How event significance, team quality, and school proximity affect secondary market behavior at March Madness.
Rishe, Patrick J. ; Mondello, Michael ; Boyle, Brett 等
Introduction
There is considerable research which discuss how sport ticketing
professionals engage in inelastic ticket pricing (Fort, 2004; Krueger,
2001), and that such behavior is not in contrast with
profit-maximization because it enables organizations to bolster other
non-ticket sources of revenue (Coates & Humphreys, 2007; Krautmann
& Berri, 2007). Separately, there is long-standing evidence from the
marketing literature of a strong correlation between the price of a
product and perceptions of product quality (Monroe & Krishnan, 1985;
Scitovsky, 1945; Tsao, Pitt & Caruana, 2005).
The secondary market for sports tickets affords an opportunity to
(1) confirm whether organizations systematically engage in inelastic
ticket pricing, and (2) explore the relationship between perceptions of
quality and secondary price markups over face values. Academic research
regarding the secondary ticket market has become more voluminous in
recent years but is still relatively underdeveloped. This is
particularly the case as it pertains to collegiate sporting events and
specifically college basketball, where there has been tremendous growth
in secondary market revenues. For example, the estimated dollar volume
of total secondary market sales increased from $2.6 billion in 2011 to
$4.0 billion in 2013, with increases from $420 million to $676 million
over the same span for college sports alone, and increases from $100
million to $160 million for March Madness alone. (1)
Studying secondary market behavioral trends at March Madness (i.e.,
the Division I Men's College Basketball tournament) presents a
unique opportunity to examine whether an organization (in this case, the
National Collegiate Athletic Association) is able to price tickets in
the inelastic range of consumer demand when dealing with two unique
anomalies not typically confronted by sports organizations. First, there
are several micro-events (e.g., four Regional Finals and the Final Four)
contained within the overall tournament, none of which are held annually
in the same city. Second, there is greater consumer uncertainty
regarding perceptions of event quality for these micro-events because
the event participants (i.e., the qualifying teams) are not known until
5-6 days before the event occurs. Due to the added travel and logistical
costs for non-local fans associated with last-minute travel (e.g.,
airfares), greater event uncertainty could impact event demand, and
subsequently, secondary markups associated with tickets for the event.
Therefore, the primary goals of this paper are: 1) examine the
secondary price behavior associated with March Madness to determine
whether the afore-mentioned anomalies impact the NCAA's ability to
set their face prices where demand is inelastic; and 2) determine which
factors are the most influential in determining the magnitude of
secondary markups. In terms of theoretical contributions, we extend the
pricing literature by examining pricing behavior within the unique
consumer product class of sports entertainment, focusing on secondary
ticket pricing where the "product" being priced (i.e., the
ticket) is not the primary revenue driver. Another key contribution is
the revelation that the sports organization may not always be able to
price tickets in the inelastic range of demand when there is
considerable a priori uncertainty and lack of control regarding an
event's quality. Lastly, the evidence supports the notion that
consumers' perceptions of event significance and team quality, as
well as logistical travel considerations which impact fans' total
cost of attending an event, directly impacting the magnitude of
secondary markups.
Tournament and Ticketing Logistics of March Madness
March Madness admits 68 teams in a single-elimination tournament
taking place over three weeks from mid-March through early April.
Thirty-two of the 68 teams automatically qualify by winning their
end-of-season conference tournament, while the other 36 teams are
at-large selections chosen by a Selection Committee mostly comprised of
senior-level Division I athletic administrators (e.g., athletic
directors, conference executives). Apart from selecting the at-large
teams, the Selection Committee is charged with the arduous tasks of (1)
dividing the teams into four similarly competitive Regionals, (2)
seeding the teams within each Regional, and (3) determining which teams
should play where during the first weekend of the tournament. (2)
Achieving these "selection goals" simultaneously in a way
that promotes both ticket sales and competitive balance is a challenging
endeavor. Consider that The Ohio State University (located in Columbus,
OH) was selected as the #2 seed in the West Region for the 2013
tournament. The reason the Buckeyes were placed in the West Region was
to add more competitive depth to that region, though doing so made it
less likely for Buckeye fans to make a last-minute cross-country trip to
Los Angeles for the Regional. However, the Selection Committee placed
Ohio State in nearby Dayton, OH for their first two tournament games for
the likely reason of boosting ticket sales for early-round games (given
the close proximity of Dayton to Columbus). In short, it would appear
that the NCAA Selection Committee's primary scheduling goal in the
early rounds of the tournament is to maximize ticket sales by ensuring
that the better teams within a particular pod are reasonably close to
their home city, while the primary goal of the Regional allocation of
teams is to minimize the disparity in competitive balance across the
Regionals (with proximity being an added bonus if it comes to fruition
as was the case with Louisville playing the 2013 Midwest Regional in
nearby Indianapolis).
Regarding ticket allocation for both the Regionals and Final Four,
the NCAA allocates tickets to a variety of groups including the Local
Organizing Committee, the general public, the participating schools, the
National Association of Basketball Coaches (NABC), the NCAA's
corporate partners, and a general allocation to member NCAA schools. One
key distinction is that the NCAA only sells "All-Session"
ticket packages to the Final Four (i.e., a three-game package including
both semifinals and the finals). Conversely for the Regionals, the NCAA
sells All-Session packages as well as separate tickets for just Session
1 (the Regional semifinals) and Session 2 (the Regional Finals).
