Optimum design of electronic communities as economic entitiesorman.johnson.cornell.edu/orman/publications/communitiesISF.pdf · Optimum design of electronic communities as economic

Inf Syst Front (2006) 8:179–194

DOI 10.1007/s10796-006-8778-9

Optimum design of electronic communities as economic entitiesLevent V. Orman

Received: 15 July 2004 / Revised: 23 March 2005 / Accepted: 7 April 2005C© Springer Science + Business Media, LLC 2006

Abstract Electronic communities can be designed to or-

ganize consumers, to pool their purchasing power, and to

guide their purchasing decisions. Such commercial elec-

tronic communities have the potential to facilitate the creation

of novel marketplaces, and even radically change the buyer-

seller interaction, as physical communities did throughout

the history. Commercial electronic communities are groups

of consumers that participate in the marketplace as a sin-

gle unit. In addition to bargaining power gained from such

bundling, such communities can expand markets by reducing

market uncertainty, and they have the potential to drastically

reduce consumers’ transaction costs, by facilitating group

transactions and bulk purchasing. Communities are charac-

terized by their size, their pricing strategy, and their member-

ship characteristics. Analytical models and numeric analysis

is utilized to compute the optimum size of a community for

given market characteristics. Two major community pricing

strategies are analyzed to improve the community design, and

the conditions are derived where one dominates the other.

Finally, market segmentation techniques are introduced to

control the membership characteristics of the community to

further improve the design.

Keywords Electronic communities . Electronic markets .

Economics of markets . Market design . Consumer surplus .

Electronic business models

1. Electronic communities

Electronic communities are virtual gathering places for

people sharing common interests. They have been studied

L. V. Orman (�)Cornell University, Sage Hall, Ithaca NY 14853e-mail: [email protected]

extensively as a social phenomenon, and there is increas-

ing interest in their commercial potential (Rheingold, 1993;

Suttles, 1972). Early Internet vendors have established com-

munities to facilitate interaction among their customers, in

an effort to engender loyalty and a sense of community

among their customers, and to engage customers to extend

the length of stay at their site (Williams and Cothrel, 2000).

Such vendor-based communities are useful, but limited in

scope and functionality as commercial enterprises. They are

limited in scope to one vendor’s products, and they are lim-

ited in functionality to serve only the commercial interests

of one particular vendor.

More recently, there has been an increasing interest in

communities as stand-alone commercial enterprises indepen-

dent of any particular vendor. These communities narrowly

define their audience in terms of a common interest, and act

as an information clearinghouse and a community center, in

addition to being a commercial enterprise with affiliated ven-

dors, product recommendations, and endorsements. These

interest-based communities are more customer-centric com-

pared to the vendor-based communities, and they range in

focus from parenting to gardening as in parents.com and gar-

den.org, and range in audience from dairy farmers to gourmet

chefs as in dairyfarmer.net and gourmet.org. These commu-

nities are more ambitious in their objectives, attract a more

loyal and focused customer base, but they often fail to orga-

nize the consumers to increase their negotiating power with

vendors. Instead, they often act as a narrowly focused niche

retailer, or as an agent of vendors and distributors, rather than

entering the market as an agent of consumers and negotiat-

ing on their behalf (Armstrong and Hagel, 1996; Mougayar,

1998).

There have been efforts to build communities that par-

ticipate in the marketplace as agents of consumers. They

act as buying cooperatives by organizing consumers and

Springer

180 Inf Syst Front (2006) 8:179–194

aggregating their demand, and then negotiating on their be-

half with vendors, as in merkata.com, ewinwin.com, online-

choice.com and letsbuyit.com. However, the success of such

transaction-based communities has been limited, since their

membership is very ephemeral, and limited to a single pur-

chasing decision (Hagel and Brown, 2002; Williams and

Cothrel, 2000).

There is a need to combine the organized membership of

interest-based communities with the commercial potential

of transaction-based communities. Such electronic commu-

nities designed to organize consumers, pool their purchas-

ing power, and guide their purchasing decisions have the

potential to facilitate electronic marketplaces, and even rad-

ically change the nature of buyer-seller interaction, as phys-

ical communities throughout the history have often changed

the nature of physical marketplaces (Armstrong and Hagel,

1996; Lechner and Hummel, 2002). We will refer to such

electronic communities as commercial electronic communi-

ties, and their potential and optimum design is the subject

of this article. We will show that communities can benefit

consumers significantly by pooling their purchasing power.

But, the optimum design of communities to achieve those

benefits is complex, and requires a careful analysis of a num-

ber of factors including community size studied in Section 3,

transaction costs in Section 4, pricing strategy in Section 5,

and community membership in Section 6.

Commercial electronic communities in their most general

form are communities that participate in the marketplace as

a single unit. Members join a community and pay a fixed

membership fee which entitles them to all the goods and

services provided by the community. Such communities are

often limited in the physical world due to the intensive in-

formation processing requirements in forming the commu-

nity. There are some examples of such communities in the

physical world, but in their effort to reduce the information

processing requirements to manageable levels, they tend to

limit their size and their scope, and even then they are of-

ten criticized for being heavy handed and non-responsive

to consumer needs. Consequently, such communities in the

physical world are considered feasible only when the trans-

action costs are very high and the goods are very complex,

as in health care delivery, public infrastructure, and complex

financial instruments (Porter and Scully, 1987). A typical ex-

ample is mutual funds where investors join funds, and a fund

manager engages in transactions on behalf of all the mem-

bers of the fund. All members participate in all transactions

by virtue of their membership, and the fund manager aggre-

gates the demand from all members, resulting in significant

reduction in transaction costs.

