Sunteți pe pagina 1din 21
Transportation Research Part E 40 (2004) 443–463 www.elsevier.com/locate/tre Management emphasis and performance in the
Transportation Research Part E 40 (2004) 443–463 www.elsevier.com/locate/tre Management emphasis and performance in the

Transportation Research Part E 40 (2004) 443–463

Transportation Research Part E 40 (2004) 443–463 www.elsevier.com/locate/tre Management emphasis and performance in the

www.elsevier.com/locate/tre

Management emphasis and performance in the airline industry: an exploratory multilevel analysis

Sveinn Vidar Gudmundsson *

CERMAS Research Centre and Department Strategy, Toulouse Business School, 20 Boulevard Lascrosses, BP7010, 31068 Toulouse Cedex 7, France

Abstract

Carrying through decisions in organisations is inherently micro-level, while the aggregated performance implication of decisions is a macro-level phenomena. This paper determines through exploratory factor analysis, factors associated with airline performance using a two-level bottom–up hierarchical approach. The determinant factors are used to test, through regression analysis, the relationship across lower and higher level factors and between the higher level factors and performance (distress versus non-distress). The results showed that airlines having higher relative score on productivity and brand image were less likely to be under financial distress, while airlines having higher relative emphasis on market power were more likely to be under financial distress. 2004 Published by Elsevier Ltd.

Keywords: Multilevel analysis; Business performance; Business distress; Airlines; Business success; Airline strategy; Airline performance

* Tel.: +33 5 61 29 48 43; fax: +33 5 61 29 49 94. E-mail address: s.gudmundsson@esc-toulouse.fr

1366-5545/$ - see front matter 2004 Published by Elsevier Ltd.

doi:10.1016/j.tre.2004.08.004

444 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

1. Background

1.1. Introduction

Performance indicators are a source of debate as to which one provide the most effective feed- back on business performance. What is more, the link between variation in the indicators and company performance has been difficult to establish. Quantitative failure prediction models (see Altman, 1993; Gudmundsson, 1998, 1999, 2002) based on financial ratios are a case in point. These serve their purpose in segregating companies into the failed and non-failed groups, but do not produce underlying explanatory relationships behind performance variation. Eccles (1991) and Argenti (1976) argued that alternative methods of performance measurement was needed, stressing the point that traditional financial ratios are too simplistic and backward-looking to serve their purpose. Many theories have attempted to explain variation in business performance. Leadership theories are associated with the belief that variance in the firm s performance can be associated with the leadership qualities of top managers. Preisendo¨ rfer and Voss (1990) argued, using human capital theory, that in small organisations a considerable proportion of variance in organisational activ- ities and outcomes could be associated with individuals. Another study by Westerberg et al. (1997), using multidimensional performance constructs of market performance and financial per- formance, showed that CEO characteristics had considerable impact on performance. However, a stream of early research (Salancik and Pfeffer, 1977) has claimed that organisations are driven by other more important factors than top managers. Another school of thought claims that market orientation holds the key to the firm s success. Narver and Slater (1990), for example, state that market orientation is associated with the crea- tion of superior value for buyers and therefore superior performance of the firm. Appiah-Adu and Ranchhod (1998) found positive and significant association between market orientation and three out of four performance measures examined. Market dominance is, however, segregated in terms by assuming superior market-share, while superior market-share can be achieved with or without market orientation. For this reason a non-market orientated firm under distress can have superior market-share, achieved through deep price cutting or mergers and acquisitions (Buzzel and Gale, 1987). A similar line of thought is the economies of scale factor (Smith, 1955) that is considered by some as being the essential element in the firm s success: a firm should focus on achieving critical mass and benefit from economies of scale as a consequence. Levine (1987) showed that in the air- line industry there are further economies that have been demonstrated to play an even greater role than economies of scale: economies of scope, which denotes the reach of a route network; and economies of density that depicts the combination of small distributed loads from a number of points of origin into a larger more economical load through a centre to a final destination. Yet the advent of so called low-cost, direct service carriers, has demonstrated that superior perform- ance can be achieved with relatively low density economies (point to point services versus hub and spoke) and high performance can be maintained with relatively limited scope economies and slow growth. 1

1 Southwest Airlines is an example of this.

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

445

The organisational structure factor (Leavitt and Bass, 1964; Lawrence and Lorsch, 1967) holds that the way the firm is organised is essential for success. Factors such as decentralisation and em- ployee autonomy play role in the organisational structure factor. The firm should be organised in such a way that all components work in harmony to reach the firm s mission. However, as Eilon (1992) points out there have been a debate as to the impact of centralisation versus decentralisa- tion often leading to contradictory results when associated to performance. One side argues that centralisation leads to better utilisation of resources and lower costs, while the other side argues that decentralisation leads to faster reaction times and greater innovation. The productivity factor is another important line of thought that attempts to explain variation in performance from input cost of resources as ratio to output value of goods and services. Most researchers agree that one-factor productivity per se is too simplistic on its own to explain vari- ation in performance and suggest multifactor measures to capture the impact on profitability (Heap, 1992; Schefczyk, 1993; Oum and Yu, 1998). Schefczyk (p. 304) argues that even multifac- tor productivity measures focusing on return to scale do not reflect overall performance, especially in terms of customer, service and competition issues. He used data envelopment analysis (DEA) to address profitability in the airline industry, which is described as a non-parametric approach to measure productivity, while Oum and Yu (1998) used total factor productivity (TFP) for the same industry. TFP is defined as the amount of aggregated output produced by a unit of aggregate in- put. These approaches do provide results as to relationships between various resource inputs and performance measures and thus heightening the interest in the underlying qualitative management and strategic causal relationships that are usually outside the scope of such studies. Finally, we have contingency theories (Wood, 1979; Miller, 1981) stating that there is an inher- ent complexity and factors explaining variation in performance may be dependent on variables such as strategy selected or industry specific factors. Support for this view can be sought in recent research such as Wouters et al. (1999) that used benchmarking approach to determine operational performance measures and their relationship with financial performance in the transportation sec- tor; and Otley (1999) researching management performance from the standpoint of managerial control, suggesting an inductive framework that goes beyond measurement and towards a holistic approach to performance management. Having briefly discussed different schools of thought, the purpose of the research that follows was to come up with a bottom-up framework, separating micro- and macro-factors, when analys- ing performance and apply it to firms in a specific industry segment.

2. Approach

2.1. Multilevel method

A multilevel method 2 allows a bottom–up approach: going from the specific to the general. The segregation of causes from symptoms of company failure was suggested by Argenti (1976). The proposition has been much cited in the literature dealing with business failure, but has lacked

2 The use of the term ‘‘multilevel’’ in this research does not imply a stringent application of the concept as described by Hox (1995) for example.

