Transportation Research Part A 97 (2017) 30–46

Contents lists available at ScienceDirect

Transportation Research Part A

journal homepage: www.elsevier.com/locate/tra

Leasing and profitability: Empirical evidence from the airline

industry q
Sylvain Bourjade ⇑, Regis Huc, Catherine Muller-Vibes
Toulouse Business School, France

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, we empirically measure the impact of aircraft leasing choices on airlines
Received 16 March 2016 financial performance. We use public data on 73 airlines operating worldwide over the per-
Received in revised form 2 December 2016 iod 1996–2011. In estimating the impact of leasing on profitability, we control for potential
Accepted 5 January 2017
endogeneity by applying robust instrumental variables estimation, while introducing a set
Available online 24 January 2017
of individual and macroeconomic factors. Our results are threefold. First, we identify a non-
monotonic and concave effect of leasing on an airline’s profit margin, suggesting decreas-
JEL codes:
ing marginal returns to leasing in this sector. This is an original finding for the industry. We
C26
G32
also derive a confidence interval for the optimal level of leasing. Second, we show that the
L93 impact of leasing on an airline’s operating profit is stronger for Low Cost Carriers than for
Full Cost Carriers: deviating from the optimal level of leasing might be more harmful for a
Keywords: LCC than for a legacy carrier. Finally, we analyze how an airline’s experience affects the
Leasing relationship between leasing and profitability.
Performance Ó 2017 Elsevier Ltd. All rights reserved.
Profit
Airline Industry

1. Introduction

Airlines’ strategies are mainly driven by decisions on their network, products, prices and resources (specifically their
fleet). As aircraft represent a long lifetime asset (25 years or more) and require a high investment, decisions on the fleet com-
position are the most engaging ones. Operating leases, as opposed to financial leases or ownership, allow airlines to recover
some flexibility when making these fleet decisions.1 Indeed, a leased aircraft can be operated for a period lower than its useful
life, with limited upfront capital requirement. With operating leases, unlike financial (or capital) leases, the ownership risks and
rewards remain with the lessor2 and the assets are off the lessee’s balance sheet. Typical operating and financial leases in the

q
We are grateful to the editor (John Rose), an anonymous referee, Catherine Casamatta, Gilles Chemla, Sveinn Gudmundsson, Alexander Guembel, Marc
Ivaldi, Andreea Mitrache, Sébastien Mitraille, David Stolin, Miguel Urdanoz, Chao Wu and Maxim Zagonov. All errors are ours.
⇑ Corresponding author.
E-mail address: s.bourjade@tbs-education.fr (S. Bourjade).
1
The typical contract length is around 6 years for operating leases and around 15 years for financial leases. See Gavazza (2010), Vasigh et al. (2015) and Oum
et al. (2000).
2
The criteria to classify leases as operating or financial are discussed in the IAS 17 and FAS13/ASC840 accounting standards. Typically, a lease is classified as a
financial lease if: (i) it runs for the major part of the asset life (75% for FAS); or (ii) the total lease payments represent essentially all of (more than 90% for FAS)
the value of the asset; or (iii) there is a bargain or automatic purchase option at the end of the lease. Any other leases are considered as operating leases. This
distinction will vanish with the implementation of the IFRS16 on January 1st, 2019 which will force lessees to recognise assets and liabilities for all leases, i.e.
consider all leases as financial leases.

http://dx.doi.org/10.1016/j.tra.2017.01.001
0965-8564/Ó 2017 Elsevier Ltd. All rights reserved.
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 31

airline industry are called ‘dry’ leases because the lessee only rents the aircraft, as opposed to ‘wet’ leases where also the crew,
maintenance and insurance are rented.3 The lessor community has long argued that leasing has tangible benefits for airlines.
Dick Forsberg, from the Irish lessor Avolon, summarizes the key strengths of lessors in three points: ‘‘They provide liquidity
to weaker credits, make metal available from their order books when manufacturers have sold out, and give airlines flexibility
in terms of their financing mix.”4 This may explain why operating leasing has become an essential means of accessing capacity
for the airline industry. According to Boeing and Centre for Aviation, the proportion of leased aircraft worldwide went from 0.5%
in 1970 to 40% in 2015 and 50% of Airbus’ 2015 deliveries are financed by lessors.
In this paper, we analyze whether operating leasing with its above-mentioned advantages truly allows airlines to improve
their financial performance.
Indeed, while aircraft lessors are thriving,5 the financial performance of the airline industry as a whole has remained rel-
atively low for decades. A McKinsey analysis published by IATA Economics Briefings (2013) shows that airlines have exhibited
the worst average Return on Invested Capital (ROIC) over the 1965–2007 period compared to all other industries. However, even
if the financial performance of the airline industry as a whole has been below average, Fig. 1 documents that a few airlines have
been successful and profitable. Moreover, it is worth noticing that they have achieved those profits while pursuing different
leasing strategies.
Despite this puzzling anecdotal evidence, the literature has only examined the determinants of airlines leasing choices
through their impact on economic efficiency.6 Our objective in this paper is to tackle this issue by analyzing the impact of leas-
ing on airlines’ operating profit margin.
For our empirical analysis we collect data including airlines’ financial information (revenues, operating profits, rental
expenses and fixed assets), as well as additional financial and macroeconomic data. Our sample covers approximately
75% of the total industry revenues for the period 1996–2011.
We characterize the impact of leasing, among other macroeconomics and business factors, on the operating profit margin
using a Two Stage Least Squares (2SLS) regression with instrumental variables. This allows us to take into account the endo-
geneity that could result from the influence of an airline’s profits on its leasing decisions, and some potential unobserved
heterogeneity between airlines.
Estimating an airline’s financial performance based on a reduced form equation is the result of a compromise between
research objectives, data availability and estimation method. It is worth noticing that our main objective is to identify the
impact of leasing on an airline’s profitability, rather than to predict profitability. In that respect, we believe that our approach
provides meaningful and robust results.
The major findings of our paper are the following. First, we identify a non-monotonic and concave relationship between
leasing and profit margin. This allows us to determine the optimal level of leasing which maximizes the operating margin:
53.4%, and the corresponding confidence interval. This result is in line with the industry statistics presented in the first para-
graph. Moreover, the concave effect we find indicates that airlines face decreasing marginal returns to capital investment in
operating lease on their profit margin. The marginal benefits of leasing are thus higher for companies with lower leasing
ratio. This is consistent with the accounting and finance literature that characterizes diminishing marginal returns to invest-
ment.7 Intuitively, firms have incentives to first undertake the most profitable investment opportunities before less profitable
ones. However, while this literature documents a non-increasing relationship between operating performance and capital
investment (and consequently operating leasing), we show that the relationship between those variables is non-monotonic.
Second, we show that the impact of leasing on an airline’s operating margin depends on its business model. We charac-
terize similar patterns (i.e. diminishing marginal returns to leasing and a non-monotonic and concave impact of leasing) for
both Low Cost Carriers (LCCs) and Full Cost Carriers (FCCs). However, the magnitude of this impact is significantly different
for both groups. We show that the benefits of leasing are more important for LCCs than for FCCs and that diminishing mar-
ginal returns to leasing are less pronounced for the latter category. Intuitively, as the aircraft expenses represent a higher
proportion of LCCs’ cost base and, as leasing allows airlines to gain flexibility in their capacity, the impact of leasing is stron-
ger for LCCs. The consequences of a deviation from the optimal level of leasing are therefore more harmful for a LCC than for
a FCC. Our model also allows us to characterize the airlines that could enhance their operating margin by choosing an opti-
mal level of leasing. This may help airlines and lessors to refine their strategy and investors to assess which airlines have the
highest chances to succeed.
Third, we analyze the effect of the combination of an airline’s leasing strategy with its experience. Leasing is more prof-
itable for younger airlines than for long-established ones. Indeed, established airlines are less dependent on their leasing
strategy as their experience allows them to better adapt their behavior to changes in the economic framework.8 Also, estab-
lished airlines may own a greater absolute number of aircraft or variety of aircraft types than younger airlines, allowing for more

3
For more details on leasing strategies, see Mancilla (2010) or Vasigh et al. (2015). Notice that ‘wet’ leases are much less frequently used than ‘dry’ leases.
4
In Air Transport: Aviation International News, 2012.
5
For instance, Ireland based Avolon has been recently acquired by China’s HNA group for $7.6 billion.
6
An exception is Oum et al. (2000). We discuss their results in the literature review.
7
Previous studies have shown that an increase in the invested capital (which partly come from operating leases activities in the airline industry) tends to
make future firm performance lower. Abarbanell and Bushee (1998), Fairfield et al. (2003) and Titman et al. (2004) have documented that when firms increase
their capital expenditures, future earnings and operating performance decrease.
8
See Ismail and Jenatabadi (2014).
32 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

