discussion above about whether online sales truly eliminate spatial boundaries in markets.
We add to Goldmanis et al.’s original data and specifications here. Figure 1 shows how
the composition of employment in the same three industries changed between 1994 and 2007.
Each panel shows the estimated fraction of employment in the industry that is accounted for by
establishments of three employment size classes: those having 1-9 employees, those with 10-49,
and those with 50 or more. In addition to the three industries studied in Goldmanis et al., the
figure also shows for the sake of comparison the same breakdown for total employment in the
entire County Business Patterns coverage frame (essentially all establishments in the private
nonfarm business sector with at least one employee).
effects among car dealers are particularly noteworthy in that auto manufacturers and dealers in
the U.S. are legally prohibited from selling cars online. Therefore any effects of e-commerce
must be channeled through consumers’ abilities to comparison shop and find the best local outlet
at which to buy their car, not through changes in the technology of car distribution. While this
technology-based channel is important in some industries, the consumer-side search channel is
the one posited in their model, and therefore new car dealers offer the most verisimilitude to the
theory from which they derive their predictions.
Panel A shows the breakdown for travel agencies. It is clear that during the early half of
the sample period, which saw the introduction and initial diffusion of e-commerce, the share of
industry employment accounted for by travel agency offices with fewer than 10 employees
shrank considerably. This lost share was almost completely taken up by establishments with 50
or more employees. After 2001, the share losses of the smallest offices stabilized, but the 10-49
employee category began to lose share to the largest establishments. These patterns are
consistent with the predictions of the theory—the largest offices in the industry benefit at the
cost of the smaller offices.
The aggregate impact observed among travel agencies resulted from the nature of the institutional shifts in
industry revenues that e-commerce caused. Responding to a shift in customers toward buying tickets online, airlines
cut ticket commissions to travel agents, which accounted for 60 percent of industry revenue in 1995, completely to
zero by 2002. These commission cuts were across the board, and did not depend on the propensity of travelers to
buy tickets online in the agents’ local markets.
County Business Patterns do not break out actual total employment by size category, so we impute it by
multiplying the number of industry establishments in an employment category by the midpoint of that category’s
lower and upper bounds. For the largest (unbounded) size categories, we estimated travel agency offices and
bookstores with 100 or more employees had an average of 125 employees; auto dealers with more than 250
employees had 300 employees. Imputations were not necessary in the case of the total nonfarm business sector, as
the CBP do contain actual employment by size category in that case.
Panel B shows the same results for bookstores. Here, the pattern is qualitatively similar,
but even more stark quantitatively. While the fraction of employment at stores with 10-49
employees is roughly stable over the entire period, the largest bookstores gained considerable
share at the expense of the smallest.
Panel C has the numbers for new car dealers. In this industry, establishments with fewer
than 10 employees account for a trivial share of employment, so the interest is in the comparison
between the 10-49 employee dealers and those with more than 50. Again, we see that the large
establishments accounted for a greater fraction of industry employment over time, with the
largest establishments gaining about 10 percentage points of market share at the cost of those
with 10-49 employees.
Finally, panel D does the same analysis for all establishments in the private nonfarm
business sector. It is apparent that the shifts toward larger establishments seen in the three
industries of focus were not simply reflecting a broader aggregate phenomenon. Employment
shares of establishments in each of the three size categories were stable throughout the period.
These predictions about the market share and entry and exit effects of introducing an
online sales channel in an industry are based on the assumption that firms behave non-
cooperatively. If e-commerce technologies instead make it easier for firms to collude in certain
markets, e-commerce technologies might actually make those markets less competitive.
Campbell, Ray, and Muhanna (2005) use a dynamic version of Stahl (1989) to show theoretically
that if search costs are high enough initially, e-commerce-driven reductions in search costs can
actually make it easier for collusion to be sustained in equilibrium, as they increase the profit
difference between the industry’s collusive and punishment (static Nash Equilibrium) states.