446 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

empirical investigation. We integrated this idea into a multilevel framework by dividing factors into two principal levels: performance factors (PF) and performance indicators (PI). Performance factors are defined as: ‘‘factors that can be directly altered by management decisions resulting in,

either individually or collectively, performance change of the whole organisation or any of its sub- components. While performance indicators are defined as: ‘‘factors that change directly or indi- rectly in consequence of change in one or more performance factors’’. The PFs and the PIs can have two reminiscent relationships with performance: positive and negative. PFs having higher emphasis by managers of non-distressed companies, are classified as PF (+); and factors, having higher emphasis by managers of distressed companies, are classified as PF ( ). For PIs the same applied, e.g. significant positive difference of the means, becomes PI (+), and negative difference, PI ( ). Furthermore, PIs, unlike PFs that are all low level factors, could be considered multilevel factors: determined by the directness of the influence enacted by

a change in a PF. The higher the level of a PI (i.e. further from direct influence of a particular

PF) the more macro and important it is in terms of association with the overall performance state of the firm. Therefore, it follows that the interrelationships become more complex as the level of the PI is higher (macro level), making it more difficult for top management to use it to track the influence of alteration in individual PF (micro-level). For example, a change in aircraft passenger load-factor on route i due to a change in frequency offered on the route would be regarded as low

level PI. The collective implications of changes in this particular PI, is then reflected in higher level PIs, such as financial ratios. It is important to emphasize, at this stage, that if a PF ( ) is emphasised by firms, that partic- ular group of firms is not necessarily more prone to failure than a group of firms not emphasizing

a factor. Rather it means that the firms observed having financial difficulties are more likely to

emphasise this factor than firms not under distress, for the better or the worse. Thus, the influence of each PF must be established in terms of this dichotomy. Looking at the PIs a similar pedagogic applies, although individual PIs can have the peculiar trait of showing positive change due to a change in a PF that has a proportionally larger adverse impact on another PI. Take for example,

a reduction in flight frequency (#departures) on route j causing positive change in the passenger

load-factor (seat capacity/seat demand) (low level PI) because of reduction in seat capacity on

a route in a stable demand situation. Overall aircraft utilisation, a higher level PI, is, however,

negatively affected, if we assume that the excess aircraft hours caused by reduction in flight fre- quency cannot be applied elsewhere in the airline route network. Hence, change in frequency could be at a greater loss than benefit if looking at aircraft utilisation opposed to the passenger load-factor in isolation. Given the coverage in this section we can postulate that by examining the multilevel relationship between the PFs, and PIs, the explanatory power for variation in performance would be superior to

a single level empirical model or a traditional failure prediction model based on financial ratios.

2.2. Sample and data gathering procedures

A questionnaire was used to gather data on 62 component items, composed of PF and PI items related to airline management. The population was new airlines 3 established after marked

3 These can be termed new-entrant airlines and include value based airlines (VBA).

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

Table 1

Response statistics

447

 

Q-I

Q-II

Total

Response rate by airlines

65% (40/26)

42% (60/25)

Performance status

Non-distress

20

25

45

Distress

25

15

40

Geographical dispersion US EU Other

22

20

42

23

16

39

0

4

4

Total

45

40

85

conditions in a domestic market had been liberalised. 4 The questionnaire was administered twice: once in 1993 to 40 airlines and 242 individuals and again in 1998 to 60 airlines and 282 indi- viduals. The surveys mailing strategy was designed in such a way that response rates by airlines would be maximised. This led to an average of six mailings to each airline in 1993 and five in 1998. Response rate by airlines (see Table 1) was 65% in 1993 and 42% in 1998. The lower response rate in 1998 was probably due to a larger number of airlines located in areas of the world not repre- sented previously. Mean testing was used to assess whether there was difference between the peri- ods and significant differences were found for 10 items and five items that were relevant to the models presented here. These items are identity marked in the tables (see Tables 3 and 4). Geographical dispersion of responses in the second survey was ranging from 50% US, 40% EU and 10% other countries, while in the first survey it was 49% US and 51% EU. In the more recent survey about 37% of the airlines had had operating losses for two or more years preceding the survey and were classified as distressed as a result. In the first survey, this was different with a ma- jor half or 55% of airlines being in distress.

2.3. Measures

The questionnaire asked respondents to indicate on a scale from 0 to 10, the importance placed on various management factors in their airline. The items in the questionnaire were all developed on the basis of literature research and comments on a pilot questionnaire. All items were segre- gated into two groups posterior to the administration of the questionnaire according to the ability of management to directly influence their outcomes. The low level items, usually determinants of airlines strategic characteristics, were labelled as PF items. Higher level items were considered those that were more a consequence of manipulating PF items.

4 Population was identified through the March 1992, and 1997 issues of Flight International that publishes brief historical and contact information for the world airlines.

448 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

2.4. Analysis

In order to reduce items to a manageable level, exploratory factor analysis was used to con- struct descriptive factors for PF items and PI items. Multiple linear regression (MLR) was used to test hypothesis pertaining to the relationship of the resulting factor PFs and PIs. On the assumption that higher level PIs are better predictors of overall performance than lower level PFs, logistic regression (LR) was used to segregate the relationship of the PI factors with the sam- ple firms dichotomous performance state: distressed or non-distressed.

3. Results

3.1. Descriptive statistics and reliability estimates

When performing the factor analysis, coefficient alpha was used to assess scale reliability for each sub-scale constructed, based on the latent factors. Reliability for sub-scales varied for alpha from 0.52 to 0.86. Nunnally (1978) suggested that alpha of 0.50–0.60 being sufficient at the outset of research and all 17 factors extracted met the minimum criteria. One factor had only one item loading and was therefore, excluded from further analysis. A number of factors had three items loading, but one item with a low loading (<.40) that was deleted from the analysis based on Churchill (1979). This deletion raised the number of factors having two items only, from four to six. Spector (1992) argues that factors of less than three items are not likely to be useful. This argument was deemed to be valid in terms of the two-item scales needing further development in future research, but not substantiating the deletion of the factors from further consideration. Correlation matrixes were constructed to see if any of the items used in the scales had non-sig- nificant correlation with all of the other items. Nunnally (1978) recommended that items with con- sistently low correlation (<.30) to be deleted. This led to a number of deletions from the item list. The Keyser–Mayer–Olkin (KMO) was used to test for sample adequacy. For the PI items the result was on the better side of middling (0.76). While for the PF items it was mediocre (0.65). As the KMO is well above the unacceptable level (0.50) it was deemed that the factor analysis should go ahead with the existing sample size. The statistic for Bartlett test of sphericity posed also nei- ther problems in the use of the factor model for PI items (p 6 .001) nor PF items (p 6 .001). Table 2 shows the means, standard deviations, correlation, and reliabilities for the study s var- iables. The table also shows acceptable interrater reliabilities for the aggregated variables.

