Professional Documents
Culture Documents
Welfare
Shivendu Shivendu
Muma College of Business, University of South Florida, Tampa, Florida 33620,
shivendu@usf.edu
Abstract
The pervasiveness of the Internet and digitization has revolutionized the delivery and
consumption of information goods. In this paper, we study the impact of digitization and the
shift in consumers’ preferences for the digital medium on outcomes including social welfare in
content markets. We consider a publisher who offers information goods in the physical and the
digital mediums and also in a bundle of physical and digital mediums in a market where
consumers are heterogeneous in both their valuations for content and their preferences for
mediums. We find that the publisher’s optimal content- medium strategy is to offer the content
only in the digital medium under some market conditions, and to offer the content in a bundle
of mediums and the digital medium, under other conditions. Interestingly, while the price of
the bundle of mediums increases with the marginal cost of the physical medium, the price of
the digital medium may decrease when the marginal cost of the physical medium is sufficiently
large. Surprisingly, we find that consumer surplus can decrease as the proportion of digital-
savvy consumers (consumers who prefer the digital medium over the physical medium)
increases, and social welfare may have a discrete decrease when the proportion of digital-savvy
consumers is sufficiently large. Counter to intuition, when more consumers prefer the digital
medium, while the digital price may have a discrete increase, the price of the bundle may have
a discrete decrease.
1. Introduction
The advent of the Internet and information technology has led to the digitization of content
industries and, this, in turn, has transformed the distribution and the consumption of
information goods. While traditionally consumers have purchased the physical medium to
access information goods (print newspapers, books, CDs, DVDs, etc.), in a digital era, an
increasing number of consumers buy and consume information goods or content in the digital
medium (digital newspapers, eBooks, album downloads, video downloads, etc.) (Sporkin
2011).
Digitization can potentially expand the consumer base by providing additional
convenience and ease of use through anytime-anywhere access (Koukova et al. 2008). On the
other hand, the digital access to information goods may serve as a substitute for access through
the physical medium, thereby cannibalizing physical sales (Kannan et al. 2009). In addition,
pricing of content over the two mediums varies both within and across content industries. For
example, both The New York Times and The Los Angeles Times offer a choice of home delivery
+ digital or digital-only access, but they do not offer a home delivery-only option. In contrast,
The Washington Post offers only digital access but no bundle. Game Informer magazine offers
print-only and digital-only options separately. Warner Music sells digital-only albums as well
as CDs that come with a digital copy. Independent record labels such as Soulection and Triple
Pop offer only digital albums and tracks (Droppo 2014). An important question for content
publishers is to determine the optimal content pricing strategies over dual-medium access
under different market conditions.
While the physical and digital mediums differ significantly in costs to produce and
distribute, publishers also must consider consumers’ heterogeneous preferences for one
medium over the other. For example, a recent consumer report finds that 38% of consumers
preferred digital access to video games, while 62% of consumers still preferred having a
physical CD of the games (NPD Group 2015). Moreover, the growth of consumption of digital
content by 157% from 2010 to 2014 indicates that the proportion of consumers who prefer the
digital medium is growing over time, and this trend is likely to continue (comScore 2015). The
heterogeneous and evolving consumer preferences toward the two mediums raise new
1
Throughout this paper, our publisher is “she” and consumer is “he.”
2 See Stremersch and Tellis (2002), and Venkatesh and Mahajan (2009) for a comprehensive review of bundling
literature.
10
11
provides content in the physical medium. The marginal cost of the physical medium includes
material costs, handling and shipping costs, and labor costs. On the other hand, the publisher
incurs zero marginal cost to serve a consumer to whom she provides content in the digital
medium (Bakos and Brynjolfsson 1999, Chen and Png 2003, Sundararajan 2004b, Aron et al.
2006, Balasubramanian et al. 2015)3. When the publisher offers content in a bundle of the
digital and the physical mediums to a consumer, then she incurs the same marginal cost as in
the case of offering the content in only the physical medium, i.e., c p . We denote the price
charged by the publisher for content in a physical medium as p p , the price for content in a
digital medium as pd and the price for content in a bundle of mediums as pb . To keep the
focus of this research on optimal content-medium pricing strategies, we assume that all
infrastructural costs for providing content in a physical or a digital medium are sunk, and the
publisher has no supply-side constraints.
2.1 Consumers
The market consists of a unit mass of consumers. Consumers’ valuations ( v ) for the content
are independent and are uniformly distributed, i.e. v U [0,1] (Lang and Vragov 2005, Dou et
al. 2013). A recent study by the Newspaper Association of America (NAA) finds that around
60% of readers read content in print, 48% of readers read content in digital formats and 34%
read content in print as well as in digital mediums, wherein each medium “provides somewhat
different functional and experiential qualities fitting some usage occasions better than others”
(Newspaper Association of America 2012). A recent consumer report finds that while some
consumers prefer video games in a digital medium, others prefer video games in a physical
medium (NPD Group 2015). These studies indicate that population consists of two segments;
a segment of consumers prefers the digital medium over the physical medium while the other
segment of consumers prefers the physical medium over the digital medium.