Economic Profit (2002-09, cumulative $ mln) Leased fleet/Total Fleet**

0 500 1000 1500 2000 2500

Emirates Emirates 56%

Ryanair Ryanair 13%

Aeroflot Aeroflot 55%

LAN LAN 40%

Copa Copa* 25%

Air Arabia Air Arabia* 86%

Easyjet Easyjet 38%

Allegiant Allegiant* 21%

Aegean Aegean* 85%

* Copa, Air Arabia and Aegean data for 2004-09 period only, Allegiant 2005-09 period
** Average value of leased fleet (estimated as operating lease expense * 8) divided by
(value of PPE+leased fleet value) over the 2002-2009 period

Fig. 1. Airlines Economic Profit vs. Propensity to Lease (the first part of Fig. 1 comes from the IATA Vision 2050 report (2011) and the second one is
generated with our data, described in Section 3.).

flexibility in the short term when allocating aircraft across routes.9 The additional flexibility brought by leasing is therefore less
profitable for long-established airlines. Moreover, these historical companies may have incentives to favor long-term debt as
they can bargain lower interest rates from financial institutions and, all other things being equal, leasing is generally more
expensive than long-term debt.
This paper is related to the financial economics literature that studies the corporate decisions to lease. The ‘traditional’
theory for leasing is based on tax benefits whereby lessees transferred assets to lessors who could depreciate them in
exchange for lower lease payments (Miller and Upton, 1976). A large part of the literature focused on determining whether
leases and debt are substitutes (Myers et al., 1976; Marston and Harris, 1988; Yan, 2006) or complements (Ang and Peterson,
1984; Lewis and Schallheim, 1992; Eisfeldt and Rampini, 2009). However, as highlighted by Graham et al. (1998), some of
these studies did not adequately differentiate between finance leases and operating leases, even though the relationship
between company debt and lease appears to be different according to the type of lease considered. They also report that
operating and finance leases respectively account for 41.2% and 6.3% of fixed claims, the remaining part being long-term
debt. In our paper, we consider operating leases separately as they represent most lease contracts and they offer more flex-
ibility to airlines with a typical contract length of around 6 years, as opposed to financial leases that run for around
15 years.10
Apart from the traditional tax based incentives, Smith and Wakeman (1985) and Vasigh et al. (2015) for the airline indus-
try discussed other incentives for leasing. If we categorize the benefits from leasing identified by these studies, a lease should
either reduce the cost of capital (generally by lowering risk), or increase returns (by increasing revenues, decreasing costs or
reducing capital requirements). In addition, companies might enter lease contracts without clear financial benefits but due to
management’s incentives. Our paper belongs to this literature that analyzes the non-tax-based incentives for firms to lease.
Despite the special feature of this industry, there are relatively few studies dedicated to leasing in the transportation man-
agement literature. Gavazza (2011) explains that one of the arguments to let specialized entities (the lessors) own the fleet,
is their ability to repossess aircraft more easily at the end of the lease and reallocate the capacity with lower friction costs.
Gavazza (2010) shows (i) that leased aircraft are shed faster than owned aircraft (ii) that high volatility firms (i.e. firms who
value flexibility more) tend to lease more than others and (iii) that leased aircraft are typically more productive than owned
aircraft.
Another leasing advantage is that it may lower business risks by providing operators with additional flexibility in their
capacity management. Leasing allows airlines to modify their network in range (for instance, an airline positioned only
on short haul routes with narrowbody aircraft could lease widebody aircraft to operate long-haul routes) or in gauge (by
changing capacity on individual routes to follow changes in demand). More generally, through leasing airlines could reduce
capacity in downturns by not renewing leases that are expiring,11 and could add capacity in upturns faster instead of facing

9
We thank an anonymous referee for suggesting us this interpretation.
10
See Footnote 1 for references.
11
For example, if an airline leases 30% of its fleet (under 6-year operating leases contracts) it can reduce its fleet by 5% each year by not renewing an expiring
lease, assuming the leases expiry dates are uniformly distributed over the coming 6 years.
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 33

longer leadtimes for new deliveries. Moreover, leasing allows airlines that suffer from capital budgeting or credit restrictions to
obtain better or faster access to new aircraft. Bjelicic (2012) shows that this is particularly true for younger companies and LCCs
as they can then access additional capacity and accelerate their development. Leasing can also contribute to reduce costs, in
particular those associated with disposing of the asset. As Shleifer and Vishny (1992, p. 1355) noted: ‘‘The institution of airline
leasing seems to be designed partly to avoid fire sales of assets: airlines can stop their leasing contract when they lose money
rather than dump airplanes on the market which has no debt capacity.”
Our paper is also related to Ismail and Jenatabadi (2014) who study the impact of airlines age on the interaction between
the economic situation, internal operations and performance. The influence of the leasing strategy which is a particularly
important internal operation variable is however not investigated in their paper. Our results are therefore complementary
to theirs as we show that the effects of leasing on performance crucially depend on their age.
Finally, while the fleet is an essential part of an airline’s business model, studies about airlines’ performance typically do
not include leasing as a variable, and relatively few studies attempt to explain the impact on financial performance. Indeed,
the literature dedicated to the analysis of airline performance12 focuses on efficiency and airlines’ productivity. Schefczyk
(1993), Gittell et al. (2004) and Barros and Peypoch (2009) attempt to explain profits, however they do not consider the type
of financing used (e.g. lease or buy). An exception is Oum et al. (2000) who compute the optimal demand for leased capacity
from a sample of 23 firms over the 1986–93 period. Our approach differs from theirs in two ways. First, while they assume that
an airline’s profit is concave in its capacity, we let the data generate this concave relationship and prove it is non-monotonic. We
thus show that an optimal level of leasing exists and we estimate a confidence interval for this optimal level. Second, since we
have access to data for 73 airlines for a longer and more recent period, we are able to characterize some additional features of
airlines’ leasing strategies.
Our paper is organized as follows. Section 2 presents the database, Section 3 describes our empirical model, Section 4
comments on the results of our estimations. Finally, Section 5 concludes and additional results are displayed in Appendices
A and B.

2. The data

2.1. Our sample

For this paper, we build a database containing financial data for 73 international airlines. Our sample covers a represen-
tative portion of the global airline industry (approximately 75% of the industry’s total yearly revenues over the period) and
includes listed and unlisted airlines from all regions, active and inactive firms (i.e. acquired or bankrupt), and a mix of the
business models known as FCCs and LCCs. FCCs correspond to the traditional airline business model, frequently called legacy
airlines, such as American Airlines or Air France. They are not necessarily ‘historical’ carriers as some FCCs entered the mar-
ket in the 1990s’, like Jet Airways. The low cost business model for airlines focusses on strongly reducing costs in order to
offer significantly lower prices to customers. For example, LCCs operate on point-to-point routes or offer limited passenger
services. The emergence of this new type of airlines has strongly increased competition in the past decades. However, this
business model does not seem to be a guarantee for success, as a great variance in performance for these LCCs is observed in
the industry. There is no standard definition for a LCC though, as it incorporates a large range of airlines operating in different
markets. Consequently, we followed the classification provided by ICAO (2014) to distinguish FCCs from LCCs in our sample.
Our database covers approximately two industry cycles (16 years, between 1996 and 2011). In total, we collected data for
868 company-years. Our database includes the following airlines13:

& North America: Air Canada, AirTran Airways, Alaska Air, Allegiant Travel, America West, American Airlines, Continental
Airlines, Delta Air Lines, Frontier, Hawaiian, JetBlue Airways, Northwest, Southwest Airlines, Spirit Airlines, United Airlines,
US Airways, WestJet
& Latin America: Aeromexico, Avianca TACA, COPA Holdings SA, GOL Linhas Aereas Inteligent, LAN, TAM Linhas Aereas,
VARIG
& Europe: Aegean Airlines, Aer Lingus, Aeroflot, Air Berlin, Air France, KLM, Alitalia, Austrian Airlines, British Airways, Iberia,
Easyjet, Finnair, Flybe, Lufthansa, Norwegian Air Shuttle, Ryanair, Scandinavian Airlines, Sky Europe, Swiss International,
Swissair, Turkish Airlines, Vueling
& Middle East: Air Arabia, El Al Israel Airlines, Emirates, Jazeera Airways, Royal Jordanian
& Africa: Ethiopian Airlines, Kenya Airways, South African Airways
& Asia Pacific: Air Asia, Air New Zealand, All Nippon Airways, Asiana, Cathay Pacific, China Airlines, China Southern, China
Eastern, EVA Airways, Japan Airlines, Jet Airways, Korean Airlines, Malaysian Airlines, Qantas, Shenzhen Airlines, Singa-
pore Airlines, Skymark, Thai Airways, Tiger Airways, Virgin Blue.