3.2. Factor extraction

The maximum-likelihood method was used for factor extraction in both factor models and var- imax was used for orthogonal rotation, which enhances the interpretability of the factors. The PF level exploratory factor analysis (see Table 3) was performed on observed items that can be directly manipulated by managers. The factor analysis resulted in 10 factors. The first factor was labelled service features (a = .86) explaining 22.4% of the variance and consisting of five items. The name of the factor was grounded on the fact that most of the items were related with actions important to serve business passengers. An airline that is high on this

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

449

Table 2 Factor means, standard deviations and Pearson correlations a,b

 

Mean

SD

Correlations b

 
 

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

1. Productivity 2. Brand image 3. Decentralisation 4. External influencing 5. Market power 6. Channel power 7. Service features 8. Operations efficiency 9. Cost competitiveness 10. Performance incentives 11. Gearing management 12. Integration orientation 13. Labour flexibility 14. Marketing orientation 15. Cost reduction 16. External growth

7.54

1.46

(.78)

 

7.70

1.61

.06

(.67)

 

6.00

1.67

.13

.08

(.69)

 

6.39

2.05

.06

.03

.04

(.63)

 

6.20

2.00

.03

.05

.02

.07

(.52)

 

7.51

1.80

.00

.14

.03

.01

.05

(.56)

 

5.87

2.41

.21

 

.35

.10

.03

.32

.37

(.85)

 

6.66

2.08

.33

.07

.44

.06

.13

.02

.01

(.67)

 

7.94

1.43

.14

.13

.41

.38

.26

.00

.01

.01

(.69)

 

5.82

2.49

.12

.19

.05

.14

.12

.08

.03

.01

.00

(.82)

 

6.69

1.86

.44

.07

.14

.09

.07

.02

.04

.06

.02

.01

(.67)

6.62

1.63

.07

.30

.13

.17

.31

.11

.00

.01

.01

.01

.03

(.71)

5.34

1.89

.32

.25

.00

.11

.01

.40

.02

.02

 

.03

.01

.04

.01

(.62)

6.48

1.74

.18

.08

.09

.03

.08

 

.24

.03

.01

.04

.01

.01

.02

.02

(.54)

 

7.41

2.03

.04

.10

.02

.12

.21

.03

.05

.05

.02

.01

.04

.01

.02

.03

(.62)

3.71

1.74

.03

.03

.10

 

.10

.25

.06

 

.05

.00

.01

.02

.03

.04

.05

.03

.03

(.55)

a Results for item correlation are available from the author.

b Reliabilities are in parentheses. For all correlation above .21, p 6 .05, one-tailed tests; for all correlation above .31, p 6 .01, one-tailed tests; for all correlations above .35, p 6 .001, one-tailed tests.

450 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

Table 3 Factor results for performance factors (PF)

Determinant

Component

Component

Eigenvalue

Loading

Expl.

factors

items a

Mean

SD

var.

Service features Alpha = .86

Frequent flyer programs (+) Business passengers ( ) Yield management system (+) Hub and spoke operations ( ) Feeder airline agreements (+)

5.26

3.41

7.61

.77

22.39

6.89

3.18

.75

7.01

2.78

.64

5.06

3.48

.46

4.42

3.17

.46

Operations efficiency Alpha = .67

Matching of aircraft size with route requirement Interlining agreements ( ) Acquisition of airport slots (+)

7.38

2.46

3.76

.88

11.07

5.76

2.94

.59

6.36

3.38

.42

Cost competitiveness Alpha = .69

Increase margins (+) b Cost control b Competitor analysis (+)

8.24

1.79

2.36

.86

6.95

8.85

1.42

.62

6.74

2.16

.47

Performance incentives Alpha = .82

Employees incentive program (+) Managers incentive program (+)

5.52

2.82

1.95

.76

5.73

5.86

2.89

.76

Gearing management Alpha = .66

Debt reduction ( ) b Acquisition of new aircraft (+) Forec. adv. eff. of the econ. on the airl. (+) Fuel costs (+)

6.11

3.08

1.92

.88

5.64

7.13

2.77

.50

6.42

2.15

.48

7.06

2.17

.43

Integration orientation Alpha = .71

Logistics systems (+) Control systems (+) Inter departmental communication

6.42

2.21

1.73

.72

5.07

6.88

1.97

.53

6.55

1.99

.41

Labour flexibility Alpha = .62

Flexible job descriptions b Job rotation ( )

6.39

2.47

1.39

.88

4.09

4.05

2.32

.55

Marketing orientation Alpha = .54

Media advertising (+) Market research ( ) Price leadership in served markets ( )

6.20

2.30

1.24

.90

3.64

6.37

2.16

.47

6.87

2.74

.43

Cost reduction Alpha = .62

Cost reduction (+) Reduction of labour costs (+)

8.14

1.94

1.14

.76

3.36

6.69

2.76

.65

External growth Alpha = .55

Merger/acquis. to gain market share ( ) Alliance with the incumbents ( ) b Diversification into other industries ( )

3.11

2.97

1.03

.63

3.02

5.34

3.05

.52

1.50

2.21

.41

a Sign in parentheses shows difference of means, if significance is <.10, then no sign is provided. (+) means a significantly higher mean for non-distressed airlines. b Significant difference between 93 and 98 surveys.

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

451

dimension will tend to believe in the importance of high yield passengers. The factor loading ran- ged from 0.46 to 0.77. The second factor was labelled operations efficiency (a = .67) and explains 11.1% of the total variance. This dimension consists of three items with factor loading ranging from 0.42 to 0.88. One item had a loading <.40 and was deleted. Items loading on this factor seemed to measure issues associated with setting up and maintaining efficient routes. An airline that scores high on this factor will tend to believe in the importance of setting up and maintaining efficient route structures with well connected flights. The third factor labelled cost competitiveness (a = .69) ex- plained 7.0% of the total variation. This dimension consists of three items with factor loading ranging from 0.47 to 0.86. One item had a loading <.40 and was deleted from consideration. An airline scoring high on this factor tends to believe in the importance of relative cost competi- tiveness, by researching competitors actions and keeping costs in check. The fourth factor was labelled performance incentives (a = .82) and explained 5.7% of the total variation. This dimension consisted of two items with approximately the same factor loading of 0.76. One item had a loading <.40 and was deleted from further consideration. An airline scoring high on this factor will emphasise incentive programs as a tool to motivate employees. The fifth factor labelled gearing management (a = .66) explained 5.6% of the variation. This dimension consisted of four items with loading ranging from 0.43 to 0.88. An airline scoring high on this factor tends to have high awareness of the impact of economic cycles and debt in the fleet acqui- sition process. The sixth factor labelled integration orientation (a = .71) explained 5.1% of the total variation. It consists of three items with loading ranging from 0.41 to 0.72. An airline that scores high on this factor tends to believe in the importance of organisational integration as well as control through information systems. The seventh factor labelled labour flexibility (a = .62) explained 4.1% of the total variation. It consists of two items with loading of 0.88 and 0.55. An airline that scores high on this factor tends to believe in the importance of flexible workforce and job rotation to motivate employees and achieve responsiveness to customers needs, e.g. service quality. The eight factor labelled market- ing (a = .54) explained 3.6% of the total variation. The factor consists of three items with loading ranging from 0.43 to 0.90. One item had loading <.40 and was therefore deleted from considera- tion. An airline scoring high on this factor places high importance on stimulating demand though advertising and price leadership that is well grounded through market research. The ninth factor labelled cost reduction (a = .62) explained 3.4% of the total variation. Two items loaded on this factor: 0.76 and 0.65. Airlines scoring high on this factor are in the process of reducing costs op- posed to keeping costs in check, as a result, high ranking implies restructuring. The tenth factor labelled as external growth (a = .55) explained 3.0% of the total variation. This factor consists of three items with loading ranging from 0.41 to 0.63. Airlines scoring high on this factor tend to seek growth through external means such as mergers and alliances. The PI level exploratory factor analysis (see Table 4) was performed on observed items that can normally not be directly manipulated by airline managers. The first factor labelled productivity (a = .78) explained 28.6% of the total variation. This fac- tor consists of six items with loading ranging from 0.40 to 0.73. Airlines scoring high on this factor tend to believe that high productivity, quality and long-term profitability is achieved through shared employee beliefs, reflected in vision and culture. The second factor labelled brand image