We model consumers’ preferences for mediums by segmenting the market wherein r
proportion of consumers prefers the digital medium (digital-savvy consumers), i.e. r (0,1) ,
3 In §4.7, we discuss the implications on our results, if this assumption were to be relaxed.
12
WTP for the access to the content, i.e., v(1 ) , if he is offered the content in his un-preferred
medium. We assume that the mismatch cost is homogeneous across all consumers. In §4.4, we
relax the assumption of the homogeneous mismatch cost for all consumers and assume
consumers have a continuously distributed heterogeneous mismatch cost. We numerically
examine the impact of such a heterogeneous mismatch cost on our results and show that our
key results continue to hold. Moreover, our conceptualization also implies that consumers’
WTPs in the two mediums are perfectly correlated. In §4.6 we relax this assumption and show
through a numerical analysis that our results hold even when the WTPs for content in the two
mediums are asymmetric and independent.
Our modeling approach of segmenting the market in two distinct groups is similar to
the one adopted in the literature in operations research. Hsiao and Chen (2014) categorize
consumers into one of two segments based on their preference for purchasing online or
purchasing in a retail store. Consumers in the first segment have a higher utility from
purchasing in a retail channel (utility v for purchasing in a physical store, and utility 1v for
purchasing online, where 1 1 ), whereas consumers in the second segment obtain a higher
utility from purchasing online (utility v for purchasing in a physical store, and utility 2 v for
purchasing online, where 2 1 ). Tan and Carrillo (2014) utilize a non-negative consumer
13
approach of bundling of substitutes in prior literature (Lewbel 1985, Venkatesh and Kamakura
2003, Armstrong 2013). Parameter can also be interpreted as the additional convenience
that consumers experience, if they have a choice of consuming the content in either medium.
For analytical tractability, we assume that the sub-additive parameter is homogeneous
across all consumers. While Venkatesh and Kamakura (2003) make this assumption to get a
closed-form solution, this abstraction is also similar to the constant value dependence
assumption in McGuire and Staelin (1983) and to the analysis of substitute goods in Bakos and
Brynjolfsson (1999). Note that if the content is offered in a bundle of mediums, then digital-
savvy as well as traditional consumers have no medium-related heterogeneity because they get
content in both mediums. Note that 1 , because the WTP for content in a bundle is sub-
additive, i.e., v(1 ) v v(1 ) . In §4.5 we relax the assumption of sub-additivity and
examine the impact on our results, when the mediums are perceived as complements or super-
additive.
14
v U[0, w1 ] . Furthermore, the valuation of the bundle is sub-additive, i.e. v(1 ) , where
. This alternative conceptualization is similar to our model setup with r 0 (i.e. all
bundle is pb (1 c p ) / 2 . Now, suppose the publisher offers content in the digital medium
in addition to the bundle, and sets the same bundle price as when she offers only the bundle
4 We use “Profile” to denote a market coverage profile in the rest of the paper.
15
off by offering the content in the digital medium in addition to the bundle compared to offering
content in the bundle only (See proof of Proposition 1 in Appendix B). This result is not
obvious because one might think that, since the bundle and the digital are partial substitutes, it
may not be optimal to offer them together. The reason for this counter-intuitive result is as
follows.
When the marginal cost of the bundle is relatively small, the traditional consumers buy
only the bundle (do not switch to buying the digital medium), but some digital-savvy
consumers who do not buy when only the bundle is offered, now buy the digital. Therefore,
the market coverage increases and the publisher gains additional profit from the digital-savvy
consumers without cannibalizing the profit from the traditional consumers, the publisher is
better off. When the marginal cost of the bundle is relatively large, the bundle price is high.
When both the bundle and the digital are available to consumers, not only the digital-savvy
consumers but also some traditional consumers who do not buy when only the bundle is
available, now buy the digital medium, and some traditional consumers who buy the bundle
when only the bundle is available, now switch to buying the digital medium. This leads to the
increase in the market coverage for both types of consumers, and the publisher also saves on
the marginal cost. The profit gain from the increased market coverage from both types of
consumers and from the savings on the cost outweighs the revenue loss from consumers who
switch from the bundle to the digital medium.
From the analysis above, we need to focus only on the two content-medium pricing
strategies, i.e., offering the digital medium and the bundle, and offering only the digital
medium. When the publisher offers content in a bundle as well as in the digital medium, the
consumers’ purchase decisions are based on their individual rationality (IR) and incentive
compatibility (IC) constraints for each of the consumer types. We derive all possible market
coverage profiles and based on the comprehensive analysis in Appendix B, we show that only
one of the following four market coverage profiles is potentially optimal under a specific
parameter space. Under Profile 1, while traditional consumers buy the bundle, some digital-
savvy consumers buy the bundle and some buy the digital ( T b and DS b & d ). Under
Profile 2, traditional consumers buy only the bundle and digital-savvy consumers buy only the
16
b Bundle d Digital
17
18
P , P , P , where P p
I
* *
II
*
III I
* d*
I 1/ 2, pIb* (1 c p ) / 2 ,
Note that optimal pricing scheme PI* corresponds to Profile 1 or 2, pricing scheme PII*
corresponds to Profile 3 and pricing scheme PIII* corresponds to Profile 4. It is easy to see
from Lemma 1 that the price of the digital medium in the pricing scheme PI* is higher than
that in pricing schemes PII* and in PIII* . On the other hand, the price of the bundle in pricing
19
in pricing scheme PIII* . In order to figure out the parameter space under which a particular
pricing scheme is optimal, we compare the feasibility conditions for Profiles 1, 2, 3, and 4 (see
proof for Proposition 2a – 2g in Appendix B).