Regarding face value ticket pricing, the NCAA sets the face value for
all March Madness tickets well before the tournament field is set. The
only factor known with certainty prior to their price setting is the
facility capacity for each venue. For the 2013 Final Four, there were
seven separate 'price points' assigned for the "All
Session" ticket packages ($5000, $420, $280, $250, $230, $190,
$175). For the Regionals, Table 1 below details the various face value
prices associated with each ticket type sold at each host venue. (3)
Given that each micro-event within the tournament takes place in
different cities and involves different groupings of teams, this
conveniently lends itself to an examination of how different "event
draws" appeal to consumers, and thus, produce differing secondary
market behavior. Based on prior conversations with NCAA representatives,
local event organizers and local tourism officials, a "quality
draw" of teams is desirable for boosting both ticket sales and
general tourism activity. (4) Furthermore, these sources note that a
"quality draw" of teams reflects some combination of (1)
highly ranked teams from that year's national polls, (2) teams with
fan bases who historically travel well, and/or (3) teams that are
located within a reasonable driving proximity to the event. (5)
Literature Review
Inelastic Sports Pricing
Though the concept of inelastic sports pricing has garnered
considerable attention for decades because of its potential implications
regarding whether sports organizations are exhibiting profit-maximizing
behavior, several papers noted that complementary revenues stemming from
products tied to the sporting event may make up for the lost gate
revenues resulting from inelastic pricing. Krueger (2001) observed that
face value prices for the 2001 Super Bowl were priced below what
consumers were willing to pay on the secondary market, likely because
the NFL earned sufficiently high revenues from media revenues. Fort
(2004) used gate and media revenue data from Major League Baseball to
show that inelastic pricing was consistent with profit maximization if
local television revenues compared favorably to the average marginal
television revenue across the rest of the league. Extending the scope of
research to include both professional baseball and basketball, Coates
and Humphreys (2007) argued that inelastic ticket pricing is consistent
with profit maximizing behavior due to the inter-related pricing
decisions on tickets, concessions, and other related goods sold at these
sporting contests. Examining the four major team sports leagues in North
America, Krautmann and Berri (2007) argued that concession revenues
allow professional sport teams to discount their ticket prices
significantly.
Correlation Between Perceptions of Product Quality and Willingness
to Pay
Scitovsky (1945) first suggested price serves as a cue for product
quality. Consumers were particularly dependent on this cue when other
more salient attributes of the product are absent (Tsao, et al. 2005).
Much of the conceptual foundation of the current study draws from Monroe
and Krishnan's (1985) price-perceived quality model, which defines
price as having an influence on both perceived quality as well as
perceived sacrifice. In turn, perceived value is the net of these two
factors, the former having a positive effect on perceived value, the
latter a negative effect. Subsequently, one's willingness to buy is
seen as directly affected by perceived value. (6) Their work is
particularly germane to the present study. Tournament seeding serves as
an initial proxy of product (team) quality while allowing the most
talented teams the best chance to advance to later rounds by pairing
them with easier early-round opponents. This creates the potential for
maximizing the perceived overall quality of later-round tournament
games, given the advancing of top teams.
However, history has shown that tournament game results rarely--if
ever--follow true-to-form with regard to seeding. Consider the 2014
tournament, where the #3 seed in the Midwest (Duke) lost in the first
round to #14 seed Mercer. Hence, the perceived value of the 2014 Midwest
Regional--which initially was high because of the potential matchups
involving a blueblood Duke program as well as 2013's national
runner-up Michigan--diminished rapidly when the better teams failed to
advance through the tournament. In short, perceived value (via perceived
quality) is highly dynamic within the context of the March Madness
Tournament. Though we do not directly introduce game results into our
analysis, nevertheless the intrigue of March Madness as a sporting event
rests heavily on the unpredictability of the Tournament, thus providing
a rich environment to study secondary price elasticities.
Intuitively, the quality of the March Madness product would also
include the location of the seats purchased in the game venue. A study
by Leslie and Sorensen (2014) using secondary ticket sales data to 56
major rock concerts found the best seats were approximately four times
more likely to be resold than low-to-mid quality tickets. Similarly,
Salaga and Winfree (2013) examined the secondary ticket market for the
NFL and specifically for personal seat licenses (PSL) and season ticket
rights (STR). High-quality seating locations were a strong predictor of
price, and higher face value ticket prices were also associated with
lower secondary market PSL and STR transaction prices.
Perceived sacrifice is also a unique aspect of consumption of the
March Madness product. Unlike single-site sporting events such as the
Super Bowl, fans of different Tournament teams are required to travel
different distances, paying correspondingly varied prices to reach their
team's game. Such travel costs no doubt play a role in a fan's
calculus of the value of a secondary market ticket. We therefore
additionally incorporate travel distance (along with proxies for
perceived quality) in assessing consumers' willingness to pay above
face value on the secondary market.
Secondary Pricing Behavior, Dynamic Pricing, and Revenue Management
Because inelastic ticket pricing in the primary market drives
secondary market pricing decisions, a review of the secondary price
literature is warranted. Using data from online bidding transactions
from the secondary ticket market for NFL playoff games, Drayer and
Shapiro (2009) found that secondary prices were significantly related to
the total number of online ticket transactions per game, the home
team's city per capita income, and the home team's city
population. In a separate study for NFL regular season games, Drayer,
Rascher, and McEvoy (2012) found the point spread, percent of capacity
sold, the presence of a new stadium, a team's comparative quality,
the face value of tickets, and current and lagged win percentages each
had a statistically significant impact on the value of secondary ticket
prices.
Additionally, several studies have addressed why dynamic pricing is
critically important for revenue maximization in today's sports
environment. Drayer, Shapiro, & Lee (2012) noted sport organizations
needed to evolve and develop ticketing strategies capitalizing on the
secondary market since other revenue sources like stadium naming rights
and sponsorship inventories are limited. Rascher et al. (2007) concluded
a general variable ticket pricing policy for the San Francisco Giants
could have increased the team's revenue by roughly 7%. Separately,
using data from the San Francisco Giants in comparing the team's
dynamic pricing policy to prices arising on the secondary market,
Shapiro and Drayer (2012) found prices from the previous year, the
current year's season ticket prices, the Giants winning percentage
as well as that of opposing teams, and how far back the Giants were from
first place all impacted the optimal dynamic price for a given game. Zhu
(2014) tested MLB primary and secondary ticket data for one franchise to
gauge how the team could increase revenue with dynamic ticketing, and
found that the franchise could increase revenue by 6.93% when dealing
with uninformed consumers but only by 3.67% when dealing with customers
more savvy with the intricacies of the secondary market.
