Printer Friendly

Market forces and aircraft safety: an extension.


Recent work by Chalk focuses on whether market forces provide safe products via stock market reaction to unanticipated events. Chalk finds a $200 million decline in McDonnell Douglas stockholder wealth related to the 1979 DC-10 crash in Chicago. That decline far exceeds reasonable estimates of regulatory and liability costs, suggesting a market penalty for unsafe products. His results, however, do not seem consistent with the notion of efficient markets, as American Airlines maintenance procedures were quickly identified as the likely cause of the crash. This study finds no resulting shareholder wealth loss for American Airlines or McDonnell Douglas.


In a recent article in this journal, Andrew Chalk [1986] attempts to test whether market forces provide safety when products are too complex to permit buyer repurchase inspection. To do so he uses the rather ingenious method of examining the abnormal returns to McDonnell Douglas stock prices following the May 25, 1979, crash of a DC-10 just outside O'Hare Airport in Chicago.

Efficient market theory predicts that perceived safety problems in products will result in a fall in the firm's stock price due primarily to a product demand effect. If this is the case, market forces would compel producers to invest in product safety even in situations when the technical safety aspects of the product are beyond the comprehension of the average consumer, as is the case with large commercial aircraft.

Chalk's study finds large and statistically significant negative abnormal returns to McDonnell Douglas stockholders on the order of $200 million. In almost every case where news is released just after the crash (both favorable and unfavorable with respect to McDonnell Douglas), Chalk finds statistically significant abnormal returns to McDonnell Douglas stock.

What is interesting about this particular event was that faulty maintenance and not a product defect was ultimately blamed as the principal cause of the crash. Initially the general safety of the DC-10 was questioned, but subsequent evidence indicated that the maintenance procedure of an American Airlines mechanic (who later moved on to Continental and taught the technique there) of removing the engine and pylon in one step instead of separately, caused stress which led to the engine tearing loose. American Airlines and Continental ended up paying fines of $500,000 and $100,000 respectively for this procedure, but American did not admit guilt for the crash in paying the fine.(1)

Given that most of the blame for the crash fell on American's maintenance procedure, the finding of significant cumulative abnormal returns by Chalk are surprising in the context of efficient markets. Evidence of faulty maintenance procedures became public information within two weeks after the crash. Because of the initial uncertainty over the cause of the crash, significant abnormal returns for particular event days are likely, but those negative abnormal returns should be offset in later periods as the true cause becomes known to investors. Thus the finding of significant cumulative abnormal returns to McDonnell Douglas would not be consistent with the semistrong form of efficient markets.

The purpose of this note is to replicate and extend the findings of Chalk's study by examining the market forces that affected the other major party involved in subsequent litigation, American Airlines Corporation. Following Chalk's argument, one would expect to see the market impose substantial costs on American if it is believed that travel on that airline is less safe than previously thought. Travelers could substitute towards other airlines where possible, reducing cash flow and hence the value of the firm. The size of the demand effect should have little to do with American's investment in aircraft safety. Rather, it should depend on brand name loyalty and flight substitutability.

Alternatively, if the airworthiness of all DC-10s was at issue, travelers could substitute away from DC-10 flights on the same or other airlines. To the extent that such substitution was possible, we would expect to see airlines with no DC-10s incurring gains relative to those flying the DC-10.(2)


To examine the impact of the crash on American Airlines, the "event study" methodology employed by Chalk was replicated. Employing the market model with its usual assumptions, the abnormal return, eit, of firm i at time t is given by

eit = Rit - ^Rit.


R is the actual return and ^R is the return predicted by the market model

|R it = a + Beta |R mt + eit.


To test for the significance of eit in equation (1) over the forecast period, the abnormal return is standardized by dividing by the estimated standard error of the forecast. The standard error must be adjusted to account for the fact that market return data is actual return data. As noted by Chalk, the resulting Z-statistic is t-distributed with N-2 degrees of freedom under the null hypothesis of zero abnormal return.(3)

Equation (2) was estimated using daily return data obtained from the Center for Research in Security Prices (CRSP) tape. The CRSP equally-weighted market index was used as a proxy for market return data.(4) The estimation period chosen was 150 trading days immediately preceding the crash.(5)

In the aftermath of the crash, all DC-10s were grounded at one point, thus affecting all airlines using those planes. To draw additional implications about the impact of the crash, a portfolio of airlines flying the DC-10 and a portfolio of airlines not using the DC-10 were included in this study to determine the market forces acting on those firms.(6) As passengers attempted to substitute towards perceived safer aircraft and airlines, the portfolio of airlines not using DC-10s should have been the primary beneficiary. Therefore, one might expect to see this portfolio experience positive abnormal returns, while American and the DC-10 portfolio experience negative abnormal returns due to crash-related events. Limited route competition may, however, severely weaken any demand effect on the residuals.