We present our results in §3.1 in §3.2 by first segmenting the parameter space on the
mismatch cost parameter and then on the subadditivity parameter , and then on the
proportion of digital-savvy consumers r and the marginal cost of the physical medium c p . The
division of the parameter space for the presentation of results is driven by the convenience of
exposition. At the end of Appendix B, we present the set of conditions in which pricing scheme
The condition on the mismatch cost parameter, i.e., 1/ (1 r ) , implies that 0.5 because
r (0,1) .
publisher adopts pricing scheme PI* for the entire feasible parameter space.
The economic reasoning for Proposition 2a is as follows. From Lemma 1, the price of
the digital decreases with and is lower in pricing schemes PII* and PIII* than in PI* , for any
(0,1) . This implies that given a large , if the publisher were to adopt PII* or PIII* , then
the price of digital has to be lower. When the publisher considers adopting pricing scheme PII*
or PIII* instead of PI* , her tradeoff is between reducing the cost by selling the digital instead of
a bundle to traditional consumers and reducing the revenue from (1) digital-savvy consumers
by reducing the digital price and (2) traditional consumers by incentivizing them to switch to
the digital medium. When the mismatch cost parameter is large, the reduction in revenue
from both segments of consumers outweighs the reduction in cost by selling the digital to
traditional consumers. Therefore, the publisher is better off by adopting PI* , rather than PII* or
20
either small or large r as long as this proportion is above the threshold, i.e. r (1 ) / .
When the mismatch cost parameter is very large, even if the proportion of digital-savvy
consumers is small and the marginal cost of a bundle is high (leading to high price of a bundle),
the publisher is better off by offering pricing scheme PI* under which the traditional consumers
buy only the bundle for which they have a higher WTP, rather than the digital, for which they
have a lower WTP.
3.2. Optimal pricing schemes when is not larger than 1/ (1 r )
Now, we describe the optimal pricing schemes when the medium-mismatch cost parameter (
) is not larger than 1 / (1 r ) which implies that can take any value between 0 and 1 . We
group the results around the value of and describe the optimal pricing scheme in the entire
PROPOSITION 2b: If (I) 0.5 1/ (1 r ) or (II) 0.445 0.5 and 1 , then the
publisher’s optimal content-medium strategy and pricing schemes are characterized in the
following table:
cp (0, c p1 ] (c p1 , c p 2 ] (c p 2 , ] ( , c p 3 ] (c p3 ,1)
r
(0, r0 ] PI* : bundle digital PII* : bundle digital PIII* : digital only
(r0 , r1 ] PI* : bundle digital PII* : bundle digital PIII* : digital only
additive parameter, i.e. 0.445 0.5 and 1 . Under these conditions, when the marginal
21
only in the digital medium and adopts the pricing scheme PIII* . Interestingly, when the marginal
cost is high and the proportion of the digital-savvy consumers is large ( c p 3 c p 1 and
r2 r 1), the publisher offers both the digital medium and a bundle of the two mediums.
the marginal cost of the bundle is high ( c p 1 ), if the publisher continues to offer PII*
, then on the one hand, due to the high price of the bundle, the demand for the bundle is low;
and on the other hand, due to the high marginal cost, the profit margin from each unit of the
bundle is low. This leads to a low profit. By shifting to the strategy of offering content only in
the digital medium, the publisher can save on the high marginal cost. Therefore, the publisher
is better off by offering only the digital medium and adopts PIII* .
) and the marginal cost of the bundle is high ( c p 3 c p 1), the publisher shifts from offering
the pricing scheme PI* to offering the pricing scheme PIII* (third row of the table in Proposition
2b) wherein she stops offering the bundle and lowers the price of the digital medium (
PIIId * PId * ). This is because if the marginal cost is high, then the publisher may be better off
by not offering the bundle. Moreover, a lower digital price ( PIIId * ) incentivizes more traditional
consumers who have lower WTP for the digital to buy the digital medium. In this case, the
profit gain from the cost saving and the increased market coverage from the traditional
consumers overweighs the profit loss from the reduced profit margin because of the lower
digital price. Therefore, the publisher is better off by offering only the digital medium to all
One might think that when the proportion of digital-savvy consumers is large and the
marginal cost of the bundle is high, then the publisher will be better off by offering only the
digital medium. Counter-intuitively, we find that when the proportion of digital-savvy
consumers is large ( r r2 ), the publisher is better off by offering a menu of the bundle and the
22
c p3 c p 1 ). The reason for this interesting result is that when there is a large proportion of
digital-savvy consumers in the market, the publisher can gain a higher profit from them by
charging a higher digital price ( 1/ 2 ) under PI* . But the higher digital price makes the digital
medium less attractive compared to the bundle, especially to the traditional consumers who
have a lower WTP for the digital. When the publisher offers only the digital medium at a high
price, then fewer traditional consumers buy the digital medium. Therefore, the publisher is
better off by offering a menu of the bundle and the digital which allows the traditional
consumers to buy the bundle for which they have a higher WTP, rather than to buy the digital
for which they have a lower WTP. Therefore, the publisher gains a higher profit from both
types of consumers. The tradeoff is that the cumulative gain of the high profit from the large
proportion of digital-savvy consumers, and from the increased demand from the traditional
consumers, outweighs the loss due to the high marginal cost of the bundle. This leads to the
optimality of offering a menu of the bundle and the digital-only with pricing scheme PI* .