A more direct mechanism through which online sales channels support collusion is that
the very transparency that makes it easier for consumers to compare products can also make it
easier for colluding firms to monitor each other’s behavior. This makes cheating harder. Albæk,
Møllgaard, and Overgaard (1997) document an interesting example of this, albeit one that
doesn’t directly involve online channels, in the Danish ready-mixed concrete industry. In 1993,
the Danish antitrust authority began requiring concrete firms to regularly publish and circulate
their transactions prices. Within a year of the institution of this policy, prices increased 15-20
percent in absence of any notable increases in raw materials costs or downstream construction
activity. The policy—one that, ironically, was implemented with hopes of increasing
competition—facilitated collusion by making it easier for industry firms to coordinate on
anticompetitive prices and monitor collusive activities. Online markets are often characterized by
easy access to firms’ prices. If it is hard for firms to offer secret discounts because of market
convention, technological constraints, or legal strictures, this easy access fosters a powerful
monitoring device for colluders.
5. Implications of Online Commerce for Firm Strategy
The fundamental effects of opening a concurrent online sales channel in an industry that
we discussed in Section 3 can have implications for firms’ competitive strategies. These strategy
choices can in turn induce and interact with the equilibrium changes we discussed in Section 4.
This section reviews some of these strategic factors.
A key factor—perhaps the key factor—influencing firms’ joint strategies toward offline
and online markets is the degree of connectedness between online and offline markets for the
same product. This connectedness can be multidimensional. It can involve the demand side: how
closely consumers view the two channels as substitutes. It can involve the supply side: whether
online and offline distribution technologies are complementary. And it can involve firms’
available strategy spaces: how much leeway firms have in conducting separate strategic
trajectories across channels, which is particularly salient as it regards how synchronized a firm’s
pricing must be across offline and online channels.
At one extreme would be a market where the offline and online channels are totally
separated. Specifically, consumers view the product as completely different depending upon the
channel through which it is sold (perhaps there are even separate online and offline customer
bases); there are no technological complementarities between the two channels; and firms can
freely vary positioning, advertising, and pricing of the same product across the channels. In this
case, each channel can be thought of as an independent market. The firm’s choices in each
channel can be analyzed independently, as there is no scope for strategic behavior that relies
upon the interplay between the two channels.
Of more interest to us here—and where the research literature has had to break new
ground—are cases where there are nontrivial interactions between online and offline channels
selling the same products. We’ll discuss some of the work done in this area below, categorizing
it by the device through which the online and offline are linked: consumer demand (e.g.,
substitutability), technological complementarities, or strategic restrictions.
5.1. Online and Offline Channels Linked Through Consumer Demand
One way the online and offline sales channels can be connected is in the substitutability
that buyers perceive between the channels. The extent of such substitutability determines two
related effects of opening an online channel in a market: the potential for new entrants into an
online channel to steal away business from incumbents, and the amount of cannibalization
offline incumbents will suffer upon opening an online segment. Not all consumers in a market
need to view this substitutability symmetrically. Distinct segments can react differently to the
presence of online purchase options. The observed substitutability simply reflects the aggregate
impact of these segments’ individual responses.
These factors have been discussed in several guises in the literature investigating the
strategic implications of operating in a market with both online and offline channels. Dinlersoz
and Pereira (2007), Koças and Bohlmann (2008), and Loginova (2009) construct models where
heterogeneity in consumers’ views toward the substitutability of products sold in the two
segments affects firms’ optimal strategies.
Dinlersoz and Pereira (2007) and Koças and Bohlmann (2008) build models where some
customers have loyalty for particular firms and others buy from the lowest-price firm they
encounter. Offline firms with large loyal segments (“Loyals”) stand to lose more revenue by
lowering their prices to compete in the online market for price-sensitive “Switchers.” Hence the
willingness of incumbents from the offline segment to enter new online markets depends in part
on the ratios of loyal customers to Switchers. This also means the success of pure-play online
firms is tied to the number of Switchers. In some circumstances, opening an online channel can
lead to higher prices in the offline market, as the only remaining consumers are Loyals who do
not perceive the online option as a substitute. Depending on the relative valuations and sizes of
the Loyals and Switchers segments, it is even possible that the quantity-weighted average price
in the market increases. In effect, the online channel becomes a price discrimination device.
