Professional Documents
Culture Documents
FAILURE?
Franziska Kupfer
Hilde Meersman
Evy Onghena
Eddy Van de Voorde
University of Antwerp
Department of Transport and Regional Economics
{franziska.kupfer ; hilde.meersman ; evy.onghena ;
eddy.vandevoorde}@ua.ac.be
ABSTRACT
During the past 30 years, air cargo has evolved from a by-product to a
potential profit centre for airlines. This paper provides an overview of some
important trends in the global air cargo market on an aggregated level.
Subsequently, an insight into these trends is gained by means of a time series
model using co-integration theory and an error correction model. In addition,
some business economic aspects of the air cargo market are investigated. By
combining several levels of the air cargo market, this paper explains part of
the economic rationality behind the air cargo market structure.
KEYWORDS:
Air cargo, market trends, time series modelling, cargo airlines
In section 3 the focus lies on the macro-economic aspect of air cargo and its
potential volume. A regression analysis is carried out to discover the
underlying factors that influence the development of air cargo.
The micro-economic aspect of air cargo is treated in section 4. Particularly,
the income and cost side of airlines are analysed on a case basis. First of all,
the development of the operating income and cost per block hour of Cargolux
and a selection of US cargo airlines is examined. Secondly, it is investigated
whether the bankruptcy of certain all-cargo airlines could have been foreseen.
In the last section the overall conclusions are summarized and the agenda for
further research is set up.
Figure 2.1: Evolution of worldwide air freight traffic in FTKs (millions), 19752007
180000
160000
140000
120000
100000
80000
60000
40000
20000
2007
2005
2003
2001
1999
1997
1995
1993
1991
1989
1987
1985
1983
1981
1979
1977
1975
Source: ICAO Journal, 1987-2006; ICAO Annual Report of the Council, 2008
Figure 2.2: Evolution of the share of all-cargo and combi traffic (freight tonnes
carried), international and domestic, scheduled services, 1976-2006
Source: IATA World Air Transport Statistics, 1981-2007, IATA Members Air Transport
Operations
Figure 2.3: Evolution of the share of all-cargo and combi traffic (freight, incl.
express tonne-kilometres performed), international and domestic, scheduled
services,
1976-2006
Source: IATA World Air Transport Statistics, 1981-2007, IATA Members Air Transport
Operations
In order to gain an insight into the evolution of both traffic types, the annual
percentage change of freight tonnes of all-cargo and combi traffic is depicted
in figure 2.4. It becomes clear from this figure that all-cargo traffic is more
volatile than combi traffic. This is due to the fact that combi traffic involves
more traditional carriers and is related to passenger traffic. Moreover, in case
of a worsening economic climate, airlines will reduce their all-cargo capacity
more easily than their combined passenger-belly capacity. An example of this
is the period 1979-1983, in which the airline industry was in crisis as a
consequence of the increase in oil prices combined with a stagnating demand
and decreasing revenues. The figure shows a decrease in all-cargo traffic in
this period, while the change in combi traffic remains positive, with a sharp
increase in 1983 as a sign of recovery from the crisis. In the second half of the
1980s the airline industry performed relatively well. However, this changed in
1990 due to a rise in oil prices which induced a crisis in the period 1990-1993.
A slowdown of the economy in different countries but especially in the US and
the UK, gave rise to a decreasing demand. The first Gulf War in 1991
worsened the situation even more and led to the bankruptcy of several
airlines. (Nolan, Ritchie and Rowcroft, 2004, p. 240)
In 1991, all-cargo traffic as well as combi traffic decreased. The strong growth
of all-cargo traffic in 1994 indicates again a recovery from the crisis. The
combi traffic decrease in 1998 was a consequence of the economic crisis in
East-Asia, resulting in a traffic decrease on intra-Asian routes and thus
problems for several Asian airlines. In addition, some European and NorthAmerican airlines saw a traffic decrease on their routes to and from East-Asia.
The figure shows a final traffic decrease in 2001 (all-cargo and combi)
induced by the crisis in the airline industry in the aftermath of 9/11. Rather
remarkable is that all-cargo traffic recovered well in 2002, while combi traffic
still decreased.
