Professional Documents
Culture Documents
iMPORTANT rESEARCH pAPER-pricing PDF
iMPORTANT rESEARCH pAPER-pricing PDF
ABSTRACT
This study examines the association between when an airline sells its passenger
seats and the pricing method (marginal cost or full cost) it employs. Prior litera-
ture suggests that when firms are able to change prices during the selling period,
the optimality of full cost pricing or marginal cost pricing depends on when
demand information is revealed during the period between capacity commitment
decisions and time of sale. Full cost-based pricing is appropriate in determining
capacity commitment and prices simultaneously, while marginal cost provides
more relevant information for pricing when capacity has been committed. Using
the price and cost data from a sample of four U.S. domestic airlines, we find that
full cost explains price variations of first-day sales robustly. The adjusted R2 of the
marginal cost pricing model is larger in the sample of sales two days prior to
departure than in the sample of first-day sales. In the analysis of the sample of
sales two days prior to departure, we find that, based on the adjusted R2 of the
full cost pricing and marginal cost pricing models, the explanatory power of mar-
ginal cost pricing is relatively weaker than full cost pricing. Our results document
the use of different cost information along the dynamic change of price and pro-
vide implications in understanding the role of cost information in setting prices.
Keywords Airline industry; Full cost; Marginal cost; Pricing
* We would like to thank CMA Canada, the Canadian Academic Accounting Association, the Schulich
School of Business, and the Villanova School of Business for their financial support.
relatives aux prix et aux coûts d’un échantillon de quatre lignes aériennes nationales
des États-Unis, les auteurs constatent que le coût complet explique en grande partie
les fluctuations de prix observées dans les ventes de la première journée. La valeur R2
ajustée du modèle des coûts marginaux est plus élevée dans l’échantillon des ventes
deux jours avant le départ que dans l’échantillon des ventes de la première journée.
Dans l’analyse de l’échantillon des ventes deux jours avant le départ, les auteurs no-
tent que, selon le R2 ajusté des modèles d’établissement des prix fondés sur le coût
complet et sur les coûts marginaux, le pouvoir explicatif de l’établissement des prix
en fonction des coûts marginaux est relativement plus faible que celui de l’établisse-
ment des prix en fonction du coût complet. Les résultats obtenus par les auteurs vien-
nent documenter l’utilisation de données différentes, en ce qui a trait aux coûts, au fil
de l’évolution dynamique des prix et contribuent à la compréhension du rôle de
l’information relative aux coûts dans l’établissement des prix.
Mots clés : coût complet, coûts marginaux, établissement des prix, industrie
aérienne
This paper investigates how service firms, when facing uncertain demand, dynami-
cally adjust service prices based on cost accounting information. The role of cost
information in pricing decisions draws extensive attention in the accounting
literature. However, the optimality of marginal cost pricing or full cost pricing is
still an unresolved debate (e.g., Banker and Hughes, 1994; Balakrishnan and Sivara-
makrishnan, 1996; Noreen and Burgstahler, 1997). Extant accounting literature sug-
gests that identifying the appropriate cost information based on business settings is
essential for achieving the optimal pricing policy (Göx, 2002; Balakrishnan and Siv-
aramakrishnan, 2002; Banker, Hwang, and Mishra, 2002). One of the important
characteristics to determine the efficiency of cost-based pricing is a firm’s ability to
adjust prices after capacity is committed (Göx, 2002; Balakrishnan and Sivarama-
krishnan, 2002). When firms face uncertain demand, price adjustment is often neces-
sary in response to the updated demand information. After firms have committed
service capacity, they need to update their prices upon receiving additional demand
information. In a wide range of industries, managers face a crucial problem: how to
sell inventories or capacity before a deadline. For example, in the fashion industry,
an off-season item typically either has to be sold at a heavily discounted price or
scrapped. Firms in service industries, such as hotels or airlines, also encounter this
problem because they have to perform their services at a specific time and need to
sell services before service capacity expires. In the hotel industry, for example, the
potential revenue from a room is lost if a room is not occupied. The lost revenue
represents an opportunity cost to the hotel. Hence, firms in industries facing prod-
ucts or service expiration have a strong incentive to sell all available products or ser-
vices prior to the expiration day. Hence, it is important to understand how firms
apply cost information in price adjustment before the capacity expires.
