Transport Policy 115 (2022) 275–285

Contents lists available at ScienceDirect

Transport Policy
journal homepage: www.elsevier.com/locate/tranpol

Identifying the drivers of profitable airline growth

Yun Shwe Yee Maung a, *, Ian Douglas b, a, David Tan a
a
School of Business, University of Wollongong, WOLLONGONG, 2522, NSW, Australia
b
School of Aviation, University of New South Wales, SYDNEY, 2052, NSW, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: While regulatory changes in the global economic environment have facilitated airline capacity growth, profit­
Airline growth ability remains as a challenging issue for airlines globally. This study examines whether the airline business
Airline profitability model has an impact on the tension between growth and profits and is the first to apply the dynamic system
Airline business models
generalised method of moments model to this relationship. We find that full-service network carriers are faced
Dynamic panel GMM
with a trade-off between growth and operating profits, whereas low-cost carriers are able to simultaneously
pursue growth-oriented strategies whilst improving profitability. We discuss the key differences in the business
models that drive these results and expect that short-term effects of COVID-19 travel constraints will impact
different airlines in distinct ways.

1. Introduction 2005). While numerous researchers studied the influence of firms’

growth on profitability in other industries, only a limited number of
While regulatory changes in the global economic environment have prior studies explored the growth-profit relationship specifically for the
facilitated airline capacity growth, profitability remains as a challenging airline industry. Empirical evidence on this relationship has been
issue for airlines globally. Poor profitability in the airline industry is inconsistent; while a positive relationship is observed between in the
driven by a number of factors including high fixed costs, overleveraged airlines growth and profits in earlier studies (Lau and Matthesiss, 1992;
balance sheets, low entry barriers and fuel price volatility (Warnock-­ Chin and Tay, 2001), a recent study by Hazel (2018) suggests an adverse
Smith, O’Connell and Maleki, 2017). IATA Economics (2019) reported impact of growth on airline profit. While the ambiguity in past results
that airlines only delivered a return on invested capital (ROIC) above the could be assigned to differences in sampling and empirical modelling,
weighted average cost of capital (WACC) in the four years from 2014 we hypothesize that this discrepancy is more likely due to the impact of
after falling continuously below the WACC during the preceding decade. airline business models on the growth-profit relation and an endoge­
Furthermore, airlines’ ROIC is the lowest among other supply chain neity problem in the previous approaches.
sectors in air transportation (IATA Economics, 2013). This highlights the The purpose of this paper is to add to the existing body of evidence by
failure of the airline sector to sustain financial health, even in a period of investigating the causal relationship between the growth and profit­
increasing air passenger traffic, fleet, and growing networks over recent ability of airlines, and in particular, study whether airline business
decades. The scale of this growth can be seen in the number of air models have a moderating impact on this relationship. We account for
passengers that grew from approximately 2.7 billion to 4.5 billion dur­ the influence of capacity growth, measured by available seat kilometres
ing the decade of 2010 to 2019 (Statista, 2020). Cost savings generated (ASKs), on an airline’s operating margins and unit profit. This paper
by new technology tended to be competed away for passenger and aims to determine whether an airline’s business model influences the
market share growth rather than improved profits. relationship between growth and profit and is the first to apply the
Several past studies examined various areas of growth and profit in dynamic system generalised method of moments (GMM) panel specifi­
the aviation industry. These include the connection between air traffic cation to the airline growth-profit relationship. The GMM estimates are
growth and economic growth (measured as GDP) (Chin and Tay, 2001; robust to simultaneity, dynamic endogeneity, and heterogeneity; all of
Profillidis and Botzoris, 2015), the influence of load factor and yield on which likely plagued the results of prior studies.
profit (Chin and Tay, 2001; Oum et al. 2005) and the impact of airline While the industry is in a period of massive disruption from COVID-
business model on profit (Collins et al. 2011; Alamdari and Fagan, 19, the forecasts from IATA (2020) indicate an anticipated return to

* Corresponding author.
E-mail addresses: [email protected] (Y.S.Y. Maung), [email protected], [email protected] (I. Douglas), [email protected] (D. Tan).

https://doi.org/10.1016/j.tranpol.2021.11.007
Received 26 March 2021; Received in revised form 7 October 2021; Accepted 9 November 2021
Available online 12 November 2021
0967-070X/© 2021 Elsevier Ltd. All rights reserved.
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

more normal traffic and operational patterns as the vaccination pro­ passenger demand could be influenced by economic and supply-related
grams expand worldwide. The shutdown of international flying from factors. Doganis (2010) argues that air travel growth over the last four
March 2020 occurred on a scale that dwarfed earlier periods of decades resulted mainly from growth in the world’s economies and a
disruption following the 2001 attacks in New York or the Global rise in personal income levels. This is further supported by IATA (2013)
Financial Crisis in 2008. Early research into the scale and structure of analysis that showed world revenue passenger kilometres (RPKs) in­
government support for the aviation sector indicates that governments crease at up to triple the rate of world GDP growth, while world freight
were faced with a trade-off between maintaining connectivity (even the tonne kilometres rose approximately four times as fast as world GDP
survival of airlines) and deregulation (Abate et al., 2020). While the growth. Since over-capacity impinges on yields (Doganis, 2006) and can
massive disruption could have provided an opportunity for ration­ obstruct the equilibrium of demand and supply, it can have negative
alisation or restructuring, governments have moved in many countries consequences on load factors, ticket prices, yields and profitability. As
to provide support approaching $US 200 billion in a mix of waived fees, such, capacity growth - that is the level of supply in the market - play a
direct cash injections, and loan funds (IATA, 2020). crucial role in determining an airline’s successful financial operation.
Gössling (2020) and Charlton (2020) identify the broader impact of
COVID-19 shutdowns on the aviation value chain, with drastic decline in 2.2. Airlines’ growth and profitability
revenue streams to airports, air navigation service providers, and leasing
companies. Gössling (2020) succinctly describes the situation by noting Growth has a downward pressure on the financial performance in the
that the world has switched within months from a debate around airline industry as capacity is associated with significant fixed cost that
over-tourism to a discussion of pathways for getting a largely grounded must precede passenger demand. Reid and Mohrfeld (1973) specified
fleet into the air and considering the structure of a more economically that airlines’ profitability cannot be improved when its size is expanded
robust industry for the future. This study examines the profitable growth beyond the level of minimum unit operating expense. The importance of
potential of airlines for different business models and hence can help cost efficiency is further addressed in the empirical analysis by Manuela
future capacity planning for the airline industry. et al. (2019). They find that unit cost efficiency is the main driver of
The paper is organised as follows: Section 2 is a brief review of the above average profitability, more so than pricing power. That is,
airline growth and profits literature; Section 3 discusses the data pro­ cost-efficient airlines have better position to control airfares. In addi­
cedure, descriptive statistics and methodology. Section 4 discusses tion, unit profit cannot be increased by means of reductions in unit
empirical results and provides managerial and policy implications; and operating expense when the size of an airline is already at or over the
Section 5 summarises the findings and concludes. maximum point of unit profit (Reid and Mohrfeld, 1973). Reid and
Mohrfeld (1973) emphasised that substantial growth above the
2. Literature review & hypothesises maximum unit profit leads to the decline in total profitability because
the rate of reduction in unit profit is higher than that of an increase in
2.1. Economic growth and Gross Domestic Product (GDP) in the airline size.
context A small number of studies investigated the direct relationship be­
tween airline growth and profitability. Lau and Matthesiss (1992)
Growth in the aviation industry is driven by deregulation, open skies employed the Markov chain to examine the asset size growth and net
agreements, growth in the economy, and rapid technological change. profit margins of the carriers in the United States. The authors forecast
Airlines have been growing capacity (ASKs) through the expansion of the survival prospect of airlines by grouping airlines into a number of
both fleet and network. This in turn supports and indeed drives increases states in terms of their total assets size and net profit margin. The results
in air passenger traffic demand in the market. It is important to under­ revealed that profitable operations are closely related to an individual
stand in this context that some airlines, dependent on their strategy, airline’s growth. Chin and Tay (2001) employed a similar methodology
grow ahead of economic growth (GDP) whereas others grow more when examining the relationship between the growth in asset size and
slowly than GDP. the profitability of airlines in the Asia Pacific market and reported a
Many studies have addressed growth. First, the multiple regression positive relationship. The authors added that profitable airlines do not
analysis of GDP, load factor and profit by Chin and Tay (2001) examines shrink in size and airlines with higher profits have larger survival
the economic cycle and the level of profit generated by Asian airlines probabilities.
over the decade of 1986–1996. A subsequent study by Profillidis and The above two studies focus on only one geographic region; the
Botzoris (2015) analysed air traffic and economic activity for various former examined the airlines in the U.S and the latter studied those in
regions across the world over the period 1980-2013. The variables used Asia. There is no prior research studying the relationship of airline
for analysis include air trips per thousand inhabitants and GDP per growth and profitability across different geographical regions. The
capita. The findings indicate that there is a minor correlation depending geographic position plays an important role in determining profits for
on the nature of markets; that is, the correlation strengthens in more airlines. O’Connell (2011) remarked on the strength of Gulf carriers,
mature markets. These studies attribute the growth in the airline in­ leveraging geographic advantage in the Middle East region with
dustry to GDP. approximately 4.5 billion people residing within an 8-hour flight dura­
The effect of GDP is further explained by Tarry (2015a) from an tion. Competing airlines in Europe deal with regulatory constraints and
airline economics point of view. GDP does not have a steady effect on high airport charges that make it harder for airlines to deliver profitable
fares but it should be considered in traffic forecasting models since there financial operations. Therefore, this study will be conducted at the
is a strong connection between GDP and underlying traffic demand global level to account for regional variations and to obtain a general­
(Tarry, 2015a). Tarry also accentuates two components, the airlines’ ised justification of the impact of growth on profits in the airline sector.
cost base and ticket prices, which determine the traffic multiplier effect,
that is, the lower the fares, the higher the multiplier. Tarry (2015b) also 2.3. Airline business models and financial performance
mentions the significance of a regular review of assumptions underlying
forecasting models because these assumptions are usually derived from The financial success of airlines can also be influenced by the choice
previous experiences. Over-pessimistic or optimistic forecasts could of business model. Within the framework of a firms’ generic business
have an impact on capacity decisions and result in negative financial strategy (Porter 1985), airlines fall into ‘cost focus’ (Low Cost Carrier)
outcomes. Hence, economic growth plays an important role in capacity and ‘differentiation’ (Full Service Network Carrier) strategies.
planning and airlines’ financial health. Full-service carriers tend to have a stronger focus more on hub and spoke
From a theoretical perspective of demand, Doganis (2010) noted that network structures and they usually focus their branding on

