Assessing the Financial Condition of the Major U.S. Passenger Airlines Over the 1993-2003 Period Using the P-Score and Z” Score Discriminant Models

Goodfriend, Jason

Abstract

Rare prior to the deregulation of the airline industry in 1978, airline bankruptcies have now become rather endemic. Since 1982, over 140 airlines have filed receivership. This number includes eight of the carriers that were formerly referred to as “trunk air carriers”; now known as “Majors. ” Major carriers are defined as those air carriers with annual revenues greater than $1.0 billion.

The purpose of this paper is to analyze the recent performance of two statistical models that are designed to predict the likelihood of severe financial distress for a corporation. One of these models is the Altman Z-Score [Altman, 2002]. This model is applicable to any type of firm. The other model is a scoring approach developed by Pilarski and Dinh that is applicable specifically to air carriers [Pilarski and Dinh, 1999].

Since the failure rate in this industry is so high, such a study should be of interest to many groups including stockholders, bondholders and other creditors, lessors, governmental agencies such as DOT and the FAA, and the flying public.

Introduction

Many indices are useful in describing the financial health of a corporation. For example, traditional ratio analysis is a frequently used tool for financial analysis. It would be of enormous benefit to financial and government analysts, however, were there to exist a single index that had both descriptive and predictive power concerning the financial risk incurred by a corporation.

This would be especially helpful for analyzing the financial situation of the airlines, as many of them are currently encountering turbulent financial conditions. The airline industry is well known to be subject to high levels of business and financial risk, even during ‘good times’ [Gritta, Freed, and Chow, 1998]. Rare prior to the deregulation of the industry in 1978, airline bankruptcies have now become rather endemic. Since 1982, over 140 airlines have filed receivership. This number includes eight of the carriers that were formerly referred to as “trunklines,” now known as “majors.”2 Major carriers are defined as those air carriers with annual operating revenues greater than $1.0 billion.

While there is no such one ‘magic’ performance measure that is designed to assess and/or predict the financial health of an airline, there are two indices that have been shown to be strongly correlated with the likelihood that an airline will enter Chapter 11 bankruptcy. One of these models is the Altman Z-Score [Altman, 2002]. This model is applicable to any type of firm. The other model is a scoring approach developed by Pilarski and Dinh that is applicable specifically to air carriers [Pilarski and Dinh, 1999]. In this paper, both of these models are applied to most of the U.S. major airlines, recent time series are formed, and the results are compared. For the most part, these models perform well in displaying the financial situation of these airlines.

Both discriminant analysis and logit models are employed in this paper for a reason. There are some differences between the two models. The independent variable coefficients in a discriminant model are not subject to the standard regression tests or interpretations. They are estimated to merely compute the discriminant score. Thus, the logit model offers a reasonable complement to any discriminant analysis. For instance, each independent variable coefficient (β) measures the change in log-odds of financial distress with respect to a unit change in the corresponding independent variable. In this sense, the logit model is preferable to the MDA, in that the partial contribution of each explanatory variable to financial health or distress may be measured. Furthermore, the standard test of significance such as the student’s t test is readily applicable in the logit model, while that is not possible for the coefficients of the discriminant analysis. However, even in the case of the logit model, the R of the model is not useful and researchers often rely on the value of the logarithm of the likelihood function.

The Altman Z-Score uses multiple discriminant analysis (MDA) to calculate a ‘Z-Score’ that is used to discriminate between those firms that are predicted to enter a state of bankruptcy in the near future, and those that are predicted not to do so in the near future. This model has been used since the early 1970s. The inputs to the model are some well-known financial ratios that have been demonstrated to be statistically significant.

Gritta et. al. have applied the Altman Z-Score to assess air carrier fitness for the years 1966-1996 [Gritta, Davalos, and Chow, 2000]. However, there are now several different versions of the model. Specifically, there is a version designed for firms for which book value of equity is available (rather than market value), and for which the firms are classified as ‘nonmanufacturers’. Passenger airlines fit into this category. In their Form 41 financial reports to the U.S. Department of Transportation (DOT), book value of equity is reported; not market value. Furthermore, a passenger airline does not sell a manufactured item. Rather, passenger airlines sell tickets for air transportation.

For this paper, we shall refer to this measure as the P-Score. Note that variables Y^sub 2^ and Y^sub 3^ are the same as the Altman variables X^sub 2^ and X^sub 4^ respectively. The number P may be interpreted as the probability that the air carrier will file for bankruptcy in the near future.