452 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

Table 4 Factor results for performance indicators (PI)

Determinant factors

Component items a

Component

Eigenvalue

Loading

Expl.

 

Mean

SD

var.

Productivity Alpha = .78

Employees productivity (+) Shared company vision (+) Company culture (+) Aircraft utilisation Service quality (+) Long-term rather than short-term profits (+) b

8.24

1.72

5.71

.73

28.56

6.66

2.59

.69

7.52

2.02

.67

8.54

1.41

.52

8.13

1.86

.52

6.18

2.74

.40

Brand image Alpha = .67

Brand image (+) Favourable attitude of travel agents (+)

7.70

2.19

2.21

.73

11.05

7.31

2.39

.60

Decentralisation Alpha = .69

Employees autonomy to take ( ) decisions Decentralised organisation ( ) structure

6.52

1.69

1.50

.69

7.52

4.81

2.69

.60

External influencing Alpha = .64

Management s external contacts (+) Influencing government policy on aviation (+)

6.33

2.25

1.38

.80

6.90

6.45

2.56

.53

Market power Alpha = .52

Market share ( ) Achieving critical mass ( ) Investors attitudes towards the airline ( )

5.91

2.68

1.22

.94

6.08

5.67

2.98

.43

7.02

2.72

.42

Channel power Alpha = .56

Computer reservation systems ( ) Passenger load factors ( )

7.26

2.61

1.00

.87

5.01

7.58

2.07

.41

a Sign in parentheses shows difference of means, if significance is <.10, then no sign is provided. (+) means a significantly higher mean for non-distressed airlines and vice versa. b Significant difference between 93 and 98 surveys.

(a = .67) explained 11.0% of the total variation. This factor consists of two items with loading of 0.73 and 0.60. A third item had a loading <.40 and was deleted from consideration. Airlines scor- ing high on this factor tend to emphasise brand image, believing that it has importance in building customer loyalty and product preference for example in the distribution network: favourable at- titude of travel agents. The third factor labelled Decentralisation (a = .69) explained 7.5% of the total variation. It consists of two items with loading ranging from 0.60 to 0.69. Airlines scoring high on this factor tend to believe in employees autonomy to achieve faster reaction times and greater innovation. The fourth factor labelled external influencing (a = .64) explained 6.9% of the total variation. The factor consists of two items with loading of 0.80 and 0.53. Airlines scoring high on this factor tend to believe in the importance of facilitating the airline s well-being through external influence: employee networking and influencing air transport policy. The fifth factor la- belled market power (a = .52) explained 6.1% of the total variation. It consists of three items with loading ranging from 0.42 to 0.94. Airlines scoring high on this factor tend to believe in the importance of size in order to attract investment in the airline. The seventh factor channel power (a = .56) explained 5.0% of the total variation. The factor consists of two items with loading of 0.87 and 0.41. Airlines scoring high on

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

453

this factor tend to believe in the importance of computer reservation systems to boost passenger load factors. This indicates an acceptance of channel power to facilitate sales of seats, despite high- er associated costs compared to direct sales: telephone and Internet.

3.3. Hypotheses

A number of hypotheses were formulated based on literature review and case studies in order to predict relationships between the extracted factors across the two aggregated factor levels: PF and PI. The factor productivity is assumed to represent how airlines believe that high productivity, qual- ity and long-term profitability is achieved through shared employees beliefs, reflected in the vision and culture. Therefore, it is assumed that productivity is influenced by first level factors that have positive impact on employees productivity, while factors such as cost reduction will either have negative or non-significant relation with productivity. This stems, from the assumption that em- ployee redundancies or other forms of cost reduction will create tensions (Doherty and Horsted, 1996) associated with reduced productivity in the respondents mind. The assumption is made in qualitative terms regardless of whether such negative productivity impact occurs or not. What is clear, however, is that cost reductions and employee redundancies will most likely affect the busi- ness culture and vision in an adverse way in the short-term. Finally, gearing management can have major impact on the firm s aggregated input–output ratio. For example, through poor timing of aircraft orders with delivery lag, frequently causing deliveries during economic downturns (Liehr et al., 1998).

H 1a : Service features, operations efficiency, performance incentives, integration orientation, labour flexibility and gearing management, will be significantly related to productivity. H 1b : Productivity is positively related to non-distress.

The factor brand image is assumed to represent airlines that believe in the importance of the brand to generate customer loyalty and preference. For airlines this factor would facilitate pene- tration in the distribution network through favourable attitude of travel agents (TAs). Good brand image works as selection advantage in a case of two similar choices. This being, not only, an advantage with the TAs, but also the passenger depending on, who exercises the decision power. The creation of brand image in the minds of these two groups is probably somewhat dif- ferent although no studies exist to support the assumption. It is expected that the factor is signif- icantly related to service features because service is an important function of brand image. Another service element is operations efficiency, which provides a match with what the TA is look- ing for on behalf of the customer, such as: good connections (function of interlining agreements and alliances), convenient departure times (function of airport landing slots) and frequency (func- tion of matching aircraft size with route requirement). The last item listed can imply two things. First, more frequency with smaller equipment is more favourable, especially in terms of business passengers. Second, the airline should be competent in adjusting aircraft size to demand to reduce the probability of passenger diversion from specific flights (involves more time spent on behalf of the TA finding a flight). Brand image is also expected to be related to labour flexibility, as positive staff willing to walk that extra mile for the benefit of the passenger has a strong impression on the

454 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

customer. This is what Carlson (1987) termed as turning the organisational pyramid upside down, empowering the front-line staff to take unprecedented decisions for the customers benefit at their specific level. Further, if the concept is expanded one can assume that flexible job descriptions in- crease the responsiveness of the organisation, a potentially important element in facilitating brand image and proactive service provision. Finally, marketing orientation is expected to be related to brand image. The factor marketing is composed of items dealing with advertising and market re- search, both of which shape the ability of the airline to build its brand image. In the past low cost carriers, with PeoplExpress being the prime example, have had brand image conflicts due to changes in strategy, such as no-frills to frills, projected to the market through advertising.