276
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

connectivity, frequency, and premium in-flight services. Airlines prac­ Table 1

tising this strategy first target high-yielding travellers in the respective Variables and definitions.
markets with the remaining seat capacity filled with low-yielding dis­ Notation Variable Name Definition
counted tickets (Pels, 2021).
OM Operating margin The ratio of operating profit
For low cost carriers (LCCs), the focus of the network is usually on (earnings before interest
point-to-point operations, and an emphasis on incremental revenue for and tax) to operating revenue.
catering, checked baggage, seat assignment and other ancillary services. UP Unit Profit Profit obtained per unit of capacity.
Features of LCCs tend to include high density single-class seating, faster Calculated as cost
per available seat kilometres (CASK)
aircraft turnaround times, and high utilisation of resources such as substracted from
aircraft and crew. Each model has its own competitive advantage and revenue per available seat kilometres
the robustness of each respective strategy plays a crucial role in the (RASK). CASK
success of an airilne’s operations. is calculated as total operating cost
divided by total
A study of the impact of effectiveness of the low cost business model
ASK and RASK is calculated as total
on profitability (Alamdari and Fagan 2005) illustrates that the operating operating
margins of airlines increase with the degree of adherence to the revenue divided by total ASK.
cost-leadership strategy. However, results from another empirical study ASK Available seat kilometres Total passenger capacity. Calculated
(Collins, Román and Chan, 2011) on the influence of the choice of airline as the number
of available seats for sale multiplied
business model on the persistence of profitability in the U.S suggests that by the stage
full-service carriers attained more persistent profit margins and asset distance flown.
turnover ratios than low cost carriers. Furthermore, a viability test of the LF Passenger load factor Utilisation of aircraft passenger
low-cost business model showed financial success is only delivered in capacity. Calculated as
the ratio of revenue passenger
large and dense markets (Francis et al., 2007; Pels, 2008). Given the
kilometres (RPK) to
differences in the influence of business model on airline profitability, available seat kilometres (ASK). RPK
business model will be incorporated into the examination of growth and is calculated as
profit relation. number of sold seats multiplied by
the stage distance
flown.
3. Data and method YIELD Passenger Yield Average fare. Calculated as total
passenger revenue
3.1. Hypotheses divided by the total revenue
passenger kilometres.
FUEL Fuel Price Jet fuel price per gallon at mid-point
The first hypothesis concerns the overall relationship between airline
of the year.
profits and growth. We posit that there is a statistically significant trade- WORLD.GDP World GDP Annual global GDP dollar value
off between growth and profits among the possible strategies that air­ reported by the
lines can pursue. World Bank.
COUNTRY. Country GDP Annual GDP dollar value for each
Hypothesis 1. Growth has a negative impact on airlines’ profitability GDP country reported by
Given the importance of the airlines’ choice of business model on the World Bank.
profitability, business model is explicitly tested within the growth- FX.RATE Foreign exchange rate Annual average exchange rate of
airline local currency
profitability relationship. with respect to USD.
HHI Herfindahl–Hirschman Competition measure calculated
Hypothesis 2. An airline’s business model impacts on its growth-
Index using the market
profitability relationship share of ASK of each region.
Both hypotheses are tested within the dynamic system GMM AGE Firm Age The age of the airline at the year of
framework. Assuming sequential exogeneity and the strength and the observation. Dummy
validity of internal instrumental variables, the GMM estimates will variables for the age categorises of i.
Younger than 10
isolate causality between airline growth and profits. These conditions
years of age, ii. Between 10 and 20
will be formally tested in the subsequent sections. years of age, and
iii. Older than 20 years of age; are
3.2. Data collection and sample created.
BM Business model dummy A dummy variable equal to 1 if the
airline is a FSNC and
The dataset of this study is manually constructed from several data equal to 0 otherwise.
sources. Airline performance measures are primarily sourced from REGION Airline’s base region A nominal indicator of the airline’s
airline annual reports, press releases and annual shareholder pre­ region: Europe,
sentations. IATA World Air Transport Statistics (IATA WATS) database North America, Middle East, Asia-
Pacific and
and Statista are also employed as supplementary sources for missing South America.
data. The economic measures such as GDP, exchange rate and jet fuel
price are retrieved from the World Bank, OFX and indexMundi respec­
tively. The measures employed in the econometric modelling are across several continents.
described in Table 1.
The final sample contains 64 major international airlines over the 3.3. Measures
period of 2010–2017, inclusive. By region, airlines are represented by 26
Asia Pacific, 16 European, 16 North American, 4 South American, and 2 3.3.1. Dependent variables – profit
Gulf carriers. By business model, these 64 airlines comprise of 41 full- Building upon the extant airline economics and financial perfor­
service and 23 low-cost airlines. The full dataset is an unbalanced mance literature, profitability measures, namely operating margin (the
panel with 405 airline-year observations. While previous attempts to proportion of profit earned on a dollar of revenue) and unit profit (profit
explore the growth-profit relation of airlines focused on only one region, earned per seat kilometre), are employed as the dependent variables in
this study is unique in its international context by examining airlines this study to analyse the growth-profit relation of airlines. Several