Figure 2.4: Annual % change of freight tonnes of all-cargo and combi traffic,
1977-2006
Source: IATA World Air Transport Statistics, 1981-2007, IATA Members Air Transport
Operations
Figure 2.5: Export/import air freight tonnage ratio4 for selected regional
markets in 2008
USA to:
Africa
Middle East excl.
Israel
2.6 : 1
14.2 : 1
Europe to:
East Africa
North Africa
1 : 3.4
1.2 : 1
1%
3%
1%
1%
22%
14%
15%
16%
25%
5%
1%
3%
1%
3%
1%
2%
16%
19%
15%
16%
3%
9%
17%
4%
1%
13%
31%
19%
3%
41%
12%
15%
24%
28%
18%
27%
8%
37%
36%
27%
26%
19%
32%
19%
15%
10%
12%
2%
15%
11%
8%
2%
3%
6%
World
15.24
AS-EU
2.79
EU-AS
1.53
AS-NA
2.98
36%
9%
35%
32%
5%
23%
21%
14%
4%
11%
2%
2%
NA-AS
1.13
Intra-Asia
1.60
EU-NA
0.80
10%
14%
3%
6%
4%
3%
NA-EU
0.67
LA-NA
0.42
NA-LA
0.39
Primary products
Intermediate materials
Capital equipment
Consumer products
Non-refrigerated foods
Refrigerated foods
Source: Based on MergeGlobal world air freight supply and demand model, MergeGlobal Value
Creation Initiative, 2008, p.36.
24%
10
3%
industrial production
merchandise exports
1981=100
400
300
200
100
0
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Sources:
air freight tkm: ICAO
GDP in constant USD of 2000: Worldbank
Industrial Production: industry value added in constant USD of 2000: Worldbank
Merchandise exports in USD of 2000: Worldbank and IMF
One of the major problems in analysing time series which all show an upward
trend, is to find out whether this trend is deterministic or stochastic6. This is
traditionally done by unit root and/or stationarity tests. Time series with a
deterministic trend are stationary and can be related to each other by simple
ordinary least squares regressions taking into account the deterministic trend.
Time series with stochastic trends and which are therefore not stationary can
only be related to each other by a regression equation if they are cointegrated,
which means that they should have a common trend. The seminal work of
Nobel Prize winners Engle & Granger treats this in detail and Grangers
Representation theorem states that cointegrated series are related to each
other by means of a very specific dynamic model, the error-correction model
Association for European Transport and contributors 2009
11
(ECM), which models the long run equilibrium relation between cointegrated
series and the short run adjustments towards this equilibrium relationship. If
nonstationary time series are not cointegrated, they cannot be represented by
a simple regression and there is no long run equilibrium relation between
them. Only their short term behaviour can be modelled in a statistically reliable
way.
The first step in the analysis is to discover whether the trend in the time series
under consideration is stochastic or deterministic. Traditionally this is tested
by means of the following tests: Dickey-Fuller (DF), augmented Dickey-Fuller
(ADF), Dickey-Fuller with detrending (DFGLS), Phillips-Perron (PP),
Kwiatkowski, Phillips, Schmidt, and Shin (KPSS), Elliot, Rothenberg, and
Stock (ERS), and Ng-Perron (NP). If the time series have a stochastic trend,
the next step is to test whether they are cointegrated or not. There are several
tests for cointegration: Engle-Granger 2-step approach, Engle-Granger-Yoo 3step approach, the dynamic ordinary least squares (DOLS) developed by
Stock & Watson, the unrestricted ECM approach, and the Johansen
cointegration test. Finally, if the series are cointegrated, their relation can be
represented by an error correction model. If they are not cointegrated, there is
no long run equilibrium relation which ties the series together.
Unit roots tests for world air freight in tkm (TKM), world GDP in constant
prices (GDP), and world industrial production in constant prices (IP) revealed
a stochastic trend in the series whether they were measured in levels or in
logarithms. Several cointegration tests were applied indicating no
cointegration between TKM and GDP, and between TKM and IP. This leads to
the conclusion that there is no long run equilibrium relation between world air
freight on the one hand and world GDP or world industrial production on the
other hand.
As air cargo consists mainly in international traffic of high value goods, the
evolution of world air freight can be better explained by an indicator for world
international trade in high value goods. This is approximated by the volume of
world merchandise exports (MERCH) as a global indicator of international
trade in combination with the share of manufactures in the total value of
merchandise exports (SHAREMANU)7. An increase of the latter can be the
result of an increasing share of manufactures in the volume of merchandise
trade, an increase of the value of the manufactures, or a combination of both.
All cointegration tests indicated that TKM, MERCH and SHAREMANU are
cointegrated. For the error correction specification, the best results were
obtained with the logarithm of TKM and MERCH and by adding a dummy
variable for the crisis in 1991 induced by the higher oil prices and the Gulf War
(DUM91) and a dummy variable for the structural impact of 9/11
(DUMBREAK).
12
with
TKM
MERCH
SHAREMANU
DUM91
Std. Error
t-Statistic
Prob.
-0.348759
0.213602
-1.632749
0.1308
LNMERCH
0.986979
0.055337
17.83581
0.0000
SHAREMANU
0.936589
0.268378
3.489818
0.0051
DUM91
-0.065071
0.030437
-2.137904
0.0558
DUMBREAK
-0.075160
0.034519
-2.177367
0.0521
0.209563
0.766090
0.273549
0.7895
DSHAREMANU(1)
DDUM91(1)
-0.002301
0.017582
-0.130859
0.8982
DDUMBREAK(1)
-0.034809
0.049727
-0.700013
0.4985
DLNXMERCH(-1)
-0.056092
0.353517
-0.158669
0.8768
DSHAREMANU(-1)
0.005442
0.507549
0.010722
0.9916
DDUM91(-1)
-0.011760
0.021071
-0.558099
0.5880
DDUMBREAK(-1)
0.014882
0.034023
0.437394
0.6703
R-squared
0.997523
5.529040
Adjusted R-squared
0.994821
0.453568
S.E. of regression
0.032641
-3.703331
0.011720
Schwarz criterion
-3.065219
Log likelihood
57.43998
Hannan-Quinn criter.
-3.534040
F-statistic
369.1756
Durbin-Watson stat
1.821909
Prob(F-statistic)
0.000000
13
The short run adjustments are estimated given the DOLS-estimates for 0, ,
4 and are reported in table 3.2.
Table 3.2: Error correction model for world air freight and world merchandise
exports
Dependent Variable: DLNTKM
Method: Least Squares
Sample (adjusted): 1984 2007
Included observations: 24 after adjustments
Coefficient
Std. Error
t-Statistic
Prob.
DLNMERCH
1.006034
0.073951
13.60399
0.0000
DSHAREMANU
0.553164
0.290538
1.903929
0.0722
DDUM91
-0.041447
0.017759
-2.333856
0.0307
DDUMBREAK
-0.060238
0.025114
-2.398598
0.0269
RESDOLS(-1)
-0.943221
0.231856
-4.068135
0.0007
R-squared
0.787872
0.063426
Adjusted R-squared
0.743214
0.049532
S.E. of regression
0.025100
-4.348839
0.011970
Schwarz criterion
-4.103411
Log likelihood
57.18607
Hannan-Quinn criter.
-4.283727
Durbin-Watson stat
1.583792
with
RESDOLS = lnTKM + 0.349 - 0.987lnMERCH - 0.937SHAREMANU + 0.075DUMBREAK
+ 0.065DUM91
The error correction model reveals that the change in world air freight is due
to the current change in world merchandise exports, the current change in the
share of manufactures and an error correction term which is an adjustment to
deviations from the long run equilibrium in the previous period. The
adjustment speed, which is given by =-0.94, is high as it close to -1.
The elasticity of air freight with respect to merchandise exports is not
significantly different from 1 neither in the long run equilibrium relation, nor in
the short run adjustment. So a one percent change in world merchandise
exports will result in a one percent change in air freight. An increase of the
share of manufactures in the value of merchandise exports with one
percentage point, will lead in the long run to a one percent increase in air
freight as 2 is not significantly different from 1. In the short run the impact will
be smaller than 1. There has clearly been a negative impact in 2001 as a
consequence of 9/11 which has led to a structural downward shift in air
freight.
Figure 3.2 gives the actual value of air freight (in logarithms) and the fitted
values calculated with the estimated error correction model. In the most recent
years, the model overshoots the actual values and so there is room for
improvement, especially when the model would be used to make forecasts.
14
88
90
92
94
96
actual
98
00
02
04
06
fitted
Therefore in a next stage the total demand for air freight can be split into the
major air freight markets (cf. Figure 2.6). Moreover, the transport price of the
best alternative, i.e. maritime shipping, can also be incorporated. At the supply
side there is room for investigating the effect of oil price changes and capacity
adjustments, and the way individual airlines adjust their strategy in order to
determine their competitive position.
15
16
products, the competition and the so called willingness to pay. On the other
hand, it is possible that the same carrier flies the two stretches of an origindestination relation with a high load factor but hardly breaks even due to a
relatively low yield. When strategically choosing the routes, the presence of
sufficient sales potential and return cargo certainly plays a decisive role.
Therefore, it is interesting to calculate the financial-economic importance of
this.
To gain an insight into the benchmarks a number of proxy-variables are
quantified in table 4.1, using figures from the largest European full-freighter
carrier, namely Cargolux. In 20089, Cargolux employed a homogeneous fleet
of 16 B747-400F airplanes and achieved an operating turnover of about $1.97
billion, with a net loss of $61 million. Despite this loss the carrier shows some
stability, especially concerning the results10.
Table 4.1: Business economic indicators Cargolux
Indicator
(Operating)
income/block
hour ($)
11
Costs /block
hour ($)
(Operating)
income/trans
ported tonne
($)
(Operating)
income/tonne
-kilometre
flown ($)
Average
distance
12
(km)
Fleet B747400F
2000
2001
2002
2003
2004
2005
2006
2007
2008
13,009
12,346
12,527
13,808
16,299
18,935
18,172
18,644
21,553
12,434
12,086
11,758
12,782
15,171
17,914
17,198
19,170
22,220
1,721
1,688
1,690
1,882
2,027
2,232
2,368
2,375
2,805
0.195
0.194
0.194
0.215
0.236
0.272
0.292
0.301
0.365
8,818
8,700
8,701
8,734
8,596
8,210
8,097
7,879
7,690
10
11
12
12
13
14
14
15
16
The calculated indicators in table 4.1 first of all indicate a great stability during
the period of 2000-2002. As from 2003 a clearly upwards movement can be
seen. For Cargolux, both a strong increase in realized income and
corresponding operating costs as well as an increase in operating and net
profit can be observed. In 2004 the airline saw a strong growth in production
(expressed in transported tonnage as well as in tonne-kilometres flown), but
also a strongly increased load factor (measured as FTKs/ATKs - freight tonnekilometres flown/available tonne-kilometres). There is a remarkable strong
increase of costs per block hour in 2007 (+ 11%) and in 2008 (+ 16%). The
operating income per block hour increases as well in 2007 (+ 3%) as in 2008
(+ 16%), but clearly not enough to achieve a positive result in both years.
The indicators in table 4.1 were calculated on the basis of aggregated and
consolidated figures13. Therefore, it is possible that the figures disguise a
number of underlying movements, e.g. concerning new commodity flows, new
customers, etc.
Association for European Transport and contributors 2009
17
Nevertheless, based on the figures calculated in table 4.1 and taking into
account the comments, mentioned above, some additional indicators can be
calculated for the B747-400F airplanes used by Cargolux:
- Based on the figures of 2008, Cargolux should pursue a turnover per
block hour of at least $21.500.
- On the basis of the cost structure of 2008, a turnover of about $0.36 to
0.37 per tonne-kilometres seems right for a B747-400F.
Each of the indicators is quite interesting as it can be compared to the
willingness to pay of the customer, but interpretation is subject to the
constraints already mentioned.
Starting from these indicators, different indicative calculations can be made for
other business cases, taking into account the average distance flown, the
block hours and the average load factor. Besides, one has to bear in mind the
willingness to pay of a shipper which is a function of e.g. the value of the
goods to be transported.
It is obvious that the figures above are calculated for a specific plane, the
B747-400F. Due to economies of scale the prices per transported tonne
should be higher for smaller aircraft types like A300, B-757 and DC-8, which
are often used for air cargo transport. In addition, it concerns calculations for
only one airline. Therefore, it is very interesting to compare the indicators with
some other carriers.
Tables 4.2 to 4.5 show business economic indicators for four American fullfreighter airlines from 2000 to 2007: Polar Air Cargo Airways, Kalitta Air,
Gemini Air Cargo Airways and Kitty Hawk Aircargo. The last two carriers,
Gemini and Kitty Hawk disappeared from the market in 2008. The indicators
were calculated based on data provided by US Department of Transportation
Bureau of Transportation Statistics (BTS). More specifically, Schedule P-12
of the Form 41 Financial Reports is used, which contains quarterly profit and
loss statements for US carriers with operating revenues of $20 million or
more. These data were not yet available for 2008Q4, which explains why the
indicators are only calculated until 2007. To obtain traffic statistics, the T1
database was used, which provides traffic and capacity data by service class
for US carriers. From this database, only the service classes G (scheduled allcargo service) and P (non-scheduled civilian all-cargo service) were taken into
account since these service classes represent the majority of traffic for the
four carriers that are investigated. Other service classes concerning
passenger/cargo services or military all-cargo services are less important or
even inexistent for these carriers. In tables 4.2 to 4.5, operating income and
costs are expressed in two ways: per ramp-to-ramp hour14 and per revenue
aircraft airborne hour15. The ramp-to-ramp hours correspond with the block
hours used in table 4.1 but are only available as from 2003. Therefore, the
operating income and costs are also calculated per revenue aircraft airborne
hour.
18
17
2000
2001
2002
2003
2004
2005
2006
2007
NA
NA
NA
14,666
11,472
13,870
15,285
15,311
14,992
13,791
15,051
15,739
12,562
15,268
16,726
16,876
NA
NA
NA
15,921
12,479
13,465
15,798
15,897
14,938
17,987
14,865
17,086
13,664
14,822
17,287
17,522
1,012
1,046
1,284
1,246
1,026
1,247
1,428
1,411
0.285
0.273
0.325
0.326
0.254
0.306
0.330
0.333
3,551
3,824
3,954
3,826
4,033
4,079
4,328
4,242
14
15
15
15
12
39
38
37
19
2000
2001
2002
2003
2004
2005
2006
2007
NA
NA
NA
13,317
13,246
13,129
20,837
22,465
NA
NA
NA
14,433
14,441
14,049
22,978
24,847
NA
NA
NA
9,904
10,740
11,508
19,850
21,406
NA
NA
NA
10,734
11,709
12,313
21,889
23,675
NA
NA
NA
1,127
973
1,051
1,543
1,695
NA
NA
NA
0.290
0.257
0.245
0.424
0.449
NA
NA
NA
3,888
3,793
4,282
3,637
3,772
NA
NA
NA
NA
NA
NA
11
20
Table 4.3 shows a huge increase in Kalittas operating income and costs per
ramp-to-ramp hour and revenue aircraft airborne hour in 2006. Expressed per
ramp-to-ramp hour, the operating income grew by 59% and the cost by 78%.
This is mainly due to a strong increase in the airlines operating income
(+29%) and operating cost (+40%). Moreover, its ramp-to-ramp hours
decreased in 2006 by 19%. In addition, the operating income per transported
tonne and per tonne-kilometre flown also increased strongly in 2006,
respectively by 47% and 73%. This is a consequence of both the increase in
Kalittas operating income and the decrease in its tonnage (-12%) and tonnekilometres (-26%) in 2006.
As mentioned above, normally, scheduled all-cargo services and nonscheduled civilian all-cargo services are taken into account concerning the
carriers traffic information. However, for Kalitta, there were no data available
about scheduled all-cargo services from 2003 to 2006, so for this period only
data referring to non-scheduled civilian all-cargo traffic are used to calculate
the indicators.
The following indicators were calculated for the B747-200F aircraft used by
Kalitta:
- On the basis of the 2007 figures, Kalitta should aim at a turnover per
ramp-to-ramp hour of at least $22,500.
- Based on the cost structure of 2007, a turnover of about $0.45 per
tonne-kilometre seems right for a B747-200F.
- Considering the average distance ranging from 3,600 km to 4,300 km
in the last five years, a price between $1,000 and $1,700 per
transported tonne seems to be a realistic indicator for the turnover.
21
2000
2001
2002
2003
2004
2005
2006
2007
NA
NA
NA
9,016
4,732
6,362
5,409
NA
7,936
8,326
7,714
9,905
5,234
7,043
5,965
NA
NA
NA
NA
9,310
4,920
6,883
5,569
NA
7,652
10,927
9,891
10,228
5,442
7,620
6,141
NA
1,185
859
746
852
391
573
501
NA
0.368
0.270
0.201
0.245
0.133
0.175
0.143
NA
3,222
3,182
3,720
3,479
2,945
3,279
3,513
3,642
NA
NA
NA
NA
NA
NA
NA
NA
Gemini Air Cargo Airways was an American ACMI cargo airline that operated
worldwide scheduled and charter services on a wet-lease basis. The airline
was faced with financial problems in 2006 but it emerged from bankruptcy
reorganization in August 2006. However, Gemini filed again for bankruptcy
protection in 2008 and disappeared from the market in August 2008. The main
reasons for the failure of Gemini were the huge oil price increase in
combination with its fuel-inefficient fleet of DC-10-30Fs.
Looking at the indicators in table 4.4, the instability of Geminis operating
income and costs is remarkable. In 2004, all the indicators decrease sharply,
followed by an increase in 2005 and a new decrease in 2006. The decrease in
2004 is due to a drop in Geminis operating expenses and operating income,
combined with an increase in its output indicators (ramp-to-ramp hours,
revenue aircraft airborne hours, tonnage and tonne-kilometres). In 2006, the
airlines operating expenses and operating income decrease, while its output
indicators also decrease.
As in the case of Kalitta, only data referring to non-scheduled civilian all-cargo
traffic were available from 2004 to 2006. Geminis failure in 2008 cannot be
predicted on the basis of these indicators since the figures for 2007 were not
available anymore.
Association for European Transport and contributors 2009
22
2000
2001
2002
2003
2004
2005
2006
2007
NA
NA
NA
236
183
163
2,411
2,019
5,648
3,445
284
274
214
192
2,824
2,392
NA
NA
NA
195
23
234
2,286
2,013
5,684
5,978
240
227
27
276
2,677
2,385
567
358
33
30
24
24
409
367
0.632
0.385
0.030
0.031
0.024
0.023
0.380
0.362
898
932
1,120
976
1,010
1,037
1,077
1,014
NA
NA
NA
NA
NA
NA
18
23
Also for Kitty Hawk, table 4.5 is based on data referring to only non-scheduled
civilian all-cargo traffic for all years, except 2002. Analysing the indicators in
table 4.5, the strong variations in both operating income and cost indicators
are remarkable. This was also the case for Gemini Air Cargo.
The material above has the advantage to be a concrete basis for analysing
the matching problem of supply and demand. An exercise like this is of
course only indicative, since airlines determine their prices not necessarily on
the basis of the average costs, whether or not within the framework of costplus pricing. The basic rule remains price-fixing based on marginal costs.
However, an important question is on which basis the marginal costs are
calculated: an additional tonne or an additional airplane. Recent research
departs from an airplane as marginal unit24, but it is clear that in practice, the
marginal cost of additional tonnage in an airplane with free capacity for new
traffic will also be used as a basis for price-setting.
In function of the product, from which cargo volume and from which prices and
which turnover can it be interesting for a full-freighter carrier (and its
customer) to fly towards a specific (cargo) airport? When defining a potential
market a carrier bases itself best on existing traffic flows that combine a high
value/weight ratio and specific characteristics (i.e. the need for fast transport
because of the time factor), but are not necessarily transported by air freight
at the moment. Furthermore, one can search for new markets on the basis of
products that show comparable characteristics such as the market of spare
parts, expensive and relatively light products, etc.
It is, however, typical that in the past a number of new niche markets were
captured by integrators. There are various reasons for this. New niches, with
limited goods flows at the start, can be included relatively easily in an existing
network, often at low marginal cost and hence at a low (introduction) price.
Moreover, the integrators offer almost automatically additional services, e.g.
concerning handling and/or warehousing.
For full-freighter carriers and airports that have freight ambitions but no
integrators as their customer, the future competitive struggle will be
determined to a large extent by how and at which speed they will anticipate
new market developments.
24
after the mid-1960s, often combine four aspects of financial well-being, i.e.
liquidity, leverage, turnover and profitability. A paper of Gritta et al. (2006)
provides a good overview of such methods especially with a view to
applications for the air transport industry.
One of the earliest bankruptcy-forecasting models is the Z-Score model of
Altman (1968), which was estimated by using stepwise multiple discriminant
regression and incorporates a liquidity ratio, two profitability ratios, a leverage
ratio and a turnover ratio. A high ratio in one category increases the Z Score
and thus lessens the danger of bankruptcy. For the bankruptcy analysis in this
paper, however, a variant of the Z Score will be applied to American fullfreight carriers, the so-called Z Score (Altman, 1983). This score was already
used by the U.S. Bureau of Transportation Statistics to analyse airline
financial health. The reason behind the use of this variant of the Z score is,
that the turnover ratio (operating revenues to total assets) can distort the
results of airlines as they usually work with operating leases. Even though
leased airplanes are not included in the total assets, the revenues resulting
from those leases are part of the total turnover. A carrier that leases airplanes
therefore would have a higher turnover ratio than a carrier that owns its fleet
and thus a higher Z Score. Due to this problem the turnover ratio is not
included in the Z Score. It is set up as follows:
Z = 6.56X1+3.26X2+6.72X3+1.05X4
With X1-X4 corresponding to the ratios of the original model (Altman, 1968),
defined as:
X1 = working capital to total assets (liquidity ratio)
X2 = retained earnings to total assets (profitability ratio)
X3 = earnings before interest and taxes to total assets (profitability ratio)
X4 = market value of equity to book value of total liabilities (leverage ratio)
Z Scores with a value of 1.1 or less indicate a high probability of bankruptcy
while a score of 2.6 and higher, points to a low probability of bankruptcy in the
near future. The range between 1.1 and 2.6 is referred to as grey zone,
where no prediction can be made due to insufficient statistical significance.
For this research the Z Score was calculated for the four American all-cargo
airlines Polar Air Cargo, Kalitta Air, Gemini Air Cargo and Kitty Hawk Aircargo.
The results are shown in table 4.6.
25
2001
2002
2003
2004
2005
2006
2007
-7.54
-7.89
-0.48
-3.36
-2.24
6.27
5.68
2.36
0.89
-2.74
-10.03
-14.72
-9.78
-13.2
-7.27
NA
NA
-0.03
NA
NA
2.88
6.51
NA
2.53
Kalitta Air
NA
NA
NA
11.45
9.33
9.43
8.9
10.89
For Polar Air Cargo, low Z scores can be observed until 2004, as its parent
company Atlas Air Worldwide Holdings was not in good financial health and
filed for bankruptcy protection in 2004. However, the scores increased
between 2000 and 2004 and Polar Air Cargo could recover and return to
profitability. This is also reflected in the Z scores, being above 2.6 since
2005.
Gemini Air Cargo had a low Z score in 2000 as well, which, however,
decreased in the following years in contrast to the score of Polar Air Cargo. In
the beginning of 2006, Gemini filed for protection against bankruptcy for the
first time, but the carrier emerged from it in the summer of 2006 as an
agreement with the largest lender was reached. The very low Z score of
Gemini in 2005 was already a sign for the bad financial state of the carrier in
the beginning of 2006. Although the Z score for 2006 increased, it was still
quite low and the carrier could not recover and ultimately stopped its
operations in the summer of 2008.
The Z Scores for Kitty Hawk Aircargo are difficult to compare with the other
carriers as not enough data were available. However, a Z score was
calculated that points to financial difficulties in 2001, a year after Kitty Hawk
filed for Chapter 11 bankruptcy. The carrier emerged from Chapter 11 in 2002
but soon entered bankruptcy protection again at the end of 2007, before
ceasing its operations in January 2008. However, the Z scores for 2004 and
2005 did not point to any financial problems and the score for the first three
quarters of 2007 (the data for the last quarter were not available) was only
slightly below the mark of 2.6 and did not specifically warn of the probability of
bankruptcy.
The only air cargo carrier, whose Z scores were above the critical value of
2.6 for the whole period between 2000 and 2007 was Kalitta Air, which points
to good financial health of the carrier.
The cases analysed above make clear that the Z score can help to forecast
financial difficulties and probable bankruptcies of all-cargo airlines. However,
as the example of Kitty Hawk Aircargo shows, the Z score is not always
Association for European Transport and contributors 2009
26
reliable as the airline still had a quite high score two years before its
bankruptcy. Another factor hindering the analysis is the scarcity of data for the
calculations. The case of Kitty Hawk Aircargo could not be sufficiently
analysed as data for a number of years were unavailable.
27
5. CONCLUSIONS
Air freight transport has developed very rapidly over the last decade. While
previously, air cargo was regarded as a by-product of passenger transport, a
lot of traditional carriers consider it now as an instrument that adds positively
to the ultimate goal of profit maximization. In addition, a number of carriers
started to specialize in the air cargo market. Their success or failure depends
on a number of factors. The future evolution of world merchandise trade is
crucial and especially the trade in high value goods needs close monitoring.
Special attention should be paid to the imbalances on trade routes, which
results in imbalances between incoming and outgoing cargo flows.
A number of indicators at business level, such as operating income and costs
per block hour and/or per tonne and tonne-kilometre show that every airline
reacts to the market differently and dissimilar results can be achieved, even
when operating the same aircraft. Moreover, it is clear from the analysis that
the yield is more important than the load factor for an all-cargo carrier aiming
at profit maximization. Finally, the Z Score, which was calculated for four US
all-cargo carriers, seems to be a good indicator for the financial health of a
cargo airline.
Further research should aim at a study of the air cargo market on a
disaggregate level, focusing on the different goods categories transported by
air cargo, the major routes and individual airlines. Concerning individual
carriers, a more in-depth analysis at business economic level should be
carried out, leading to a better understanding of the cost and organisational
structure of individual cargo airlines.
The analysis done in this paper shows that the air cargo market is extremely
volatile. The economic crisis of 2008-09, with large traffic and turnover
decreases on certain routes and in certain airports, illustrates this clearly. It is
a fine line between success and failure.
28
REFERENCES
Altman, E. (1983). Corporate Financial Distress: A Complete Guide to
Predicting, Avoiding and Dealing with Bankruptcy. New York. John Wiley &
Sons, 368 p.
Altman, E. (1986). Financial Ratios, Discriminant Analysis and the Prediction
of Corporate Bankruptcy, Journal of Finance, 23(4), 589-606.
Blauwens, G. De Baere, P. and Van de Voorde, E. (2008), Transport
Economics, Third Edition, Antwerpen, Uitgeverij De Boeck, 519 p.
Clancy, B. and Hoppin, D. (2004), After The Storm The MergeGlobal 20042008 World Air Freight Forecast,
http://www.aircargoworld.com/archives/features/1_may04.htm
Franses, P.H. (1998), Time series models for business and economic
forecasting, Cambridge University Press, 296 p.
Gritta, R.D. (2006), A Review of The History of Air Carrier Bankruptcy
Forecasting and the Application of Various Models to the U.S. Airline Industry:
1980-2005, XIV International Economic History Congress, Helsinki.
Hamilton, J.D. (1994), Time Series Analysis, Princeton University Press, 820
p.
Hayashi, F. (2000), Econometrics, Princeton University Press, 712 p.
Herman, F. and Van de Voorde, E. (2006), Bijdrage tot de economische
analyse van het full-freighter luchtvrachtvervoer, Tijdschrift
Vervoerswetenschap, 42 (4), 16-23.
MergeGlobal Value Creation Initiative (2008), End of an era? Why the super
spike in fuel prices may signal the end of super growth in air freight,
American Shipper, 50 (8), p.32-47.
Nolan, J.F., Ritchie, P. and Rowcroft, J. (2004), September 11 and the World
Airline Financial Crisis, Transport Reviews, 24 (2), 239-255.
Stock, J.H. and Watson, M. (1988), Testing for common trends, Journal of
the American Statistical Association, 83, 1097-1107.
Stock, J.H. and Watson, M. (1993), A simple estimator of cointegrating
vectors in higher Order integrated systems, Econometrica, 61, 783-820.
Verbeek, M. (2008), A Guide to Modern Econometrics, 3rd edition, John Wiley
& Sons, Ltd, 488 p.
Zhang, A. and Zhang, Y. (2002), A model of air cargo liberalization: passenger
vs. all-cargo carriers, Transportation Research Part E, 38 (3-4), 175-191.
Association for European Transport and contributors 2009
29
NOTES
1
30
19
31