H 2a : Service features, operations efficiency, labour flexibility, and marketing orientation, will be significantly related to brand image. H 2b : Brand image is positively related to non-distress.

Airlines scoring high on the factor decentralisation tend to believe in employee s autonomy (Thomas and Velthouse, 1990) to achieve higher performance through faster reaction times and better innovation. As such the factor is expected to be significantly related with labour flexibility. Labour flexibility increases the necessity to co-operate. Job rotation, as a function of decentrali- sation introduces to the employee different functions and stimulates proactive behaviour and cross-departmental knowledge (Spreitzer, 1995). Labour flexibility as an element of decentralisa- tion is also a crucial cost saving tool, by cross-training staff to perform various functions and make decisions. Hence, decentralisation is expected to be significantly related to cost competitiveness.

H 3a : Labour flexibility and cost competitiveness, will be significantly related to decentralisation. H 3b : Decentralisation is positively related to non-distress.

Airlines scoring high on the factor external influencing tend to believe in the importance of facil- itating the airline s well-being through external influence. Policy making in air transport is high on the agenda in most countries, although, one can assume that the nature of airlines influence on governments and agencies has changed in liberalised and deregulated markets. For low-cost or value based airlines, opposed to large incumbents, we can expect some variation in interest, with two issues high on the agenda: CRS biases (Beuvais, 1992) and lack of landing slots at airports (Morrell, 1998). Therefore, it is expected that operations strategy will be significantly related to external influencing. The external influencing influences costs, and governmental influence, federal or local, is often necessary when entering or protecting markets. Hence, the external influencing is expected to be positively related with cost competitiveness.

H 4a : Operations efficiency and cost competitiveness, will be significantly related to external influencing. H 4b : External influencing is positively related to non-distress.

Airlines scoring high on the factor market power tend to believe in the importance of size in order to attract investment. As a result, it is expected that factors dealing with the network, com-

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

455

petitiveness, routes, growth, integration and reputation in the financial community, are impor- tant. The following factors are expected to have significant relationship with market power: serv- ice features, operations efficiency, cost competitiveness, integration orientation, marketing orientation, cost reduction and external influencing. Service features includes items such as hub and spoke networks that are important to achieve economies of density and scope (Gimeno and Woo, 1999; Bania et al., 1998) that are usually associated with market power. Operations effi- ciency includes items such as acquisition of slots, necessary to boost market-share. Cost compet- itiveness is an important element in achieving market power, allowing the airline to offer competitive prices in order to build market share. The PIMS (Buzzel and Gale, 1987) program identified low prices as a vehicle towards market-share, without cost competitiveness and quality, as being a non-sustainable strategy. Integration orientation is necessary to keep up with growth, especially if it grows fast, which is a characteristic of market-share driven airlines. Marketing ori- entation is expected to be related to market power, even though these two can be considered quite distinctive in terms. Cost reduction, such as labour cost reduction, is especially important to show financial control (Flint, 1999) resulting in favourable image within the financial community, facil- itating capital access for expansion programs. External growth is the final factor expected to be related to market power. This factor was not emphasised highly by most respondents as can be seen from the item averages. However, alliances are most prominent and especially important for an airline wishing to achieve critical mass quickly.

H 5a : Service features, operations efficiency, cost competitiveness, integration orientation, marketing orientation, cost reduction and external growth, will be significantly related to PI market power. H 5b : Market-power is positively related to distress.

Airlines scoring high on the factor channel power tend to believe in the importance of computer reservation systems to boost passenger load factors. For low-cost or value based airlines the importance placed on this factor could be diverse as some do not participate in computer reser- vation systems (CRSs), while those that do, have had strong views on partiality of the CRS owner airlines (Feldman, 1997). It is expected, that this factor, is significantly related to service features, operations efficiency and marketing orientation. The factors service features assumes high conven- ience for passengers through ease of product access. Computer reservation systems are important to communicate information about products and to facilitate seamlessness in meeting complex customer requirements. Operations efficiency is composed of items pertaining to capacity, connec- tions, and convenient departure times and routes through availability of slots. As all of these items are communicated through the CRS as a distribution tool for the product, it is expected that oper- ations efficiency is significantly related to channel power. The final expected factor to be signifi- cantly related with channel power is marketing orientation. Advertising, market research, and price leadership that compose the factor, are all items that are communicated through the CRS, thus, impacting loads.

H 6a : Service features, operations efficiency and marketing orientation, will be significantly related to channel power. H 6b : Channel power is positively related to non-distress.

456 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

The testing of the a hypotheses was conducted through multiple linear regression analysis to explore further the strengths of the predetermined relationships reflected in the hypotheses estab- lished in this section, while the b hypotheses were tested by the means of logistic regression.

3.4. Multiple linear regression analysis

Stepwise approach was used for variable entry in the multiple linear regression (MLR) analysis (see Table 5). Colinearity was examined through tolerance and VIF analysis and no factor posed

Table 5 The relationship between PFs and PIs: MLR results

 

DR 2

b

SE

Beta

t

I. Dependent variable: productivity Gearing management Operations efficiency Labour flexibility Service features Marketing orientation R 2 .492

II. Dependent variable: brand image Service features Integration orientation Labour flexibility Performance incentives R 2 .310

.195

.409

.080

.45

5.15

***

.130

.333

.080

.36

4.18

***

.097

.288

.080

.31

3.62

***

.037

.185

.081

.20

2.29

*

.033

.164

.078

.18

2.11

*

Adj.

R 2 .455

F = 13.19 ***

.124

.319

.090

.35

3.54

***

.090

.262

 

.087

.30

2.99

**

.055

.208

.089

.23

2.34

*

.040

.173

.087

.20

2.00

*

Adj.

R 2 .270

F = 7.74 ***

III.

Dependent variable: decentralisation

 

Operations efficiency Cost competitiveness

.190

.351

.078

.43

4.51

***

.163

.327

.077

.40

4.23

***

 

R 2 .353

Adj.

R 2 .334

F = 19.33 ***

IV.

Dependent variable: external influencing

 

Cost competitiveness

 

.314

.089

.38

3.52

***

 

R 2 .147

Adj.