In this paper, Z-Scores and P-Scores are presented as time series for the major carriers. Rather that viewing the values of the Z-Scores and P-Scores as absolute predictors of bankruptcy, it may be more beneficial to use them to compare airlines and to observe trends.

Methodology

The airlines examined for this study are Alaska Airlines, Inc. (AS), America West Airlines, Inc. (HP), American Airlines, Inc. (AA), Continental Air Lines, Inc. (CO), Delta Air Lines, Inc. (DL), Northwest Airlines, Inc. (NW), Southwest Airlines, Co. (WN), Trans World Airlines, Inc. (TWA), United Air Lines, Inc. (UA), and US Airways, Inc. (US). Since American Eagle Airlines, Inc. (MQ) is owned by AMR Corporation, (the same company that owns AA), the financial accounts of MQ have been conglomerated with the financial accounts of AA for this study. American Trans Air, Inc. (TZ) has historically used a very different business model than the other major passenger airlines, due to the large role charter flights have played for this airline. Consequently, results for TZ are not included in this study. Similarly, major freight carriers (such as DHL Airways) are not included in this study, although they are defined as “majors” by DOT.

Z-Scores and P-Scores are presented in time-series format for the above list of airlines starting with the fourth quarter of 1993 and ending with the fourth quarter of 2003 (10 years). (It should be noted that Form 41 airline data for the fourth quarter of 2003 is preliminary at the time of this writing.) The time series for TWA ends in 2001, since this airline was absorbed into AA in that year.

The set of airlines presented is divided into three groups. Group 1 is the set of airlines with a Z-Score less than -1.5 for at least one quarter since the first quarter of 2002. This is referred to as the “High Risk” Group. Group 2 is the set of airlines with a Z-Score that is greater then or equal to -1.5 for all quarters since the first quarter of 2002, but for which the Z-Score is less than or equal to -1.0 for at least one quarter since the first quarter of 2002. This is referred to as the “Intermediate Risk Group”. Group 3 is the set of airlines with a Z-Score that exceeds -1.0 for all quarters since the first quarter of 2002. This is referred to as the “Low Risk Group”. It is recognized that this division of airlines into groups is somewhat arbitrary. However, it will be observed that the P-Scores reinforce this grouping.

The time series relationship between the Z-Score time series and the P-Score time series are explored. It is demonstrated that whenever the Z-Score indicates severe financial distress (i.e. a negative number with large absolute value), the P-Score tends to indicate this also (i.e. a large probability). For some of the points of these time series for which the models indicate severe financial distress, airline financial statements are examined to determine if there is additional evidence indicating such distress.

Results for the “High Risk” Group

Results for the “High Risk” Group for both the Z-Score and P-Score models are presented in this section.

Z-Score Results

Figure 1 displays the Z-Score results for the “High Risk” Group. The quarters of the occurrences of entrance into Chapter 11 bankruptcy are indicated on the graph by printing the name of the airline, along with an arrow pointing to the data point. It can be seen that steep declines in the Z-Score preceded each of the three bankruptcies. TWA’s Z-Score, already low, declined steeply in the fourth quarter of 1994, and it filed for Chapter 11 two quarters later. (It was also in very poor financial condition at the time of its purchase by AA in the second quarter of 2001, and this is reflected in its Z-Scores prior to that quarter. TWA continued reporting separately to the DOT through the fourth quarter of 2001.)

With regard to the airline’s recent financial difficulties, the Z-Scores of US began their deep descent in the third quarter of 2001; four quarters prior to that airline filing for Chapter 11 bankruptcy. The sharp drop in UA’s Z-Scores began in the first quarter of 2001; seven quarters prior to that airline filing for Chapter 11 bankruptcy. The significant slide in AA’s Z-Scores began in the first quarter of 2002. AA has avoided filing for Chapter 11 bankruptcy, although the carrier threatened to do so last year. The Z-Scores for AA never reached the extreme lows experienced by UA and US.