In the parameter regions in which it is optimal to offer a menu of the bundle and the
digital, the publisher offers PI* or PII* under different conditions. When the marginal cost is
the publisher’s optimal pricing strategy is to switch from PI* to PII* wherein the traditional
consumers buy the bundle and the digital but the digital-savvy consumers buy only the digital
(Profile 3, Figure 1). This is because the publisher’s gain from saving on the marginal cost by
incentivizing both consumer segments to buy the digital is higher than her loss from the digital-
savvy consumers who are offered the digital at a lower price.
PROPOSITION 2c: If 0.445 0.5 and 1 , then (a) when r (0, r1 ] , the publisher’s
optimal pricing strategies remain the same as presented in the first two rows of the table in
Proposition 2b; and (b) when r (r1,1) , the publisher adopts PI* if c p (0, c p 3 ] , and adopts
PIII* if c p (c p3 ,1) .
Proposition 2c describes the optimal pricing schemes when the mismatch cost
parameter is relatively large ( 0.445 0.5 ) and the sub-additive parameter is relatively
23
is relatively large ( r (r1 , 1] ), the publisher’s optimal pricing scheme is PI* if the marginal
cost is low, and is PIII* if the marginal cost is high. Pricing scheme PII* is not optimal for any
value of c p . This is because when both and are not sufficiently large, the difference in
traditional consumers’ WTP for the bundle and for the digital is not large. The same applies
for the digital-savvy consumers. Therefore, when r is relatively large, the publisher is better
off by offering PI* wherein she gains more revenue from digital-savvy consumers, if c p is not
large. On the other hand, if c p is large, then the publisher switches to PIII* and does not offer
In this scenario, we identify the regions in which pricing scheme PI* , PII* , or PIII* is optimal
when (i) the mismatch cost parameter ( ) is moderate ( ((1.5 1.25), 0.445) ), and the
The detailed analysis and regions are in Propositions 2d to 2g in Appendix B. Note that unlike
Propositions 2b and 2c, in this case we have all three pricing schemes across the entire
parameter spaces of the marginal cost ( c p ) and the proportion of digital-savvy consumers ( r ).
This is because when is small or moderate, it implies that the difference between traditional
consumers’ WTP for the digital and digital-savvy consumers’ WTP for the digital is relatively
smaller than in the case described in Propositions 2b and 2c. In other words, the two types of
consumers become closer in their WTP. This implies that the role of the proportion of one type
of consumers is not dominant enough to rule out either PII* or PIII* under any parameter space
of c p and r .
24
to large ( c p (0.4, 1) ), then the publisher is better off by offering only the digital medium at
Figure 2 (ii) illustrates the optimal content-medium strategy and pricing schemes when
the mismatch cost parameter is large ( 0.6 ). Comparing the optimal content-medium
strategy under small (Figure 2 (i)) with that under large (Figure 2 (ii)), when the marginal
cost is large ( c p (0.4, 0.9) ), the publisher is better off by offering only the digital medium
when is small, but she is better off by offering both the bundle and the digital when is
large. When the mismatch cost is large, traditional consumers have a much lower WTP for the
digital compared to that for the bundle. If the publisher continues to offer the digital medium
only, then she has to reduce the digital price significantly to incentivize some traditional
25
bundle. Since the publisher’s margin from selling the bundle is large, she is better off.
Interestingly when the marginal cost is very large ( c p (0.9, 1) ), and the proportion
of digital-savvy consumers is large ( r (0.55, 1) ), the publisher is still better off by offering
the bundle and the digital medium. This is counter-intuitive because one may expect that the
publisher would not offer the bundle when the marginal cost of the bundle is very high. The
intuition for this surprising observation is as follows.
The digital-savvy consumers have a higher WTP for the digital than the traditional
consumers have. When the proportion of digital-savvy consumers in the market is large, the
digital-savvy consumer segment is relatively more important to the publisher. The publisher is
better off by charging a higher digital price, which she can do under pricing scheme PI* ,
wherein the digital price is higher than that under PIII* . Thus, the publisher gets a higher profit
from the large proportion of digital-savvy consumers. However, when the price of the digital
medium is high, very few traditional consumers buy the digital because they have a low WTP
for the digital (because the mismatch cost is high). Therefore, in this situation, the publisher is
better off by offering the bundle along with the digital medium, wherein the traditional
consumers buy the bundle. At the margin, the gain from the large proportion of digital-savvy
consumers due to the higher digital price, and the gain from the increased demand from the
traditional consumers outweigh the loss due to the marginal cost of offering the bundle.
Therefore, when the mismatch cost is relatively large and the proportion of digital-savvy
consumers is large, the publisher’s optimal content-medium strategy is to offer a bundle as well
as the digital at the pricing scheme PI* even when the marginal cost is very high.
The optimal pricing schemes in the - c p space are illustrated in Figure 3. Figure 3
(ii) illustrates the optimal pricing schemes when r is large. By comparing Figure 3 (i) and
Figure 3 (ii), we can see that as r increases, the region in the - c p space in which the
publisher implements optimal pricing scheme PII* becomes smaller, and the region in which
26
publisher implements PII* or PIII* , then the loss from digital-savvy consumers is not only large
but also increases with (because the digital price is lower and decreases as increases).
Hence, the optimality of PII* becomes more restrictive and that region shrinks, and PIII* remains
Figure 3: Optimal pricing schemes in mismatch cost ( ) and marginal cost ( c p ) space,
0.3 , (i): when the proportion of digital-savvy consumers ( r ) is small ( r 0.2 ), (ii):
when r is large ( r 0.7 ).