Secondary Market Behavior in Collegiate Sports and March Madness
Though there has been little research regarding the dynamics of the
secondary market for collegiate sporting events, Rishe (2014a) analyzed
a sample of 285 online transactions involving 750 tickets sold to the
2013 Final Four. He found that (1) the NCAA under-prices tickets to the
Final Four, (2) the magnitude of the secondary price markups differed
across 'Session Type' (e.g., Session 1 tickets had
significantly higher markups than All Session packages or Session 2
tickets), (3) markups per ticket decreased as consumers bought more
tickets within a single transaction, (4) tickets in the upper-bowl
experienced lower markups than all other tickets, and (5)
centrally-located tickets experienced greater markups than all other
tickets. However, by only analyzing secondary price trends at the Final
Four, this analysis failed to consider how secondary market behavior
might differ across micro-events (Regional Finals, Final Four) that
differ in terms of team-specific factors (e.g., quality, proximity) and
local factors (incidence of locals buying tickets, facility capacity).
Theoretical Model and Data
The primary investigative task is to examine whether secondary
market behavior significantly differs across the micro-events of March
Madness which differ in terms of both event significance and the
attractiveness of an event's draw. The model specifications to
follow are founded on the following hypotheses. First, there is a direct
correlation between perceived product quality and willingness to pay.
Second, there is an inverse correlation between the ancillary costs of
attending a sporting event (such as distance to travel) and willingness
to pay. Third, there is a direct correlation between seat quality and
willingness to pay. Fourth, secondary markups will differ across
"session type" and the volume of tickets purchased within a
transaction as found in Rishe (2014a).
Hence, the initial model specification employed to examine March
Madness secondary market behavior is: MU = a + b(ALL) + c(CHAMP) +
d(LOW) + e(HIGH) + f(CENT) + g(QTY) + h(QTY2) + i(STATE) + j(TIME) +
k(LOCAL) + l(EAST) + m(MIDWEST) + n(SOUTH) + o(wEST),
(Model 1)
where
* MU = the size of the markup (i.e., difference between secondary
and face value price) as a percentage of the face value price. (7)
* ALL = 1 if an All Session ticket, 0 if a Session 1 (i.e.,
Semifinals) ticket.
* CHAMP = 1 if a Session 2 (i.e., Finals) ticket, 0 if a Session 1
ticket.
* LOW = 1 for tickets in the lower bowl (100 level sections) of a
facility, 0 for tickets in the middle bowl (200-level sections).
* HIGH = 1 for tickets in the upper bowl (300 level sections), 0
for tickets in the middle bowl (200-level sections).
* CENT = 1 for centrally located seats located in sections between
either baseline, 0 otherwise.
* QTY = the number of tickets in a given online transaction.
* QTY2 = QTY squared.
* STATE = 1 if the zip code associated with the online buyer was
from one of the four states represented in a particular regional or
Final Four, and 0 otherwise. (8)
* TIME = the difference (in days) between when the online purchase
was made and the tip-off of the event.
* EAST, MIDWEST, SOUTH, and WEST are dummy variables representing
each Regional, and have a value of one for tickets to that Regional, and
zero otherwise. Hence, tickets purchased for the Final Four is the
control group.
* LOCAL = 1 if the consumer has a zip code which is local to the
host city of the event in question, and zero otherwise.
Given that the micro-events analyzed herein differ in terms of
event significance (the Final Four is the crescendo of the overall
tournament), the aggregate quality of participating teams, the
collective proximity of participating schools from each event's
host city, and event attendance as a percentage of capacity, an
alternative model which potentially offers greater specificity regarding
the source of fluctuations in secondary markup differentials is:
MU = a + b(ALL) + c(CHAMP) + d(LOW) + e(HIGH) + f(CENT) + g(QTY) +
h(QTY2) + i(STATE) + j(TIME) + k(LOCAL) + l(RPI) + m(PROX) + n(ATTCAP) +
o(FF).
(Model 2)
In Model 2, the Regional dummy variables are replaced by RPI, PROX,
ATTCAP, and FF. RPI stands for Ratings Percentage Index, and is measured
as the median RPI of the four participating schools at a particular
micro-event. The expected sign is negative because larger RPIs imply a
less attractive draw of teams which should lower markups. (9) PROX is
the median distance in miles between the participating schools and the
host city of the micro-event in question. The expected sign is negative
based on the hypothesis that teams playing further away from home will
be less likely to boost event demand as their fans will be less likely
to make the trip due to the added costs associated with farther
distances to travel. ATTCAP is attendance as a percentage of capacity,
and the expected sign on this coefficient is positive based on the
hypothesis that greater attendance as a percentage of capacity implies
greater scarcity of tickets, which should increase secondary prices and
corresponding markups. Lastly, FF is a dummy variable with a value of
one for tickets to the Final Four and zero for tickets to any of the
Regionals. The expected sign on this coefficient is positive given that
the Final Four represents the crescendo of the tournament.
Regarding the data employed, TicketCity.com (TC) provided data for
831 secondary market transactions which resulted in 2,160 tickets sold
during the 2013 tournament. (10) Of the 831 total secondary market
transactions, 117 transactions were for the East Regionals, 102 were for
the Midwest Regionals, 231 were for the South Regionals, 96 were for the
West Regionals, and 285 were for the Final Four. Table 2 below breaks
down the number of consumer transactions for each micro-event by
session.
Results
Descriptive Summary Statistics
Table 3 below summarizes the teams that competed in each Regional
as well as the Final Four. Focusing on comparisons of the median RPI
ranking of the teams in each Regional and the median distance teams
traveled to reach the host city where their Regional was held, the
evidence suggests Indianapolis had the most attractive draw (e.g., the
lowest median RPI and lowest median distance to travel), Los Angeles had
the least attractive draw (e.g., the highest median RPI and highest
median distance to travel), and the East Regional had a better quality
draw than the South Regional. Not surprisingly (as shown in Table 4
below), the Regionals with the better draws (East, Midwest) had higher
attendances as a percentage of capacity than the Regionals with less
attractive draws (South, West).