The behavior of US’ Z-Scores, starting in the first quarter of 2003, reveals a limitation of the model (and perhaps any financial model based on financial ratios). US emerged from Chapter 11 bankruptcy in the first quarter of 2003. The behavior of the Z-Scores would indicate that the company had returned to strong financial health. Indeed, US had a net income of $1.6 billion for 2003, and the Z-Scores were strongly influenced by this profit. However, this surge in net income was really the result of a one-time $1.9 billion “net reorganization item” that was a result of the “Discharge of Liabilities” [US Airways Form 10-Q, March, 2003]. This, of course, was directly related to the emergence of the airline from Chapter 11 bankruptcy. It does not indicate a ‘turnaround’ for the company. The subsequent continuing sub-par performance of this carrier is indicative of this reality.

The other interesting item concerning Figure 1 is the low Z-Score values for US during the early years of the figure; especially between the fourth quarter of 1994 and the second quarter of 1995. US did not file for Chapter 11 bankruptcy protection during that period. However, US did incur a high level of financial distress during this period. This is indicated in its 10-K for the fiscal year ended December of 1994 [US Air Group, 10-K, 1994]. It was admitted that “…The growth of the operations of low cost, low fare carriers in US Air’s markets in the eastern U.S. continues to represent an intense competitive challenge for USAir, which has higher operating costs than its competitors”. It is also stated in this 10-K that “…Therefore, the Company believes it must reduce its cost structure substantially in order to survive”.

P-Score Results

Figure 2 displays the P-Score results for the “High Risk” Group. As in Figure 1, the quarters of the occurrences of entrance into Chapter 11 bankruptcy are indicated on the graph by printing the name of the airline, along with an arrow pointing to the data point.

Recall that a high P-Score is designed to indicate the prediction of severe financial distress. One advantage of the P-Score is that it is ‘normalized’ to always be between O and 1. Note that the sharp decline in Z-Scores for TWA in the first quarter of 1995 is matched by a steep rise in the P-Score in the same quarter. Also, the ‘peak’ P-Score for TWA occurring in the first quarter of 2001 matches the ‘valley’ Z-Score for the carrier.

Like the Z-Score model, the P-Score model displays the dramatic financial distress faced recently by UA, US, and AA. Also, both models display an improved situation for the scores for AA in the last three quarters. Like the Z-Score, the P-Score shows dramatic improvement in the scores for US. As discussed above this displays a model deficiency, since the model can be very much influenced by the “earnings before interest and tax (EBIT)”, which is much inflated due to the terms of the emergence from Chapter 11 bankruptcy. Finally, the P-Score for US during the first quarter of 1995 indicates financial distress, as did the Z-Score for the same period.

Results for the “Intermediate Risk” Group

Results for the “Intermediate Risk” Group for both the Z-Score and P-Score models are presented in this section.

Z-Score Results

Figure 3 displays the Z-Score results for the “Intermediate Risk” Group. The Z-Scores for all three airlines began declining in 2001. For the last three quarters of the graph, Z-Scores for HP and CO have been increasing. This is consistent with the documented improvement in the situation for HP. For example, in the third quarter of 2003, HP “posted a profit of $33 million, or $.60 per share, double the amount Wall Street analysts were expecting. In the same period last year it lost $50 million” [Arizona Republic, 2003].

The model displays low Z-Scores for HP in 1993-1994. This is consistent with the financial difficulty that the airline was experiencing during this period. In fact, HP emerged from bankruptcy in 1994.

The model also displays very low Z-Scores for CO during 1994-1995. Again, this is consistent with known poor financial performance. For example, the following is stated in CO’s 10-K of 1995 [[Continental Airline, Inc., 10-K, 1995]:

“During the first and second quarters of 1995, in connection with negotiations with various lenders and lessors, Continental ceased or reduced contractually required payments under various agreements, which produced a significant number of events of default under debt, capital lease, and operating lease agreements.”

P-Score Results

Figure 4 displays the P-Score results for the “Intermediate Risk” Group. The scores for CO are consistent with the Z-Scores; worsening situation in 2001-2002, followed by improvements during the last three quarters of the graph. The P-Score results for DL are also consistent with the Z-Scores during this period. However, the Z-Scores for HP are not in concert with the corresponding P-Scores. Whereas the Z-Scores are poor for many recent quarters (with the exception of the last three quarters), the P-Scores for HP make a big turn for the worse only in the third quarter of 2002. Then, these scores quickly return to ‘healthy values’. Finally, the P-Scores are consistent with the difficulties mentioned for HP in 1993-1994 and CO in 1994-1995.

Results for the “Low Risk” Group

Results for the “Low Risk” Group for both the Z-Score and P-Score models are presented in this section.