Balakrishnan and Sivaramakrishnan (2002) suggest that when firms are able to
adjust prices, full cost-based pricing policies are more justifiable for setting list
prices than for setting tactical prices. Göx (2002) suggests that the optimality of
full cost pricing or marginal cost pricing depends on when demand information is
revealed during the period between capacity commitment decisions and time of
sale. To examine how the point at which a firm obtains demand information
during a pricing process affects the use of full cost pricing versus marginal cost
pricing, Göx classifies three pricing scenarios, each with different levels of price
adjustability and different levels of uncertainty about demand information in
setting capacity. The results show that when a firm has no demand information
while setting capacity, but is able to modify prices, full cost provides the most
appropriate information. On the other hand, in a partial uncertainty scenario
where a firm obtains updated demand information after setting capacity, marginal
cost provides sufficient information for pricing.
In this study, we examine how firms use cost information to decide and adjust
the prices during the period between committing the capacity by beginning to sell
the services and the end of the selling period. Specifically, using price and cost data
collected from the airline industry, we investigate the association between two
types of costs (full cost and marginal cost) and pricing decisions at two points of
time (when the sale of a service is initially offered to the market and just prior to
the scheduled flight date).
We choose to investigate the airline industry because we are able to obtain the
data of the changes in ticket prices over time. We choose four companies among
the top 10 domestic airline companies in the United States. Cost information is
obtained from traffic and financial statistics of Form 41 reports submitted by the
U.S. airlines to the Federal Aviation Administration (FAA) and the Department
of Transportation (DOT). We choose two transaction times: the first day of sales
and sales two days prior to departure (Figure 1). The first day of sales is when the
airline commits the capacity to a certain route; the airline sets the price with little
demand information and is able to adjust the price when receiving updated
demand information. The sales two days prior to departure represent a partial
uncertainty scenario as in Göx (2002), when a firm obtains updated demand infor-
mation after setting capacity and changes prices.
We find that full cost provides more information compared to marginal cost in
explaining the price variation of first-day sales and prices two days prior to depar-
ture. Marginal cost is not associated with price of first-day sales and is positively
associated with prices two days prior to departure. Marginal cost provides greater
explanatory power of ticket prices than full cost; the adjusted R2 of the marginal
cost pricing model is larger in the sample of sales at two days prior to departure
than in the sample of first-day sales (80.12 percent versus 73.59 percent).
1. Using the airline industry as an example, economic researchers have developed theories on
whether price discrimination or advance-purchase discounts exist and how these practices
achieve economic efficiency (e.g., Gale and Holmes, 1992; Dana, 1998). This stream of research
centers on the relation between demand and advance-purchase discount and does not examine
the role of cost information in advance-purchase discount.
2. Operation management literature emphasizes the analysis of allocation of resources for each
fare product and demand forecast (see a review by McGill and van Ryzin, 1999).
3. We make the hard-capacity assumption because our main research question is about the
dynamic pricing process for service capacity or perishable assets. To deliver services, firms need
to commit capacity, which would be not changed at a short period. Hence, we assume that
firms face hard-capacity constraints.
4. Balakrishnan and Sivaramakrishnan (1996) argue that full cost pricing does not achieve optimal
decisions for a firm with hard-capacity constraints. Göx (2002) documents that the type of
capacity constraint influences the level of capacity.
To summarize, full costing provides more information on the prices of the ser-
vices in the beginning of sales.