12
For a recent summary of the literature related to airline performance seeLi et al. (2015)
13
Airlines identified as LCCs are italicized.
34 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

In our analysis, we examine the effect of leasing on an airline’s financial performance. As a measure of airline’s profits we
focus on EBITDAR (Earnings Before Interest Taxes Depreciation Amortization and Rentals) margin, i.e. the ratio of EBITDAR
divided by total revenues. Since assets under an operating lease are a substitute (operationally) to owned aircraft, we need to
consider profits before the aircraft cost is deducted (Depreciation Expenses or Rentals). Thus, we use EBITDAR as our main
measure of profits since it captures the profits from operations before the buy or lease decision affects the financial variables.
It is also worth noticing that practitioners frequently use EBITDAR to assess the operating performance of a company. More-
over, as EBITDAR is affected by the size of a company, we normalize it by dividing it by revenues.14 This ratio therefore allows
us to consider a measure of operating efficiency and profitability that is not affected by the choice of aircraft financing alterna-
tives or the size of the airlines. For each airline, we gather information from the income statement and notes to the financial
statements on revenues, depreciation and amortization (D&A) expenses, EBIT (Earnings Before Interest and Taxes), interest
expenses and operating lease expenses (Rentals). We calculate EBITDAR by taking the sum of EBIT, D&A expenses and rental
expenses.
To assess the importance of leasing in an airline’s fleet we estimate the proportion of leased assets in the total assets by
dividing the leased fleet value by the sum of a company’s property plant and equipment (PP&E) and the value of the leased
fleet. Indeed, for airlines, the value of other assets in the PP&E is negligible relative to the value of the purchased aircraft.
Therefore, we believe we can confidently use PP&E to identify the value of the purchased fleet in the leasing ratio. To esti-
mate the value of a leased fleet, finance practitioners multiply the annual rental expenses by a capitalization factor. This cap-
italization factor depends on the length of the depreciation period and the financing rate. If we apply the standard
depreciation period and financing rates for this industry (respectively 15–20 years and 6–9%), this capitalization factor
stands between 7 and 8.15 We choose to apply a factor of 8, as recommended by Moody’s (2006).16
We collect financial items in local currency from financial databases (Bloomberg, Infinancials and Thomson Reuters
Eikon) as well as companies’ annual reports and DOT Form 41. We use the exchange rate at the end of the year (December
31) to convert local currency values into United States Dollars (USD). A rigorous analysis has been performed to reconcile the
discrepancies between the different sources and find the missing items. The reported EBIT has been adjusted when it
included non-operating items (e.g. one off items such as gains from asset sales, financial items. . .) to obtain a consistent
operating profit. Moreover, when operating lease expenses were not reported explicitly in the income statement (or in
the notes detailing the operating expenses) we used the amount corresponding to the following year’s operating lease
commitments (available in the notes, as part of the ‘‘other commitments” section). As rental expenses and minimum lease
commitments are not often available in the financial databases, using annual reports is almost the only option to gather this
information.
Macroeconomic data for each of the countries in our sample come from the US Department of Agriculture Research
service (GDP growth, population). The corporate marginal tax rates are from the KPMG corporate tax surveys and finally
we collect companies foundation dates from companies’ websites.

2.2. Descriptive statistics

Some descriptive statistics are presented below, underlying the heterogeneous behavior of airlines with respect to their
leasing decisions, depending on their experience or business model. Appendix A provides some detailed information on our
dataset while Table 1 shows descriptive statistics, for carrier-year observations, of our measures of operating margin and
leasing ratio, by business model.
From Table 1, we observe that there is a significant variation of both profitability and leasing within each business model
and across the business models. To obtain a visual impression of the differences across the two business models, we plot the
distribution of leasing for both LCCs and FCCs in Fig. 2.
We clearly see on these box-plots that more LCCs tend to choose higher levels of leasing than FCCs and that these leasing
amounts are more spread around the mean for LCCs than for FCCs. We run a mean comparison t-test on these two subsam-
ples and we find that the means for leasing in the two groups are significantly different from each other.17 This result moti-
vates our analysis to better assess what the impact of leasing decisions is on airlines’ profitability. We also distinguish the
distribution of leasing according to age groups, defined as the quartiles from our sample, and observe some heterogeneity in
the leasing choices. This is depicted in Fig. 3:
We observe that the youngest airlines choose higher levels of leasing. This could be explained by their need to access
capacity to operate a network, while not being able to purchase a large proportion of their fleet. In addition, oldest airlines
tend to use more homogeneous leasing strategies, which might be due to organizational rigidities.
Fig. 4 the distribution of airlines by age within each business model. We see that companies older than the median are
FCCs, whereas companies younger than the median are a homogenous mix of FCCs and LCCs.

14
This also has the advantage of neutralizing the size effect of mergers.
15
For a detailed explanation of this capitalization factor, see Koller et al. (2010).
16
We perform a robustness check on the choice of the capitalization factor, by running the same estimations with a value of 7, and the results do not change.
17
The cross on both box-plots represents the mean of each subsample.
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 35

Table 1
Descriptive statistics for profit margin and leasing ratio.

Business Model Mean Standard Deviation Minimum Maximum 1st Quartile Median 3rd Quartile
EBITDAR/Sales FCCs (21 airlines) 17% 8%
13% 45% 12% 17% 22%
Lease 42% 22% 2% 97% 26% 40% 57%
EBITDAR/Sales LCCs (52 airlines) 20% 11% -32% 48% 14% 19% 27%
Lease 54% 31% 0% 100% 27% 50% 82%

Fig. 2. Distribution of Leasing for LCCs and FCCs.

Fig. 3. Distribution of Leasing across Airlines Experience (Age).

36 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

Fig. 4. Distributions of Airlines Age per Business Model.

3. Empirical model

In this section, our objective is to identify and measure the impact of airlines’ leasing strategies on their profitability, con-
trolling for the main features of worldwide air transport, at the airline level, for a 15-year period. The above mentioned lit-
erature identifies diminishing marginal returns to investment. As in the airline industry, a significant share of the
investments is financed through operating leases, we can expect diminishing marginal returns to leasing in this industry.
Additionally, even if this literature shows this effect is non-increasing, because of the existing trade-off between the cost
premium of lease and its flexibility benefits, we investigate whether the impact of leasing on profitability is non-
monotonic and concave, with an optimal level for leasing. This discussion leads to the following hypothesis:

Hypothesis 1. The impact of leasing on profitability is non-monotonic and concave.

Moreover, the descriptive statistics highlight differences in leasing strategies among business models and airlines’ expe-
rience. We are thus interested in testing whether these different strategies lead to distinct impacts of leasing on profitability.
We therefore test the two following hypotheses:

Hypothesis 2. The impact of leasing on profitability is different depending on an airline’s business model.

Hypothesis 3. The impact of leasing on profitability is different depending on an airline’s experience.

In the remainder, we first introduce the regression analysis based on a reduced form profit margin equation in which the
yearly profit margin for each airline is related to the amount of leasing and a set of control variables of interest. Then we
propose alternative specifications to test the aforementioned hypotheses.