277
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

researchers (see, for example, Merkert and Swidan, 2019; Zou and Chen, premium seats and low-fare discount seats is crucial for precluding
2017) employed operating margin to measure the airlines’ financial either high yield or low yield spills.3 Revenue spills from suboptimal
outcome. While net profit margin has been used in the prior literature on allocation of fares for the different seat classes will have a negative
airlines’ growth-profit relation (Lau and Matthesiss, 1992; Chin and Tay, impact on an airline’s financial outcome. Further, empirical findings by
2001), operating margin is the variable of choice in the current study as it Oum et al. (2005) emphasised the importance of pricing and yield
measures the efficiency and financial viability of the airlines’ business management strategy on the airlines’ financial success.
operations. Furthermore, operating margin sidesteps issues of inconsis­
tent approaches to depreciation and ownership under differing tax iii Gross Domestic Product (GDP)
regimes.
In robustness tests, unit profit is employed in our analysis as an The airline business is susceptible to the strength of economic ac­
alternate measure of profitability. It is a common profit measure in the tivity of a country. It has been empirically shown that the air traffic
airline industry as it captures the supply of air travel capacity; that is, growth rate is positively associated with GDP growth rate (Chin and Tay,
both the number of seats and stage distance flown by the airline. 2001). Increased capacity in parallel with GDP growth can induce a rise
in airfares whist maintaining load factors and profits (Stalnaker et al.,
3.3.2. Regressor of interest – growth 2019). Since GDP has an influence on airline’s profitability, both (airline
The growth variable considered in this study is the airlines’ output home) country GDP and world GDP will be included as regressors.
measure known as available seat kilometres (ASKs). The number of ASKs
is defined as the total capacity available for carriage of passengers on iv Exchange Rate
each flight sector by stage distance flown (Doganis, 2010). We examine
the operational measure (ASK) rather than the accounting measure The volatility of the exchange rate is of importance in air trans­
(asset size value) as used in prior studies (Lau and Matthesiss, 1992; portation as airline revenue is earned across multiple currencies.
Chin and Tay 2001) as it better represents not only the level of supply of Generally, a devaluation of a country’s currency is prone to prevent
air travel capacity in the market, but also the dynamics of airline op­ outbound travel and improves the attraction of inbound travel, and vice
erations that cannot be captured by asset size alone. Empirical results of versa. In a study of the impact of exchange rate on air travel, Day (1986)
the U.S market showed an inverse relation between airline capacity and showed that U.S citizens are more likely to travel overseas while for­
unit revenue (Hazel, 2018). As such, capacity (ASK)1 is employed as the eigners are less likely to travel to the United States when there is an
metric of airline growth. increase in the value of the U.S dollar. As such, an airline’s local ex­
change rate will likely have an impact on its profitability. In this study,
3.3.3. Control variables the airline’s home country exchange rate with respect to the US dollar
Several airline and industry specific measures are included as re­ value will be included as a regressor as most of the airline transactions
gressors due to their significant impact on airlines operating profit­ are undertaken in USD, such as aircraft leases and the cost of fuel.
ability.2 The rationale for inclusion is described below.
v Fuel Price
i Passenger load factor
The variation in fuel price can affect the profitability of airlines as
Load factor, the proportion of output (seats) sold, is a measure of fuel is a major cost component. On average, 28.7% of airlines’ expenses
utilisation of passenger aircraft capacity and is influenced by the char­ are from fuel and the reduction in fuel price lowered industry-wide
acteristics of travel demand, such as business or leisure. If the level of break-even load factors and improved airlines’ financial performance
passenger traffic does not follow with a rise in capacity (ASKs), pas­ in 2015 (IATA, 2016). As such, the impact of fuel price is highly relevant
senger load factors will fall. Theoretically, ticket prices are reduced to to the financial viability of airline operations. Moreover, Kumari (2012)
drive load factors when the underlying traffic growth is insufficient in stressed that airlines are highly sensitive to fuel prices and upward
absorbing capacity growth (Holloway, 2008). This decline in ticket movements in fuel price can easily obstruct profitable operations.
prices can impinge on the airline’s financial performance. In addition,
there is abundant empirical evidence that load factor is related to airline vi Herfindahl–Hirschman Index (HHI)
profitability (see, for example, Chin and Tay, 2001; Kalemba et al.,
2017). The HHI is calculated at the region level using airline capacity
(ASKs). This metric is a proxy for the movement in cross-sectional and
ii Passenger Yield temporal variations in competition. Studies by Borenstein (2011) and
Clougherty and Zhang (2009) have found that competition has had an
Passenger yield, also known as the average fare, is an indicator of the impact on airline financial performance. In fact, competition measures,
revenue obtained per unit of output sold. The amount of yield earned is such as HHI, have been employed as instruments due to their relation­
determined by fare allocation. The optimal balance between high-fare ship with pricing and profitability (Mumbower, 2014).

vii Firm Age

1
While great-circle distance data was used as a readily available basis for Based on the life-cycle theory of the firm, a firm’s decision to re-
ASK values, variations will occur in day-to-day operations, impacted by invest and expand will be a function of its age as it is reflective of its
weather, congestion, the selection of routes with more favourable winds, or the growth opportunities (Mueller, 1972). Moreover, a well-established
avoidance of military actions. The actual route flown will often be longer than airline may be able to extract greater economic rent from its cus­
great circle distance, and this will add some amounts of cost in fuel burn that tomers than a younger airline. A study by Ismail and Jenatabadi (2014)
could not be directly accounted in the empirical analysis. Hence, these varia­
finds that the age of an airline is a moderating factor on the relationship
tions are contained in the residuals and assumed orthogonal to the relationship
of interest.
2
Note that in this study, operating profits (not market returns) are examined.
3
As such, covariates that typically linked to market returns, such as leverage and Low-yield revenue spill refers to the loss of earning from too few low-fare
systematic risk, are excluded as they do not impact the operating environment seats allocation and high-yield revenue spill refers to the loss from too few
of the business. high-fare seats (Wensveen, 2011).

278
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

between financial performance and growth. Table 3

Descriptive statistics of airline performance measures by business model.

3.4. Summary statistics Low Cost Carriers - LCCs (obs = 137, airlines = 23)

Variable Mean Std.dev Minimum Maximum

Tables 2 and 3 contain the descriptive statistics for the full sample OM (%) 9.47 9.85 − 15.91 38.06
and two subsamples defined by business models, respectively. Notice­ UP (¢) 0.62 5.72 − 47.69 46.13
ably from Table 2, on average, global operating margin lies at 6.37% and ASK (billion) 47.55 49.06 10.25 247.64
the unit profit is 0.78 cents per ASK. This highlights the difficulty in LF (%) 83.12 4.66 69.88 94.00
Yield (¢) 6.87 2.14 3.45 11.86
delivering profits in the airline industry. The average passenger load
AGE (years) 20.52 11.61 3.00 52.00
factor is 80.45% and the average yield attained is 8.75 cents per revenue Full Service Carriers - FSCs (obs = 268, airlines = 41)
passenger kilometre. With regards to concentration, the Asian region OM (%) 4.82 7.18 − 35.02 23.19
had the most competition in 2013 with a HHI value of 614 while the UP (¢) 0.86 1.32 − 2.45 7.47
Middle East had the greatest amount of concentration (HHI = 6589) ASK (billion) 119.95 109.28 8.31 445.15
LF (%) 79.08 4.11 65.75 90.60
during 2017. In terms of business model from Table 3, while average Yield (¢) 9.71 2.38 5.61 2.06
unit profit of FSCs is higher than that of LCCs (0.86 compared to 0.62 AGE (years) 60.56 27.18 5.00 98.00
cents), average operating margin of FSCs is lower than that of LCCs
All currency values are in USD.
(4.82% compared to 9.47%).
The notation is defined in Table 1.
LCCs obtain less unit profit since they charge lower fares as a result of
their “no frills” service.4 On the other hand, FSCs earn a lower operating
slope coefficients for control variables X; εit is the error term of airline i
margin that is likely driven by the complexity of their premium service
at time t.
and the additional operating expenses associated with it. This is because
A random-fixed effects panel regression5 is first applied to (1) to
one cent of unit profit earned by a FSC and that earned by a LCC cannot
provide a baseline model that is representative of prior studies (Hidayat
create the same level of profits due to the differences in unit revenue and
and Abduh, 2012; Odeck and Brathen, 2012). The random-effects model
unit cost. For example, while an LCC with 5 cents revenue per ASK on 4
assumes that airline effects are uncorrelated with the regressors, and
cents cost per ASK obtains 20% margin, FSC with 9 cents revenue per
thus can be captured entirely in the error term of (1). Importantly, the
ASK on 8 cents cost per ASK can only obtain an 11% margin although
random-effects estimator requires the assumption of strict exogeneity in
both earned one cent of profit per ASK. As expected, average load factor
the regressors, which is likely problematic for variables such as ASK, LF,
of LCCs (83.12%) is greater than that of FSCs (79.08%). Yield figures are
and YIELD. For example, YIELD and PROFITS are likely to be simulta­
also as expected where the average lies at 6.87 for LCCs in contrast to
neously determined; similarly, LF is likely to be influenced by the shocks
9.71 cents for FSCs. LCCs are approximately 40 years younger than FSCs,
to the airline industry that also impact profitability. The strict exoge­
on average.
neity assumption prevents the inclusion of a dynamic autoregressive
(AR) term in (1) that would capture any persistence in profitability
3.5. Methodology across adjacent periods.
The Arellano and Bover (1995) dynamic system GMM model will
We analyse the relation between airline profitability and growth by also be applied to the estimation of (1). The GMM estimator accounts for
estimating the following equation: dynamic endogeneity, heterogeneity, and simultaneity (Wintoki et al.,
PROFITit = β0 + β1 GROWTHit + Xit B + εit (1) 2012). Under the assumption of sequential exogeneity, the GMM esti­
mator exploits the careful use of lags as instrumental variables to tease
where: out causality in equation (1). This model has been successfully applied to
PROFITit is the profitability measure, OM and UP as defined in profitability at the firm level in the corporate finance literature (Schultz
Table 1, for airline i at time t; β0 is the intercept term; GROWTHit is the et al., 2010; Goddard et al., 2005), and will be applied similarly in the
growth measure, ASK as defined in Table 1, for airline i at time t; β1 is the current study to the airline industry.
slope coefficient for the GROWTH variable; Xit is a vector of control The dynamic system GMM studies the first-difference of (1) which
variables for airline i at time t as outlined in Section 3.3.3; B is a vector of eliminates any potential airline fixed-effects. Second, the inclusion of
lags as instrumental variables accounts for potential simultaneity that
Table 2
may exist in (1). Under the assumption of sequential exogeneity, care­
Descriptive statistics of airline performance and economic measures. fully selected lags6 of the regressors are orthogonal with future errors yet
are likely correlated with contemporaneous values of regressors which
All Airlines (obs = 405, airlines = 64)
makes them ideal instruments. This is predicated on the absence of serial
Variable Mean Std.dev Minimum Maximum correlation in the errors which must be tested in the GMM model di­
OM (%) 6.37 8.45 − 35.02 38.06 agnostics. Finally, an autoregressive term is included in (1) to account
UP (¢) 0.78 3.49 − 47.69 46.13 for the dynamic process of PROFITS.
ASK (billion) 95.46 99.39 8.31 445.15
LF (%) 80.45 4.70 65.75 94.00
Yield (¢) 8.75 2.66 3.45 20.60
FUEL ($) 2.27 0.74 1.27 3.13
WORLD.GDP (trillion $) 75.91 3.09 65.96 80.89
COUNTRY.GDP (trillion $) 4.86 6.93 0.12 20.17
FX.RATE ($) 0.57 0.50 0.00 1.65
AGE (years) 47.01 29.89 3.00 98.00
HHI 1408.00 1130.00 612.00 6589.00 5
The Hausman test indicates that a random-effects model is appropriate and
All currency values are in USD. The notation is defined in Table 1. more efficient than its fixed-effects counterpart.
6
Current and lag values of the differenced predetermined and exogenous
variables are valid IVs as they are uncorrelated with errors at time t. Lags of the
differenced endogenous variables are valid IVs as they are uncorrelated with
errors at t+1. See Koo et al. (2018) and Roodman (2009) for discussions on the
4
Separate charges for in-flight catering and other services. appropriate lag selection of IVs in the dynamic panel GMM model.

