “less likely” to fly Northwest increased by 4 percentage points (12.7 percent). The difference between the two groups, which was significant in the fall of 1998, was no longer significant in the summer of 1999. The results under Definition 2 displayed even stronger convergence, with the percentage of metro Detroit respondents who stated they were “less likely” to fly Northwest declining by 8.4 percentage points (17.3 percent), while the percentage of out-state residents who stated they were “less likely to fly Northwest” increased by 5.5 percentage points (15.9 percent).

      Taken together, these results suggest that over time, economic rationality may have replaced anger. Residents in metropolitan Detroit display an increased willingness to fly Northwest, despite the strike, perhaps reflecting their lack of choices; and outstate residents appear to display a decreased willingness to fly Northwest, perhaps reflecting their greater choices. This difference is most pronounced for Definition 2, which limits the respondents in “metro Detroit” to those closest to the Detroit airport.⁸

Conclusions

      The results of this study should be considered very preliminary and experimental, as we are unaware of any other attempt to survey potential retail consumers regarding the effects of labor conflict on their views toward purchasing struck goods or services. The confluence of these two strikes in Michigan along with the quarterly administration of SOSS, however, provided an opportunity to determine if sensible results could be obtained on this question.

      We believe the results are interesting. They suggest that companies and unions should consider the consequences of labor conflict for consumer behavior before embarking on that path. The results suggest that a strike generates negative attitudes on the part of consumers toward purchasing the struck product or service, that consumers act on these negative attitudes, at least in the short run, and that these negative attitudes may persist for a substantial period of time after the strike ends. The results also suggest that a product that is essentially undifferentiated from the product of its competitors, with a high elasticity of substitution, such as air travel, will suffer more than a product that may be seen as differentiated, with a relatively low elasticity of substitution, such as automobiles.

      The results using region as a measure of differences in consumer substitution possibilities for the same service, air travel, generated results that were surprising but explainable. We hypothesized that there was an “anger effect” toward Northwest shortly after the strike from those who are dependent on Northwest, which may be associated both with short-term frustration about the strike and long-term frustration associated with an absence of choice in air travel. This “anger effect,” however, may have started to dissipate 9–10 months after the strike, with consumer responses perhaps tending to be based more on economic rationality and elasticities of substitution than on anger toward Northwest.

      We have not developed a full model of post-strike consumer response to labor conflict. Moreover, there may be bias due to the nature of the question. Reminding respondents about a strike months after the strike may have encouraged respondents to express a negative attitude toward the companies, even if the negative view had dissipated. Despite these caveats, we believe this paper provides a useful first look at the post-strike response of retail consumers to labor conflict and suggests that future research in this area could be fruitful.⁹

Acknowledgments

      This research was supported by the Institute for Public Policy and Social Research, Michigan State University. The survey data on which this paper was based are available at <www.ippsr.msu.edu/soss>.

Endnotes

1. This statement is based on data provided by Detroit Metropolitan Wayne County airport via fax dated November ', 2001. A copy of the fax is available upon request.

2. Product differentiation for Northwest (and other airlines) is likely to be largely based on implicit price reductions through the awarding of frequent flier miles. Other things equal, the implicit price of a ticket with an identical nominal price on another airline would be higher than on Northwest for a high-demand Northwest customer (who was not high-demand on the other airline) because of the value of the additional Northwest frequent flier miles to the high-demand Northwest customer. We did not obtain data on the accumulation of Northwest frequent flyer miles of each respondent.

3. Documentation of SOSS is available at <http://www.ippsr.msu.edu/SOSS/SOSS.HTM>.

4. The fall 1998 SOSS contained a question about whether the respondent was an employee of GM. As GM employees are always likely to purchase GM products because of price incentives, we computed the percentages excluding the 22 GM employees in the sample. The percentages of non–GM employee respondents who stated they were less likely to purchase a GM product as a result of the strike was 25.7 percent, almost identical to the 26 percent from the all respondents.

5. These data, obtained from all 17 Michigan airports with commercial air service in 1998 and 1999 and from the Michigan Department of Transportation, are available on request.

6. For a county map of Michigan, go to <http://midata.msu.edu/index01.html>.

7. Respondents in Genesee and Lapeer counties were considered to live outside the Detroit metropolitan area (“outstate”) for the purposes of this paper because of their proximity to the airport in Flint (Genesee County).

8. These results did not change when logistic regressions were run, including variables for union membership and income as controls. The regression results are available on request.

9. Also contrary to the inference of negative post-strike consumer effects is the evidence that the market share of Northwest did not decline from 1998 to 1999, staying at about 47 percent outstate and 76 percent in Detroit. (See endnote 6 for information on data.) But, this says nothing about what would have happened to market share had there been no strike.

References

Becker, Brian E., and Craig A. Olson. 1986. “The Impact of Strikes on Shareholder Equity.” Industrial and Labor Relations Review, Vol. 39, no. 3 (April), pp. 425–38.

Christian, Nichole M. 1998. “3,400 Strike G.M. Plant; Assembly Put at Risk.” The New York Times, June 6, p. 7.A.

DeFusco, Richard A., and Scott M. Fuess, Jr. 1991. “The Effect of Airline Strikes on the Struck and Nonstruck Carriers,” Industrial and Labor Relations Review, Vol. 44, no. 2 (January), pp. 324–33.

Kramer, Jonathan K., and Geraldo M. Vasconsellos. 1996. “The Economic Effects of Strikes on Shareholders of Nonstruck Competitors.” Industrial and Labor Relations Review, Vol. 49, no. 2 (January), pp. 213–22.

Neumann, George R. 1980. “The Predictability of Strikes: Evidence from the Stock Market.” Industrial and Labor Relations Review, Vol. 33, no. 4 (July), pp. 525–35.

Neumann, George R., and Melvin Reder. 1984. “Output and Strike Activity in U.S. Manufacturing: How Large Are the Losses?” Industrial and Labor Relations Review, Vol. 37, no. 2 (January), pp. 197–211.

“UAW Announces Outcome of Flint Ratification Vote.” 1998. PR Newswire, July 29.

“UAW Strikes at Delphi East in Flint, Michigan.” 1998. PR Newswire, June 11.

Zuckerman, Laurence. 1998a. “Northwest Pilots Strike as Talks Fail on Contract.” The New York Times, August 29, p. 1.A.

Zuckerman, Laurence. 1998b. “Northwest Workers Returning to Work.” The New York Times, September 14, p. 12.A.