3.1. Leasing and diminishing marginal returns

The equation we first fit on this dataset is a reduced form that is based on an economic model even though it does not
explain the full behavior of airlines. The theoretical foundations of our model are the following. When airlines make their
lease or purchase decisions, they face a trade-off between the gains from more flexibility through leasing and the cost pre-
mium of leases. These decisions are equivalent to making a dedicated or flexible irreversible investment or hire a certain
proportion of temporary and permanent workers. There are relatively few studies dedicated to formal models of leasing
in the transportation management literature, but the literature on irreversible investment and temporary work has more
extensively studied this trade-off (see Dixit and Pindyck, 1994, Boeri and Garibaldi, 2007, or Goyal and Netessine, 2007
among others). We derive our reduced form model from this literature. Airlines should purchase aircraft when there is little
uncertainty (which is true for some of their routes), as this is less costly and no flexibility is required. However, for the
remaining routes on which there is higher uncertainty, leasing becomes more attractive even though more expensive, as
it provides more flexibility to airlines. Airlines should then only own part of their aircraft because of uncertainty.
Our specification is based on these economics models. It results from the selection of variables that are reasonable can-
didates for explaining the levels of operational profit margins. We thus design our model to allow for a potential concave and
non-monotonic impact of leasing on profitability.
Note that we focus here on the specific impact of leasing while controlling for other relevant factors affecting profitability.
Our model is written in Eq. (1) below:

pmit ¼ a þ bL Leaseit þ bL2 Lease2it þ X it b þ eit ð1Þ

-212090050800-2242185125095-1778000123825-361505595885-198818550165where the subscripts i and t refer to an
airline and a period respectively. pmit is the profit margin defined as the ratio of EBITDAR/Sales. EBITDAR is a measure of
profits before including depreciation, amortization and rentals. Lease measures the proportion of leasing expenses and is
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 37

computed as the share of leased assets to the total assets. The profit margin and leasing variables introduced in this model
are described in details in Section 2. We model potential diminishing marginal returns to leasing by introducing the variable
Lease2it . bL and bL2 are parameters to be estimated. We then control for firm specific characteristics, macroeconomic factors,
demographic conditions and time effects, included in the set of variables X it . b is a vector of parameters to be estimated. The
variables in X it are the following:
 Low Cost Carrieri, a dummy variable taking the value of 1 if the airline is operating under the Low Cost business model
and 0 otherwise. As suggested in the literature, LCCs tend to have different leasing behaviors, and their performance tends
to be higher.
 Taxit, which stands for the marginal corporate tax rate in the origin country. The inclusion of this variable is motivated by
the academic literature18 justifying the use of leasing based on tax incentives.
 Competitionit, which measures the density of competitors within the following six geographical zones: Europe, Asia-
Pacific, Latin America, North America, the Middle East and Africa. It is computed as the number of airlines in our dataset
based in the same region as airline i at period t.19
 Trendt, a time trend, capturing the constant increase of worldwide traffic.
 Crisist, a dummy variable taking the value 1 for the crisis years 2001, 2002, 2003, 2008, 2009, and 0 otherwise. This vari-
able allows us to model the industry’s cyclicality and the abrupt profit variations during crises.
 Chapter 11it, a dummy variable taking the value 1 for the airlines under the protection of Chapter 11, e.g. US bankruptcy
reorganization. This variable helps us explain temporary lower performance, especially for the relatively large US major
carriers.
 GDP Growthit, represents the country annual variation of the GDP, in percent. We control for GDP growth as there might
be periods when renting is more attractive because of market conditions, such as periods of high economic growth.
 Oil Pricet, the annual average oil price in USD per barrel.
 Europei, Asia-Pacifici, Latin Americai, North Americai, Middle-Easti, a set of dummy variables taking the value 1 if an
airline is based respectively in Europe, the Asia-Pacific region, Latin America, North America or the Middle East, and 0
otherwise. Note that the omitted dummy here is the one referring to airlines based in Africa.

The parameters of our regressions are estimated by applying the Two Stages Least Squares technique. This is the appro-
priate estimation method to control for potential endogeneity in one or more of the regressors. In our setting, the variable
Lease might not be fully exogenous as we cannot reject the assumption that firms would derive their leasing strategy based
on their profitability. In the financial economics literature, Ang and Peterson (1984), who document that debt and leasing are
complements, state that firms with low profits and high growth should lease more. The technique of instrumental variables
is also appropriate for correcting the bias resulting from potential omitted variables in our model, for instance individual
airline effects.20 We thus solve for this issue of endogeneity and insure the causal relationship specified in our model by using
instrumental variables for the potential endogenous right-hand side variables Lease and Lease2. To be valid candidates, the
instruments should be correlated to the operational profit margin, but only through the suspected endogenous explanatory
variable Lease, not directly. The instrumental variables we use in our estimations are first macroeconomic indicators. Country
population and country population square capture country development and traffic, hence the demand for leased aircraft.
Indeed, the demand for leased aircraft is increasing with traffic, which itself depends on a country’s population and propensity
to fly.21 We introduce a dummy variable characterizing companies younger than 5 years of age to isolate the specific necessity
for new entrants to rent a fleet of aircraft to launch their business. We also build an instrument that is computed as the average
level of the variable Lease at period t. In order to avoid spurious correlations between our explained variable and this instru-
ment, we compute it for each airline i as the average over all other airlines at period t, worldwide. This variable is assumed
to pick up leasing supply shocks resulting for instance from the availability of leased aircraft fleet, macroeconomic and risk fac-
tors, as well as lessors and aircraft manufacturers’ strategies, which would impact all airlines’ leasing decisions without affect-
ing directly the profitability of airline i.
In order to validate this set of instruments we need to check they satisfy two requirements: they must be correlated with
the endogenous explanatory variables and they must be exogenous (uncorrelated with the structural error). We first check
the explanatory power of the instruments by looking at the F-statistic of the first stage regressions, namely the regressions of
the endogenous right-hand side variables on the instruments. The 2SLS estimator is considered reliable when the F-statistic
exceeds 10 as suggested in Staiger and Stock (1997). The F-statistics for the first stage regressions of Lease and Lease2it on the
instruments are respectively 14.04 and 12.71 suggesting that weak identification should not be a problem here. We also run
a test based on the Kleibergen-Paap Wald F-statistic,22 which is more appropriate in settings with several endogenous
explanatory variables and the presence of heteroscedasticity.23 The value of this statistic for our model estimation is 11.91,

18
See Miller and Upton (1976), Myers et al. (1976) or Ang and Peterson (1984) for instance.
19
With no additional information on airlines’ networks, it would not be relevant here to compute more sophisticated proxies for competition, like an HHI
index.
20
Note that this issue is further addressed at the end of this section. We present a robustness check that consists in running a fixed effects regression.
21
See Vasigh et al. (2012) and Airbus GMF (2016) for a more detailed explanation of this relationship.
22
SeeKleibergen and Paap (2006).
23
We suspect the presence of heteroscedasticity and correct for it, as explained in the following paragraph.
38 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

and by applying the critical values compiled by Stock and Yogo (2005), we can be confident we are not in the presence of so-
called weak instruments (the IV relative bias being comprised between 5% and 10%). Second, not only the instruments have to
be correlated with the endogenous right-hand side variables, but also they need to be exogenous, that is correlated with the
explained variable but only through the endogenous explanatory variables. We perform the appropriate Hansen J test for overi-
dentifying restrictions to check whether the instruments are uncorrelated with the structural error term. In other words, we test
whether none or at least one instrumental variable from outside the model is invalid (because not exogenous). For this test, the
null hypothesis is that all instrumental variables are uncorrelated with the error term and the test statistic is equal to 1.550 and
should be compared to the value of a v2q , with q the number of instruments minus the number of endogenous regressors. The p-
value for this test at the 5% level is equal to 0.4606 so we cannot reject the null hypothesis of our instruments being uncorrelated
with the error term, which means our set of instruments is valid for the estimation of our model. Third, deviating from this set of
instruments leads to statistically weaker results and less meaningful economic interpretation for the parameters. Note that
these tests are also run for the additional estimations we present and they validate the chosen set of instruments each time.
Lastly, the assumption of homoscedasticity may not be appropriate for our operating margin specification, as some unob-
servable factors like network size and topology or intensity of competition at the route level are included in the error term.
The error variance may not be constant. In theory, the presence of heteroscedasticity does not affect too much the magnitude
of the estimates which are obtained with the assumption of constant variance. It may affect the standard deviation of these
estimated parameters more significantly though. We therefore propose an alternative estimation procedure which calculates
robust standard errors in order to circumvent the potential presence of heteroscedasticity. This procedure is known as the
heteroscedastic consistent-covariance matrix estimator (HCCME) recommended by MacKinnon and White (1985).
The data analysis and estimations are conducted with the Stata software. In the remainder, we show that the leasing and
ownership trade-off can be further decomposed according to the business model and experience of an airline. We introduce
some alternative specifications of our model aiming at identifying these industry specificities regarding leasing strategies.