Table 5 presents descriptive statistics for mean and median
secondary prices, face value prices, and markups (both in dollars and
percentage terms) for each Regional and the Final Four, broken down by
Session Type. Consistent with Rishe (2014a), it would appear that markup
percentages are consistently higher for Session 1 compared to
All-Session or Session 2. However, in a significant contrast to the
Final Four findings, average and median markups are negative for each
Regional irrespective of Session Type (i.e., the central trend in
secondary prices is below the face value at the Regionals). Furthermore,
the degree to which secondary prices fell below face value prices was
less severe for the East and Midwest Regionals compared to the South and
West Regionals. Additionally, Table 6 shows that secondary markups were
far more common for Final Four purchases than Regional purchases, and
that the incidence of Regional markups was more common for the East and
Midwest Regionals.
In short, the descriptive statistics suggest that (1) consumers
place a significant value premium on the Final Four compared to the
Regionals, and (2) Regional markups significantly differ from each
other. The next step is to use regression analysis to further examine
the root
of these Regional markup discrepancies. (11)
Regression Results
Table 7 below present regression results for Model 1 using ordinary
least squares (OLS) estimation corrected for heteroskedasticity (H/S)
using White's heteroskedastic-consistent standard errors (HCSE).
(12,13) The overall adjusted R-squared is 0.467, suggesting that 46.7%
of the variation in secondary markups can be explained by the joint
variation of the variables in Model 1. Focusing on the Regional
coefficients, ceteris paribus, the Final Four secondary markup
percentages were approximately 54 percentage points higher than for the
East Regional, 60 percentage points higher than the Midwest Regional,
104 percentage points higher than the South Regional, and 92 percentage
points higher than the West Regional. (14)
Furthermore, all the coefficients had their expected signs while
only two variables (TIME, STATE) were statistically insignificant.
Ceteris paribus, secondary markups were: (1) 11.9 percentage points
lower for local consumers compared to non-locals, (2) 12.8 percentage
points higher for lower-tier seats and 34.8 percentage points lower for
upper-bowl seats compared to middle-tier seats, (3) 39.4 percentage
points higher for centrally-located seats, and (4) 20.0 percentage
points lower for All Session tickets and 42.0 percentage points lower
for Session 2 tickets compared to Session 1 tickets. Lastly, the signs
on QTY and QTY-squared suggest there is a non-linear inverse
relationship between the number of tickets purchased and the size of the
secondary markup per ticket.
For greater specificity regarding what may be driving differences
in secondary markups across micro-events, Table 8 below presents
regression results for Model 2 (also estimated using OLS with
White's HCSE). The overall adjusted R-squared for Model 2 (0.475)
is significantly greater than for Model 1. Furthermore, each of the
newly introduced event-specific variables has their expected sign and is
statistically significant. The coefficient on RPI suggests secondary
markups increased by 1.1 percentage points as the median RPI of the
teams within a pod fell (i.e., as the median quality of the teams
improved) by one ranking point. The coefficient on PROX suggests
secondary markup percentages increased by 1.01 percentage points for
each 100 miles closer the median proximity of the schools competing in a
pod were to the event's host city. A one percentage point increase
in ATTCAP increased secondary markups by 4.7 percentage points. Lastly,
secondary markups for the Final Four were 77.3 percentage points higher
than for the Regionals collectively. All other model coefficients had
their expected signs, while only two variables (TIME, STATE) were
statistically insignificant.
Separately, the data was partitioned and a Chow test was conducted
to determine if the regression coefficients are statistically different
when comparing secondary market behavior between the Final Four and the
Regionals. These results are presented in Table 9 below. Note that some
variables had to be dropped from the earlier specifications to ensure
the model is of full rank. The null hypothesis of the Chow test is that
the regression coefficients obtained from running separate regressions
are identical, implying that the Final Four and Regionals data can be
pooled. However, the F-statistic associated with the Chow falls well
into the rejection region, suggesting the underlying secondary markup
behavior is structurally different when comparing the Final Four to the
Regionals. Though not presented for space reasons, similar conclusions
(i.e., the appropriateness of disaggregating the data) are drawn when
running separate regressions for each Regional. Collectively, these
results imply the secondary market behavior of each micro-event within
March Madness has its own unique dynamic which is partially driven by
differences in event significance and the quality of each event's
draw.
Other variables were considered but ultimately excluded for a
variety of reasons. For example, population and income--two variables
commonly included to proxy local demand factors--were excluded because
their inclusion introduced a significant degree of multicollinearity
which impacted the statistical significance of RPI and PROX, and neither
variable was statistically significant on its own. Practically speaking,
the size and affluence of the city hosting these micro-events may be
less impactful upon secondary prices if an overwhelming majority of
event attendees are not from the host city's MSA. (15)
Separately, alternate proxies for team quality--such as national
rankings and tournament seeding--were considered. However, national
rankings are an inferior proxy of quality compared to RPI because such
rankings are determined subjectively through human voting by media
members and coaches (depending on the poll used), whereas RPI is
formulaic based on the win percentages of the team in question, its
opponents, and its opponents' opponents. Also, national polls only
rank the Top 25 teams, whereas every team in Division I college
basketball has an RPI rank. Regarding a team's tournament seed,
this too is subjectively determined by the Selection Committee, and the
values of the variable itself are more constrained (1 through 16) than
is true for RPI. (16)
Another factor considered when assessing the appeal of a particular
draw was the brand appeal of the participating schools at an event. This
is subtly different from current team quality because brand appeal is
largely built upon past success. Schools with greater brand appeal
should attract greater interest, and thereby, drive up secondary prices
and markups. One objective proxy for brand appeal is the revenue a
school's college basketball program generates, since present levels
of ticket sales and alumni donations are positively correlated with past
on-court success. (17) However, when "brand appeal" was
incorporated into Model 2 (either in place of or with RPI), it was not
statistically significantly nor did the adjusted R-square improve.