As the point of capacity expiration draws closer, firms face an increasingly
higher risk that the services will not be sold. A pricing adjustment after capacity is
committed can be viewed as a sequential decision to capacity planning. Göx (2001)
examines pricing heuristics in a two-period capacity-planning model. He finds that
with respect to pricing in the second period (i.e., after the capacity has been set
and cannot be changed), full cost pricing does not reach an optimal solution. To
extend this finding, Göx (2002) demonstrates that after capacity has been commit-
ted, marginal cost pricing is optimal for firms to adjust prices upon the arrival of
additional demand information. Based on his results, most firms facing a
hard-capacity constraint should include only marginal cost to adjust the prices
once they receive additional demand information. Marginal cost, which measures
the cost associated with the sale of an additional unit of service, provides sufficient
information for setting the price for this additional sale. Hence, upon capacity
expiration, marginal cost is the only relevant information about the cost associated
with the additional sale. Furthermore, pricing decisions close to the date of capac-
ity expiration are viewed as short-term decisions.
We examine the pricing process at two stages: (a) the beginning of a selling
period and (b) the period between the arrival of new demand information and
the expiration of service. We examine how the cost information influences the
pricing decision for these two stages (Figure 1). At the beginning of a selling
period, a firm decides capacity commitment and sets service prices. Based on the
review of prior literature, full cost provides sufficient information for setting
pricing and capacity decisions (Banker and Hughes, 1994; Balachandran et al.,
1997; Balakrishnan and Sivaramakrishnan, 2002; Göx, 2002). Our first hypothesis
is as follows:
HYPOTHESIS 1. There is a stronger association between price and full cost than
between price and marginal cost when setting price of the first-day sale.
After capacity has been committed, marginal cost pricing is optimal for firms
to adjust prices upon the arrival of additional demand information (Göx, 2002),
because marginal cost measures the incremental costs associated with additional
sales. Hence, marginal cost plays an important role in pricing at the date when
capacity is close to expiration. Our second hypothesis is as follows:
HYPOTHESIS 2. There is a stronger association between price and marginal cost than
between price and full cost when capacity is close to expiration.
METHODOLOGY
To test our hypotheses, we choose four companies among the top 10 domestic
airlines in the U.S. market and survey the offered ticket prices in the U.S. domestic
airline market. We choose the airline industry for the following reasons. First, in the
TABLE 1
Summary traffic data from sample airports (2007)a
Passengers (000)
Arrival 13,581 12,253 21,458 11,756 11,999 15,891 11,676
Departure 13,581 12,242 21,431 11,694 12,058 15,719 11,648
Scheduled flights
Departures 149,528 134,068 245,670 127,399 180,794 151,202 167,093
a
Data from Bureau of Transportation and Statistics.
domestic ticket prices. Airlines are required to disclose total operation expenses,
which include fuel, nonfuel flying operating expenses, and traffic service and pro-
motion costs. Nonfuel flying operating expenses include the cost of flying opera-
tions labor, passenger services labor, maintenance labor, maintenance materials
and overhead, general overhead, and ground property and equipment. Traffic ser-
vice and promotion costs include expenses of aircraft and traffic servicing labor,
and promotions and sales labor.
We estimate cost models for each airline to get the firm-specific cost esti-
mates that refer to the cost information used for decision making. Based on
the operating expense categories of the financial data of the airlines, Banker
and Johnston (1993) estimate cost drivers for ten cost categories. They identify
volume-based drivers, including the number of passenger and capacity seat-
miles, and operation-based drivers, including aircraft types, average stage length,
density, and hub concentration. We modify the firm-specific estimation model
by including only the number of passengers, capacity seat-miles, and number of
departures. We exclude operation-based drivers such as aircraft type because the
airlines do not differentiate between different types of airplanes when making
pricing decisions for the same route. We also exclude average stage length, den-
sity, and hub concentration because these drivers do not vary significantly
within the same firm over time. Because we estimate cost models for each firm,
operation-based factors that represent the difference between airlines are not
applicable to our models.
To obtain the marginal cost per passenger, we estimate the cost drivers of fly-
ing expenses, which include fuel costs and nonfuel flying expenses. Banker and
Johnston (1993) document that fuel costs vary with capacity seat-miles and the
number of departures. Nonfuel flying operation costs are mixed costs and are also
driven by capacity seat-miles and the number of departures respectively. Traffic
and promotion expenses are costs associated with ground service activities. Banker
and Johnston (1993: 589) find that traffic and promotion expenses are positively
associated with the number of passengers and negatively associated with hub
operations. Because we sample flights through the hubs, we do not include traffic
and promotion expenses in the measure of marginal costs.