4. Comparative Statics and Numerical Extensions
In this section, we analyze the impact of changes in the marginal cost or in the proportion of
digital-savvy consumers, on the prices, publisher’s profit, consumer surplus and social welfare.
We also provide extensions via numerical analysis wherein we relax some of our model
assumptions.
4.1 The impact of the marginal cost on the optimal prices
PROPOSITION 3a: While the optimal price of the bundle always increases with the marginal
cost, the optimal price of the digital has a discrete decrease under some conditions.
Proposition 3a highlights the impact of the marginal cost on the digital price as well as
on the bundle price (Figure 4). Note that when the publisher changes the pricing scheme from
PI* to PII* , (i) the price of the bundle has a discrete increase though the rate of price increase
27
Figure 4: The optimal prices with the marginal cost of the bundle ( c p ):
0.15, r 0.35, 0.55
This result is counter-intuitive because prior literature (Bakos and Brynjolfsson 1999,
Milgrom and Strulovici 2009) suggests that as the price of one good increases, the price of
another substitutable good also increases. In our context, the digital medium and the bundle
are partial substitutes (Kannan et al. 2009, Koukova et al. 2008), but an increase in the marginal
cost of the bundle leads the publisher to switch from PI* to PII* wherein the price of the bundle
increases but the price of the digital decreases. The intuition for this surprising result is as
follows. When the proportion of digital-savvy consumers is relatively small, if the marginal
cost of the bundle is sufficiently large, the publisher is better off by incentivizing some
traditional consumers to buy the digital instead of the bundle to save on the cost. However,
when the mismatch cost is large, the traditional consumers have a low WTP for the digital
medium and in order to incentivize them to buy the digital medium, the publisher increases the
bundle price, and at the same time, reduces the digital price (switch from PI* to PII* ). Here the
trade-off is such that the gain from saving on the cost and increasing the market coverage
outweighs the loss of the profit margin from consumers who buy the digital medium because
of the reduced digital price.
4.2. The impact of the proportion of digital-savvy consumers on the optimal prices and
the optimal market coverage
28
when r is relatively large, the publisher switches to PI* . When the publisher switches from
PII* to PI* , the price of the digital has a discrete increase and the price of the bundle has a
discrete decrease.
. Given the large marginal cost, the publisher is better off by incentivizing some traditional
consumers to buy the digital rather than the bundle (Profile 3, Figure 1 (ii)). Further, in pricing
scheme PII* , both the digital price and the bundle price increase with r . This is because as r
increases, the publisher is better off by increasing the digital price to gain a higher revenue
from digital-savvy consumers who buy the digital medium. Moreover, the relative
attractiveness of the bundle increases as the digital price increases, so the publisher increases
the bundle price to gain a higher revenue from traditional consumers who buy the bundle.
As r increases and becomes sufficiently large, i.e. at some threshold, the publisher is
better off by switching to the pricing scheme PI* from PII* . This implies that the publisher
29
bundle price ( pIb* , discrete decrease). By charging a higher digital price, the publisher can gain
the higher profit margin from the digital-savvy consumers. However, since the digital price has
now become high, the bundle becomes more attractive not only to the traditional consumers,
but also to the digital-savvy consumers. Therefore, at the threshold, the publisher is better off
by reducing the bundle price, so that some high-valuation digital-savvy consumers switch to
buying the bundle, for which they have a higher WTP than for the digital medium. All
traditional consumers who purchase buy only the bundle. Therefore, by switching to PI* from
PII* , the publisher’s gain from both types of consumers increases, and the publisher is better
off. This finding suggests that as more consumers become digital-savvy, the publisher may be
better off by increasing the digital price and decreasing the bundle price, which implies that
the difference between the bundle price and the digital price may become smaller.
At the threshold value of r where the publisher switches from PII* (or PIII* ) to PI* , the
market coverage has a discrete decrease. This is because the digital price has a discrete increase
when publisher switches to PI* , and the increased digital price leads to a lower market coverage
for digital-savvy consumers. In addition, after the publisher switches to PI* , the traditional
consumers who purchase buy only the bundle, leading to lower market coverage for the
traditional consumers. After the threshold, the market coverage for the digital-savvy
consumers and the total market coverage increase continuously as the proportion of the digital-
savvy consumers ( r ) increases.
4.3. The impact of the proportion of digital-savvy consumers on the optimal profit,
consumer surplus, and social welfare
PROPOSITION 4a: The publisher’s optimal profit increases with the proportion of digital-
savvy consumers under all pricing schemes.
The publisher’s profit increases with the proportion of digital-savvy consumers in
pricing scheme PI* . This is because the total market coverage increases with r while profit
margins from the bundle ( pI c p ) and digital ( pId * ) are independent of r . The profit gain
b*
from digital-savvy consumers outweighs the profit loss from traditional consumers, leading to
30
profit margin from both the bundle ( pII c p ) and the digital ( pIId * ) increase with r since the
b*
bundle and digital prices increases with r and 2) the increase in market coverage from digital-
savvy consumers is greater than the decrease in market coverage from traditional consumers.
The publisher’s profit increases with r under the pricing scheme PIII* ; while the total market
coverage under pricing scheme PIII* is independent of r , the digital price increases with r ,
when pricing scheme PII* or PIII* is optimal. Social welfare can decrease with r under some
conditions.