Discussion
This paper marks the first time that real-time secondary price data
is employed to examine secondary market behavior across multiple stages
of the largest collegiate sporting event (March Madness). The empirical
results reveal distinctively higher secondary markups for Final Four
tickets relative to tickets for any of the Regionals. Additionally,
secondary market behavior across the Regionals is sensitive to the
perceived quality of an event's draw, which itself is impacted in
part by the quality of the teams at the event and the proximity of those
schools to the event's host site. In short, the results suggest an
organization's ability to engage in inelastic ticket pricing may be
compromised when setting prices for several micro-events whose drawing
power is not known with certainty until just days before the event.
It is not altogether surprising that "locals" pay lower
markups than "non-local" secondary ticket consumers for two
reasons. First, local consumers have the luxury of making last-minute
decisions to attend an event, and thus, can exhibit greater patience
(i.e., price sensitivity) to see if secondary prices eventually fall to
their reservation price. Second, local fans may not exhibit particularly
strong allegiances to watch any of the participating schools since it is
uncommon that a Regional or Final Four participant is playing in its
home city. Hence, this too would make local fans more price-sensitive
compared to non-local fans that are traveling to the host site with a
strong rooting interest for one of the participating teams.
In terms of application, these results have the potential to
provide strategic justification for the NCAA to employ a dynamic pricing
model for future March Madness tournaments. Though evidence of sizable
secondary markups over face values at the Final Four suggests the NCAA
prices those tickets in the inelastic region of the demand curve,
contrasting evidence of mostly secondary price "markdowns" at
the Regionals suggests the NCAA may be systematically overpricing the
face values for the Regionals. Given that the attractiveness of a
particular Regional's draw is partially dependent upon the
collective quality of the teams at the Regional as well as their
proximity to the host site, any future algorithm used to dynamically
price such tickets would be warranted to account for these factors.
Ultimately, however, the NCAA's motivation to more dynamically
price tickets for the purpose of boosting ticket revenue might be
relatively weak given the massive media and advertising revenues
generated from March Madness. (18)
Similar to Rishe (2014a), one limitation of this analysis is the
lack of supply-side data regarding the number of tickets available on
the secondary market for each of these micro-events. TicketCity was not
able to provide information about the number of tickets available for
resale at the time of each consumer's online transaction. Though
capacity of facilities (imbedded in the ATTCAP variable) partially
captures the supply side of the primary pricing market, the number of
tickets available for resale on the secondary market would add further
illumination to supply-side considerations. Future research should
strive to obtain more supply-side ticket data in order to better account
for potential simultaneity issues.
Another issue to examine in future research is whether conclusions
regarding secondary market behavior for a particular event are
consistent across different ticket resellers. Though the present
research is based strictly on secondary sales data from TicketCity due
to data availability, one of the authors is currently engaged in a
secondary price analysis of the 2014 Rose Bowl and the 2014 BCS National
Championship Game (also contested at the Rose Bowl) using data from two
ticket resellers (TicketCity and SeatGeek). The preliminary results
suggest that secondary price markups are not significantly different
across these resellers. Though not always possible to obtain proprietary
secondary sales data from ticket resellers, future studies should strive
to obtain data from at least two different ticket resellers to enhance
the robustness of conclusions drawn.
One part of the tournament not examined herein was secondary market
behavior at the early rounds of the tournament. Given that the Selection
Committee tries to boost early round ticket sales by placing the top
Regional seeds at sites which minimize the distance traveled by fans,
this could yield higher secondary demand for these games compared to the
Regionals. Daily hotel metrics provided by Smith Travel Research show
that Orlando and St Louis--which both hosted early-round tournament
games in March 2014--experienced a net MSA hotel occupancy rate increase
of approximately 14% relatively to historical norms. (19) Conversely, a
recent study of twenty-four separate Regional Finals between 2008 and
2013 found that the median net increase in occupancy rates was less than
2% higher than historical norms (Rishe, 2014c). Thus, it would be
illuminating to see if the secondary market behavior mirrors the
differences in net occupancy spikes observed when comparing early-round
pods to Regional Finals.
References
Chapman, J. (1987). The impact of discounts on subjective product
evaluations. Unpublished doctoral dissertation, Marketing Department,
Virginia Polytechnic Institute and State University, Blacksburg, VA.
Chapman, J., & Wahlers, R. (1999). A revision and empirical
test of the extended price-perceived quality model. Journal of Marketing
Theory and Practice, 7(3), 53-64.
Coates, D., & Humphreys, B. (2007). Ticket prices, concessions,
and attendance at professional sporting events. International Journal of
Sport Finance, 2, 161-170.
Crupi, A. (2013, March 6). Show me the moneyball: March Madness
generates $1 billion in ad sales. Ad Week. Retrieved from
http://www.adweek.com/news/television/show-me-moneyball-marchmadness-generates-1-billion-ad-sales-147732
Dodd, W., & Monroe, K. (1985). The effect of brand and price
information on subjective product information. In E. Hirschman & M.
Holbrook (Eds.), Advances in consumer research (Vol. 12, pp. 85-90).
Provo UT: Association of Consumer Research.
Drayer, J., Rascher, D., & McEvoy, C. (2012). An examination of
underlying consumer demand and sport pricing using secondary market
data. Sport Management Review, 15, 448-460.
Drayer, J., & Shapiro, S. (2009). Value determination in the
secondary ticket market: A quantitative analysis of the NFL playoffs.
Sport Marketing Quarterly, 18, 5-13.
Dwyer, B., Drayer, J., & Shapiro, S. L. (2013). Proceed to
checkout? The impact of time in advanced ticket purchase decisions.
Sport Marketing Quarterly, 22, 166-180.
Fort, R. (2004). Inelastic sports pricing. Managerial and Decision
Economics, 25, 87-94.
Gujarati, D. (2011). Econometrics by example. Hampshire, UK:
Palgrave Macmillan.
Krautmann, A., & Berri, D. (2007). Can we find it at the
concessions? Understanding price elasticity in professional sports.