Z-Score Results

Figure 5 displays the Z-Score results for the “Intermediate Risk” Group. The relatively high Z-Scores for these airlines are consistent with the well-known fact that these airlines are not in financial danger. They also correctly indicate the dominance of WN in terms of financial health.

P-Score Results

The P-Score values for these airlines are so low, that it is not practical to graph them on the scale as all of the P-Scores of the other airlines. In fact, during the entire time period studies, the highest such P-Score was for NW for the second quarter of 2003; the P-Score was 0.1. NW’s P-Score stayed at 0.1 in the next quarter, and then it dropped considerably. AS’ P-Scores were always below 0.06. As with the Z-Scores, WN usually had the best P-Scores. For example, WN’s Z-Score for the fourth quarter of 2003 was 0.0009.

Conclusion

Both the Z-Score and the P-Score models are good of indicators impending financial distress of in the airline industry. The rate of failure in air transport is abysmal, and both the Z Score and P Score models clearly demonstrate the fact that the risk of failure is still very high. Both models are thus tools useful to a wide audience involved with the air transport industry, including stockholders, bondholders, banks, lessors and other creditors, and governmental agencies that need to be able to gauge financial stress and the likelihood of future problems. But these models alone cannot be totally relied upon as infallible predictors of bankruptcy. The reason is that the industry is in such a weaken condition that it is hard to predict the next to fail. The analysis in this paper has also shown that both models are vulnerable to accounting practices that may mask the true financial state of an airline in the short-run. Finally, the models can be of aid to one other group not mentioned above. That group is airline management. The models show the variables that are the keys to successful financial performance. Management can thus center on actions that will improve the variables key to reversing the low and negative trends in the ratios, at least in part due to managerial mistakes in the areas of financial leverage, liquidity and profitability.

2 These 8 carriers are: America West, Braniff, Continental, Eastern, PanAm, TWA, United, and US Airways.

By: Jason Goodfriend, Ph.D

Richard D. Gritta, Ph.D.

Bahram Adrangi, Ph.D.

Sergio Davalos, Ph.D.

1 The views and opinions expressed in this paper are those of the authors and do not reflect the official position of the US Department of Transportation or any part therein. A prior version of this paper was presented at the 9th Annual Meeting of the Air Transport Research Society. World Conference on Transportation Research, in Istanbul, Turkey, on July 3, 2004.

Jason Goodfriend (Ph.D.-University of Virginia in Systems Engineering) is Transportation Specialist in the Office of Advanced Studies of the Bureau of Transportation Statistics, a branch of the U.S. Department of Transportation in Washington, D.C.

Richard Gritta (Ph.D.-University of Maryland) is a Professor of Finance and Transportation at the R.B. Pamplin Jr. School of Business, University of Portland. His research interests include airline financial patterns, air carrier bankruptcy forecasting, and risk and return in air transportation.

Bahram Adrangi (Ph.D.- University of Oregon) is a Professor of Economics at the R.B. Pamplin Jr. School of Business, University of Portland. His areas of research interest are transportation economics, financial economics and international economics

Sergio Davalos (Ph.D.-University of Arizona) is an Assistant Professor in the Milgard School of Business at the University of Washington-Tacoma. His research interests include developing knowledge and information comprehension support through computer-based systems, application of machine learning techniques for financial problems, application of agent technology for organizational computation modeling, and use of biological mechanisms for addressing organizational computing needs

Copyright Credit Research Foundation Fourth Quarter 2004

Provided by ProQuest Information and Learning Company. All rights Reserved

You May Also Like

Real benefits to business of electronic bill and invoice presentment and settlement, The

real benefits to business of electronic bill and invoice presentment and settlement, The Fensterstock, Albert Abstract Present…

Dynamics of risk, liquidity and information in commercial credit granting applications, The

dynamics of risk, liquidity and information in commercial credit granting applications, The Hastings, Richard D Abstract Rece…

Assessing the Financial Condition of the Major U.S. Passenger Airlines Over the 1993-2003 Period Using the P-Score and Z” Score Discriminant Models

Assessing the Financial Condition of the Major U.S. Passenger Airlines Over the 1993-2003 Period Using the P-Score and Z” Score Discriminant Mod…

The Effect on International Credit Management

German Consumer Perceptions of the Euro Conversion: The Effect on International Credit Management Feinberg, Martin Abstract At…