Direct tests of these models are difficult, but they do imply that if we compare two firms,
the one with the higher price will have more loyal consumers than the other. We can conduct a
rough test of this in the bookselling industry using the Forrester Technographics data. In it,
How to C#: Basic SDK Concept of XDoc.PDF for .NET
example, you may easily create, load, combine, and split PDF file(s), and add, create, insert, delete, re-order, copy, paste, cut, rotate, and save PDF page(s extract page from pdf; cut paste pdf pages
VB.NET PDF: Basic SDK Concept of XDoc.PDF
example, you may easily create, load, combine, and split PDF file(s), and add, create, insert, delete, re-order, copy, paste, cut, rotate, and save PDF page(s cutting pdf pages; delete pages of pdf online
consumers are asked whether they have shopped either online or offline at Amazon, Barnes &
Noble, or Borders in the previous thirty days. Clay et al. (2002) found that Amazon set prices
higher than Barnes & Noble, which in turn set prices higher than Borders. Thus the models
predict that Amazon’s customers will be more loyal than Barnes & Noble’s, who are themselves
more loyal than Borders’. In our test, this implies that of customers of these sellers, Amazon will
have the highest fraction of exclusive shoppers, followed by Barnes & Noble and Borders.
The results are in Table 6. In the first row, the first column reports the fraction of
consumers who purchased a book in the past three months and shopped only at Amazon. The
second column gives the fraction of customers who purchased a product from Amazon as well as
from Barnes & Noble or Borders. If we take the first column as a crude measure of the fraction
of Amazon’s loyal customers and the second column as a measure of those willing to shop
around, Amazon’s customer base is roughly split between Loyals and Switchers. While the
models would predict that, given the observed price difference, Barnes and Noble’s Loyals-to-
Switchers ratio should be lower, this is not the case in the data, as reflected in the second row.
However, Borders’ low ratio of Loyals to Switchers is consistent with them having the lowest
prices. A caveat to these results, however, is that they could be confounded by internet use. The
models’ predictions regard the loyalty of a firm’s online customers. If many of Barnes & Noble’s
loyal customers are offline, our measure might overstate the loyalty of Barnes & Noble’s online
consumers. We address this in the second panel of Table 6 by recalculating the fractions after
conditioning on the consumer having purchased a book online. Now the evidence is exactly in
line with the predictions of Dinlersoz and Pereira (2007) and Koças and Bohlmann (2008): the
rank ordering of the firms’ prices is the same as the ordering of the Loyals-to-Switchers ratio.
In Loginova (2009), consumers’ ignorance of their valuations for a good shapes the
nature of the link between online and offline markets. Consumers in her model differ in their
valuations for the market good, but do not realize their valuations until they either a) visit an
offline retailer and inspect the good, or b) purchase the good from an online retailer (no returns
are allowed). Under certain parameter restrictions, there is an equilibrium where both channels
are active and all consumers go to offline retailers and learn their valuations. Upon realizing their
utility from the good, they decide either to immediately purchase the good from the offline
retailer or to go home and purchase the product from an online retailer while incurring a waiting
cost. This creates an equilibrium market segmentation where consumers with low valuations buy
from online stores and high-valuation consumers buy immediately at the offline outlet they
visited. The segmentation lets offline retailers raise their prices above what they would be in a
market without an online segment. The imperfect substitutability between online and offline
goods segments the market and allows firms to avoid head-on competition.
These papers focus on the extent to which goods sold on online and offline channels are
substitutes, but it is possible in certain settings that they may be complements. Empirical
evidence on this issue is relatively sparse. Gentzkow (2007) estimates whether the online edition
of the Washington Post is a substitute or complement for the print edition. The most basic
patterns in the data suggest they are complements: consumers who visited the paper’s website
within the last five days are more likely to have also read the print version. However, this cross
sectional pattern is confounded by variation in individuals’ valuations from consuming news. It
could be that some individuals like to read a lot of media, and they often happen to read the
online and offline versions of the paper within a few days of one another. But conditioning on
having read one version, that specific individual may be less likely to read the other version. This
is borne out in a more careful look at the data; instrumenting for whether the consumer has
recently visited the paper’s website using shifters of the consumer’s costs of reading online,
Gentzkow finds the two channels’ versions are rather strong substitutes. Using a different
methodology, Biyalogorsky and Naik (2003) look whether Tower Records’ introduction of an
online channel lifted or cannibalized its offline sales. They find cannibalization, though it was
modest, on the order of 3 percent of the firm’s offline sales. Given that brick-and-mortar record
stores have clearly suffered from online competition since this study, their result suggests that
much of the devastation was sourced in across-firm substitution rather than within-firm
5.2. Online and Offline Channels Linked Through Technological Complementarities
Wang (2007) ties the online and offline channels with a general complementarity in the
profit function that he interprets as a technological complementarity. His model treats the
introduction of e-commerce into an industry as the opening of a new market segment with lower
entry costs. The model’s dynamic predictions are as follows. Taking advantage of the new, lower
entry costs, pure-play online sellers enter first to compete with the brick-and-mortar incumbents.