Journal of Sports Economics, 8, 183-191.
Krueger, A. (2001, Feb 1). Seven lessons about Super Bowl ticket
prices. New York Times. Retrieved from
http://www.nytimes.com/2001/02/01/business/economic-scene-sevenlessons-about-super-bowl-ticket-prices.html
Monroe, K., & Krishnan, R. (1985). The effect of price on
subjective product evaluations. In J. Jacoby & J. Olson (Eds.),
Perceived quality: How consumers view stores and merchandise (pp.
209-232). Lexington, MA: D.C. Heath.
Morganteen, J. (2014, January 27). The economics of Super Bowl
tickets: TiqIQ CEO. CNBC. Retrieved from
http://www.cnbc.com/id/101366720
O'Toole, T. (2010, April 22). NCAA reaches 14-year deal with
CBS/Turner for men's basketball tournament. USA Today. Retrieved
from http://content.usatoday.com/communities/campusrivalry/post/2010/04/ncaa-reaches-14-year-deal-with-cbsturner/1#.UmWCDhCzLMo
Rascher, D., McEvoy, C., Nagel, M., & Brown, M. (2007).
Variable ticket pricing in Major League Baseball. Journal of Sport
Management, 21, 407437.
Rao, A., & Monroe, K. (1988). The moderating effect of prior
knowledge on cue utilization in product evaluation. Journal of Consumer
Research, 15, 253-264.
Rishe, P. (2014a). Pricing insanity at March Madness: Exploring the
causes of secondary price markups at the 2013 Final Four. International
Journal of Sport and Society, 4, 67-78.
Rishe, P. (2014b, January 6). Distance, stadium capacity explain
cheaper secondary prices for 2014 BCS National Championship Game.
Forbes. Retrieved from
http://www.forbes.com/sites/prishe/2014/01/06/distance-stadium-capacity-explain-cheaper-secondary-prices-for-2014-bcs-national-championship/
Rishe, P. (2014c, October). Do mobile sporting events produce net
increases in tourism, local hotel revenues, and overall economic impact?
Evidence using STR data. Mustang Journal of Business and Ethics.
Rishe, P. (2012, January 6). Dynamic pricing: The future of ticket
pricing in sports. Forbes. Retrieved from
http://www.forbes.com/sites/prishe/2012/01/06/dynamic-pridng-thefuture-of-ticket-pricing-in-sports/
Rovell, D. (2014, January 6). BCS title tickets selling on the
cheap. ESPN.com. Retrieved from
http://espn.go.com/collegefootball/bowls13/story/_/id/10250044/2014-vizio-bcs-national-champi-onship-tickets-cheapest-14-year-history
Salaga, S., & Winfree, J. A. (2013). Determinants of secondary
market sales prices for National Football League personal seat licenses
and season ticket rights. Journal of Sports Economics.
Scitovsky, T. (1945). Some consequences of the habit of judging
quality by price. The Review of Economic Studies, 12, 100-105.
Shapiro, S., & Drayer, J. (2013). An examination of dynamic
ticket pricing and secondary market price determinants in Major League
Baseball. Sport Management Review, 17, 145-159.
Tsao, H., Pitt, L., & Caruana, A. (2005). The formation of
'high price-high quality' inferential belief: A study among
young buyers of mobile phones and notebooks. Asia Pacific Journal of
Marketing and Logistics. 17, 50-60.
Zeithaml, V. (1988, July). Consumer perceptions of price, quality,
and value: A means-end model and synthesis of evidence. Journal of
Marketing, 52, 2-22.
Zhu, J. D. (2014). Effect of resale on optimal ticket pricing:
Evidence from Major League Baseball tickets. Working paper. Texas
A&M University.
Endnotes
(1) These dollar value estimates for various segments of the
secondary market were obtained through an email communication between
one of the authors and an executive from a national ticket resale firm.
The executive asked for anonymity.
(2) Input regarding the Selection Committee's duties,
responsibilities, and challenges was obtained from prior informal
conversations with Doug Elgin and Greg Shaheen. Mr. Elgin is the
long-standing Commissioner of the Missouri Valley Conference and has
five years of experience on the Selection Committee. Mr. Shaheen is a
former Senior Vice President at the NCAA who for a decade oversaw the
operational logistics of March Madness.
(3) Information on ticket allocation and pricing was provided to
one of the authors via a May 2013 email communication with an NCAA
executive working within the NCAA's Championships and Alliances
team.
(4) One of the authors is the director of a consulting firm that
has conducted economic impact studies at multiple Final Fours and
Regionals since 2000. As such, he has cultivated business relationships
with numerous civic leaders, collegiate sports executives, and NCAA
officials who have shared informal in-person feedback regarding the
operational logistics of both bidding upon and hosting Regional Finals
and Final Fours.
(5) Rovell (2014) and Rishe (2014b) cite evidence that the
secondary prices for the 2014 BCS National Championship game between
Florida State and Auburn at the Rose Bowl in Pasadena CA were among the
lowest in over a decade in part because of the distance the team's
fans had to travel. Also, Morganteen (2014) quotes TiqlQ CEO Jesse
Lawrence as saying that distance was a factor in why secondary prices
for 2014's Super Bowl XLVIII between the Denver Broncos and Seattle
Seahawks at MetLife Stadium near New York City were among the lowest for
a Super Bowl in the last decade.
(6) Monroe and Krishnan's model has been validated and
extended in a number of subsequent studies (e.g., Dodd and Monroe 1985;
Zeithaml, 1988; Chapman, 1987; Rao and Monroe, 1988; Chapman and
Wahlers, 1999).
(7) Markup percentage was chosen as the dependent variable in order
to focus on the issue of how much fans were willing to spend over face
in comparative terms across observations. "Markups in dollars"
was deemed a weaker proxy since the same dollar markup (say $25) could
represent a different 'percentage markup' across different
face values (e.g., a $25 markup over a face of $175 is a larger
percentage markup than if the face price was $225). Secondary sales
prices did not include transaction fees.