The results of the market model estimation are reported in Table I for American and the two portfolios. The abnormal returns for specific days are reported in Table II. Table II is similar to that found in Chalk except that it is oriented towards news concerning American Airlines. Most of the major events surrounding the crash are included in that table.(7)

What is most striking about the results in Table II is the lack of significant abnormal returns. Because the crash occurred after the market had closed, the significant negative abnormal return to American Airlines on May 25 cannot be attributed to the crash. American did announce on that date a deep discount plan similar to United's, and the observed return may be a response to that announcement.

The one date with significant abnormal returns to almost all firms with correctly predicted signs was June 6. The Federal Aviation Agency (FAA) grounded all DC-10s on that date. Both American and the DC-10 portfolio experienced large negative abnormal returns, while the non-DC-10 portfolio experienced a positive, though insignificant, abnormal return.


Estimates from the Market Model
 DC-10 Non-DC-10
 American Portfolio Portfolio
a -0.0017 -0.0007 -0.0010
 (0.002) (0.76) (1.17)
B 1.6296 1.4486 1.5725
 (8.36) (15.62) (17.99)
R square 0.3165 0.6198 0.6843

Note: Absolute values of associated t-statistics are in parentheses.


Major Events Surrounding the DC-10 Crash of May 25, 1979, and Corresponding Abnormal Returns
Date Event
5/25/79 Friday McDonnell Douglas DC-10 owned by American
 Airlines crashes in Chicago shortly after take-off
 from O'Hare Airport (4:04 p.m. EDT).
5/28/79 Monday FAA issues Administrative Directive (AD)instructing all
 U.S. DC-10 operators to inspect engine mounting bolts
 (10:58 a.m. EDT)
5/29/79 Tuesday FAA grounds all DC-10s for further inspections by
 issuing a second AD (8:59 p.m. EDT)
5/30/79 Wednesday Some DC-10s return to service after second
5/31/79 Thursday "Bolt" theory dismissed. FAA tallies results of the
 second inspection. Safety defects were found in 37 of
 138 U.S. registered DC-10s (revealed after close).
6/1/79 Friday The Airline Passengers Association (APA) announces it wil
 file suit to ground all DC-10s until the problem(s) that
 caused the crash of Flight 191 can be found and corrected

Abnormal Return(%)
American DC-10 Non-DC-10
Airlines Portfolio Portfolio
-4.02 -0.66 -0.31
(1.69) (0.58 (0.29)
-2.08 -1.71 -0.06
(0.87) (1.51 (0.06)
 0.42 -1.37 1.45
(0.17) (1.21) (1.36)
 2.25 1.03 -0.12
(0.95) (0.91) (0.11
-1.27 -0.27 -0.23
(0.53) (0.24) (0.22)
6/4/79 Monday National Transportation Safety Board (NTSB) issues its s
 recommendations publicly. FAA announces that the second
 inspection it ordered may have caused more damage. Some
 DC-10s are regrounded at 9:00 p.m. EDT.
6/5/79 Tuesday Federal judge orders all DC-10s grounded until the cause
 of the Chicago crash is found and the problem cured
 (4:45 p.m. EDT). FAA wins a stay of the order (9:30
 p.m. EDT) to contest it in court the following day.
6/6/79 Wednesday At 3:00 a.m. EDT the FAA decides to ground all DC-10s
 and starts informing operators immediately. McDonnell
 Douglas learns of it at 5:45 a.m. EDT. At 9:00 a.m. EDT
 the FAA announces that it does not object to the tempora
 restraining order. At 2:00 p.m. EDT the FAA explains its
 position at a press conference.
6/7/79 Thursday Two DC-10 customers rumored to be ready to cancel orders
 for ten DC-10s worth $400 million.
6/8/79 Friday Maintenance emerges as the principal cause of the crash.
 Court battles to lift the grounding figure prominently.