279
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

4. Results and discussion between growth and profitability. However, the RE panel model as­
sumes strict exogeneity which is a condition likely to be violated in the
4.1. Empirical results data. A Durbin-Wu-Hausman test concludes that there is significant
endogeneity in (1) and that IVs should be employed for inference
Table 4 below contains the results of the RE and GMM panel esti­ testing.7 As such, we now turn to model (b) which employs the dynamic
mation for the full sample of airlines. Note that non-stationary variables panel GMM specification.
are first-differenced until stationary. In the RE specification for OM The dynamic system GMM specification employ one-period lags of
(model (a)), we can see that LF, YIELD, FUEL, WORLD.GDP, and the first-differences of the endogenous variables: LF, ASK, YIELD and the
COUNTRY.GDP are statistically significant. Notably, capacity (ASK) is dependent variable.8 Model (b) in Table 4 contains the GMM results
not statistically significant which suggests that there is no trade-off when OM is the profitability measure. Some of the statistically signifi­
cant regressors in (a) are statistically insignificant in (b), indicative that
these variables are spuriously correlated with, and do not have any
Table 4
Results of growth and profitability relation (All airlines). causal influence on, profitability. Notably, load factor and yield both
have a statistically significant positive causal impact on OM. ASK re­
Dependent variable: OM Dependent variable: UP
mains statistically insignificant in (b). WORLD.GDP exhibits a signifi­
Regressor (a) RE (b) GMM (c) RE (d) GMM cantly negative coefficient in (a) and (b), consistent with the law of
ASK 0.0029 − 0.1104 0.0003*** − 0.0005 diminishing marginal returns. The coefficient of BM is statistically
(0.0199) (0.3694) (0.0001) (0.0006) insignificant across all models in Table 4. This suggests that airlines of
LF 0.8940*** 1.5274* 0.004 − 0.0003 different business models have no notable fixed differences in OM and
(0.2349) (0.8501) (0.0006) (0.0021)
YIELD 1.7852*** 3.4274* 0.0020*** − 0.0030
UP. Competition (HHI) has a positive impact on OM which suggests that
(0.5878) (1.9238) (0.0006) (0.0038) less competitive routes are more profitable.
FUEL − 6.8897*** – − 0.0205 – The Sargan J-statistic for (b) is statistically insignificant which lends
(1.2264) – (0.0154) – support to the validity of the GMM instruments.9 The Arellano-Bond test
WORLD.GDP − 0.3395** − 1.2765*** − 0.0051 − 0.0003
statistic for serial correlation of order 2 in the errors is statistically
(0.1581) (0.3442) (0.0050) (0.0006)
COUNTRY.GDP 0.0841*** 0.1628 − 0.0001 0.0011 insignificant which indicates that the lags of the endogenous variables
(0.2914) (0.4159) (0.0001) (0.0011) are suitable instruments. Finally, in unreported results, the first-stage
FX.RATE − 4.5976 − 10.0309 − 0.0397 0.0865** regressions are statistically significant at the 5 percent level10. Note
(5.5920) (17.4305) (0.0363) (0.0405) that FUEL is excluded in the GMM specification due to collinearity.
BM 0.6970 2.6692 0.0080 0.0073
(0.4557) (7.7405) (0.0068) (0.0133)
Models (c) and (d) are the RE and GMM estimations of (1) when unit
HHI 0.0014*** 0.0031** 0.0000 0.0000 profit is the dependent variable. We see a similar pattern with the RE
(0.0004) (0.0014) (0.0000) 0.0000 estimates reporting some statistical significance while the GMM results
AGE2 − 1.0686 − 7.8547 0.0035 0.0042 are largely insignificant. The R-Square statistic of model (c) is 3 percent,
(0.9515) (5.0195) (0.0041) (0.0522)
which suggests that (1) is unlikely to be a suitable specification for
AGE3 − 1.8075** − 9.0044 − 0.0048 0.0075
(0.8850) (5.8073) (0.0033) (0.0460) modelling unit price. In comparison, the R-Square in model (a) is 34
REGION2 − 1.0361* 3.2891 0.0056 0.0021 percent.
(0.6094) (7.7830) (0.0049) (0.0424) Table 5 presents the estimation results of (1) with the addition of an
REGION3 − 9.5318*** 1.2783 − 0.0330 0.0196 interaction term of BM and ASK (ASK.BM). Recall that BM is equal to 1 if
(2.4988) (23.6604) (0.0302) (0.0597)
the airline is a full-service network carrier, and 0 if it is a low-cost car­
REGION4 0.3496 3.2955 0.0053 0.0053
(0.4725) (3.9581) (0.0051) (0.0277) rier. We posit that an airline’s business model may play a role in the
REGION5 − 3.5884*** − 6.5531 − 0.0072 0.0280 growth-profit relationship. ASK.BM allows for differences in the ASK
(1.2955) (4.8207) (0.0102) (0.0569) coefficients across business models.
Observations 337 273 340 212
Model (e) is similar to model (a) of Table 4 in terms of statistically
No. Instruments N/A 51 45
No. Airiines 64 64 64 62 significant terms. Interestingly, model (f) reports a statistically signifi­
R-squared 0.34/0.40/ N/A 0.03/0.12/ N/A cant relationship between ASK and OM, for both LCCs and FSCs. For
0.34 0.03 LCCs (BM = 0), a one-unit increase in ASK will lead to a 0.6926 increase
J-Statistic N/A 32.47 N/A 21.76 in operating margin, ceteris paribus. This suggests that LCCs have been
Arellano-Bond AR N/A − 0.41 N/A − 1.56
able to capture both growth and profit simultaneously. In contrast, for
(2)
FSCs, a one-unit increase in ASK reduces operating margin by 0.0390
The notation is defined in Table 1. *, **, and *** denote statistical significance (sum of the coefficient estimates of ASK and ASK.BM). As in Table 4,
level of 10, 5 and 1 percent, respectively In the random effects panel specifi­ YIELD and LF both have statistically significant effects on OM in (f). The
cation, robust clustered standard errors (airline level) are applied and reported
significant positive coefficient for BM suggests that FSC have a signifi­
in the parentheses and the R-squared values contain within-group, between-
cantly higher OM than LCCs, once we account for the differences in the
group and overall, respectively. For GMM specifications, robust two-step stan­
dard errors, incorporating the Windmeijer (2005) estimation, are included in the
parentheses. The J-Statistic follows a chi-squared distribution with (K-p) degrees
of freedom, where K is the number of moment conditions and p is the estimated
7
parameters; the null hypothesis is that the moment conditions are correctly The results of the Durbin-Wu-Hausman test for endogeneity are omitted for
specified. The Arellano-Bond test statistic follows an asymptotic normal distri­ the sake of brevity. Please contact the corresponding author if interested.
8
bution with the null hypothesis is that no autocorrelation of order q in the dif­ This is the instruments list for the levels equation of the system GMM
ferenced errors. Region dummies are reported. Year dummies are unreported for specification. The differenced equation employs the two-period lags of the
the sake of brevity. Model (d) employed deeper lags than Model (b) and an endogenous variables.
9
additional AR(2) term to expunge inherent serial correlation in UP. If the instruments are endogenous, and thus produce inconsistent parameter
estimates, the overidentified moment restrictions may be systematically
violated. The J-statistic follows a chi-squared distribution. The null hypothesis
states that the instruments are orthogonal to the errors and in this case, it
cannot be rejected at the 5 percent level.
10
Interested readers can contact the corresponding author for the first-stage
regression results.