(8) Alternatively, "miles" was considered as an
independent variable, and was measured as the distance between the event
site and the consumer's zip code. However, whether used in place of
"State" or with "State" in either Model 1 or Model
2, the coefficient on "miles" was strongly insignificant and
did not improve the adjusted R-squared.
(9) A team's RPI is weighted as 25% on the team's winning
percentage, 50% on the team's opponents' winning percentage,
and 25% on the opponents' opponents' winning percentage. For
more, visit http://en.wikipedia.org/wiki/Ratings_Percentage_Index#Basketball_Formula.
(10) TicketCity is an Austin, TX based ticket broker founded in
1990. They have formal relationships with professional teams (San
Francisco 49ers of the NFL), college athletic programs (including
numerous schools from the Pac-12 Conference), and numerous college bowl
games (including the Beef O'Brady's Bowl and BBVA Compass
Bowl).
(11) The sample data from Table 5 indicates that fans were much
more likely to buy Session 1 tickets than either All Session or Session
2 tickets for the Regionals. For the Final Four, the most commonly
transacted ticket was the All Session pass. These results likely reflect
the method of ticket allocation mentioned earlier (i.e., the NCAA only
sells All-Session tickets for the Final Four, but sells each different
type of Session for the Regionals ... and Session 1 appears to be the
most popular among the 3 Sessions given its consistently higher
markups).
(12) When H/S is present, OLS coefficient estimates are still
unbiased but the OLS assumption of having an error term with constant
variance is violated, thereby causing the standard errors and thus
t-tests of individual variable significance to be unreliable.
(13) Given the presence of H/S, various remedial approaches were
attempted following Gujarati's textbook prescriptions (Gujarati,
2011). The log transformation of the dependent variable did remove H/S
but led to the sample size falling considerably given that many markup
percentages were negative. Weighted least squares (WLS) using the
predicted value of the dependent variable as the weight did not remove
H/S. Conversely, White's corrected standard errors did remove H/S.
(14) To illustrate the importance of accounting for differences
across micro-events, note that the adjusted R-squared was only 0.230
when Model 1 was run without the Regional dummies.
(15) Based on both first-hand consulting research that one of the
authors has conducted for 3 Final Fours and 2 Regional Finals, coupled
with informal conversations with tournament organizers who have
collected data on the origin of event attendees at these events,
approximately 80-95% of fans who typically attend these events come from
outside the host site's MSA.
(16) Despite the noted limitations of "seeding" as a
proxy for quality, it was tested. When used in place of RPI, the
coefficient on "seeding" was negative and significant, but the
adjusted R-squared fell by over one percentage point compared to the
results reported in Table 9. Lastly, RPI and "seeding" could
not be used in the same specification due to multicollinearity issues.
(17) Data for 2012-13 basketball revenues for the sixteen schools
under consideration was obtained from the U.S. Department of
Education's Office of Postsecondary Education
(http://ope.ed.gov/athletics/).
(18) The NCAA has a 14-year, $10.8 billion media rights deal with
CBS and Turner Sports, assuring the NCAA roughly $771 million annual in
rights fees alone (O'Toole, 2010). And according to research by
Kantar Media, CBS and Turner jointly netted over $1 billion in
tournament ad sales revenue during March Madness in 2012, more than the
respective playoffs and championship series for the National Football
League, National Basketball Association, Major League Baseball and
National Hockey League (Crupi, 2013).
(19) This information was provided by Smith Travel Research at the
behest of one of the authors who sought hotel occupancy data during two
separate presentations in April 2014 to civic groups in Orlando and St
Louis, respectively, regarding the ability of early-round
"pods" to increase net tourism activity.
Patrick J Rishe, PhD, is a professor of business and economics in
the Business Department at Webster University. His research interests
include ticket pricing and economic impact.
Michael Mondello, PhD, is an associate professor in the Department
of Management at the University of South Florida. His research interests
include stadium financing, contingent valuation method (CVM), employee
compensation, analytics, and ticket pricing.
Brett Boyle, PhD, is an associate professor in the Department of
Marketing at Saint Louis University. His research interests include
sports consumer issues, sports sponsorship, and economic impact.
Table 1
Face Value Prices for 2013 Regionals
EAST
Price Points All-Session Session 1 Session 2
1 $600 $340 $320
2 $240 $130 $125
3 $200 $110 $105
4 $175 $98 $93
MIDWEST
Price Points All-Session Session 1 Session 2