Abnormal Return(%)
American DC-10 Non-DC-10
Airlines Portfolio Portfolio
-0.05 0.20 -0.33
(0.02) (0.18) (0.31)
-1.47 -0.54 -0.09
(0.61) (0.48) (0.08)
-5.31 -2.87 1.12
(2.22) (2.54) (1.05)
 0.44 -2.68 0.01
(0.18) (2.37) (0.09)
-0.32 -0.72 -0.61
(0.13) (0.63) (0.57)
Date Event
6/11/79 Monday Congressional hearing start.
6/15/79 Friday Judge Robinson issues order staying further proceedings.
6/18/79 Monday The NTSB tells a Congressional hearing that the
 American Airlines DC-10 that crashed was probably
 damaged during maintenance in March.
 European airlines agree on a maintenance plan that would
 allow their DC-10s to fly again, but not in U.S. airspac
6/19/79 Tuesday FAA clears the Boeing 747, Airbus A300B and Lockheed
 L-1011 of faults similar to the DC-10s.
 European airlines resume non-U.S. DC-10 flights.
6/25/79 Monday At hearings before an administrative law judge of the NT
 the FAA and McDonnell Douglas agree to a one-week
 postponement to try to reach an out-of-court settlement.
7/479 Wednesday FAA announces it will tighten its supervision of carrie
 maintenance procedures.

Abnormal Return (%)
American DC-10 Non-DC-10
Airlines Portfolio Portfolio
-2.43 -0.94 0.2
(1.02) (0.83) (0.28)
-2.44 -0.31 -0.30
(1.02) (0.27) (0.28)
-0.62 -1.58 0.18
(0.26) (1.40) (0.27)
-1.12 0.82 0.48
(0.47) (0.73) (0.45)
-2.44 -1.13 -1.08
(1.02) (1.01) (1.01)


Cumulative Average Residuals (Percentages) for Selected Dates
 DC-10 Non-DC-10 McDonnell
Date American Portfolio Portfolio Douglas
7/13 -5.24 -6.25 -0.08 -7.06
 (0.34) (0.97) (0.01) (0.61)
7/17 -6.27 7.46 -1.02 -3.00
 (0.39) (1.10) (0.15) (1.07)
7/30 0.60 4.74 4.22 -13.84
 (0.09) (0.63) (0.61) (1.03)

On Friday, June 8, maintenance emerged as the principal cause of the crash. No significant abnormal returns were indicated, however, for American or the two portfolios. In fact, no significant abnormal returns for any of the event days after the June 6 announcement were found.

The overall impact of the crash on a firm's stock price is measured by the cumulative abnormal return. This is the sum of the abnormal returns over a chosen period immediately after the crash. Those results are reported in Table III. The dates used in Table III are those used in the study by Chalk. The cumulative abnormal returns for American and the airline portfolios, along with those of McDonnell Douglas, are reported. As can be seen, no statistically significant cumulative abnormal returns are found for any of the firms or portfolios in the study. In addition, the cumulative residuals for McDonnell Douglas are only about half the size of those reported by Chalk and these are not statistically significant. All other aspects of the Chalk study were easily replicated.(8)


The major hypothesis in the Chalk study is that market forces can compel an investment in product safety by producers even in those cases where the product is too complex for consumers to fully understand all the technical aspects of the product. The threat of a capital loss of a magnitude of that found for McDonnell Douglas following the DC-10 crash would force safer aircraft to be built in the future.

The cumulative abnormal return results of the Chalk study, however, raise questions about the efficiency of capital markets, rather than providing evidence towards the investment-in-product-safety hypothesis. The early speculation about McDonnell Douglas's liability proved incorrect. One would therefore expect no significant cumulative abnormal returns to McDonnell Douglas shareholders, rather than the $200 million decline found by Chalk.(9) The abnormal returns should seemingly have been associated with American Airlines, the party ultimately found liable in the crash.

The evidence from this study indicates that American Airlines experienced no significant abnormal returns even after it became widely accepted that the airline's maintenance procedures were at fault. The portfolio of airlines flying the DC-10 did not accumulate significant negative abnormal returns and the portfolio of airlines not using DC-10s did not accumulate significant positive abnormal returns. The results of this study also suggest that the cumulative abnormal returns to McDonnell Douglas shareholders were not statistically significant and were much smaller in magnitude than those found by Chalk. The corrected cumulative abnormal returns of McDonnell Douglas following the crash appear more consistent with the semistrong form of efficient markets. REFERENCES Chalk, A. "Market Forces and Aircraft Safety: The Case of the DC-10." Economic Inquiry, January

1986, 43-60. Hoffer, G. E., S. W. Pruitt and R. J. Reilly. "The Impact of Product Recalls on the Wealth of

Sellers: A Reexamination." Journal of Political Economy, June 1988, 663-70. Jarrel, G. and S. Peltzman. "The Impact of Product Recalls on the Wealth of Sellers." Journal

of Political Economy, June 1985, 512-36.