280
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

Table 5 findings of Merkert and Morrell (2012) who report that airlines oper­
Results of growth and profitability relation (All airlines). ating with available seat kilometres greater than 100 billion are found to
Dependent variable: OM Dependent variable: UP be too large to operate efficiently and remain financially viable. As the
average ASKs of full-service airlines lie at approximately 120 billion
Regressor (e) RE (f) GMM (g) RE (h) GMM
(Table 3), many FSCs in our sample fall within this category of attracting
ASK − 0.1338 0.6926** 0.0015 0.0024 a financial penalty when pursuing growth opportunities.
(0.1282) (0.3450) (0.0014) (0.0017)
ASK.BM 0.1408 − 0.7316** − 0.0013 − 0.0032*
For LCCs, more aggressive pricing is also likely to be employed to
(0.1267) (0.3392) (0.0013) (0.0019) drive volume, but this volume has a strong prospect of also driving a
LF 0.8886*** 1.7452*** 0.0004 − 0.0002 parallel increase in ancillary charges for baggage, booking fees, seat
(0.2304) (0.6183) (0.0006) (0.0020) assignment, meals, and drinks. Unlike base airfares, these ancillary
YIELD 1.7446*** 3.0914** 0.0024*** − 0.0021
components are rarely discounted, and are also more likely to apply as
(0.5694) (1.3333) (0.0007) (0.0020)
FUEL − 6.8711*** – − 0.0206 – fares fall.
(1.2208) – (0.0156) – In an alternative specification, we employ the average fuel price
WORLD.GDP − 0.3282** − 1.2543*** − 0.0052 − 0.0006 instead of the annual mid-point of fuel prices. This is contained in
(0.1573) (0.2948) (0.0051) (0.0007) Table A2 of the Appendix. Table A3 of the Appendix presents the
COUNTRY.GDP 0.0874*** 0.0004 − 0.0001 0.0008
(0.0291) (0.3017) (0.0001) (0.0009)
reproduction of Table 5 using alternative regional categories, specif­
FX.RATE − 3.7467 − 12.7547 − 0.0475 0.0655** ically: Europe, North America, Central & Caribbean, South America,
(5.6447) (12.9062) (0.0444) (0.0307) Middle East, North East Asia, and Asia Pacific. Both of these robustness
BM 0.0464 6.9448*** 0.0137 0.0184 tests do not alter the main findings of this study. Additionally, in unre­
(0.7747) (2.5735) (0.0122) (0.0136)
ported results, we include a quadratic growth term as a regressor whose
HHI 0.0013*** 0.0023* 0.0000 0.0000
(0.0004) (0.0014) (0.0000) (0.0000) coefficient estimate is statistically insignificant and did not alter our
AGE2 − 1.1987 − 7.9107* 0.0047 0.0174 findings.11
(0.8836) (4.0814) (0.0054) (0.0464)
AGE3 − 1.7100* − 9.8885** − 0.0056 0.0172
(0.8718) (4.1118) (0.0040) (0.0430)
4.2. Managerial and policy implications
REGION2 − 1.1335* 3.6117 0.0063 − 0.0098
(0.6402) (5.0797) (0.0054) (0.0229) Even before the emergence of the COVID-19 pandemic, airline fleet
REGION3 − 9.2675*** − 2.2112 − 0.0353 − 0.0107 orders were focusing more strongly on narrowbody aircraft. The market
(2.5055) (13.8448) (0.0325) (0.0499)
forecast for 2016–2035 from Boeing predicts the number of new de­
REGION4 0.2128 2.6736 0.0065 − 0.0012
(0.5083) (3.1861) (0.0060) (0.0151) liveries for single-aisle airplanes to be approximately three times higher
REGION5 − 3.6630*** − 3.6221 − 0.0066 0.0052 than that of widebody airplanes, with over 28,100 and 9,100 respec­
(1.2272) (4.2500) (0.0102) (0.0370) tively (Boeing, 2016). In addition, the Farnborough airshow in 2016 saw
Observations 337 273 340 212 almost 50% of the 279 aircraft orders made for A321 large single-aisle
No. Instruments N/A 61 N/A 54
No. Airiines 64 64 64 62
aircraft and only a handful for widebodies. The pandemic has brought
R-squared 0.35/0.40/ N/A 0.03/0.18/ N/A an additional focus on fleet structure, with carriers in Europe, Asia and
0.34 0.03 the Americas retiring significant numbers of large widebody aircraft.
J-Statistic N/A 40.93 N/A 26.31 Most of the remaining Boeing B747-400 fleet, Airbus A380s from Eu­
Arellano-Bond AR N/A − 0.44 N/A − 1.58
ropean and Asian carriers, and older Boeing B777-200 and Airbus A340
(2)
aircraft from many operators underline this resetting of fleet strategy.
The notation is defined in Table 1. *, **, and *** denote statistical significance The emerging pattern of the recovery of operations is a return to
level of 10, 5 and 1 percent, respectively In the random effects panel specifi­ domestic and short-haul regional flying before intercontinental routes.
cation, robust clustered standard errors (airline level) are applied and reported
In 2021 H1, IATA reports that the proportion of revenue passenger
in the parentheses and the R-squared values contain within-group, between-
kilometres (RPKs) reached approximately 70% of 2019 levels in do­
group and overall, respectively. For GMM specifications, robust two-step stan­
dard errors, incorporating the Windmeijer (2005) estimation, are included in the
mestic markets whereas it is only 15% in international markets (IATA,
parentheses. The J-Statistic follows a chi-squared distribution with (K-p) degrees 2021). These domestic and regional markets are the core business of
of freedom, where K is the number of moment onditions and p is the estimated low-cost carriers, but also represent a significant proportion of the total
parameters; the null hypothesis is that the moment conditions are correctly ASKs generated by large network airlines that are based in countries
specified. The Arellano-Bond test statistic follows an asymptotic normal distri­ with large domestic markets, including the United States, China, and
bution with the null hypothesis is that no autocorrelation of order q in the dif­ Australia. The combination of aircraft retirements, rebalanced networks,
ferenced errors. Region dummies are reported. Year dummies are unreported for and a shift in the strategic balance from hub specialists to shorter haul
the sake of brevity. Model (h) employed deeper lags than Model (f) and an operators will impact the earlier patterns of growth rates and
additional AR(2) term to expunge inherent serial correlation in UP. profitability.
The varying responsiveness of different airline models can be seen in
profitability-growth relation across business models. Models (g) does Suau-Sanchez et al. (2020) where substantial capacity (ASKs) reductions
not report any statistical significance except for YIELD, which is were evident from FSCs as early as February 2020 due toconstraints on
consistent with the results of Table 4. Similarly, the overall R-Square international traffic, while a reduction is not evident for LCCs until late
statistic in (g) is 3 percent which again suggests that unit profit may not March 2020. Early evidence of recovery points to short haul and
be appropriately estimated by (1). Model (h) exhibits a statistically regional flying returning well ahead of intercontinental routes has im­
significant negative coefficient for ASK.BM which supports the findings plications for the major Gulf hubs where home market traffic makes up a
of model (f). small part of total load.
This divergence in outcomes for the two business models is not un­ This asymmetrical return of the market will likely leave some car­
expected. FSCs pursuing a growth strategy are more likely to find that riers dependent on further government support while others resume
their volume increases have to be driven by more aggressive low yield
pricing, aimed at stimulating increased leisure travel demand. Demand
for higher yielding business travel tends to be more inelastic and that 11
We thank an anonymous reviewer for their suggestions for these tests of
segment is less likely to respond to capacity growth, particularly where robustness that have strengthened the study. Please contact the corresponding
growth is deployed in existing markets. This is consistent with the author for the unreported results.