1 $600 $340 $320
2 $160 $90 $85
3 $130 $75 $70
4 $90 $55 $50
SOUTH
Price Points All-Session Session 1 Session 2
1 $600 $340 $320
2 $200 $110 $105
3 $150 $85 $80
4 $100 $60 $55
5 $70 $45 $40
WEST
Price Points All-Session Session 1 Session 2
1 $600 $340 $320
2 $250 $135 $130
3 $200 $110 $105
4 $180 $90 $85
Table 2
TicketCity Transactions by Event and Session
Regional All-Session Session 1 Session 2 Total
East 20 57 40 117
Midwest 0 53 49 102
South 43 130 58 231
West 15 49 32 96
Final Four 107 90 88 285
Total 185 379 267 831
Table 3
Regional and Final Four Participants at 2013 March Madness
RPI Distance
East Regional (Verizon Center, Washington DC)
Syracuse 14 376
Marquette 12 788
Indiana 8 639
Miami 4 1050
Sum 38 2853
Median 10 713.5
Midwest Regional (Lucas Oil Stadium, Indianapolis)
Louisville 3 115
Oregon 47 2280
Duke 1 606
Michigan State 9 255
Sum 60 3256
Median 6 430.5
South Regional (AT&T Stadium, Arlington TX)
Kansas 5 526
Michigan 1200
Florida 7 1000
Florida Gulf Coast 93 1250
Sum 105 3976
Median 7 1100
West Regional (Staples Center, Los Angeles)
Wichita State 37 1380
Ohio State 10 2250
LaSalle 46 2710
Arizona 15 373
Sum 108 6713
Median 26 1815
Final Four (Georgia Dome, Atlanta)
Syracuse 14 967
Michigan 21 709
Louisville 3 421
Wichita State 37 960
Sum 75 3057
Median 17.5 834.5
Table 4
Facility Capacity and Attendance at Regionals and Final Four
Regional Capacity Session 1 AttCap1%
East 20,308 19,731 97.2%
Midwest 36,000 35,520 98.7%
South 42,614 40,639 95.4%
West 18,997 18,232 96.0%
Final Four 80,000 75,350 94.2%
Regional Session 2 AttCap2%
East 19,801 97.5%
Midwest 34,657 96.3%
South 36,585 85.9%
West 17,998 94.7%
Final Four 74,326 92.9%
Table 5
Average and Median Secondary Prices, Face Prices, and Markups
Means Metric East MW
All Sessions MU% -12.6% n/a
FV $191 n/a
SP $173 n/a
MU -$18 n/a
Session 1 MU% -6.7% -8.1%
FV $113 $70
SP $110 $66
MU -$3 -$5
Session 2 MU% -16.5% -35.8%
FV $110 $71
SP $95 $46
MU -$15 -$25
Medians Metric East MW
All Sessions MU% -32.3% n/a
FV $175 n/a
SP $120 n/a
MU -$57 n/a
Session 1 MU% -24.5% -23.6%
FV $110 $75
SP $87 $57
MU -$27 -$14
Session 2 MU% -40.6% -44.0%
FV $105 $85
SP $66 $40
MU -$43 -$30
Means South West Final Four
All Sessions -27.6% -40.4% 47.0%
$160 $213 $226
$118 $129 $384
-$42 -$84 $158
Session 1 -25.1% -43.8% 79.3%
$78 $127 $120
$58 $74 $226
-$20 -$53 $106
Session 2 -59.1% -44.7% 36.6%
$89 $123 $124
$34 $68 $192
-$55 -$55 $67
Medians South West Final Four
All Sessions -36.0% -50.0% 12.4%
$150 $200 $230
$106 $106 $223
-$52 -$106 $29
Session 1 -31.4% -49.4% 58.5%
$85 $135 $125
$50 $63 $174
-$25 -$58 $62
Session 2 -65.0% -57.9% 12.6%
$105 $130 $120
$33 $55 $136
-$64 -$69 $16
Table 6
Percentage of Transactions Resulting in Positive Markups
Regional All-Session Session 1 Session 2 Average
East 30.0% 29.8% 20.0% 26.6%
Midwest n/a 32.1% 10.2% 21.2%
South 20.9% 16.9% 6.9% 14.9%
West 6.7% 12.2% 9.4% 9.4%
Final Four 57.9% 72.2% 56.8% 62.3%
Table 7
Model 1 Estimation with OLS Corrected Using White's HCSE
Variable Coeff t-value p-value Beta Coeff
Intercept 1.681 13.8 <.0001 0.000
Low 0.128 2.2 0.0253 0.072
High -0.348 -7.5 <.0001 -0.201
Central 0.394 6.4 <.0001 0.225
East -0.541 -9.6 <.0001 -0.220
Midwest -0.596 -8.8 <.0001 -0.229
South -1.035 -12.5 <.0001 -0.542
West -0.918 -15.7 <.0001 -0.343
Qty -0.466 -7.4 <.0001 -0.702
Qty2 0.028 3.7 0.0003 0.377
Time 0.001 1.3 0.2021 0.031
State 0.003 0.1 0.9588 0.002
Local -0.119 -2.9 0.0038 -0.070
All Sess -0.200 -3.1 0.0022 -0.097
Session 2 -0.420 -8.7 <.0001 -0.229
F-value 52.830
Pr > F 0.0001
Adjt R-squared 0.467 n = 831
Table 8
Model 2 Modified with Team/Event Specific Variables
Variable Coeff t-value p-value Beta Coeff
Intercept -3.350 -4.34 <.0001
All Sess -0.112 -1.68 0.097 -0.055
Session 2 -0.354 -4.87 <.0001 -0.193
Low 0.147 2.52 0.0119 0.083
High -0.329 -7.19 <.0001 -0.190
Central 0.396 6.18 <.0001 0.226
Qty -0.468 -7.35 <.0001 -0.705
Qty2 0.028 3.63 0.0003 0.377
Time 0.001 1.32 0.188 0.033
State -0.004 -0.05 0.956 0.002
Local -0.132 -3.24 0.001 -0.077
AttCap 0.047 5.6 <.0001 0.165
RPI -0.011 -2.25 0.025 -0.086
Prox -0.000110 -2.43 0.015 -0.051
FF 0.773 12.16 <.0001 0.429
F-value 37.54
Pr > F 0.0001
Adjt R-squared 0.475 n = 831
Table 9
Testing the Equality of Coefficients:
Final Four vs Regional Data
FULL REGIONALS
Variable Coeff p-value Coeff p-value
Intercept 0.999 <.0001 0.811 <.0001
All Sessions 0.093 0.258 -0.148 0.001
Session 2 -0.283 <.0001 -0.218 <.0001
Low 0.217 0.004 0.065 0.161
High -0.222 <.0001 -0.083 0.025
Central 0.178 0.002 0.035 0.268
Qty -0.522 <.0001 -0.572 <.0001
Qty2 0.034 <.0001 0.051 <.0001
Time 0.001 0.208 0.003 <.0001
State 0.275 <.0001 0.151 0.0004
n 831 546
Adjusted R-sqr 0.230 0.417
SSE 462.9 67.4
Chow Stat 63.62
Chow p-value 0.001
FINAL FOUR
Variable Coeff p-value
Intercept 2.192 <.0001
All Sessions -0.269 0.021
Session 2 -0.705 <.0001
Low 0.292 0.022
High -0.718 <.0001
Central 0.932 <.0001
Qty -0.667 <.0001
Qty2 0.037 0.0001
Time -0.003 0.055
State 0.102 0.328
n 285
Adjusted R-sqr 0.454
SSE 192
Chow Stat
Chow p-value