(*) Associate Professor, Finance Department, University of Nebraska-Lincoln, Lincoln, Nebraska. I am grateful to Andrew Chalk, Richard DeFusco, Otis Gilley and Richard Sweeney for helpful comments. I am responsible for any remaining errors.

1. While American Airlines did not admit guilt, Chalk points out that the courts, in plaintiff's cases, split the liability: approximately 85 percent to American Airlines and 15 percent to McDonnell Douglas. A formal liability sharing agreement between American and McDonnell was reached in later cases but the details were not announced to the public. The legal and public perception thus appeared to hold American at fault.

2. Two articles in the Wall Street Journal addressed this issue to some extent: "Travelers Trauma: Some Passengers to Avoid the DC10," June 5, 1979, and "Some Travelers Still Go to Great Lengths to Avoid the DC10s," August 30, 1979. Immediately after the crash there appeared to be a great number of travelers trying to rebook flights. They were typically hindered by the limited number of non-DC-10 flights to certain destinations. The later article noted that in August some travelers still appeared to consciously avoid the DC-10, although the number had substantially declined. Several other travelers interviewed actually preferred the DC-10 because they felt the added inspections required after the crash insured a safer plane.

3. See Chalk [1986] for the detailed discussion on the estimation of the market model parameters. Also note the typographical error in Chalk's equation (2) defining the abnormal return.

4. The value-weighted index was also used to estimate equation (2) with no significant differences in the results reported in this study.

5. The 150 day estimation period is consistent with that employed by Chalk. As in Chalk, the market model was estimated over a period of 150 days before and after the crash, a period 100 days immediately before the crash, and a 100-day period ending 50 days before the crash. The results were similar, although the magnitude of the cumulative residuals varied widely with the estimation period chosen. The cumulative residuals were not significant in any of the periods, however.

6. The DC-10 portfolio included United, Continental, National Northwest, KLM, World, and Western Airlines. The non-DC-10 portfolio included Alaskan, Braniff, Delta, Eastern, Frontier, Hawaiian, Ozark, PSA, Pan Am, Piedmont, Republic, Southwest, Texas, TWA, and US Air Airlines. Alternative portfolios were constructed by eliminating the small regional carriers or those involved in an impending merger, with no significant difference in the reported results.

7. An expanded version of Table II, identical to that found in Chalk except for the companies, is available from the author on request.

8. Chalk indicates that the results reorted in that table are generated for the model estimated 150 days before the crash and 150 days following the last prediction period. I obtained similar results for McDonnell Douglas for the model estimated 150 days before and after the crash. The cumulative residuals for the three dates in Table III are -13.49 percent, -19.42 percent, and -21.27 percent with respective Z-scores of 0.68, 1.07, and 1.08. Thus, I agree with Chalk that the cumulative returns are sensitive to the estimation period. Interestingly, the cumulative residuals for McDonnell Douglas using the estimation period 100 days before the crash are +1.99 percent, +2.09 percent, and +1.03 percent (all statistically insignificant) for the three dates in the table. The cumulative residuals from the CRSP excess returns tape are also insignificant.

9. The cumulative abnormal returns found by Chalk could be considered similar in nature to those found by Jarrel and Peltzman [1985] in their study of automobile recalls. However, a recent paper by Hoffer, Pruitt and Reilly [1988] replicates the Jarrel and Peltzman study with a revised event study methodology. They find little significant evidence that securities markets penalize shareholders for an automobile recall. As noted, the cumulative returns for McDonnell Douglas are not significant in this case either.
COPYRIGHT 1989 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:includes comment
Author:Chalk, Andrew J.; Karels, Gordon V.
Publication:Economic Inquiry
Date:Apr 1, 1989
Previous Article:Investment and the nominal interest rate: the variable velocity case.
Next Article:SOVPLAN: an in-class simulation of central planning.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters