Sei sulla pagina 1di 7

Journal of Business Research 60 (2007) 161 167

Airline liberalization effects on fare: The case of the Philippines

Wilfred S. Manuela Jr.
National College of Public Administration and Governance, University of the Philippines, Diliman, Quezon City 1101, Philippines


This article explores the impact of liberalization on airfare using a framework that builds on previous studies but adapted to the peculiarities of
the Philippine airline industry. The data consist of ten routes with varying market characteristics for the period 19812003, while the model
consists of three equations estimated simultaneously using the generalized method of moments based on the Newey-West covariance estimator.
The findings indicate that airfare per kilometer is 10% lower, on average, on route with at least two airlines after liberalization. Twenty-three
routes, representing more than 90% of domestic airline passengers, have at least two airlines by 2003, indicating that most passengers benefit from
lower fares due to the prevalence of discounts and promos as a result of competition.
2006 Elsevier Inc. All rights reserved.

Keywords: Airline industry; Liberalization; Airfare

1. Introduction Philippines) now compete in major routes while two airlines

(Asian Spirit and SEAir) serve minor and short-distance routes.
The Philippine government liberalized the country's domestic By 2003 passengers in 23 airline markets have at least two
airline industry in 1995 under Executive Order (EO) 219, airlines to choose from, giving them more choices on fare,
reducing regulations on tariffs and fares, as well as regulations on departure time, and service quality.
the entry into and exit from the airline industry. Prior to A number of studies document the impact of deregulation
liberalization, only one airlinethe Philippine Airlines (PAL) and liberalization on the airline industry in the US and Europe.
operated scheduled domestic flights due to the government's Previous studies on the Philippine airline industry, however,
one-airline policy. Under the provisions of EO 219, the tend to be descriptive and do not use econometric analysis of
Philippine government privatized PAL, removed restrictions on available data. To estimate the impact of liberalization on
number of routes served and number of departures on each route, airfare, this article builds on the empirical framework employed
and reduced controls over fare in markets served by at least two by Dresner and Tretheway (1992), Maillebiau and Hansen
airlines. (1995), Marin (1995), Jorge-Calderon (1997), and Rietveld
The liberalized domestic airline industry attracted five new et al. (2002).
entrants at one time but the number of airlines has dwindled to
four following the failures in 1998 of both Grand Air and 2. The impact of deregulation on the airline industry
Corporate Air's Mindanao Express. The demise of new entrants
in a relatively short period is comparable to the experience of Competition under regulation usually revolves around
the domestic airline industry in the United States (US) in the service quality, encouraging overcapacity that tends to inflate
early 1980s (see Kahn, 1988; Borenstein, 1992). With the entry price (fare). Under deregulation, competition shifts toward price
of South East Asian Airlines (SEAir) to the scheduled airline (Douglas and Miller, 1974; Graham et al., 1983). In the US
sector in 2003, three airlines (PAL, Cebu Pacific, and Air domestic airline industry, deregulation has resulted in lower
fares and higher load factors.
To determine who has benefited from regulation, Olson and
PO Box 260, University of the Philippines, Diliman, Quezon City 1101, Trapani (1981) examine the policies of the US Civil
Philippines. Tel.: +63 926 629 1667. Aeronautics Board (CAB) by developing a method that
E-mail address: evaluates the board's policies and analyzes the distribution of
0148-2963/$ - see front matter 2006 Elsevier Inc. All rights reserved.
162 W.S. Manuela Jr. / Journal of Business Research 60 (2007) 161167

benefits due to regulation. They use an econometric model 3. The impact of deregulation and liberalization on fare
specified and estimated utilizing cross-sectional samples of city-
pair markets in 1971, 1976, and 1977, and report that the pricing Graham et al. (1983) test two hypotheses that are central to
policies of the US CAB did not serve the consumers interests. the arguments for deregulation by analyzing the US domestic
In 1971 the US CAB set fares too high, shifting the benefits of airline industry before and after deregulation. The first
regulation from consumers and airlines to aircraft manufacturers hypothesis argues that fare regulation promotes service
and other industry suppliers. These findings tend to support the competition among airlines by employing excess capacity
results of Keeler (1972). (Douglas and Miller, 1974). The second hypothesis contends
Moore (1986) analyzes the effects of airline deregulation that potential competition will keep fares at competitive levels,
on passengersbusiness and personal trips, first class and even in highly concentrated markets, which rests on the idea
coachand explores how deregulation affects capital and labor. that capital is highly mobile in the airline industry and that
He examines five markets to show the effects of deregulation on airlines compete in a contestable market (Bailey, 1981). Graham
long-haul major markets, medium-haul markets, and short-haul et al. report that routes shifted toward a single airline, and large
markets in 1976 and 1983, and estimates the fare equation using and medium hubs had more flights in 1981 than in 1978, the
ordinary least squares (OLS) with dummy variables for markets year the US ratified the Airline Deregulation Act. The entry of
with five or more carriers. Using two-stage least squares (2SLS) new carriers after 1978 resulted in the decline in market
to determine the relationship of the number of passengers with concentration on some routes. While the proponents of
density, Moore reports that deregulation results in a substantial deregulation predicted that fares would decrease as competition
increase in the number of passengers, especially in the tourist intensifies, empirical findings suggest mixed results. Fares
market and those traveling on discount fares. Capital seems to increased in short-haul markets, while fares tended to decline as
benefit from deregulation either through more efficient use due distance and market density (number of passengers) increased.
to higher load factors and fewer staff per flight or through the Competition also increased the frequency of giving discount
increase in the share prices of major airlines. Deregulation, fares to passengers, further reducing average fares, which
however, appears to result in lower wages despite initial gains supports the notion that deregulation benefits consumers
in employment in the newer airlines. through lower airfares.
Borenstein (1992) reports that the Herfindahl Index for the Dresner and Tretheway (1992) use a method derived from
US domestic airline industry fell five years after deregulation the neoclassical maximization theory to measure the effect on
(0.106 in 1977 and 0.093 in 1982) and increased steadily since prices of a change in government policy, which in turn leads to a
then, reaching 0.121 in 1990. The four-firm concentration ratio change in market conduct (see Brander and Zhang, 1990;
also increased from 56% in 1977 to almost 62% in 1990, while Marin, 1995). The authors apply the methodology on a number
the eight-firm concentration ratio increased from 81% to almost of international routes from 1976 to 1981 to determine the
91% in the same period. The increase was partly due to the effects of market structure (conduct) on fares and study the
boom in entry following deregulation and later reversed as new benefits to consumers due to bilateral liberalization, and
entrants either merged with incumbents or declared bankruptcy estimate the empirical model using two-stage indirect least
and ceased operations. Airfares fell substantially in most city- squares due to the endogeneity of the passenger variable. The
pairs, especially on long-distance routes, while real prices on findings tend to indicate that the change in policy has a
shorter routes did not fall as much and even increased in some significant effect, reducing discount airfares by an average of
markets (see Graham et al., 1983). Nevertheless, airfares on 35%. The policy change, however, has no statistically
markets with two active airlines were, on average, 8% lower significant impact on the undiscounted fare that first class,
than on monopoly routes in 1990, while a third airline resulted business class, and other non-leisure travelers use.
in an additional 8% drop in fares. Maillebiau and Hansen (1995) study the impact on
Bailey (1992) observes that while the number of airlines consumer benefit arising from bilateral liberalization in North
shrunk by more than half, the degree of competition has Atlantic routes for the years 196989 using a model that
intensified in the post-regulation period. Deregulation allowed incorporates the service accessibility variable (number of
airlines to focus on their markets and operations, which gateways with airline service) on the demand side and the price
produced new core operating capabilities such as the hub-and- variable on the supply side to estimate how liberalization
spoke delivery systems, as well as computer reservation and affects the generalized cost (a function of monetary cost and
yield management systems. Industry observers thought that service accessibility). The resulting change in consumer
deregulation was responsible for the decline in the number of surplus helps evaluate liberalization against the objective of
airlines through merger, acquisition, and bankruptcy. Deregu- the major proponents of deregulation and liberalization, which
lation may not be the major cause, however, because a is to benefit consumers. They estimate the model using the
proliferation of firms is typical in the early years of Yule-Walker method due to autocorrelation and report that
deregulation, followed by rapid consolidation. Passengers also liberalization results in a 35 to 45% reduction in fare, while
contributed to the current structure of the industry because accessibility increased by 38%.
the average passenger usually selects lower-priced services Kahn (2002) suggests that the two most important benefits of
with fewer frills, making low-cost airlines a success under deregulation are lower fares and higher productivity. The
deregulation. average fare that passengers actually paid declined by 30% in
W.S. Manuela Jr. / Journal of Business Research 60 (2007) 161167 163

real terms between 1970 and 1990, although the decline may be financial crisis on fare is positive because the prices of aviation
partly due to the introduction of jumbo jets even before 1978. fuel and other imported inputs increased in peso terms. In
The best estimates suggest that deregulated fares are lower by addition, the impact on fare of the September 11, 2001 terrorist
10 to 18%, on average, compared with the fare levels under attack on the US is positive since oil prices surged in the
regulation. The lower fares redound to passenger savings from aftermath of the attack. The additional cost of enhancing
US$ 5 billion to US$ 10 billion per year since more than 90% of security at airports and in aircraft cabins following the attack
all the passenger miles traveled in 1990 were on discount also contributes to higher fares. Therefore, the fare equation has
tickets. Kahn reports that fare per mile is much higher on thinly the following specifications:
traveled routes than on high-density markets, while fares on
routes served by the eight most concentrated hubs averaged lnFare xy b0 b1 lnPSGR xy b2 lnFREQ xy
almost 19% higher compared with similar markets served by b3 lnINCM xy b4 lnDIST xy
other airports after adjusting for differences in distance and b5 lnVCST xy b6 LIBR xy b7 FCRI xy
traffic density. b8 TERR xy e1; xy;
Rietveld et al. (2002) analyze the consumer benefits
associated with airline liberalization on selected intra-European where for each route x and year y,
routes using data on 34 routes from 1988 to 1992. The sample
routes represent different traffic densities and stage lengths, FARE the average fare per kilometer in pesos
while the period between 1988 and 1992 represents various
degrees of liberalization in the European airline market. They PSGR the number of enplaned passengers
estimate the empirical model using 2SLS due to the endogeneity
of the fare and frequency variables in the passenger equation FREQ the number of two-way flights
and report that economy fares and departure frequency on fully
liberalized routes are, on average, 34% lower and 36% higher, INCM the mean per capita regional gross domestic product
respectively, than fares and number of departures on routes with (RGDP) of endpoint cities
less degree of liberalization.
DIST the two-way distance between airports in kilometers
4. Empirical framework
VCST the average variable cost per passenger kilometer
The econometric model consists of three equations since a
system of equations is a more realistic depiction of the LIBR a dummy variable, which takes a value of 0 between
underlying theory on airline demand and is usually more 1981 and 1994, 1 between 1995 and 2003
appropriate when modeling the data generation process (Judge
et al., 1988) on the number of passengers, the number of flights, FCRI a dummy variable representing the impact of the
and the level of fares in one or more airline markets. 199798 Asian financial crisis, which takes a value of
Furthermore, supply and demand simultaneously determine 1 in 1998; 0 otherwise
the values of fare, the number of passengers, and the number of
departures whether under regulation or liberalization. System TERR a dummy variable representing the impact of the
methods are also more efficient than single-equation methods September 11, 2001 terrorist attack on the US, which
because system methods use all available information in takes a value of 1 in 2001; 0 otherwise
parameter estimation and take into account correlations
among residuals and cross-section equation restrictions, result- the error term.
ing in more efficient estimators (Greene, 1997).
Fare, expressed as average airfare per kilometer, is a function Demand, expressed as the number of passengers, is a
of endogenous variables (number of passengers and departure function of two endogenous variables (fare and frequency), four
frequency), an exogenous cost variable, two dummy variables exogenous variables, and the liberalization dummy (Marin,
that measure the impact of shocks on the industry, exogenous 1995; Jorge-Calderon, 1997; Rietveld et al., 2002). The
geo-economic variables, such as income and distance (Jorge- passenger equation has the following specifications:
Calderon, 1997), and the liberalization dummy (Rietveld et al.,
2002). Fare should respond positively with income since lnPSGR xy b0 b1 lnFARE xy b2 lnFREQ xy
airlines tend to inflate prices in more affluent markets where b3 lnPOPN xy b4 lnCPCY xy
passengers are less price-sensitive, while fare should decrease b5 lnDIST xy b6 ALTM xy
with distance because airlines are able to distribute their fixed b7 LIBR xy e1; xy;
costs over a longer distance. Moreover, airfare should decrease
with frequency since more flights given a particular level of where for each route x and year y,
demand tend to result in lower fares, while fare should increase
with cost since airlines are likely to set fares at levels that enable POPN the mean of the provincial populations of endpoint
them to recover their operating costs. The impact of the Asian airports
164 W.S. Manuela Jr. / Journal of Business Research 60 (2007) 161167

CPCY the average number of passenger seats per two-way frequency, cost, capacity, distance, and number of operators)
flight come from the CAB, while the data on income, population, and
consumer price index come from the National Statistical
ALTM the average fare per kilometer of land or sea transport. Coordination Board's Philippine Statistical Yearbook 2003.
The Land Transport Franchising and Regulatory Board and the
The other variables are as defined in the fare equation. Maritime Industry Authority provided the data on bus and
Frequency, expressed as the number of two-way flights, is a passenger ship fares, respectively.
function of the endogenous passenger variable and two The econometric model consists of three equations due to the
exogenous variables (Rietveld et al., 2002), and the terror- endogeneity of the passenger and frequency variables in the fare
attack dummy. The frequency equation has the following equation. In this study, the author estimates the three equations
specifications: simultaneously because a system of equations estimated one
equation at a time results in biased and inconsistent estimators
lnFREQ xy b0 b1 lnPSGR xy b2 lnCPCY xy due to the inclusion of endogenous variables among the
b3 lnOPTR xy b4 TERR e1; xy; explanatory variables (Intriligator, 1978).
where for each route x and year y, The researcher observes statistically significant autocorrela-
tion when using OLS to estimate the model. The literature deals
OPTR the number of airlines. with the problem of autocorrelation in a number of ways.
Maillebiau and Hansen (1995) use the Yule-Walker method to
The other variables are as defined in the fare and passenger correct the OLS estimates, while Keeler (1972) uses the
equations. This author expressed all continuous variables in Balestra-Nerlove estimator. Rietveld et al. (2002) report first-
natural logarithms to be able to interpret the coefficients as order autocorrelation but do not correct the 2SLS estimates.
elasticities (see Dresner and Tretheway, 1992; Maillebiau and They simply argue that the Durbin-Watson coefficients are in
Hansen, 1995; Rietveld et al., 2002). the indeterminate range. In this study, the author estimates the
model using the generalized method of moments (GMM)
estimator based on the Newey-West covariance estimator,
5. Method which is robust to both heteroskedasticity and serial correlation
(Verbeek, 2000). The NeweyWest estimator provides a way to
Maillebiau and Hansen (1995) consider the fare variable as calculate consistent covariance matrices in the presence of both
exogenous and estimate both the demand and fare equations autocorrelation and heteroskedasticity (Johnston and DiNardo,
using OLS, while Dresner and Tretheway (1992) treat the 1997), while the GMM allows one to drop the assumption of
passenger variable as endogenous in the fare equation and homoskedasticity and apply White's estimator for heteroske-
estimate the fare equation using two-stage indirect least squares. dasticity of unknown form (Greene, 1997). Since the GMM
Marin (1995) estimates the demand and fare equations estimator based on the NeweyWest covariance estimator is
separately using instrumental variables for the endogenous robust both to heteroskedasticity and autocorrelation, the
variable in both equations, whereas Jorge-Calderon (1997) estimation method used in this study is an improvement over
considers fare as exogenous and frequency as endogenous in the the estimation methods used in previous studies on air transport.
demand equation and uses weighted 2SLS to estimate the Rietveld et al. (2002) use a fixed-effects model in estimating
model. Rietveld et al. (2002) consider the passenger, fare, and their equations to control for the unobserved heterogeneity
frequency variables as endogenous and use 2SLS to estimate the between years and country-pairs. In this paper, the researcher
econometric model. also estimates the econometric equations using the fixed-effects
The data for analysis in this study come from 230 model, treating the systematic time and space variation in the
observations per equation or 690 observations for the three data as fixed effects, in order to control for the unobserved
equations representing ten city-pairs with varying market heterogeneity between years and city-pairs. Johnston and
characteristics for the period 19812003. The researcher uses DiNardo (1997) argue that the fixed-effects estimator is robust
balanced panel data (Wooldridge, 2002) because relative to pure to the omission of any relevant time-invariant explanatory
cross-section or time-series data, panel data allow better control variables and more efficient than the random-effects estimator
for individual heterogeneity; offer more informative data, more unless one is certain that all the time-invariant factors possibly
variability, less collinearity among variables, more degrees of correlated with the other explanatory variables are measurable.
freedom, and more efficiency; enable better identification and
measurement of effects; and have better ability to eliminate 6. Estimation results and analysis
biases resulting from the aggregation over airlines at the route
level (Baltagi, 2001). The ten airline markets that comprise the The measurement of the fare variable is quite problematic
sample attracted entry at one time. They represent the short-haul because liberalization has intensified the practice of third-degree
(up to 350 km), medium-haul (351 to 700 km), and long-haul (at price discrimination among airlines. One solution proposed
least 701 km) markets; have at least one competing mode of by Rietveld et al. (2002) is the use of the full economy fare since
transport; and have Metro Manila as the other end of the city- the average of business class and discount fares tends to
pair. The airline-related data (number of passengers, fare, flight approximate the full economy fare. Maillebiau and Hansen
W.S. Manuela Jr. / Journal of Business Research 60 (2007) 161167 165

(1995), on the other hand, use discount fares since most to result in more flights, the NEWCERT variable serves as a
passengers fly using discount fares under liberalization. This proxy for the frequency variable.
study uses the average of the full economy fare and the discount The coefficient of the income variable (INCM) is highly
fare because only one airline in the Philippines offers business significant. A percent increase in per capita RGDP results in a
class for domestic passengers on a number of routes; all major 0.86% increase in airfare, which tends to indicate that airfares
airlines offer discounts and promos throughout the year. Since are higher in markets with a higher per capita RGDP.
the data from CAB do not include the number of passengers for Although the coefficient of the cost variable (VCST) is
each level of fare, this paper uses the simple average of the significant at the 5% level, the cost variable does not behave as
full economy fare and the discount fare. The measurement of expected despite the use of identifiable variable costs. The
the frequency variable also poses some problems because the negative sign suggests that airfare per kilometer declines as cost
records from CAB are fraught with errors. Information on per passenger kilometer increases. Marin (1995) and Rietveld
flight cancellations and additional flights are rarely available, et al. (2002) report highly significant coefficients for the proxies
while data on flight schedules for a number of years are and the cost variable, respectively. Dresner and Tretheway
missing for some airlines. Consequently, the author derives the (1992) and Maillebiau and Hansen (1995), however, do not
frequency variable using the number of passenger seats divided include a cost variable in their price equation due to the lack of
by the average aircraft size. The cost variable is also difficult to airline cost data to derive the marginal cost.
measure. Rietveld et al. use total operating expenses per The airlines practice of milking their most profitable
available ton-kilometer to construct the cost variable using routes (usually the longer routes like ManilaCebu City,
data on large flag carriers, while Marin (1995) uses airport ManilaCagayan de Oro, and ManilaDavao City) to subsidize
presence and wages as proxies for cost. In this study, the unprofitable routes (usually shorter routes like ManilaBaguio
airlines identifiable variable costs constitute the cost variable and ManilaLegaspi) may help explain the negative sign of the
as defined in the literature (see Pindyck and Rubinfeld, 1998; cost variable in the price equation. Keeler (1972) reports that
ABN-AMRO, 2002) and as reported in the airlines income high-density markets in the US are more profitable than the
statements, computed as variable cost per passenger kilometer. aggregate airline rates of return, arguing that airlines use the
Table 1 presents the result of the fare equation estimation. excess profits to cross-subsidize airline service on less
The number of passengers (PSGR) is highly significant. As profitable, low-density markets. Airlines operating with low
expected, a higher demand results in higher prices. With a load factors have tremendous incentives to reduce fares
coefficient of 0.45, however, fare is relatively inelastic with (Besanko et al., 2004). Thus, the argument that Philippine
respect to demand, which seems in line with the results of carriers are reluctant to pass the average cost per kilometer on to
Dresner and Tretheway (1992) and Marin (1995) who report passengers on some routes that have little demand for fear of
highly significant coefficients of 0.74 and 0.08, respectively. further depleting passenger traffic is not unreasonable. Since
The frequency (FREQ) variable is highly significant and, as longer routes have lower cost per kilometer because airlines are
expected, has a negative impact on airfare per kilometer, since a able to distribute their fixed costs over a longer distance, the
higher supply leads to a lower price when demand is constant. negative sign may mean that Philippine carriers inflate airfares
With a coefficient of 0.51, however, frequency is inelastic in in longer routes. The airlines income statements indicate that
relation to price. The results of this study conform to those of the gross passenger revenue exceeds the variable component of
Graham et al. (1983) who use NEWCERT, a variable direct operating expenses across airlines. The airlines also
representing the presence of new carriers in a route in their report positive gross operating income, which suggests that
fare equation, with estimates ranging from 0.27 to 0.18, all each airline is able to cross-subsidize routes with little traffic by
highly significant. Since additional carriers serving a route tend charging more per kilometer on longer and high-density routes,
indicating that cross-subsidization is profitable.
Table 1 Graham et al. (1983) argue that the US CAB set fares below
Fare equation
cost in short-haul markets and above cost in long-haul markets
Variable Coefficient Standard error T-statistic Probability during regulation. The reluctance of Philippine carriers to
PSGR 0.45 0.06 7.932053 0.0000 increase fares to reflect costs in routes with relatively less traffic
FREQ 0.51 0.08 6.525119 0.0000 for fear of reducing demand, and their practice of charging
INCM 0.86 0.04 21.329210 0.0000
higher fares in routes where demand is relatively price inelastic
DIST 0.51 0.08 6.477909 0.0000
VCST 0.13 0.06 2.097952 0.0363 tend to indicate that some form of cross-subsidization persists.
LIBR 0.16 0.04 3.654001 0.0003 Besides cross-subsidization, heavy discounting during lean
FCRI 0.09 0.02 4.085985 0.0000 seasons may have caused the negative relationship between cost
TERR 0.13 0.04 3.175316 0.0016 and fare. Since the government deregulated fare setting in
Adjusted R2 0.9266
markets with at least two airlines, some airlines are likely to set
Included observations 230
fares below average cost during off-peak periods to induce
Dependent variable: airfare per kilometer (FARE). demand and increase market share at the expense of rivals.
Estimation method: generalized method of moments.
Instruments: INCM, POPN, DIST, CPI, OPTR, LIBR, FCRI, TERR. Besanko et al. (2004) argue that airlines offer big discounts
Significant at the 1% level (highly significant). during lean seasons because selling a seat near marginal cost but
Significant at the 5% level. below average cost is better than not being able to sell the seat.
166 W.S. Manuela Jr. / Journal of Business Research 60 (2007) 161167

fare = ( / (ln y + )), where is the coefficient of LIBR and y

is the average airfare under liberalization. Borenstein (1992)
reports similar findings in markets with at least two active
airlines. The reduction in airfare under liberalization means that
regulation is likely to result in higher fares as observed by
Keeler (1972), Olson and Trapani (1981), Graham et al. (1983),
Kahn (1988), and Rietveld et al. (2002). In addition, Moore
(1986) reports that markets with more than four carriers
experienced a reduction in airfares by 24 to 29% in 1983
when compared with their 1976 levels. Thus, although the
maximum number of airlines serving a number of routes in the
Philippines is only three after the failure of Grand Air,
Fig. 1. Average fares for routes served by Grand Air. Source of basic data: Civil
liberalization results in lower average fares on routes with at
Aeronautics Board. CEB stands for ManilaCebu City, DVO stands for
ManilaDavao City, TAC stands for ManilaTacloban, CGY stands for least two airlines (see Borenstein, 1992). This observation is
ManilaCagayan de Oro, ILO stands for ManilaIloilo City, and PPS stands consistent with findings that indicate prices are lower in markets
for ManilaPuerto Princesa. with two sellers instead of only one and that price competition
intensifies when the number of sellers increases from two to
The airlines practice of heavy discounting in the early years three (Besanko et al., 2004).
of liberalization in the Philippines appears to be the major cause The dummy variables representing the Asian financial crisis
of Grand Air's demise. The fare war that ensued between the (FCRI) and the terrorist attack (TERR) on the US behave as
upstartsGrand Air, Cebu Pacific, and Air Philippinesto expected. Specifically, the Asian financial crisis results in over
grab market share probably resulted in airfares below average 5% increase in the average fare per kilometer between 1997
costs. Fig. 1 shows that average fares on routes served by Grand and 1998, computed as fare = ( / (ln y + )), where is the
Air in 1998 are lower than in 1997, despite the fall in the value coefficient of FCRI and y is the average airfare in 1997. The
of the Philippine peso during the same period, which increased terrorist attack on September 11, 2001 results in almost 9%
the cost of jet fuel and other imported inputs in peso terms. increase in the average fare per kilometer from the previous
Since the price elasticity of demand for jet fuel is inelastic, year, computed as fare = ( / (ln y + )), where is the
ranging from 0.00 to 0.15 (Pindyck and Rubinfeld, 1998), coefficient of TERR and y is the average airfare in 2000. The
Grand Air's operating cost most likely increased during this two shocks tend to result in higher prices as airlines adjust
period, exacerbating the airline's already dire financial fares to reflect the increasing cost of operations. These shocks
condition. The 0.13 coefficient reported in Table 1 means that appear to reduce the downward impact of airline liberalization
a cost increase does not fully translate into an increase in airfare. on fare.
Marin (1995) and Rietveld et al. (2002) report similar findings.
In the Philippines, regulatory lags that are inevitable in the 7. Conclusion and policy implication
review process, which persisted until 1995, may partly explain
why increases in cost do not fully translate into fare increases. This paper has empirically explored the impact of airline
Moreover, the ensuing practice of CAB to regulate fares in liberalization on fare using a sample of ten routes with varying
markets with only one airline probably contributed to the low market characteristics and state of competition for the period
coefficient of the cost variable. Since only ManilaCebu City 19812003. The results of the estimated fare equation indicate
and ManilaDavao City attracted entry in 1995, most of the that the average fare per kilometer on routes served by at least
routes included in the study remained regulated until 1996 and two airlines is, on average, 10% lower. Twenty-three routes,
three of the city-pairs in the sample reverted to single-airline representing more than 90% of total domestic passengers in
markets by 2001. Documents from CAB indicate that airlines 2003, have at least two airlines, implying that most passengers
still use distance-based formula for fare setting, which is useful benefit from lower fares. Therefore, although deregulation and
in helping airlines decide on the level of airfare that may liberalization encourage airlines to price discriminate to
stimulate demand while minimizing the impact of discount fares improve profitability and increase demand (Kahn, 1988),
on profits. Philippine domestic airline passengers are better off under
The distance (DIST) variable has the expected signairfare liberalization because of the downward pressure on fare as a
per kilometer falls with distanceand is highly significant and result of the prevalence of discounts and promos due to com-
inelastic with a coefficient of 0.51, which seems in line with the petition (Graham et al., 1983; Moore, 1986; Rietveld et al.,
0.40 and 0.33 reported by Rietveld et al. (2002) and Marin 2002).
(1995), respectively. Graham et al. (1983) obtained similar
results in their study of the US domestic airline industry References
average fare per mile declines up to 1000 miles then remains
ABN-AMRO. Costs key to a runway success. ABN-AMRO; 2002 [Online]
almost the same up to 2000 miles, and then declines again. available at, accessed on May 9, 2005.
The liberalization of the Philippine domestic airline industry Bailey EE. Contestability and the design of regulatory and antitrust policy. AEA
results in a 10% reduction in the average fare, computed as Pap Proc 1981;71(2):17883.
W.S. Manuela Jr. / Journal of Business Research 60 (2007) 161167 167

Bailey EE. Airline deregulation: confronting the paradoxes. Regul Mag 1992;15 Kahn AE. Airline deregulation. In: Henderson DR, editor. The concise
(3):1-13 [Online] available at encyclopedia of economics. Indianapolis: Liberty Fund, Inc.; 2002. [Online]
reg15n3-bailey.html; accessed on December 17, 2004. available at;
Baltagi BH. Econometric analysis of panel data. 2nd ed. West Sussex, England: accessed on December 17, 2004.
John Wiley and Sons; 2001. Keeler TE. Airline regulation and market performance. Bell J Econ Manage Sci
Besanko D, Dranove D, Shanley M, Schaefer S. Economics of strategy. 3rd ed. 1972:399424 [Autumn].
New York: John Wiley and Sons; 2004. Land Transport Franchising and Regulatory Board for air-conditioned fares of
Borenstein S. The evolution of US airline competition. J Econ Perspect 1992;6 representative bus companies.
(2):4573. Maillebiau E, Hansen M. Demand and consumer welfare impacts of
Brander JA, Zhang A. Market conduct in the airline industry: an empirical international airline liberalization: the case of the North Atlantic. J Transp
investigation. Rand J Econ 1990:56783 [Winter]. Econ Policy 1995;29(2):11536.
Civil Aeronautics Board for airline-related data. Marin PL. Competition in European aviation: pricing policy and market
Douglas GW, Miller JC. Quality competition, industry equilibrium, and structure. J Ind Econ 1995;43(2):14159.
efficiency in the price-constrained airline market. Am Econ Rev Maritime Industry Authority for sea transport fares of representative passenger
1974;64:65769. shipping companies.
Dresner M, Tretheway MW. Modeling and testing the effect of market structure Moore TG. U.S. airline deregulation: its effects on passengers, capital, and labor.
on price: the case of international air transport. J Transp Econ Policy J Law Econ 1986:128 [April].
1992;26(2):17184. National Statistical Coordination Board, Philippine Statistical Yearbook. Makati
Graham DR, Kaplan DP, Sibley DS. Efficiency and competition in the airline City, Philippines, 2003.
industry. Bell J Econ Manage Sci 1983:11838 [Spring]. Olson CV, Trapani III JM. Who has benefited from regulation of the airline
Greene WH. Econometric analysis. 3rd ed. Upper Saddle River, New Jersey: industry? J Law Econ 1981:7593 [April].
Prentice Hall International; 1997. Pindyck RS, Rubinfeld DL. Microeconomics. 4th ed. Upper Saddle River, New
Intriligator MD. Econometric models, techniques, and applications. Upper Jersey: Prentice Hall; 1998.
Saddle River, New Jersey: Prentice Hall; 1978. Reitveld P, Schipper Y, Nijkamp P. Estimating welfare effects of European
Johnston J, DiNardo J. Econometric methods. 4th ed. New York: McGraw-Hill; airline liberalization. World Conference on Transportation Research Paper
1997. No. 1354. Amsterdam, the Netherlands: Department of Spatial and
Jorge-Calderon JD. A demand model for scheduled airline services on Environmental Economics, Free University; 2002. [March].
international European routes. J Air Transp Manag 1997;3(1):2335. Verbeek M. A Guide to modern econometrics. New York: John Wiley and Sons;
Judge GJ, Hill RC, Griffiths WE, Luetkepohl H, Lee T-C. Introduction to the 2000.
theories and practice of econometrics. 2nd ed. New York: John Wiley and Wooldridge JM. Econometric analysis of cross section and panel data.
Sons; 1988. Cambridge, MA: The MIT Press; 2002.
Kahn AE. Surprises of airline deregulation. Am Econ Rev 1988:316 22 [May].