One may expect that consumer surplus would monotonically increase with r under the
pricing scheme PII* or PIII* because both the digital price and the bundle price increase with r
and the total market coverage is independent of r . Surprisingly, we find consumer surplus first
increases and then decreases with r under the pricing scheme PII* or PIII* (Figure 6a). Figure
6a shows that the publisher’s optimal strategy is to offer both the bundle and the digital medium
and charge PII* under Scenario 1 (blue curve); and the publisher’s optimal strategy is to offer
only the digital medium and charge PIII* under Scenario 2 (green curve). Under either scenario,
we find that the consumer surplus first increases and then decreases as r increases. The
intuition for this surprising result is as follows. We know the digital-savvy consumers have
higher WTP for the digital medium than the traditional consumers, hence, given the same
digital price, the digital-savvy consumers’ surplus is higher than that of the traditional
consumers. As r increases, the market coverage for the digital-savvy consumers becomes
larger and the market coverage for the traditional consumers becomes smaller, though the total
market coverage remains constant (i.e. 1/ 2 ). Therefore, the total consumer surplus increases.
On the other hand, as the price of the digital increases with r , the surplus decreases for both
types of consumers who purchase the digital medium. When r is sufficiently large (beyond a
threshold), the surplus gain from the increased market coverage from the digital-savvy
31
Figure 6: (a) consumer surplus ( CS * ) with r under different parameter scenarios; (b)
social welfare ( SW * ) with r under different parameter scenarios. Superscript of CS ,
SW refers to the scenarios. Subscript of CS , SW refers to the pricing schemes.
Scenario 1: { 0.1, cp 0.4, 0.35} ; Scenario 2: { 0.1, cp 0.7, 0.35} ;
Scenario 3: { 0.1, cp 0.7, 0.7} ; Scenario 4: { 0.1, cp 0.7, 0.55} .
Figure 6b shows that the social welfare increases with r when PII* or PIII* is optimal
in the feasible region of r ; and the social welfare has a discrete decrease when the publisher
switches from PII* to PI* or from PIII* to PI* in the feasible region of r . The intuition for a
discrete decrease in the social welfare is that when the publisher changes its optimal pricing
strategy from PII* (or from PIII* ) to PI* as r becomes sufficiently large, the market coverage
from both types of consumers has a discrete drop, and the total cost increases because more
consumers from both types of consumers buy the bundle. The decrease in market coverage and
the increase in cost lead to a discrete decrease in social welfare.
4.4. Continuous distribution of the mismatch cost parameter
In this section, we generalize our model by relaxing the assumption of the homogeneous
medium-mismatch cost parameter ( ) for all consumers to verify the robustness of our results.
When there is a continuous heterogeneity among consumers in their preferences over
the two mediums, a consumer has no mismatch cost ( 0 ) when he receives the content in
his preferred medium, and has some mismatch cost ( 0 ) when he receives the content in
his un-preferred medium. is continuous and characterizes the degree of inconvenience
32
we create 10,201 different scenarios wherein market parameters c p and r range from 0 to 1
with an increment of 0.01. For each scenario, we determine the optimal bundling strategy by
comparing the optimal profit from offering the bundle and the digital with that from offering
the digital medium only. We adopt a grid search procedure to search the optimal prices that
generate the maximum profit for each case of offering. Note that in the computation of profit,
we make sure the IC and IR constraints are satisfied. After getting the optimal prices, we
compute the optimal demand, consumer surplus, and social welfare for each scenario. Below,
we present the results of numerical analysis.
Result 1: The publisher’s optimal bundling strategies are consistent with those reported in §3.
Figure 7 presents the region plot of the optimal bundling strategies in the space of the marginal
cost and the proportion of digital-savvy consumers. We find that when the marginal cost is not
high, offering both the bundle and the digital medium is optimal; when the marginal cost is
high and the proportion of the digital-savvy consumers is relatively large, offering both the
5
We set the support for from 0.2 to 0.7 to have better visualization of our results. We have checked that our results
hold for the support of in any range from 0 to 1.
33
Figure 7: Region plot for the optimal bundling strategies with marginal cost and the
proportion of digital-savvy consumers, 0.1
Result 2: The optimal price of the digital medium can have discrete decreases as the marginal
cost of the bundle increases (consistent with Proposition 3a). The optimal price of the bundle
can have discrete decreases as the proportion of digital-savvy consumers increases (consistent
with Proposition 3b).
We examine the impact of the marginal cost on the publisher’s optimal prices and the
impact of the proportion of digital-savvy consumers on the optimal prices. Figure 8a shows
that when the proportion of digital-savvy consumers is relatively small and the marginal cost
of the bundle is sufficiently large, while the bundle price has discrete increases, the price of the
digital incurs discrete decreases. This result is consistent with Proposition 3a and the intuition
is similar to that discussed in §4.1. The reason for the multiple discrete decreases is that in this
generalized model with a continuously heterogeneous mismatch cost, the publisher has
multiple optimal pricing schemes for different values of the marginal cost ( c p ). Hence, as c p
increases and reaches a threshold, the publisher switches from one optimal pricing scheme to
another optimal pricing scheme, which leads to a discrete drop in the digital price and a discrete
increase in the bundle price. Figure 8b shows that as the proportion of digital-savvy consumers
34
Figure 8: Optimal prices with (a) the marginal cost, ; (b) the proportion of digital-savvy
consumers, c p 0.4, 0.1
Result 3: The consumer surplus can decrease as the proportion of digital-savvy consumers
increases, and social welfare may have a discrete decrease when the proportion of digital-
savvy consumers is sufficiently large. This result is consistent with Proposition 4b.