During the sample period, fuel prices vary at an unusually large range due to
fluctuations in oil prices. Historical fuel costs do not provide sufficient information
for our estimation in cost. We use fuel in gallons to estimate fuel consumption
drivers and estimate the cost model with 10-year data of each sample company as
follows:
TOTAL FUEL ðGALLONÞ ¼ a1 CAPACITY SEAT-MILES þ a2
DEPARTURES þ e ð1Þ:
Airlines provide both passenger and freight services. Because of a high cor-
relation between seat-miles and ton-miles, we combine them and calculate avail-
After estimating operating expenses for each firm, we calculate unit marginal
cost per seat-mile. Based on the estimated coefficients, we calculate the marginal
cost of fuel and non-fuel flying operation. The marginal cost of fuel (MC_FUEL)
is equal to a^1 * OIL FUTURE. We use the quote of oil future at the transaction
day because it is the most relevant information of fuel cost while firms set ticket
prices. The marginal cost of nonfuel flying operations per seat-mile (MC-NONFUEL
FLYING) is b ^ . We sum up these two components to obtain the measure of marginal
1
cost per seat-mile (MC).
In (3) and (4), MILEAGE equals the number of miles of each sampled route. We
compare the adjusted R2 to evaluate how the association between price and unit
cost information changes in the samples of first-day sales and sales two days prior
to departure.5
RESULTS
We collected transaction prices of 128 direct one-way routes for 14 days during-
June and July 2007. Table 2 presents descriptive statistics of our sample. The mean
of ticket prices at first-day sales is $393.53 and the mean of prices of sales at two
days prior to departure is $406.82. The mean of total full cost per seat-mile is
$0.040, and the mean of mileage per route is 1340. The marginal cost per seat-mile
is $0.009.
Table 3 shows the regression estimation of the association between prices and
unit full cost and marginal cost. We find that first-day sales depend more on full
cost pricing than on marginal cost pricing; the adjusted R2 is larger in the full cost
pricing model than in the marginal cost pricing model (76.79 percent versus 73.59
percent). Specifically, the coefficient on MC is insignificant, which indicates that
airlines do not consider marginal costs in setting the price on the first-day sales.
Hence, Hypothesis 1 is supported.
Using the sample of the two-day prices to analyze the models of full cost
pricing and marginal cost pricing, we find that coefficients on both MC and FC
are significant, which suggests that both full cost and marginal cost have been used
in setting the prices of two-day sales. The adjusted R2 for the full cost pricing
model is higher than that in the marginal cost pricing model (85.24 percent vs.
80.12 percent), that is, full cost provides better information in explaining price
variations of sales at two days prior to departure than marginal cost. Our Hypoth-
esis 2 is not supported.
Full costs have commonly been used in product pricing decisions (Govindara-
jan and Anthony, 1983; Shim and Sudit, 1995; Stahl, 2005). Although marginal
cost pricing theory suggests the use of incremental cost associated with additional
sales for pricing after capacity has been set (Göx, 2001), airline companies seem to
be more likely to use full cost pricing than marginal cost pricing in setting the
prices when capacity is close to expiration.
5. Capital market-based accounting research has widely applied the adjusted R2 comparisons to
evaluate relative information content of various accounting measures (Dechow, 1994; Biddle,
Seow, and Siegel, 1996; Ely and Waymire, 1999; Lev and Zarowin, 1999; Davis, 2002; Francis
and Schipper, 2003). For instance, Lev and Zarowin (1999) have used the adjusted R2s as the
primary metric to evaluate changes in the value-relevance of accounting measures, earnings,
and book values.