R 2 .135

F = 12.40 ***

V. Dependent variable: market power

 

Service features Integration orientation Cost competitiveness External growth Cost reduction

.103

.298

.097

.30

3.06

**

.097

.292

.094

.30

3.09

**

.067

.258

.095

.26

2.72

**

.052

.240

.099

.24

2.42

*

.041

.197

.094

.20

2.08

*

 

R 2 .360

Adj.

R 2 .313

F = 7.65 ***

VI.

Dependent variable: channel power

 

Labour flexibility Service features Marketing orientation R 2 .336

.159

.373

.094

.39

3.98

***

.128

.343

.095

.35

3.60

***

.048

.207

.092

.22

2.25 * F = 11.79 ***

Adj.

R 2 .307

* p 6 0.05. ** p 6 0.01. *** p 6 0.001.

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

457

problems in the analysis. The MLR models explanation power of the dependent variables ranged from 7% to 49%. The outcomes of productivity. The results of the stepwise MLR analysis supported hypotheses

H 1a and the factors in the model explained 49% of the variation in the depended variable. The

results showed that productivity was significantly related to gearing management (b = .41,

p 6 .001), operations efficiency (b = .33, p 6 .001), labour flexibility (b = .29, p 6 .01) and service

features (b = .19, p 6 .05). Another unexpected factor that entered the model was marketing ori-

entation (b = .16, p 6 .05). Variable contribution to the model was detected through change in R 2 (gearing management, DR 2 = .195, p 6 .001; operations efficiency, DR 2 = .130, p 6 .001; labour flexibility, DR 2 = .097, p 6 .01; service features, DR 2 = .037, p < .05; and marketing orientation, DR 2 = .033, p 6 .05). Two expected factors performance incentives and integration orientation did not enter the model. The outcomes of brand image. The results of the MLR analysis partially supported hypotheses

H 2a and the factors in the model explained 31% of the variation in the depended variable. The

results showed that brand image was significantly related to service features (b = .32, p 6 .01) and labour flexibility (b = .21, p 6 .05). However, two other factors contributed to the explained variation in brand image, namely integration orientation (b = .26, p 6 .01) and performance incentives (b = .17, p 6 .1). The change in R 2 was examined to detect each factor s contribution

to the explained variation in R 2 (service features, DR 2 = .124, p 6 .01; integration orientation, DR 2 = .090, p 6 .01; labour flexibility, DR 2 = .055, p 6 .05; performance incentives, DR 2 = .040,

p 6 .1). Two expected factors did not enter the model, operations efficiency and marketing orien-

tation, while two unexpected factors entered: integration orientation and performance incentives. The outcomes of decentralisation. The results of the MLR analysis partially supported hypoth- eses H 3a and the factors in the model explained 35% of the variation in the depended variable. The results showed that decentralisation was significantly related to operations efficiency (b = .35,

p 6 .001) and cost competitiveness (b = .33, p 6 .001). The change in R 2 was examined to detect

each factors contribution to the explained variation in R 2 (operations efficiency, DR 2 = .19,

p 6 .001; cost competitiveness, DR 2 = .16, p 6 .01). Unexpected factor operations efficiency entered

the model instead of the expected labour flexibility. The outcomes of external influencing. The results of the stepwise MLR analysis supported hypotheses H 4a and the factors in the model explained 15% of the variation in the depended var- iable. The results showed that external influencing was significantly related to cost competitiveness

(b = .31, p 6 .01). No other factors entered the model. The outcomes of market power. The results of the MLR analysis supported hypotheses H 5a and the factors in the model explained 36% of the variation in the depended variable. The results showed that market power is significantly related to service features (b = .30, p 6 .01), integration orientation (b = .29, p 6 .01), cost competitiveness (b = .26, p 6 .01), external growth (b = .24,

p 6 .05) and cost reduction (b = .20, p 6 .05). Variable contribution in the model was detected

through change in R 2 (service features, DR 2 = .103, p 6 .01; integration orientation, DR 2 = .097,

p 6 .01; cost competitiveness, DR 2 = .067, p 6 .05; external growth, DR 2 = .052, p 6 .05; cost reduc- tion, DR 2 = .041, p 6 .05). Two expected factors did not enter the model, operations efficiency and

marketing orientation. The outcomes of channel power. The results of the stepwise MLR analysis supported hypotheses

H 7a and the factors in the model explained 34% of the variation in the depended variable. The

458 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

Table 6 The relationship of PIs with performance: LR results a

Independent variables

Coeff. (b)

S.E.

Wald

R

Productivity

.618

*

.315

3.840

.140

Brand image

.841

*

.378

4.945

.177

Decentralisation

.517

.348

2.214

.048

External influencing

.416

.367

1.284

.000

Market power

.641 *

.308

4.326

.157

Channel power

.345

.294

1.379

.000

Constant

.098

 

.278

.1236

2 Log Likelihood Nagelkerke R 2 Goodness-of-fit test b

77.36

 
 

.293

.325

* p 6 0.05.

a Dependent variable = financial performance (1 = distress, 0 = non-distress).

b Hosmer and Lemeshow.

results showed that channel power was significantly related to labour flexibility (b = .37, p 6 .001), service features (b = .34, p 6 .01) and marketing orientation (b = .21, p 6 .05). Variable contribu- tion to the model was detected through change in R 2 (labour flexibility, DR 2 = .159, p 6 .001; serv- ice features, DR 2 = .128, p 6 .01; marketing orientation DR 2 = .048, p 6 .05). An unexpected factor entered the model, labour flexibility instead of the expected operations efficiency.

3.5. Results of logistic regression

The non-stepwise logistic regression model (see Table 6) had significant model chi-square (p 6 .001) and good Nagelkerke R 2 (0.29). Hypotheses H 1b was confirmed,productivity has ex- pected negative sign, which indicates positive relationship (p 6 .05) with non-distress. Hypotheses H 2b was confirmed, brand image has expected negative sign, which indicates positive relationship (p 6 .05) with non-distress. Hypotheses H 3b was not confirmed, decentralisation has unexpected positive sign, which indicates positive relationship with distress. The relationship was, however, non-significant. Hypotheses H 4b was not confirmed, external influencing has expected negative sign, which indicates positive relationship with non-distress. However, the relationship was non-significant. Hypotheses H 5a was confirmed, market power has expected positive sign, which indicates positive relationship (p 6 .05) with distress. Hypotheses H 6a was not confirmed, channel power has unexpected positive sign, which indicates positive relationship with distress. The rela- tionship was however non-significant.