3.2. Leasing and business model

As presented in the introduction, airlines can be characterized by two business models called the Low Cost and Full Cost
models. To categorize companies as LCCs we follow the generally accepted ICAO (2014) classification and include low fares
companies that mainly operate single aisle aircraft in point to point domestic or regional markets. We construct dummy
variables characterizing these two groups and we introduce them in Eq. (2) to form interaction terms with the variables Lease
and Lease2. This specification allows us to highlight separate effects of the leasing strategy on profitability, depending on the
airline business model. The model is written in Eq. (2):

pmit ¼ a þ bLFCC Leaseit FCC it þ bL2FCC Lease2it FCC it þ bLLCC Leaseit LCC it þ bL2LCC Lease2it LCC it þ X it b þ eit ð2Þ
The variable LCC takes the value 1 if the airline is a LCC and 0 otherwise; the variable FCC takes the value 1 if the airline is a
FCC and 0 otherwise. The variables in X are the same as in the previous estimations (only the variable LCC has been removed).
Note that we use the same set of instruments as in the previous regression plus some additional ones as the number of
endogenous explanatory variables increases. We use the same instrument on average of leasing for all other airlines as in
the first regression except we split it between LCCs and FCCs, as it is common practice when an endogenous right-hand side
variable is interacted with categorical dummies. We also use the average level of the variable Lease for all other airlines but
this time in the same geographical zone (as they were defined in the geographical dummies), at period t, and also interacted
with the business model dummies. Finally, we add a variable measuring GDP per capita that influences airlines leasing
behavior through a country’s traffic and propensity to fly.24

3.3. Leasing and experience

In a third regression, we confirm that the share of leased assets over total assets has a different impact according to the
development stage of an airline. The rationale for distinguishing airlines based on their age is that it might have an impact on
organizational efficiency (for instance older organizations can be more prone to exhibit higher labor costs especially in a
unionized industry). Building on the results of our first regression, we test the hypothesis that when long-established airlines
have a choice between owning and leasing (e.g. access to financial resources) and have sufficient scale, leasing should be less
profitable for them. Younger airlines are expected to rent because leasing is their only alternative to access new capacity, no
matter the return to leasing investment. One conventional way of modelling this effect of age would be to introduce cross
effects between leasing variables and a variable measuring the age of each airline, with either a linear or a quadratic function.
The estimation of such a model does not lead to concluding results, which is not surprising given we do not expect the effect
of age on profitability to be continuous but rather more like a step function. We are then interested in distinguishing the
effect of leasing for different categories of airlines classified according to their maturity. We distinguish between long-
established airlines and younger ones to account for loan accessibility and risk level for financial institutions. We thus sep-
arate our sample into two groups according to the median for age (as in Ismail and Jenatabadi, 2014), which is 45 years-old.

24
See Footnote 22 for appropriate references.
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 39

The categories are modelled with dummy variables and are interacted with the variable Lease and Lease2 in our equation to
identify separate effects of leasing on profitability depending on whether an airline is younger or established. We estimate
the model specified in Eq. (3):

pmit ¼ a þ bLYoung Leaseit Young it þ bL2Young Lease2it Young it þ bLEst Leaseit Establishedit þ bL2Est Lease2it Establishedit þ X it b
þ eit ð3Þ
The variables Young and Established are two dummy variables categorizing airlines according to their maturity. The vari-
able Young takes the value 1 if the airline entered the market less than 45 years before t and 0 otherwise; the variable Estab-
lished takes the value 1 if the airline entered the market more than 45 years before period t and 0 otherwise. As in Eq. (1), we
design the model to allow for potential diminishing marginal returns to leasing, by interacting the dummies with the vari-
ables Lease and Lease2. The variables in X and the instruments are the same as in Eq. (2), except we interact them with age
dummies and not business model dummies. We introduce the average of leasing for all other airlines within a business
model, also interacted with the age dummies, and one additional instrument computed as the lease expenses for airline i
at period t divided by the sum of deliveries from Airbus and Boeing in year t. This variable is an indicator for the aircraft
supply constraint on the market.
We now present and discuss the results, before introducing some robustness checks.

4. The results

Three main results validating our hypotheses emerge clearly. First, the regressions we perform provide evidence of a glo-
bal significant effect of leasing on airlines profit margins, showing that the industry is characterized by diminishing marginal
returns to leasing expenses. We are able to estimate a confidence interval for the average amount of leasing maximizing
profit. Second, our estimations show that the effect of leasing is characterized by different magnitudes depending on the
business model chosen by an airline, all else being equal. More specifically, LCCs seem to benefit more from leasing than
FCCs. However, diminishing marginal returns to leasing are shown to be stronger for the LCCs. Third, we find that the impact
of leasing on an airline’s profitability depends on its operational experience.
Our results for the estimations of Eqs. (1)–(3) are presented in Table 2.

4.1. Leasing and diminishing marginal returns

The focus of the first regression is the impact of leasing on profitability. The parameters for Lease and Lease2 are signif-
icant, implying that leasing is not only a solution to access additional aircraft capacity, but also a profit-maximizing variable.
The quadratic specification of the relationship between these variables and the signs of the estimated parameters allow us to
conclude that the impact of leasing on profit margin is non-monotonic and concave. This strongly supports the presence of
diminishing marginal returns to leasing. Airlines with a lower leasing ratio should therefore have more incentives to lease.
Indeed, leasing may lead to increased margins. This suggests that the rented capacity is operated profitably, despite its higher
cost, and therefore that the decision to rent additional aircraft is economically motivated. Intuitively, leasing provides flex-
ibility by allowing airlines to add and remove capacity when needed (e.g. by not renewing a lease agreement, the lessee has
the option to reduce its capacity without incurring costs linked to the disposal of its own capacity). This may also explain
why the leasing industry has developed tremendously in the last twenty years. However, increasing the proportion of leased
assets becomes marginally less and less profitable, up to the point where increasing the leasing ratio may even reduce the
operating margin. This result is consistent with previous studies25 which also characterize a diminishing marginal return to
the invested capital (and therefore leasing) due to the incentives for companies to invest as a priority in the most profitable
opportunities. However, while these studies have documented this effect was non-increasing, we show that it is non-
monotonic in the airline industry.
We are therefore able to compute an optimal level of leasing that maximizes the operating margin, equal to 53.4%, as well
as a 5% confidence interval of [0.468; 0.601] for this optimal level:

b
Max pm () dpm
¼ 0 () Lease ¼ bL
¼ 53:4%
2b
Lease dLease
b L2

Confidence Interval for Lease : IC Lease ¼ ½Lease  t 1
a;df s:e:ðLease Þ ¼ ½0:534  1:96  0:0339
IC Lease ¼ ½0:468; 0:601
The standard error for this non-linear function of estimated parameters is computed by applying the Delta Method. Inter-
estingly, this confidence interval is above the 40% average leasing ratio that we observe for the airline industry. Moreover,
the 50% of Airbus deliveries to lessors belongs to this interval. This result is consistent with Oum et al. (2000) who suggest
that the optimal leasing ratio for airlines should range between 40% and 60% of their total fleet.

25
See Abarbanell and Bushee (1998), Titman et al. (2004) and Fairfield et al. (2003) for instance.
40 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

Table 2
2SLS Regression estimates, operating leases expenses.

Model Operating leases Operating leases and business Operating leases and carriers’
model experience
Parameter estimate t-value Parameter estimate t-value Parameter estimate t-value
***
Constant 0.044 0.88 0.086 1.53 0.106 3.06
Lease 0.892*** 3.22 – – – –
Lease2
0.835***
2.98 – – – –
LeaseFCC – – 0.680** 1.94 – –
Lease2FCC – –
0.624*
1.70 – –
LeaseLCC – – 1.151*** 3.08 – –
Lease2LCC – –
1.064***
3.13 – –
LeaseYoung – – – – 0.665*** 3.63
Lease2Young – – – –
0.645***
3.31
LeaseEstablished – – – – 0.485** 2.50
Lease2Established – – – –
0.440**
1.99
Low Cost Carrier 0.085*** 4.76 – – 0.054*** 3.60
Competition
0.009***
4.74
0.009***
5.12
0.008***
5.09
Tax
0.216***
3.27
0.203***
2.61
0.164***
2.99
Trend 0.006*** 2.93 0.006*** 2.81 0.006*** 3.61
Crisis
0.023***
3.40
0.022***
3.08
0.026***
4.06
Chapter 11
0.104***
4.72
0.104***
4.66
0.106***
4.99
GDP Growth 0.467*** 4.22 0.539*** 4.79 0.426*** 4.21
Oil Price
0.001***
4.02
0.001***
4.23
0.001***
4.39
Europe 0.099*** 3.39 0.102*** 3.50 0.094*** 3.62
Asia-Pacific 0.181*** 5.70 0.178*** 5.40 0.156*** 5.68
Latin America 0.063*** 2.60 0.048 1.63 0.049** 2.09
North America 0.100*** 3.76 0.098*** 3.86 0.099*** 4.29
Middle-East 0.072*** 3.44 0.068*** 2.92 0.052*** 3.00
*
Significant at 10% level.
**
Significant at 5% level.
***
Significant at 1% level.

Our result also suggests that airlines, as soon as capacity can be operated profitably across the cycle, should purchase a
high enough proportion of their fleet instead of renting it since leasing is typically more expensive. However, this proportion
should not be too high because leasing provides valuable flexibility for operating the most risky routes. Thus when airlines
choose their leasing strategy there is a trade-off between the fleet costs and the flexibility leasing provides.

4.2. Leasing and business model

Consistently with the findings of our first model, the results from the estimation of Eq. (2) on the distinction between
business models confirm that diminishing marginal returns to leasing are present for both categories of airlines. The param-
eters for these new interacted variables are statistically significant.26 We first compute the optimal level of operating leasing
for both categories of airlines and we find they are not significantly different from each other.27 However, the impact of oper-
ating leases on the profit margin is significantly different for both groups, at the 5% level. This is confirmed by the Wald tests
comparing bLFCC with bLLCC and bL2FCC with bL2LCC .
Interpreting and comparing the magnitudes of the parameters is not straightforward as the quadratic specification
implies the effect of the variable Lease on the profitability is a function of the variable Lease itself. In order to better explain
and present this effect, we draw in Fig. 5 the relationship between the operating margin and the proportion of operating
lease expenses, according to our estimated parameters.
Intuitively, LCCs may be more vulnerable to macroeconomic shocks due to their business model. Moreover, due to these
differences in cost structures between LCCs and FCCs, the aircraft expenses represent a higher proportion of their cost base
and therefore have a higher relative impact on their profit margin. The leasing benefits of flexibility are then more decisive
for them than for FCCs. Our model predicts that the difference in the impact of leasing between LCCs and FCCs is stronger
when they both choose their optimal level of leasing. Behaving optimally is relatively more profitable for LCCs. We also find
that the curve for FCCs is flatter: FCCs are less sensitive to changes in the level of leasing. Diminishing marginal returns to
leasing are less pronounced for them. Our results imply the strategic stakes of leasing decisions are higher for LCCs, in terms
of financial performance. They should therefore make those decisions very carefully.
In order to illustrate this result, we compute the number of airlines for which the leasing ratio in 2011 was outside a 1%
confidence interval for the optimal leasing ratio. This confidence interval is [46.1%; 62.0%] for LCCs and [36.5%; 72.5%] for
FCCs. Reducing the proportion of leasing expenses may not be a credible option for airlines with a leasing ratio above the

26
The parameter for the variable Lease2 FCC is significant at the 10% level when the others are all significant at the 5% or 1% level.
27
Wald test performed with a confidence level of 5%.
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 41

Operating Margin

0 100 Proportion of Operating

Leases Expenses

Fig. 5. Effect of leasing on Operating Margin for LCCs and FCCs.

highest bound of the confidence interval. Indeed, purchasing aircraft through debt may be too costly or even impossible for
them because of their financial situation. However, it seems easier for airlines with a low leasing ratio to increase this ratio.
In our sample, 10 out of 17 LCCs and 18 out of 39 FCCs have a leasing ratio in 2011 below the lowest bound of the confidence
interval (46.1% and 36.5% respectively). Those airlines, which represent 50% of our sample in 2011, could thus get an addi-
tional margin by increasing the proportion of leasing expenses up to the estimated optimal levels.

5. Leasing and experience

In our third regression estimating the interaction of leasing on experience, the values of the parameters bLYoung , bLEst ,
bL2Young and bL2Est allow us to interpret the combined effect of experience and leasing on the operational profit margin. First,
these parameters are significant at the 5% level. Moreover, as bLYoung and bLEst are positive and bL2Young and bL2Est are negative,
our result confirms a concave and non-monotonic relationship between leasing and profitability for both younger and estab-
lished airlines. The Wald tests comparing bLYoung with bLEst confirm this difference is statistically significant at the 5% level.
Note that the parameters for the squared terms are not statistically different from each other.28 Finally, as bLEst presents a
lower magnitude than bLYoung , the impact of operating leases on the profit margin is significantly lower for established compa-
nies than for younger ones.
This result is in line with the literature on the effect of age on performance. Indeed, through organizational rigidities, age
may increase the costs of flexibility and change investment strategies.29 In our sample, long-established firms were created
before the emergence of the leasing industry, which may explain partly the lower impact of leasing on profit margin for these
airlines. Moreover, Pástor and Pietro (2003) show that firms benefit from learning effects with age and therefore face lower
uncertainty on future profits. The flexibility benefits of leasing for long-established airlines are therefore lower compared to
younger companies. Another interpretation for this finding is that long-established airlines typically have a larger and more
diverse fleet. This may allow them to benefit from more flexibility for short-term expansion or substitution of aircraft across
routes without having to lease additional aircraft types. Younger airlines, with smaller and less-diverse fleets, may need to rely
more on leasing to quickly respond to changes in demand or economic conditions. Moreover, as younger airlines usually have
lower operating costs due to less-senior crew and staff, the probability of leasing an aircraft more profitably increases. Addition-
ally, these historical airlines may have access to relatively better financing conditions from banks and may benefit from deals
with aircraft manufacturers that are more favorable. Younger airlines do not benefit from the same reputation effects. Acquiring
aircraft through debt may be more complex and expensive for them, in particular when they are in a poor financial situation or
at early stages of their operations. For similar reasons, even if the leasing costs are likely to be higher, this could be the only
option for them to launch their business. The marginal impact of leasing is therefore smaller for them.
Finally, as we find that leasing is more profitable for LCCs and for younger airlines, and as LCCs are young airlines in our
specification (see Fig. 4), we examine the relative importance of these two effects. We run a similar regression as the ones
presented in Section 3, in which we introduce the dummy variables characterizing both the experience and the business
model, all interacted with the Lease and Lease2 variables.30 These variables are statistically significant. This suggests that both
the experience and the business model (LCC) separately affect returns to leasing. The effects of these two airline features are
then distinct and not confounding.31 Thus, increasing the share of leased aircraft benefits both the LCCs and the younger airlines.

28
We run an alternative regression with only one variable Lease2 common to all airlines and the results do not change.
29
See Leonard-Barton (1992) and Tripsas and Gavetti (2000) among others.
30
We omit the variable FCC for collinearity reasons.
31
The results for this estimation are available upon request.
42 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

This makes the global impact of leasing on the LCCs profit margin more important, as it results from the cumulative effect of
experience and the fact that they belong to the LCC business model.32
The most important result from these estimations highlights the fact that it is not necessarily more profitable for airlines
to rent the highest possible share of their fleet. Indeed, airlines are trading off between the flexibility provided by leasing
some of their capacity and the lower cost of owning this capacity.

5.1. Additional results

Some additional results are worth noting here. The negative sign for the variable Competition indicates that the higher the
number of competitors in the same geographical zone, the smaller the profit margin, all else being equal. This result is in line
with economic theory and the fact that competition is likely to become fiercer with the number of firms on a market, leading
to lower profits.
The impact of Tax, the marginal corporate income taxes, on the airlines operating margin is negative and significant at a
<1% level. This effect may seem surprising as we consider an operating margin before taxes. However, this may be explained
in the following ways. First, companies that operate in countries with a high marginal corporate tax rate may also incur
higher operating costs. This may consequently reduce their operating margin. Moreover, we also think that there is an incen-
tive effect of taxes. When taxes increase, companies’ incentives to exert effort in order to increase their operating margin are
weaker. These effects may explain the negative impact of taxes on the operating margin. Note that this variable is likely to
reflect some more general country specific political behavior and the interpretation of this parameter should be conducted
with caution.