But the complementarity between the online sales and distribution technology and the offline
technology gives offline incumbents incentive to expand into the online channel. It also gives
these firms an inherent advantage in the online market, as they are able to leverage their offline
assets to their gain. As a result, many of the original online-only entrants are pushed out of the
industry. Thus a hump-shaped pattern is predicted in the number of pure-play online firms in a
product market, and a steady diffusion of former offline firms into the online channel.
This is a reasonably accurate sketch of the trajectory of the online sector of many retail
and service markets. The online leaders were often pure-play sellers: Amazon, E-Trade, Hotmail,
pets.com, and boo.com, for example. But many of these online leaders either eventually exited
the market or were subsumed by what were once offline incumbents. Some pure-play firms still
exist, and a few are fabulously successful franchises, but at the same time, many former brick-
and-mortar sellers now dominate the online channels of their product markets.
Jones (2010) explores a different potential complementarity. The notion is that the online
technology is not just a way to sell product, but it can also be an information gathering tool.
Specifically, the wealth of data generated from online sales could help firms market certain
products to individuals much more efficiently and lead to increased sales in both channels.
5.3. Online and Offline Channels Linked Through Restrictions on Strategy Space
Liu, Gupta, and Zhang (2006) and Viswanathan (2005) investigate cases where the online
and offline channels are tied together by restrictions on firms’ strategy spaces—specifically, that
their prices in the two channels must be a constant multiple of one another. In the former study,
this multiple is one: the firm must price the same whether selling online or offline. Viswanathan
(2005) imposes that the price ratio must be a constant multiple, though not necessarily unity.
While it might seem unusual that these pricing constraints are exogenously imposed instead of
arising as equilibrium outcomes, it is true that certain retailers have faced public relations and
sometimes even legal problems due to differences in the prices they charge on their websites and
in their stores. Liu, Gupta, and Zhang remark that many multichannel firms report in surveys that
they price consistently across their offline and online channels (e.g., Forrester Research (2004)).
Liu, Gupta, and Zhang (2006) show that, when the equal pricing restriction holds, an
incumbent offline seller can deter the entry of a pure-play online retailer by not entering the
online market itself. This seemingly counterintuitive result comes from the uniform price
requirement across channels. An incumbent moving into the online channel is restricted in its
ability to compete on price, because any competition-driven price decrease in the online market
reduces what the incumbent earns on its inframarginal offline units. This limit to its strategy
space can actually weaken the incumbent’s competitive response so much that a pure-play online
retailer would be more profitable if the incumbent enters the online segment (and therefore has to
compete head-to-head with one hand tied behind its back) than if the incumbent stays exclusively
offline. Realizing this, the incumbent can sometimes deter entry by the pure-play online firm by
staying out of the online channel in the first place. The link across the online and offline channels
in this model creates an interesting situation in which the offline firm does not gain an advantage
by being the first mover in to the online channel. Instead, it may want to abstain from the online
Viswanathan (2005) models the online and offline models as adjacent spatial markets.
Consumers in one market cannot buy from a firm in the other market. However, one firm at the
junction of the two markets is allowed to operate as a dual-channel supplier, but it must maintain
an exogenously given price ratio of k between the two markets. Viswanathan shows that in this
setup, the price charged by the two-channel firm will be lower than the offline-only firms’ prices
but higher than the pure-play online sellers.
The emergence of online channels in a market can bring substantial changes to the
market’s economic fundamentals and, through these changes, affect outcomes at both the market
level and for individual firms. The potential for such shifts has implications in turn for firms’
competitive strategies. Incumbent offline sellers and new pure-play online entrants alike must
account for the many ways a market’s offline and online channels interact when making pricing,
investment, entry, and other critical decisions.
We have explored several facets of these interactions in this chapter. We stress that this is
only a cursory overview, however. Research investigating these offline-online connections is
already substantial and is still growing. This is rightly so, in our opinion; we expect the insights
drawn from this literature to only become more salient in the future. Online channels have yet to
fully establish themselves in some markets and, in those where they have been developed, are
typically growing faster than bricks-and-mortar channels. This growing salience is especially
likely in the retail and services sectors, where online sales appear to still have substantial room
Documents you may be interested
Documents you may be interested