281
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

commercially viable and profitable operations. The short haul return their profitability, whereas FSCs experience a trade-off between growth
will be tempered in some markets by environmental policies that have and profits. The results control for region, currency and fuel price var­
brought a focus on sustainability. Where policymakers have required iations, load factor, yield, and economic growth.
airlines to reduce flying on specific routes where a less carbon intensive This study is the first to apply the dynamic system GMM model to the
substitute (high-speed rail) is available, networks will be adjusted to­ airline growth-profitability relation and overcome the endogeneity is­
wards a longer average sector length. sues that have plagued prior regression-based studies on this topic. The
The spread of COVID19 has underlined how directly and rapidly GMM estimates are robust to dynamic endogeneity, simultaneity, and
governments are involved in airline decision making. The current heterogeneity. The diagnostic tests indicate that the instruments are
downturn in traffic, and particularly in international traffic, has driven strong and valid and hence causality in the growth-profit relation can be
government intervention in, and the bailout of the airline sector, espe­ inferred.
cially for flag carriers. In some cases the support has been direct subsidy The outcome of the differences between business models is an
or waived costs, in others it has been the provision of loan funds. interesting result. A decline in average total revenue per passenger for a
Where governments have provided financial support, it has provided network carrier is caused by diluting its yield mix when its strategy is
opportunities for the imposition of policy direction, in some cases aimed focused on pursuing growth. This decline is greater than the average
at environmental outcomes, and arguably less often aimed at improved total revenue decline for a low-cost carrier, which is able to leverage the
longer term economic sustainability. The policy impositions can take the opportunity of capturing ancillary sales opportunities from the
form of ‘carrot or stick’. That is, airlines can be offered state aid as an increased load. While some FSCs have moved to capture additional fees
incentive to reduce carbon emissions, for example abandoning flying on and charges, the bulk of ancillary revenue for these airlines tends to be
routes where a more emissions-efficient substitute (high-speed rail) is derived from financial services rather than service add-ons (Sorensen,
available. A notable case example is the green condition attached in 2019). These findings offer guidance on the limited potential of a strong
French government support for Air France-KLM group. These re­ growth strategy for a traditional network airline.
quirements include the commitment to utilise two percent of fuel from Despite the current COVID-19 induced chaos in the airline industry, a
sustainable sources by 2025 and to achieve fifty percent overall reduc­ new equilibrium will eventually emerge when normalcy in travel ar­
tion in carbon dioxide emissions per passenger-kilometre in 2030 based rangements resume and restrictions ease. The results of this study will
on 2005 levels as well as a separate 2024 target for domestic flights apply when such an equilibrium state of the industry is achieved as the
(FlightGlobal, 2020). fundamental economics of LCCs and FSCs remain. As long as any al­
Establishing limitations on short domestic (including intra-Europe) terations in travel procedures post-pandemic are applied broadly across
routes where high speed rail is presents two policy opportunities. The the industry, then the relationships (and tension) between growth and
direct one of emission reductions to one seventh of the carbon emissions profitability reported in this study remain relevant. However, during the
per passenger kilometre of an aircraft (Strauss et al., 2021) is supple­ transition back to normalcy, we may observe a divergence in the effi­
mented by the reduction of congestion and release of capacity at ciency of the recovery of LCCs and FSCs. This will likely be due to reg­
slot-constrained major airports. The paradox of imposing restrictions on ulatory constraints such as international border closures and quarantine
short-haul journeys during the recovery from COVID19 flight cutbacks is rules and their asymmetric impact on the FSCs’ hub-and-spoke network
that the market segment that is most able to recover early is short-haul, relative to the LCCs’ more flexible point-to-point model.
and particularly domestic flying where a single state is managing re­ Finally, while the present study examines airline financial perfor­
strictions. In the short run the imposition of environmental constraints mance and growth based on the operational perspective by analysing
may lengthen the time that airlines require economic support. various elements of airline revenue management, other factors such as
governance characteristics at the airline-level may have an influence on
5. Summary and conclusion airline investment decisions for capacity planning and its subsequent
impact on financial performance. Thus, future research could explore
Airline management approaches to network and capacity planning the impact of governance structure, for instance, board independence,
vary substantially across business models with some airlines expanding board size and compensation.
at a higher rate while others practicing a steady growth strategy. There
is debate as to whether pursuing a growth strategy pursued is a deter­ CRediT authorship contribution statement
minant of an airline’s already slim profitability. This study re-examines
the airline growth and profitability relationship by considering airline Yun Shwe Yee Maung: Software, Validation, Formal analysis, Data
business models and employing instrumental variables to tease out curation, Investigation, Writing – original draft. Ian Douglas: Concep­
causality. We explore whether growth can be a successful management tualization, Supervision, Writing – original draft, Writing – review &
strategy as the extant literature on this issue is inconclusive in the airline editing. David Tan: Methodology, Supervision, Investigation, Writing –
context. original draft, Writing – review & editing.
When airline business models are pooled, the growth-profitability
relationship is statistically insignificant. Once accounting for airline Acknowledgement
business model, there is a statistically significant relationship between
profits (as measured by operating margin) and growth (measure by The authors are grateful to two anonymous reviewers, Xiaowen Fu
available seat kilometres). LCCs have been able to pursue growth op­ and Richard Wu for their detailed and thoughtful comments. All errors
portunities and expand their ASK whilst at the same time increasing are our own.

Appendix

282
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

Table A1
List of Airlines

Aegean Airlines (A3) China Airines (CI) LATAM Airlines (LA)

Aer Lingus (EI) China Eastern Airlines (MU) Lufthansa (LH)
Aeroflot Russian Airines (SU) China Southern Airlines (CZ) Malaysia Airlines (MH)
Aeromexico (AM) Delta Air Lines (DL) Norwegian * (DY)
Air Canada (AC) easyJet * (DS) Pegasus Airlines * (PC)
Air China (CA) Emirates (EK) Qantas Airways (QF)
Air France-KLM (AF/KL) Etihad Airways (EY) Ryanair * (FR)
Air India (AI) EVA Air (BR) SAS Scandinavian Airines (SK)
Air New Zealand (NZ) Finnair (AY) Singapore Airines (SQ)
AirAsia * (AK) Frontier Airlines * (F9) Southwest Airlines * (WN)
AirAsiaX * (D7) Garuda Indonesia (GA) SpiceJet * (SG)
Alaska Airlines (AS) Gol Transportes Aèreos * (G3) Spirit Airines * (NK)
All Nippon Airways (NH) Hawaiian Airlines (HA) Thai Airways (TG)
Allegiant Air * (G4) Iberia (IB) Transavia * (HV)
American Airlines (AA) IndiGo * (6E) Turkish airlines (TK)
Asiana Airlines (OZ) Interjet * (4O) United Airlines (UA)
Avianca (AV) Japan Airlines (JL) Virgin America * (VX)
Azul Brazilian Airlines (AD) Jet Airways (9W) Virgin Australia (VA)
BAH (Scoot & Tiger) * (TR) JetBlue * (B6) Volaris * (Y4)
British Airways (BA) Jetstar Airways * (JQ) WestJet * (WS)
Cathay Pacific Airways (CX) Korean Air (KE) Wizz Air * (W6)
Cebu Pacific * (5J)
*
Denotes low-cost carrier (LCC). The remaining airlines are full-service carriers (FSCs)Airline IATA codes are included in
the parentheses.