4. Discussion

4.1. Summary of findings

The major findings (see Fig. 1) are that a PI factor labelled productivity has the strongest rela- tionship with PFs labelled gearing management, operations efficiency, labour flexibility, service fea-

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463 Service features -.21 +.33 +.35
S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463
Service features
-.21
+.33
+.35
Operations efficiency
Productivity
Cost competitiveness
Brand image
+.19
+.41
Performance incentives
+.44
+.38
External Decentralization influencing
Non-distress
Gearing management
+.44
+.25
-.30
Integration orientation
Decentralization
+.26
Distress
+.31
+.32
+.32
Labor flexibility
+.40
Market power
+.18
Marketing orientation
+.24
+.37
+.21
Channel power
Cost reduction
+.25
External growth

459

Fig. 1. Results of regression analysis: MLR and LR a . a Lines show significant relationships in the models (p 6 0.05, or better), broken lines show non-significant relationships. Direction (+/ ) of relationship is shown with correlation coefficient where appropriate.

tures and marketing orientation. The first factor gearing management that explains comparatively most of the variation in productivity, can be related to the necessity to manage gearing in order to prepare for industry downturn and preserve the overall airline productivity. The second factor operations efficiency deals with the matching of resources with requirements, binding together the various components of a route to provide optimum service given the resources available. As such, it is quite logical that operations efficiency has positive relationship with productivity. Labour flexibility is another clearly related factor, as has been demonstrated by so many low cost airlines, such as Southwest Airlines that is able to achieve lower costs through greater staff flex- ibility. Service features had a negative relation with productivity. A justifiable result having in mind that the factor service features is composed of items related to business passengers that re- quire more service. The last factor market-demand deals with the generation of demand through advertising but also through research and price leadership. As such, it has a spill over effect on how the employees view the airline they work for––a well researched marketing campaign rein- forces the employee s sense of purpose. What is more such campaign is essential to generate de- mand to maintain utilisation of resources, especially at the outset of opening new routes. There were four PFs related to the PI factor brand image: service features, integration orienta- tion, labour flexibility, and performance incentives. Service features deals with what product the air- line is offering, therefore, logically associated with brand image. Positive brand image has

460 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

important impact on demand, whose interaction with the airline is in most cases through travel agents. The pull of brand image is therefore through systems that must be well integrated to rein- force the image created through an effective service features. Labour flexibility is crucial for cre- ating a good service impression. Flexibility allows employees to react quickly and effectively to acute situations, creating a good impression with the passenger, e.g. employees ability to react to special situations will either reinforce or destroy a good brand image during the consumption stage. Finally, performance incentives contributed to explaining variation in brand image. Here again the expected relationship is the motivation of employees to provide good impression in the process of their jobs––to walk that extra mile to make a difference. Two PFs contributed to the explanation of the variation in the PI factor decentralisation: oper- ations efficiency and cost competitiveness. Operations efficiency should have a relationship with decentralisation assuming that it increases labour flexibility. Cost competitiveness should have an even stronger relationship as we expect that decentralisation will have positive impact on the airline s overall effectiveness. The factor external influencing was related to one PF, cost competitiveness. Here it seems that management s external contacts and ability to influence government policy can have cost implica- tions for the airline. It was expected that there would be difference between European and US car- riers on the items composing the factor external influencing, but a t test showed a non-significant difference between the two groups. As such the factor must, therefore, be viewed as having com- munality between both regions and constitute an element in achieving cost competitiveness. Five PFs contributed to explaining the variation in the PI factor market power: service features, integration orientation through systems, cost competitiveness, external growth and cost reduction. Market power represents the emphasis the airline levies on size and investor s attitudes towards the airlines. The investors play a crucial role in raising capital necessary for market-share building. One vehicle towards market power is broad target market through many service features. Integra- tion orientation becomes increasingly important as the organisation grows quickly in a market power culture. Thus, it is not surprising to see a relationship between market power (size emphasis) and integration orientation (organisational integration). Cost competitiveness is another important vehicle towards market power as it is unlikely that the airline will reach size without cost compet- itiveness. This is more relevant to low-cost airlines as such airlines emerge from size disadvantage (subject to the market power of incumbents) and must therefore achieve substantial advantage on this factor. External growth is an external means to an end, where the airline emphasizes quick expansion through alliances and mergers. Diversification into other industries one of the items in the sub-scale was not emphasised much by any of the airlines, but showed nevertheless corre- lation with the underlying factor. Cost reduction was the fifth and final factor that was related with the factor market power. Here it was assumed that cost reduction is a vehicle towards positive im- age in the financial community to maintain capital influx for further expansion and increased stay- ing power during industry recession. Three PFs contributed to the explanation of the variation in the PI factor channel power, labour flexibility, service features and marketing orientation. An obvious theoretical explanation for the explanatory power of labour flexibility to channel power was not found. However, service features are an important element in the channel strategy as such. This stems from the fact that distribu- tion system presence is not enough, there has to be a sellable product in order for travel agents to search the CRS and sell the airline. The third and last item, marketing orientation is related, as it

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

461

deals with targeted advertising of the airline s products, with most of the selling going through the TA s and the CRS. The relationship of the identified PI s with performance was as expected with two exceptions channel power and decentralisation which signs indicated positive relationship with distress, but both had non-significant coefficients. However, variables with significant coefficients showed rela- tionships as expected. Thus, the results show that airlines placing relatively high emphasis on pro- ductivity and brand image are less likely to be distressed, while airlines placing relatively high emphasis on market power are more likely to be distressed. As market power is composed of items that imply market-share building and mass, the findings support Appiah-Adu and Ranchhod (1998) that distinguished between market orientation and market dominance and found the for- mer associated with better relative performance. Furthermore, brand image can also be associated with market orientation, explaining the positive relationship with non-distress. External influenc- ing had expected negative sign, but a non-significant coefficient.

4.2. Limitations, strengths and future directions

The results should be viewed in the light of the data s limitations. The population of new-en- trant airlines necessitates a combination of longitudinal and cross-sectional approach to boost the number of cases due to both high entry and exit rates, but relatively short life-cycles. To val- idate the scales with higher degree of certainty still more cases are needed, although, this study almost reached the recommended minimum of 100 cases (Spector, 1992) and was well above the minimum criteria of the KMO test for sampling adequacy. The longitudinal approach necessitates high degree of concurrence between responses over time. The study showed that similarity in rating strengths is de facto for the two studies as few items showed statistical difference of the means, implying consistency in the constructs. For items, showing significant difference the explanation appears to be that the former study was conducted during an industry recession and the second study during an industry up-turn. Hence, items asso- ciated with the financial aspects of airline management were the only ones showing a significant difference between the two studies. The exploratory approach rather than confirmatory approach was selected due to the novelty of the approach and lack of related research on the industry under observation. A logical next step would be to improve the weakness posed by the scales having two items and use a confirmatory approach to establish further the theoretical relationships. Although the questionnaire instrument administered to the airline managers may still need improvement it is important as an initial step to identify useful constructs and underlying factors that can provide for a standard qualitative scale measuring airline management. What is impor- tant to note is that the scale takes external approach in order to make the association of actual variation in strength applied to the items easier to associate with the airline s performance. This approach was undertaken to reduce the impact of social desirability on the rating strength, e.g. ratings according to what is generally accepted in the industry or to make one look good exter- nally. However, it must also be made clear that, as with all factor-analytical, studies there may be a number of different factors possible based on the data. This fact does, however, not under- mine the value of factor analytical studies, as a factor that is proved to be stable in separate stud- ies and measure what it allegedly is supposed to measure is valuable to researchers.