5.2. Robustness checks

To validate our methodology, we perform some robustness checks on our models. First, in order to further account for the
potential presence of unobserved heterogeneity in this set-up and potential omitted variables bias, and in addition to our
instrumental variable estimation, we exploit the panel structure of the data to also include fixed effects. We compute a
within estimator for the regression depicted in Eq. (1). We need to remove the LCC and geographical dummies from our
equation. To reach the same statistical global quality we need to introduce the instrument computed as the lease expenses
for airline i at period t divided by the sum of deliveries from Airbus and Boeing in year t. The results from this estimation are
displayed in Appendix B and they are very close to the results presented in Table 2. The parameters for Lease and Lease2 in
this fixed-effects regression are tested against their estimated values from the model in Eq. (1) and they are not significantly
different. We are thus confident that our estimated parameters do not suffer from omitted variables bias.
Second, we estimate the model with the EBIT adjusted for operating leases33 (instead of EBITDAR) normalized by sales as
an alternative for financial performance. Our results barely change. Other variables were introduced as exogenous controls: alli-
ance membership, the interest rate on 10 years US Government bonds, inflation, sales, interest coverage ratio (as a proxy for
credit rating) but none of these variables turned out to have a significant impact. As the financial crisis led to a substantial
decrease in new loans issued by financial institutions, one may think that airlines could find themselves relying more on lease
financing. We therefore also introduced interacted terms between leasing and crisis but they were not statistically significant.
Third, we confirm that our results are robust to variations in the threshold categorizing younger and established airlines.
We also check whether this dynamic effect is partly driven by the timing of deregulation processes in the US (1978) and the
EU (1992). The results do not hold if we do not distinguish the airlines with respect to their age but to the fact they entered
the market before or after the local deregulation.
Fourth, we test the source of non-linearities in our model by introducing logarithms for the profitability and leasing vari-
ables as an alternative to the quadratic specification of the relationship. The estimation gives very poor statistical results, and
confirms the presence of non-monotonic and concave effect of leasing on profitability in the industry.

6. Conclusion

This paper examines whether the use of operating leases significantly improves airlines financial performance by empir-
ically analyzing a representative sample of airlines capturing approximately 75% of the total industry revenues over the per-
iod 1996–2011.
We first show that the impact of leasing on operating margin is concave. This indicates that airlines face a decreasing
marginal return to leasing on their profit margin. It also allows us to characterize the airlines’ optimal leasing level. Intu-
itively, when airlines have the ability to get the necessary funds, they should purchase their fleet to operate the safest part
of their business instead of leasing the aircraft. However, leasing becomes relatively more profitable for the riskiest, more

32
We thank an anonymous referee for this point.
33
In order to compute the adjusted EBIT, we remove the implicit lease interest expense from the EBIT. See Koller et al. (2010). Indeed, as operating leases are
off balance sheet (this will be modified in 2016), companies that choose to lease a substantial part of their assets will have an artificially low EBIT. This allows us
to reflect the real economics of operating leases.
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 43

uncertain markets. The impact of leasing is therefore positive at first, when airlines have a low proportion of leased aircraft,
but becomes negative when this proportion exceeds some threshold.
Second, we examine how an airline’s business model affects the impact of leasing on its profitability. Our estimations
point out that the benefits of leasing are more important for Low Cost Carriers than for Full Cost Carriers. We also compute
how airlines’ observed leasing strategies differ from the estimated optimal level in the industry, during the period under
consideration.
Third, we analyze how airlines’ experience affects our results. We show that leasing is more profitable for younger
companies.
Even if we estimate the impact of leasing on airlines’ profitability through a reduced form model, our rich sample over
16 years enables us to draw interesting policy conclusions that do not appear to be artefacts of the model’s simplifications.
In particular, our results suggest that 50% of the companies in our sample would have enhanced their profit margin in 2011
by choosing the optimal leasing level predicted by our model. Moreover, the optimal leasing level that emerges from our
estimations is significantly higher than the proportion of leased aircraft in the worldwide global fleet. The increasing trend
in the use of operating leasing may therefore continue for years and it may benefit both airlines and lessors.
Other factors such as competition mechanisms, shareholder base and governance may also affect financial performance.
In order to capture these effects and make some more accurate predictions on individual optimal strategies improving prof-
itability, we would need to design a structural model of airlines’ performance, requiring data collection on competition at the
route level, operators’ characteristics, airlines’ ownership (government, private, mixed. . .) and governance indicators (num-
ber of independent directors, size of the board of directors. . .). This is left for further research.

Appendix A

Airlines Average Leasing and Profit Margin over the Period 1996–2011

Airline Years in Sample EBITDAR/Sales Leasing LCC

Aegean Airlines 8 16% 85% 0
Aer Lingus 14 16% 47% 0
Aeroflot 14 22% 53% 0
Aeromexico 16 14% 82% 0
Air Arabia 8 30% 77% 1
Air Asia 8 38% 27% 1
Air Berlin 9 14% 60% 1
Air Canada 11 11% 43% 0
Air France-KLM 16 13% 32% 0
Air New Zealand 15 18% 41% 0
AirTran Airwaysy 16 18% 58% 1
Alaska Air 16 19% 45% 0
Alitaliay 13 9% 47% 0
All Nippon Airways 16 20% 53% 0
Allegiant Travel 7 19% 20% 1
America Westy 12 18% 78% 0
American Airlines (AMR) 16 12% 35% 0
Asiana 16 22% 44% 0
Austrian Airlinesy 11 12% 10% 0
AviancaTACA 15 12% 83% 0
British Airways 16 18% 17% 0
COPA Holdings SA 8 29% 24% 0
Cathay Pacific 16 22% 16% 0
China Airlines 16 16% 16% 0
China Eastern 15 24% 27% 0
China Southern 16 26% 29% 0
Continental Airlinesy 16 17% 58% 0
Delta Air Lines 16 17% 41% 0
EVA Airways 9 19% 42% 0
Easyjet 13 15% 35% 1
El Al Israel Airlines 14 10% 18% 0
Emirates 12 25% 53% 0
Ethiopian Airlines 15 20% 40% 0

(continued on next page)

44 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

Appendix A (continued)

Airline Years in Sample EBITDAR/Sales Leasing LCC

Finnair 16 13% 41% 0
Flybe 16 13% 79% 1
Frontiery 16 21% 79% 1
GOL Linhas Aereas Inteligent 9 24% 70% 1
Hawaiian 6 21% 68% 0
Iberia 16 15% 69% 0
Japan Airlines 16 10% 22% 0
Jazeera Airways 6 23% 17% 1
Jet Airways 15 24% 46% 0
JetBlue Airways 11 22% 31% 1
KLMy 11 15% 38% 0
Kenya Airways 14 23% 51% 0
Korean Airlines 16 19% 22% 0
LAN 11 22% 38% 0
Lufthansa 14 14% 28% 0
Malaysian Airlines (MAS) 16 13% 61% 0
Northwest Airlinesy 15 12% 38% 0
Norwegian Air Shuttle 10 11% 85% 1
Qantas 15 16% 19% 0
Royal Jordanian 12 18% 63% 0
Ryanair 15 32% 12% 1
SAir/Swissair Groupy 8 17% 40% 0
Scandinavian Airlines (SAS) 16 9% 43% 0
Shenzhen airlines 7 29% 43% 0
Singapore Airlines 16 24% 16% 0
SkyEuropey 9 1% 91% 1
Skymark 12 16% 92% 1
South African Airways (SAA) 14 12% 65% 0
Southwest Airlines 16 21% 27% 1
Spirit Airlines 3 35% 94% 1
Swiss Internationaly 8 6% 34% 0
TAM Linhas Aereas 9 18% 66% 0
Thai Airways 16 24% 25% 0
Tiger Airways 6 24% 67% 1
US Airways 16 13% 65% 0
United Airlines 16 11% 42% 0
VARIGy 9 15% 84% 0
Virgin Blue 7 19% 41% 1
Vueling 8 6% 99% 1
WestJet 14 24% 29% 1
y
Ceased to operate during our sample period.

Appendix B

Fixed Effects 2SLS Regression Estimates, Operating Leases Expenses

Variables Parameter estimate t-value

Lease 1.204** 2.68
Lease2
1.063**
2.39
Competition
0.007**
2.47
Tax
0.047
0.38
Trend 0.003* 1.81
Crisis
0.029***
4.49
 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46 45

Appendix B (continued)

Variables Parameter estimate t-value

Chapter 11
0.064**
2.05
GDP Growth 0.167 1.59
Oil Price
0.001***
3.22
*
Significant at 10% level.
**
Significant at 5% level.
***
Significant at 1% level.