Table A2
Results of growth and profitability relation (all airlines) using average fuel price

Dependent variable: OM Dependent variable: UP

Regressor (e) RE (f) GMM (g) RE (h) GMM

ASK − 0.1338 0.6926** 0.0015 0.0024

(0.1282) (0.3450) (0.0014) (0.0017)
ASK.BM 0.1408 − 0.7316** − 0.0013 − 0.0032*
(0.1267) (0.3392) (0.0013) (0.0019)
LF 0.8886*** 1.7452*** 0.0004 − 0.0002
(0.2304) (0.6183) (0.0006) (0.0020)
YIELD 1.7446*** 3.0914** 0.0025*** − 0.0021
(0.5694) (1.3333) (0.0007) (0.0020)
FUEL − 7.6473*** – − 0.0230 –
(1.3587) – (0.0174) –
WORLD.GDP − 0.0039 − 1.2543*** − 0.0042 − 0.0005
(0.1550) (0.2948) (0.0044) (0.0007)
COUNTRY.GDP 0.0874*** 0.0004 − 0.0001 0.0008
(0.0291) (0.3017) (0.0001) (0.0009)
FX.RATE − 3.7467 − 12.7547 − 0.0475 0.0655**
(5.6447) (12.9062) (0.0444) (0.0307)
BM 0.0464 6.9448*** 0.0137 0.0184
(0.7747) (2.5735) (0.0122) (0.0136)
HHI 0.0013*** 0.0023* 0.0000 0.0000
(0.0004) (0.0014) (0.0000) (0.0000)
AGE2 − 1.1987 − 7.9107* 0.0047 0.0174
(0.8836) (4.0814) (0.0054) (0.0464)
AGE3 − 1.7100* − 9.8885** − 0.0056 0.0172
(0.8718) (4.1118) (0.0040) (0.0430)
REGION2 − 1.1335* 3.6117 0.0063 − 0.0098
(0.6402) (5.0797) (0.0054) (0.0229)
REGION3 − 9.2675*** − 2.2112 − 0.0353 − 0.0107
(2.5055) (13.8448) (0.0325) (0.0499)
REGION4 0.2128 2.6736 0.0065 − 0.0012
(0.5083) (3.1861) (0.0060) (0.0151)
REGION5 − 3.6630*** − 3.6221 − 0.0066 0.0052
(1.2272) (4.2500) (0.0102) (0.0370)
Observations 337 273 340 212
No. Instruments N/A 61 N/A 54
No. Airiines 64 64 64 62
R-squared 0.35/0.40/0.34 N/A 0.03/0.18/0.03 N/A
J-Statistic N/A 40.53 N/A 26.31
Arellano-Bond AR(2) N/A − 0.44 N/A − 1.58
The notation is defined in Table 1. *, **, and *** denote statistical significance level of 10, 5 and 1 percent, respectively In the random effects
panel specification, robust clustered standard errors (airline level) are applied and reported in the parentheses and the R-squared values contain
within-group, between-group and overall, respectively. For GMM specifications, robust two-step standard errors, incorporating the Windmeijer
(2005) estimation, are included in the parentheses. The J-Statistic follows a chi-squared distribution with (K-p) degrees of freedom, where K is the
number of moment onditions and p is the estimated parameters; the null hypothesis is that the moment conditions are correctly specified. The

283
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

Arellano-Bond test statistic follows an asymptotic normal distribution with the null hypothesis is that no autocorrelation of order q in the dif­
ferenced errors. Region dummies are reported. Year dummies are unreported for the sake of brevity. Model (h) employed deeper lags than Model
(f) and an additional AR(2) term to expunge inherent serial correlation in UP.

Table A3
Results of growth and profitability relation (all airlines) using alternative regional categories

Dependent variable: OM Dependent variable: UP

Regressor (e) RE (f) GMM (g) RE (h) GMM

ASK − 0.1315 0.5081 0.0015 0.0018

(0.1308) (0.5373) (0.0014) (0.0017)
ASK.BM 0.1422 − 0.6742* − 0.0013 − 0.0025
(0.1291) (0.4081) (0.0013) (0.0019)
LF 0.8909*** 1.9444*** 0.0004 − 0.0005
(0.2305) (0.5730) (0.0005) (0.0019)
YIELD 1.7043*** 3.0438* 0.0023*** − 0.0015
(0.5749) (1.7237) (0.0006) (0.0025)
FUEL − 6.8186*** – − 0.0203 –
(1.2334) – (0.0155) –
WORLD.GDP − 0.3364** − 1.2505*** − 0.0052 − 0.0006
(0.1576) (0.3565) (0.0051) (0.0007)
COUNTRY.GDP 0.0789** − 0.4085 − 0.0000 0.0006
(0.0400) (0.6038) (0.0001) (0.0015)
FX.RATE − 2.7434 − 11.9223 − 0.0440 0.0590
(5.7880) (15.4031) (0.0437) (0.0360)
BM 0.1696 4.2266 0.0142 0.01448
(0.7817) (4.1805) (0.0123) (0.0121)
HHI 0.0013*** 0.0029** 0.0000 0.0000
(0.0004) (0.0013) (0.0000) (0.0000)
AGE2 − 1.1155 − 7.5056 0.0051 0.0132
(0.8846) (5.0559) (0.0056) (0.0440)
AGE3 − 1.6112* − 10.1790** − 0.0047 0.0108
(0.8869) (4.7878) (0.0036) (0.0416)
REGION2 − 0.8666 11.8344 0.0061 − 0.0130
(0.8088) (11.3711) (0.0053) (0.0381)
REGION3 − 1.6449*** 3.9087 0.0054 − 0.0088
(0.5453) (4.6651) (0.0053) (0.0240)
REGION4 − 3.4922*** − 3.3553 − 0.0056 − 0.0055
(1.2068) 5.8784 (0.0097) (0.0330)
REGION5 − 9.0219*** 0.4236 − 0.0339 − 0.0209
(2.4455) (22.5213) (0.0315) (0.0530)
REGION6 − 3.3465 12.4471 0.0033 − 0.0017
(0.4820) (9.7734) (0.0046) (0.0201)
REGION7 0.5236 2.9770 0.0082 − 0.0045
(0.6154) (4.1444) (0.0067) (0.0155)
Observations 337 273 340 212
No. Instruments N/A 61 N/A 64
No. Airiines 64 64 64 62
R-squared 0.35/0.45/0.35 N/A 0.03/0.20/0.04 N/A
J-Statistic N/A 34.88 N/A 24.10
Arellano-Bond AR(2) N/A − 0.53 N/A − 1.68
The notation is defined in Table 1. *, **, and *** denote statistical significance level of 10, 5 and 1 percent, respectively In the random effects
panel specification, robust clustered standard errors (airline level) are applied and reported in the parentheses and the R-squared values contain
within-group, between-group and overall, respectively. For GMM specifications, robust two-step standard errors, incorporating the Windmeijer
(2005) estimation, are included in the parentheses. The J-Statistic follows a chi-squared distribution with (K-p) degrees of freedom, where K is the
number of moment onditions and p is the estimated parameters; the null hypothesis is that the moment conditions are correctly specified. The
Arellano-Bond test statistic follows an asymptotic normal distribution with the null hypothesis is that no autocorrelation of order q in the dif­
ferenced errors. Region dummies are reported. Year dummies are unreported for the sake of brevity. Model (h) employed deeper lags than Model
(f) and an additional AR(2) term to expunge inherent serial correlation in UP.