462 S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

The research has attempted to link together the PFs and the PIs as single level factors. The PIs could in future research be segregated into several levels on priori. In addition, using the PF con- cept on purely quantitative data at the PF level and several PI levels, as demonstrated in the exam- ples, is definitely a viable area for comparative qualitative and quantitative studies.

4.3. Conclusion

The general finding of this study is that airlines under distress are more likely to have pursued market power tactics in an attempt to achieve fast growth and superior market-share, while non- distressed carriers have placed greater importance on productivity and brand image. The last two being, in prior research, associated with sustainable superior performance. As a consequence, this study supports prior research that has showed similar relationships in other industries pertaining to market orientation versus market power orientation. This study, however, enhances our under- standing of the productivity factor in industries with high fixed costs and a perishable product. Productivity emphasis was shown in this study to distinguish between distressed versus non-dis- tressed airlines in the model. It is hoped that this attempt at identifying factors for qualitative research into airline manage- ment using the PF approach will spark interest in further research into more detailed qualitative and quantitative multilevel performance measurement.

References

Altman, E.I., 1993. Corporate Financial Distress and Bankruptcy: A Complete Guide to Predicting and Avoiding Distress and Profiting from Bankruptcy, second ed. Wiley, New York. Appiah-Adu, K., Ranchhod, A., 1998. Market orientation and performance in the biotechnology industry: an exploratory empirical analysis. Technology Analysis & Strategic Management 10 (2), 197–210. Argenti, J., 1976. Corporate Collapse: The Causes and Symptoms. McGraw-Hill, New York. Bania, N., Bauer, P.W., Zlatoper, T.J., 1998. US air passenger service: a taxonomy of route networks, hub locations and competition. Transportation Research E 34 (1), 53–74. Beuvais, G.A., 1992. Prepared statement in: Hearing on the Airline Competition Enhancement Act of 1992. US Government Printing Office, Washington, DC. Buzzel, R.D., Gale, B.T., 1987. The PIMS Principles: Linking Strategy to Performance. Free Press, New York. Carlson, J., 1987. Moments of Truth: New Strategies for Today s Customer-driven Economy. Ballinger, New York. Churchill, G.A., 1979. A paradigm for developing better measures of marketing constructs. Journal of Marketing Research 16, 64–73. Doherty, N., Horsted, J., 1996. Re-engineering people––the forgotten survivors. Business Re-engineering 3 (1), 39–46. Eccles, R., 1991. The performance manifesto. Harvard Business Review, January–February, pp. 131–137. Eilon, S., 1992. Management Practice and Mispractice. Routledge, London. Feldman, J.M., 1997. Sticker shock. Air Transport World, April, pp. 67–71. Flint, P., 1999. Saving for the lean years: in the midst of prosperity, United readies for the next recession. Air Transport World, 46–48. Gimeno, J., Woo, C.Y., 1999. Multimarket contact, economies of scope, and firm performance. Academy of Management Journal 43 (3), 239–259. Gudmundsson, S.V., 1998. Flying too close to the sun: the success and failure of the low cost airlines. Ashgate, Aldershot.

S.V. Gudmundsson / Transportation Research Part E 40 (2004) 443–463

463

Gudmundsson, S.V., 1999. Airline failure and distress prediction: a comparison of quantitative and qualitative models. Transportation Research E: Logistics and Transportation Review 35 (3), 155–182. Gudmundsson, S.V., 2002. Airline distress prediction using non-financial indicators. Journal of Air Transportation 7 (2), 3–24. Heap, J., 1992. Productivity Management. Cassell, London. Hox, J.J., 1995. Applied Multilevel Analysis. TT-Publikaties, Amsterdam. Lawrence, P.R., Lorsch, J.W., 1967. Organization and Environment. Harvard University Press. Leavitt, H.J., Bass, B.M., 1964. Organizational psychology. Annual Review of Psychology 15, 371–399. Levine, M.E., 1987. Airline competition in deregulated markets: theory, firm strategy, and public policy. Yale Journal on Regulation 4, 393–494. Liehr, M., Gro¨ ßler, A., Klein, M., 1998. Understanding business cycles in the airline market. Working paper, Industrieseminar der Universita¨t Mannheim. Miller, D., 1981. Towards a new contingency approach: the search for organizational, Gestalts. Journal of Management Studies, 1–26. Morrell, P., 1998. Air transport liberalization in Europe: the progress so far. Journal of Air Transportation World Wide 3 (1), 42–60. Narver, J., Slater, S., 1990. The effect of market orientation on business profitability. Journal of Marketing 54, 20–35. Nunnally, J.C., 1978. Psychometric Theory. McGraw-Hill, New York. Otley, D., 1999. Performance management: a framework for management control systems research. Management Accounting Research 10, 363–383. Oum, T.H., Yu, C., 1998. Winning Airlines: Productivity and Cost Competitiveness of the World s Major Airlines. Kluwer, Norwell. Preisendo¨ rfer, P., Voss, T., 1990. Organizational mortality of small firms: the effects of entrepreneurial age and human capital. Organization Studies 11 (1), 107–129. Salancik, G.R., Pfeffer, J., 1977. Who gets power and how they hold on to it: a strategic contingency model of power. Organizational Dynamics 5, 3–21. Schefczyk, M., 1993. Operational performance of airlines: an extension of traditional measurement paradigms. Strategic Management Journal 14, 301–317. Smith, C.A., 1955. Survey of the empirical evidence on economies of scale: in Universities National Bureau Committee for Economics Research, Business Concentration and Price Policy. Princeton University Press, pp. 213–230. Spector, P.E., 1992. Summated rating scale construction: an introduction, Sage University paper series. Qualitative application in the social sciences, series no. 082. Sage, Newbury Park, CA. Spreitzer, G.M., 1995. Psychological empowerment in the workplace: dimensions, measurement, and validation. Academy of Management Journal 38, 1442–1465. Thomas, K.W., Velthouse, B.A., 1990. Cognitive elements of empowerment: an interpretive model of intrinsic task motivation. Academy of Management Review 15, 666–681. Westerberg, M., Jagdip, S., Ha¨ckner, 1997. Does the CEO matter: an empirical study of small swedish firms operating in turbulent environments? Scandinavian Journal of Management 13 (3), 251–270. Wood, S., 1979. A reappraisal of the contingency approach to organisation. Journal of Managment Studies, 334–354. Wouters, M., Kokke, K., Theeuwes, J., Donselaar, K.van., 1999. Identification of critical operational performance measures––a research note on benchmarking study in the transportation and distribution sector. Management Accounting Research 10, 439–452.