Figure 9 verifies that the consumer surplus and social welfare can decrease with r
under some conditions, when the proportion of digital-savvy consumers ( r ) is sufficiently
large. The multiple discrete decreases are due to the switching between multiple optimal
pricing schemes in the feasible regions of r wherein the digital price has discrete increases.
This leads to a decreased consumer surplus because of the lower market coverage and the lower
surplus for consumers who purchase the digital medium. The social welfare also incurs discrete
decreases because of the decreased consumer surplus when the publisher switches its pricing
schemes.
35
and the digital under the optimal pricing scheme PI* gives the publisher a higher profit,
compared to the case where the publisher offers only the bundle (refer to the proof for
Proposition 1 for the details wherein the condition 1 plays no role). When the marginal
cost of the bundle is large, the bundle becomes costly, and the publisher has more incentive to
offer the digital together with the bundle so that some low-valuation consumers who previously
do not purchase under the case where only the bundle is available now will buy the digital
medium. This leads to the higher total market coverage and the publisher is better off.
The above analysis implies that even when two mediums are complements, we have
only two content-medium strategies to choose from: (1) offering content in the bundle and in
the digital medium, and (2) offering content only in the digital medium. When the two mediums
are complements, i.e. 1 , and since by model setup, cp (0,1) , we have cp . We
36
content only in the digital medium) is not optimal. Therefore, when the two mediums are
complements, there is no feasible region in which a publisher is better off by offering content
only in the digital medium. Hence, when the two mediums are complements, offering both the
bundle and the digital medium is optimal in the entire feasible region.
4.6. When consumers’ valuations for the two mediums are independent
In our conceptualization, we assume that consumers’ willingness to pay (WTP) for the content
in the physical medium and for that in the digital medium are correlated. This assumption is
consistent with prior literature (Venkatesh and Chatterjee 2006, Hsiao and Chen 2014, Tan and
Carrillo 2014), and reflects the nature of our context wherein the publisher offers the same
content in the two mediums. However, much of the bundling literature assumes that the
valuations of the two components or products are independent (McAfee et al. 1989, Armstrong
1999, Bakos and Brynjolfsson 1999, Nalebuff 2004). In order to compare our results with the
extant literature in bundling, we numerically explore (since the analytical solution is
intractable) the implication of a setting where consumers’ WTPs for the two mediums are
independent.
Let a consumer’s valuation for the content in the physical medium be v1 , where
v1 U [0, w1 ] , and his valuation for the content in the digital medium be v2 , where
mediums, and (0,1) . This is similar to the assumption of the partial substitutability in
37
content in the digital medium, v2 . However, the key difference is that in the generalized setup
in §4.4, a consumer’s valuations are ordered. For example, for a traditional consumer, his
valuation for the physical medium is greater than that for the digital medium, that is v v(1 )
for any draw of the mismatch cost parameter . In the setup of the independent valuations, a
consumer’s valuations for the two mediums are not ordered. Hence, there are no consumer
segments. This also implies that there is no notion of mismatch cost parameter in this setup.
Under this setting, offering the content in the physical medium is dominated by offering
the content in the bundle and the logic is the same as described in §2.2; and offering the content
in the bundle only is weakly dominated by offering the content in the bundle and in the digital
medium. Now we verify if our optimal content-medium strategy results still hold under the
independent-valuation setup, i.e. whether offering only the digital medium or offering both the
bundle and the digital medium is optimal.
Figure 10: Region plot for the optimal strategies with the marginal cost ( c p ) and the
partial substitutability ( )
We adopt the numerical analysis similar to the one described in §4.4 by generating
synthetic consumers and creating a number of market scenarios from different values of the
marginal cost c p and the partial-substitutability factor . For each scenario, we determine the
optimal content-medium strategy. The region plot in Figure 10 verifies that under some
parameter space the publisher’s optimal strategy is to offer the content in a bundle of mediums
38
bundle of the two mediums dominates offering only the physical medium. Furthermore,
offering a bundle and the digital medium dominates offering only the bundle (see Appendix B
for the illustrative proof). The intuition is similar to that discussed for Proposition 1 in §2.2.
Figure 11 illustrates the upper bound on the ratio of the marginal cost of the digital to
that of the physical for which our results hold as the marginal cost of the physical medium
increases. Note that the market parameters for Figure 11 are such that the WTP for the physical
is close to that for the bundle (i.e. 0.1 ), the proportion of digital-savvy consumers is small
(i.e. r 0.1 ), and their WTP for the physical is close to that for the digital (i.e. 0.1 ). This
implies that under the market conditions such that the WTP for the bundle is close to that for
39
of the digital to that of the physical is relatively large (i.e. greater than 30%); when the marginal
cost of the physical medium is large (i.e. c p 0.8 ), the upper-bound ratio is relatively small,
but still greater than 10%. Therefore, our results in §3 are robust to the zero marginal cost of
the digital medium as long as it is relatively small compared to the marginal cost of the physical
medium.
Figure 11: The upper-bound ratio of the marginal cost of the digital to the marginal cost
of the physical as the marginal cost of physical ( c p ) increases ( 0.1, 0.1, r 0.1 ).