References m/sites/andrewcharlton5/2020/03/18/aviation-industry-not-just-airlines-faces
-meltdown/?sh=20eac183496c.
Chin, A.T., Tay, J.H.(, 2001. Developments in air transport: implications on investment
Abate, M., Christidis, P., Purwanto, A., 2020. Government support to airlines in the
decisions, profitability and survival of Asian airlines. J. Air Transport. Manag. 7 (5),
aftermath of the COVID-19 pandemic. J. Air Transport. Manag. 89.
319–330.
Alamdari, F., Fagan, S., 2005. Impact of the adherence to the original low-cost model on
Clougherty, J.A., Zhang, A., 2009. Domestic rivalry and export performance: theory and
the profitability of low-cost airlines. Transport Rev. 25 (3), 377–392.
evidence from international airline markets. Can. J. Econ. 42, 440–468.
Arellano, M., Bover, O., 1995. Another loot at the instrumental variable estimation of
Collins, D.L., Román, F.J., Chan, H.C., 2011. An empirical investigation of the
error-components models. J. Econom. 68 (1), 29–51.
relationship between profitability persistence and firms’ choice of business model:
Boeing, 2016. Current Market Outlook 2016-2035. Retrieved from. https://www.boeing.
evidence from the US airline industry. J. Manag. Account. Res. 23 (1), 37–70.
com/resources/boeingdotcom/commercial/about-our-market/assets/downloads/
Day, A.E., 1986. Impact of exchange rates on air travel. Int. J. Hospit. Manag. 5 (3),
cmo_print_2016_final_updated.pdf.
115–119.
Borenstein, S., 2011. Why can’t US airlines make money? Am. Econ. Rev. 101 (3),
Doganis, R., 2006. The Airline Business. Psychology Press.
233–237.
Doganis, R., 2010. Flying off Course IV: Airline Economics and Marketing. Routledge.
Charlton, A., 2020. Beyond airline crisis, aviation’s infrastructure faces meltdown amid
coronavirus pandemic. Forbes 18 March 2020, accessed at. https://www.forbes.co

284
Y.S.Y. Maung et al. Transport Policy 115 (2022) 275–285

FlightGlobal, 2020. How European states stepped in to support grounded airlines in Mumbower, S., Garrow, L.A., Higgins, M.J., 2014. Estimating flight-level price
2020. Retrieved from. https://www.flightglobal.com/strategy/how-european-states elasticities using online airline data: a first step toward integrating pricing, demand,
-stepped-in-to-support-grounded-airlines-in-2020/141681 (article). and revenue optimization. Transport. Res. Pol. Pract. 66, 196–212.
Francis, G., Dennis, N., Ison, S., Humphreys, I., 2007. The transferability of the low-cost O’Connell, J.F., 2011. The rise of the Arabian Gulf carriers: an insight into the business
model to long-haul airline operations. Tourism Manag. 28 (2), 391–398. model of Emirates Airline. J. Air Transport. Manag. 17 (6), 339–346.
Goddard, J., Tavakoli, M., Wilson, J.O., 2005. Determinants of profitability in European Odeck, J., Brathen, S., 2012. A meta-analysis of DEA and SFA studies of the technical
manufacturing and services: evidence from a dynamic panel model. Appl. Financ. efficiency of seaports. A comparison of fixed and random-effects regression models.
Econ. 15 (18), 1269–1282. Transportation Research Part A: Policy and Practice 46 (10), 1574–1585.
Gössling, S., 2020. Risks, resilience, and pathways to sustainable aviation: a COVID-19 Oum, T.H., Fu, X., Yu, C., 2005. New evidences on airline efficiency and yields: a
perspective. J. Air Transport. Manag. 89, 1–4. comparative analysis of major North American air carriers and its implications.
Hazel, R., 2018. Airline capacity discipline in the US domestic market. J. Air Transport. Transport Pol. 12 (2), 153–164.
Manag. 66, 76–86. Pels, E., 2008. Airline network competition: full-service airlines, low-cost airlines and
Hidayat, S.E., Abduh, M., 2012. Does financial crisis give impacts on Bahrain Islamic long-haul markets. Res. Transport. Econ. 24 (1), 68–74.
banking? A panel regression analysis. Int. J. Econ. Finance 4 (7), 79–87. Pels, E., 2021. Optimality of the hub-spoke system: a review of the literature, and
Holloway, S., 2008. Straight and Level: Practical Airline Economics. Ashgate Publishing. directions for future research. Transport Pol. 104, A1–A10.
IATA Economics, 2021. COVID-19 Airline industry financial outlook update. Retrieved Porter, M.E., 1985. Competitive Advantage: Creating and Sustaining Superior
from. https://www.iata.org/en/iata-repository/publications/economic-reports/air Performance.
line-industry-economic-performance—april-2021—presentation/. Profillidis, V., Botzoris, G., 2015. Air passenger transport and economic activity. J. Air
IATA Economics, 2013. Profitability and the air transport value chain. Retrieved from. Transport. Manag. 49, 23–27.
https://www.iata.org/en/iata-repository/publications/economic-reports/profita Reid, S.R., Mohrfeld, J.W., 1973. Airline size, profitability, mergers and regulation. J. Air
bility-and-the-air-transport-value-chain/. Law Commer. 39, 167.
IATA Economics, 2019. Economic performance of the airline industry. Retrieved from. Roodman, D., 2009. How to do Xtabond2: an introduction to difference and system GMM
https://www.iata.org/contentassets/6240530b9bcb40a6911262588432fbdf/airlin in stata. STATA J. 9 (1), 86–136.
e-industry-economic-performance-jun19-report.pdf. Schultz, E.L., Tan, D.T., Walsh, K.D., 2010. Endogeneity and the corporate governance-
IATA, 2016. Supporting Airline Industry Achievement of Sustained Financial Health. performance relation. Aust. J. Manag. 35 (2), 145–163.
IATA Financial Committee. Retrieved from. https://www.iata.org/contentassets/ea Sorensen, J., 2019. CarTrawler Worldwide Estimate of Ancillary Revenue for 2019.
db6ae846fa4e608bd66772bc684921/iata-financial-committee-white-paper.pdf, Retrieved from. https://www.cartrawler.com/ct/ancillary-revenue/worldwide-anc
2016. illary-revenue-2019.
Ismail, N.A., Jenatabadi, H.S., 2014. The influence of firm age on the relationships of Stalnaker, T., Alport, G., Buchanan, A., Taylor, A., 2019. Airline Economic Analysis
airline performance, economic situation and internal operation. Transport. Res. Pol. 2018-2019. Oliver Wyman. Retrieved from. https://www.oliverwyman.com/content
Pract. 67, 212–224. /dam/oliver-wyman/v2/publications/2019/apr/APRIL262019_Airline_Economic
Kalemba, N., Campa-Planas, F., Hernández-Lara, A.B., Sánchez-Rebull, M.V., 2017. _Analysis_2018-2019vFweb.pdf.
Service quality and economic performance in the US airline business. Aviation 21 Statista, 2020. Number of Scheduled Passengers Boarded by the Global Airline Industry
(3), 102–110. from 2004 to 2020. Retrieved from. https://www.statista.com/statistics/564717/air
Koo, T.T.R., Tan, D., Duval, D.T., 2018. The effect of levels of air service availability on line-industry-passenger-traffic-globally/.
inbound tourism demand from Asia to Australia. In: Airline Economics in Asia Strauss, J., Li, H., Cui, J., 2021. High-speed Rail’s impact on airline demand and air
(Advances in Airline Economics, vol. 7. Emerald Publishing Limited, Bingley, carbon emissions in China. Transport Pol. 109, 85–97.
pp. 145–167. Suau-Sanchez, P., Voltes-Dorta, A., Cugueró-Escofet, N., 2020. An early assessment of the
Kumari, U., 2012. Managerial strategies for enhancing shareholder value in Indian impact of COVID-19 on air transport: just another crisis or the end of aviation as we
aviation firms. Int. J. Manag. Res. Rev. 2 (10), 1841. know it? J. Transport Geogr.
Lau, R.S., Matthesiss, T.H., 1992. A Markov model of the growth and profitability of the Tarry, C., 2015a. Discerning the GDP multiplier effect. Airl. Bus. 31 (7), 52–55.
US airline industry. Logist. Transport Rev. 28 (2), 189. Tarry, C., 2015b. Whether peak or trough, keep it real. Airl. Bus. 31 (5), 52–55.
Manuela Jr., W.S., Rhoades, D.L., Curtis, T., 2019. Market power at the seattle-tacoma Warnock-Smith, D., O’Connell, J.F., Maleki, M., 2017. An analysis of ongoing trends in
international airport: the case of Alaska airlines. Transport Pol. 76, 90–99. airline ancillary revenues. J. Air Transport. Manag. 64, 42–54.
Merkert, R., Morrell, P.S., 2012. Mergers and acquisition in aviation-Management and Wensveen, J.G., 2011. Air Transportation: a Management Perspective, sixth ed. Ashgate
economic perspectives on the size of airlines. Transport. Res. E Logist. Transport. Publishing.
Rev. 48 (4), 853–862. Wintoki, M.B., Linck, J.S., Netter, J.M., 2012. Endogeneity and the dynamics of internal
Merkert, R., Swidan, H., 2019. Flying with(out) a safely net: financial hedging in the corporate governance. J. Financ. Econ. 105 (3), 581–606.
airline industry. Transport. Res. E Logist. Transport. Rev. 127, 206–219. Zou, L., Chen, X., 2017. The effect of code-sharing alliances on airline profitability. J. Air
Mueller, D.C., 1972. A life cycle theory of the firm. J. Ind. Econ. 20 (3), 199–219. Transport. Manag. 58, 50–57.

285