TABLE 2
Descriptive statistics
Notes:
a
FC PER SEAT-MILE: the sum of the average fuel costs per seat-mile and the average nonfuel
operation expenses; the average fuel costs per seat-mile equals FUEL (GALLON)t1 * OIL
FUTURE/ CAPACITY SEAT-MILESt1 and the average nonfuel operation expenses per seat-
mile equals the total NON-FUEL OPERATIONS EXPENSESt1/CAPACITY SEAT-MILESt1.
b
MC PER SEAT-MILE: the sum of the marginal cost of fuel per seat-mile (MC_FUEL) and the
marginal cost of nonfuel flying operations per seat-mile (MC-NONFUEL FLYING).
MC_FUEL is equal to â1 * OIL FUTURE where â1 is the coefficient on CAPACITY-SEAT-MILES
from the model, TOTAL FUEL (GALLON) = a1* CAPACITY SEAT-MILES + a2 *
DEPARTURES + e.
MC-NONFUEL FLYING is equal to b^1 where b^1 is the coefficient on CAPACITY SEAT-MILES
from the model, NONFUEL FLYING OPERATIONS = b0 + b1* CAPACITY SEAT-MILES
+ b2* DEPARTURES + e.
TABLE 3
Regression estimation of the association of prices and cost information
Variables Prices of first-day sales Prices of sales two days prior to departure
Variables Prices of first-day sales Prices of sales two days prior to departure
*,**,*** significant at 10 percent, 5 percent, and 1 percent levels, respectively, using a two-tailed t-
test; t-values shown in parentheses.
CONCLUSION
Accounting researchers have documented the importance of identifying appropriate
cost information in the pricing process (Göx, 2001, 2002; Balakrishnan and Sivara-
makrishnan, 2002). Using a sample of four U.S. domestic airlines with 1,265 ticket
transactions, we find that firms use full cost but not marginal cost in setting prices
at the beginning of a selling period when the firm decides capacity commitment.
Since marginal costs measure the costs directly associated with an additional sale,
firms increase the weight of marginal cost in setting short-term prices. However,
full costs still provide more explanatory power in explaining the variation of short-
term prices. This study highlights the role of full costs in setting service prices. The
limitations of this study could provide opportunities for future research. First, we
have sampled only the airline industry; airlines have a large proportion of costs
committed to fixed capacity resources compared to many other industries. Future
studies using different industries could examine the role of cost information in pric-
ing. Second, this study examines the association between prices and costs using
archival data. Field-based studies may reveal how managers with updated demand
information identify costs during a pricing process and provide better understand-
ing of the influences of cost information on pricing. Future research could also
examine whether senior management’s incentives play a role in the cost informa-
tion considered in pricing.
REFERENCES
Balachandran, B. V., R. Balakrishnan, and K. Sivaramakrishnan. 1997. On the efficiency
of cost-based decision rules for capacity planning. The Accounting Review 72 (4): 599–
619.
Balakrishnan, R., and K. Sivaramakrishnan. 1996. Is assigning capacity costs to products
really necessary for capacity planning? Accounting Horizons 10 (3): 1–11.
Balakrishnan, R., and K. Sivaramakrishnan. 2002. A critical overview of the use of full-
cost data for planning and pricing. Journal of Management Accounting Research 14 (1):
3–31.
Banker, R. D., and J. S. Hughes. 1994. Product costing and pricing. The Accounting
Review 69 (3): 479–94.
Banker, R. D., I. Hwang, and B. K. Mishra. 2002. Product costing and pricing under
long-term capacity commitment. Journal of Management Accounting Research 14 (1):
79–97.
Banker, R. D., and H. H. Johnston. 1993. An empirical study of cost drivers in the U.S.
airline industry. The Accounting Review 68 (3): 576–601.
Belobaba, P. P. 1989. Application of a probabilistic decision model to airline seat inven-
tory control. Operations Research 37 (2): 183–97.
Biddle, G. C., G. Seow, and A. Siegel. 1996. Relative versus incremental information con-
tent. Contemporary Accounting Research 12 (1): 1–23.
Bitran, G., and R. Caldentey. 2003. An overview of pricing models for revenue manage-
ment. Manufacturing and Service Operations Management 5 (3): 203–29.