4.7.2. Decision for investment in the infrastructure for the digital medium
Furthermore, our research has focused on the optimal content-medium pricing strategies and
the implications of key market parameters on the publisher, consumers and society given that
a monopolistic publisher has already made investments in infrastructure to distribute content
in the physical as well as in the digital medium (infrastructure investments are sunk). However,
in this subsection we study the digital medium investment decision of a publisher who has the
infrastructure for the physical distribution (fixed cost for the physical medium is sunk) but is
exploring the possibility of creating infrastructure for the digital medium. The publisher will
make the investment in the digital infrastructure only if the fixed cost of the digital medium is
40
the digital medium and in a bundle of the two mediums at the pricing scheme PI* such that the
demand profile 1 emerges, we find that offering the content in a bundle and in the digital
medium generates higher profit for the publisher than offering the content only in the physical
1 r 1 r 1
medium. The difference in the two profits is: (1 c2p ( ) ) . If
4 2
1 1 r
the fixed cost of the digital infrastructure ( c f ) is lower than this threshold, then the publisher
should invest in creating the digital infrastructure and offer the content in both the standalone
digital medium and a bundle of the two mediums.
5. Discussion and Conclusion
The impact of digitization on publishers and consumers is an imperative question for content
industries in the digital era. Prior analytical studies on information goods pricing over dual-
medium access provide limited insights because they either have not abstracted the unique
characteristics in the content market and provided analytically tractable pricing solutions, or
they have not analyzed the impact of changes in key market characteristics on the publisher’s
prices and profit, on the consumer surplus and the social welfare. In this paper, we bridge this
gap in literature by building an analytical model with heterogeneous consumers’ valuations and
preferences, and an asymmetric cost structure for partially substitutable physical and digital
mediums. Our model lends itself to the closed-form analytical solutions for optimal pricing
schemes, and enables us to analyze the impact of changes in the market characteristics on the
publisher as well as on consumers.
We find that offering only the digital medium is optimal under some market conditions,
while offering a bundle of mediums and the digital medium is optimal under other market
41
6
Most recently The Wall Street Journal offers the “print only” which actually includes the digital access to WSJ.com and
has a slightly lower price than the price for the print + digital option.
42
43
44
7
The generation of people born during or after the rise of digital technologies (Degraff 2014).
45
46
References
Adams, W. J., & Yellen, J. L. (1976). Commodity bundling and the burden of monopoly. The
Quarterly Journal of Economics, 475-498.
Agility PR Solutions. (2017). Top 15 U.S. newspapers by circulation,
https://www.agilitypr.com/resources/top-media-outlets/top-15-daily-american-newspapers/
Armstrong, M. (1999). Price discrimination by a many-product firm. The Review of Economic
Studies, 66(1), 151-168.
Armstrong, M. (2013). A more general theory of commodity bundling. Journal of Economic
Theory, 148(2), 448-472.
47
Droppo, D. (2014, October 9). The five coolest digital-only labels out right now.
http://www.complex.com/music/2014/10/five-coolest-digital-only-labels-out-right-now/
Eckalbar, J. C. (2010). Closed‐form solutions to bundling problems. Journal of Economics &
Management Strategy, 19(2), 513-544.
Esterl, M. (2005, May 17). New York Times sets an online fee.
http://www.wsj.com/articles/SB111630351421435538
48
Koukova, N. T., Kannan, P. K., & Kirmani, A. (2012). Multiformat digital products: How design
attributes interact with usage situations to determine choice. Journal of Marketing Research, 49(1),
100-114.
Koukova, N. T., Kannan, P. K., & Ratchford, B. T. (2008). Product form bundling: Implications for
marketing digital products. Journal of Retailing, 84(2), 181-194.
Laffont, J. J., & Martimort, D. (2009). The Theory of Incentives: the Principal-agent Model.
Princeton University Press.
Lahiri, A., & Dey, D. (2013). Effects of piracy on quality of information goods. Management
Science, 59(1), 245-264.
49
Perez-Pena, Richard. (2007). Times to stop charging for parts of its web site,
https://www.nytimes.com/2007/09/18/business/media/18times.html
50
Sundararajan, A. (2004a). Managing digital piracy: Pricing and protection. Information Systems
Research, 15(3), 287-308.
51
Wei, X. D., & Nault, B. R. (2014). Monopoly versioning of information goods when consumers
have group tastes. Production and Operations Management, 23(6), 1067-1081.
Williams, C. (2013). Lloyd's List abandons print after 279 years,
https://www.telegraph.co.uk/finance/newsbysector/mediatechnologyandtelecoms/media/1033313
6/Lloyds-List-abandons-print-after-279-years.html
Wu, S. Y., & Chen, P. Y. (2008). Versioning and piracy control for digital information goods.
Operations Research, 56(1), 157-172.
Yang, N. (2012). Swipe, share, sell: What publishers need to know about digital editions,
http://www.editorandpublisher.com/feature/swipe-share-sell-what-publishers-need-to-know-
about-digital-editions/
Zhang, Z., Nan, G., Li, M., & Tan, Y. (2016). Duopoly Pricing Strategy for Information Products
with Premium Service: Free Product or Bundling? Journal of Management Information
Systems, 33(1), 260-295.
Zhou, J. (2017). Competitive bundling. Econometrica, 85(1), 145-172.
52