References

Abarbanell, Jeffery S., Bushee, Brian J., 1998. Abnormal returns to a fundamental analysis strategy. Account. Rev. 73 (1), 19–45. http://dx.doi.org/10.2139/
ssrn.40740.
Airbus, 2016. Global Market Forecast: Mapping Demand 2016–2035.
Ang, James, Peterson, Pamela P., 1984. The leasing puzzle. J. Finance 39 (4), 1055–1065. http://dx.doi.org/10.1111/j.1540-6261.1984.tb03892.x.
Barros, Carlos P., Peypoch, Nicolas, 2009. An evaluation of european airlines’ operational performance. Int. J. Prod. Econ. 122 (2), 525–533. http://dx.doi.org/
10.1016/j.ijpe.2009.04.016.
Bjelicic, Borislav, 2012. Financing airlines in the wake of the financial markets crisis. J. Air Transp. Manage. 21 (July), 10–16. http://dx.doi.org/10.1016/
j.jairtraman.2011.12.012.
Boeri, Tito, Garibaldi, Pietro, 2007. Two tier reforms of employment protection: a honeymoon effect? Economic J. 117 (521), 357–385. http://dx.doi.org/
10.1111/j.1468-0297.2007.02060.x.
Dixit, Avinash K., Pindyck, Robert S., 1994. Investment under Uncertainty. Princeton University Press, Princeton, New Jersey.
Eisfeldt, Andrea L., Rampini, Adriano A., 2009. Leasing, ability to repossess, and debt capacity. Rev. Financial Stud. 22 (4), 1621–1657. http://dx.doi.org/
10.1093/rfs/hhn026.
Fairfield, Patricia M., Scott Whisenant, J., Yohn, Teri.Lombardi., 2003. Accrued earnings and growth: implications for future profitability and market
mispricing. Account. Rev. 78 (1), 353–371. http://dx.doi.org/10.2308/accr.2003.78.1.353.
Gavazza, Alessandro, 2010. Asset liquidity and financial contracts: evidence from aircraft leases. J. Financ. Econ. 95 (1), 62–84. http://dx.doi.org/10.1016/j.
jfineco.2009.01.004.
Gavazza, Alessandro, 2011. Leasing and secondary markets: theory and evidence from commercial aircraft. J. Political Econ. 119 (2), 325–377. http://dx.doi.
org/10.1086/659412.
Goyal, Manu, Netessine, Serguei, 2007. Strategic technology choice and capacity investment under demand uncertainty. Manage. Sci. 53 (2), 192–207.
http://dx.doi.org/10.1287/mnsc.1060.0611.
Graham, John R., Lemmon, Michael L., Schallheim, James S., 1998. Debt, leases, taxes, and the endogeneity of corporate tax status. J. Finance 53 (1), 131–162.
http://dx.doi.org/10.1111/0022-1082.55404.
Gittell, Jody H., Von Nordenflycht, Andrew, Kochan, Thomas A., 2004. Mutual gains or zero sum? labor relations and firm performance in the airline industry.
Ind. Labor Relat. Rev. 57 (2), 163–180. http://dx.doi.org/10.1177/001979390405700201.
ICAO, 2014. ‘‘List of LCCs based on ICAO definition. Economic Ananlysis and Policy Section.” <http://www.icao.int/sustainability/Documents/LCC-List.pdf>.
IATA, 2011. ‘‘Vision 2050 Report.” <https://www.iata.org/pressroom/facts_figures/Documents/vision-2050.pdf>.
IATA, 2013. ‘‘Profitability and the air transport value chain.” Economics briefing 10. <https://www.iata.org/whatwedo/Documents/economics/profitability-
and-the-air-transport-value%20chain.pdf>.
Ismail, Noor A., Jenatabadi, Hashem S., 2014. The influence of firm age on the relationships of airline performance, economic situation and internal
operation. Transp. Res. Part A: Policy Practice 67 (September), 212–224. http://dx.doi.org/10.1016/j.tra.2014.06.010.
Kleibergen, Frank, Paap, Richard, 2006. Generalized reduced rank tests using the singular value decomposition. J. Econom. 133 (1), 97–126. http://dx.doi.
org/10.1016/j.jeconom.2005.02.011.
Koller, Tim, Goedhart, Marc, David, Wessels, 2010. Valuation: Measuring and Managing the Value of Companies. Wiley, Hoboken, N.J.
Leonard-Barton, Dorothy, 1992. Core capabilities and core rigidities: a paradox in managing new product development. Strateg. Manage. J. 13 (S1), 111–125.
http://dx.doi.org/10.1002/smj.4250131009.
Lewis, Craig M., Schallheim, James S., 1992. Are debt and leases substitutes? J. Financial Quant. Anal. 27 (4), 497–511. http://dx.doi.org/10.2307/2331137.
Li, Ye, Wang, Yan-zhang, Cui, Qiang, 2015. Evaluating airline efficiency: an application of virtual frontier network SBM. Transp. Res. Part E: Logistics Transp.
Rev. 81 (September), 1–17. http://dx.doi.org/10.1016/j.tre.2015.06.006.
MacKinnon, James G., White, Halbert, 1985. Some heteroskedasticity-consistent covariance matrix estimators with improved finite sample properties. J.
Econom. 29 (3), 305–325. http://dx.doi.org/10.1016/0304-4076(85)90158-7.
Mancilla, D., 2010. Aircraft asset management: Tools and airline strategies during a world market downturn. ATRS World Conference 2010.
Marston, Felicia, Harris, Robert S., 1988. Substitutability of leases and debt in corporate capital structures. J. Account. Audit. Finance 3 (2), 147–164. http://
dx.doi.org/10.1177/0148558X8800300205.
Miller, Merton H., Upton, Charles W., 1976. Leasing, buying, and the cost of capital services. J. Finance 31 (3), 761–786. http://dx.doi.org/10.1111/j.1540-
6261.1976.tb01922.x.
Moody’s, 2006. Guideline Rent Expense Multiples for Use with Moody’s Global Standard Adjustment to Capitalize Operating Leases.
Myers, Stewart C., Dill, David A., Bautista, Alberto J., 1976. Valuation of financial lease contracts. J. Finance 31 (3), 799–819. http://dx.doi.org/10.1111/
j.1540-6261.1976.tb01924.x.
Oum, Tae Hoon, Zhang, Anming, Zhang, Yimin, 2000. Optimal demand for operating lease of aircraft. Transp. Res. Part B: Methodol. 34 (1), 17–29. http://dx.
doi.org/10.1016/S0191-2615(99)00010-7.
Pástor, L’uboš, Pietro, Veronesi, 2003. Stock valuation and learning about profitability. J. Finance 58 (5), 1749–1789. http://dx.doi.org/10.1111/1540-
6261.00587.
Schefczyk, Michael, 1993. Operational performance of airlines: an extension of traditional measurement paradigms. Strateg. Manage. J. 14 (4), 301–317.
http://dx.doi.org/10.1002/smj.4250140406.
Shleifer, Andrei, Vishny, Robert W., 1992. Liquidation values and debt capacity: a market equilibrium approach. J. Finance 47 (4), 1343–1366. http://dx.doi.
org/10.1111/j.1540-6261.1992.tb04661.x.
Smith, Clifford W., Wakeman, Lee, 1985. Determinants of corporate leasing policy. J. Finance 40 (3), 895–908. http://dx.doi.org/10.1111/j.1540-6261.1985.
tb05016.x.
Staiger, Douglas, Stock, James H., 1997. Instrumental variables regression with weak instruments. Econometrica 65 (3), 557–586. http://dx.doi.org/10.2307/
2171753.
46 S. Bourjade et al. / Transportation Research Part A 97 (2017) 30–46

Stock, James H., Motohiro, Yogo, 2005. Testing for Weak Instruments in Linear IV Regression. In: Stock, James H., Andrews, Donald, W.K. (Eds.), Identification
and Inference for Econometric Models: Essays in Honor of Thomas J Rothenberg, Cambridge University Press, New York, doi:http://dx.doi.org/10.1017/
CBO9780511614491.006.
Titman, Sheridan, John Wei, K.C., Xie, Feixue, 2004. Capital investments and stock returns. J. Financial Quant. Anal. 39 (04), 677–700. http://dx.doi.org/
10.1017/S0022109000003173.
Tripsas, Mary, Gavetti, Giovanni, 2000. Capabilities, cognition, and inertia: evidence from digital imaging. Strateg. Manage. J. 21 (10–11), 1147–1161. http://
dx.doi.org/10.1002/1097-0266(200010/11)21:10.
Vasigh, Bijan, Fleming, Kenneth, Humphreys, Barry., 2015. Foundations of Airline Finance: Methodology and Practice. Routledge, New York, NY.
Vasigh, Bijan, Reza, Taleghani, Kenneth, Fleming, 2012. Aircraft Finance: Strategies for Managing Capital Costs in a Turbulent Industry. J. Ross Publishing.
Yan, An, 2006. Leasing and debt financing: substitutes or complements? J. Financial Quant. Anal. 41 (03), 709–731. http://dx.doi.org/10.1017/
S0022109000002593.