The critical issues in designing electronic communities

range from determining their feasibility and their optimum

design, to the impact of such communities on the markets

and the economy. Communities can be viewed as bundles

of consumers, and some of the analytical techniques used to

study bundles of goods are also applicable to communities.

Unfortunately, the bundling literature is not directly applica-

ble to communities, since bundles are formed by suppliers to

maximize profits, but communities are formed by consumers

to maximize consumer surplus. The symmetry is not perfect

since the suppliers also have control over prices (Bakos and

Brynjolfsson, 2000; Bakos and Brynjolfsson, 1999; Hanson

and Martin, 1990).

Consumer cooperatives are also similar to communities

in their basic objective, however their emphasis has been on

the acquisition of bargaining power in a zero sum game en-

vironment (Ireland and Law, 1983; Porter and Scully, 1987).

This literature largely ignores the benefits of a well orga-

nized market both to buyers and sellers by expanding the

market. The benefits to sellers from an organized market can

be significant especially if the organization reduces transac-

tion costs. However, reduction in transaction costs was not

studied as an objective of consumer organization until the

recent advent of electronic markets, since any reduction in

transaction costs would have been easily offset by the cost

of organizing consumers (Anand and Aron, 2003). Internet

has changed the economics of market design by reducing

the cost of organizing consumers, and making more com-

plex structures economically feasible (Burnett and Buerkle,

2004; Hagel and Brown, 2002; Lechner and Hummel, 2002).

However, the optimum design of these new structures is very

poorly understood.

One literature that directly studied consumer organiza-

tion involved natural aggregation of consumers like families

where the cost of organization is negligible. There is a con-

siderable literature on family purchases, and family decision

making as a consumer group. Unfortunately, the size of a

family unit is often much smaller than a community. More

importantly, size was never a design variable in this literature,

since families were never designed as consumer groups, as we

intend to do with communities. We will show that the size of

a community is a critical design variable, and the optimum

size can be considerably larger than a family (Burnett and

Buerkle, 2004). The theory of clubs considered size as a crit-

ical design variable. However that literature focused on the

economics of sharing resources, and the optimum resource

allocation when sharing and congestion are the critical is-

sues. Communities do not share resources, but merely share

a transaction, and hence create no resource sharing issues

(Sandler and Tschirhart, 1980).

Section 2 introduces the concept of community and pro-

vides a numerical example to further motivate the novelty

and the economic justification of communities. Section 3

develops a mathematical model for the economic feasibil-

ity of communities under a large number of simplifying

assumptions, and shows that larger communities are not nec-

essarily better for consumers. It shows that community size

Springer

Inf Syst Front (2006) 8:179–194 181

is a critical design variable, and it computes the optimum

community size. The subsequent sections progressively relax

the simplifying assumptions, and build progressively more

elaborate models. Section 4 introduces the transaction cost

into the formulation, and shows that higher transaction costs

make communities more desirable both for consumers and

suppliers. Section 5 introduces demand function into the de-

sign of communities, and the effect of demand function on

the economic viability of communities. Demand function of

the community is shown to be an important design variable,

even more critical than the size of the community. Two types

of communities are introduced, in terms of how individual

demand functions are converted into a community demand

function. One type of community is shown to dominate the

other everywhere except for very small costs. Section 6 stud-

ies market segmentation as a strategy to optimize the mem-

bership of communities. Communities can carefully select

their members, rather than building communities out of ran-

dom collections of consumers. Segmentation is shown to

be a useful strategy that benefits communities greatly at the

expense of suppliers, and it also drastically changes the opti-

mum structure of the community. The optimum community

structure is computed for a simple 3-way segmentation, and

shown to be the opposite of the optimum structure without

segmentation.

2. Economic feasibility of electronic communities

A pure community is one where all members are entitled to

all goods and services procured by the community operator,

all bundled in one membership price. A pure community is

a mere transaction processor, and it does not store, own, or

distribute products. A pure community is rare in the physical

world, due to its intensive information processing require-

ments, but quite feasible in electronic commerce with its

reduced communication and information processing costs,

and it is our subject.

Communities can have economic benefits both for con-

sumers and for businesses. The obvious benefit to consumers

is the negotiating power gained by pooling their purchasing

power. The obvious benefit to the vendors is a well described

and focused market that is available for precise targeting.

The advantage to both parties is a significant reduction in

transaction costs, due to bulk purchasing for the whole com-

munity, and better-targeted marketing. A less intuitive advan-

tage of communities is their ability to expand markets and

generate a surplus above and beyond what could be achieved

without communities. Such an economic surplus can make

all parties better off, providing a strong justification for the

creation of communities.

However, an economic justification for communities does

not require both parties to be better off, but only the

community. A community organizer can form a community

as a separate economic entity that enters the marketplace on

behalf of its members. Such a community would be econom-

ically feasible to the extent that it can extract a surplus for the

community from the marketplace, and hence makes all of its

members better off. That surplus can be viewed as additional

consumer surplus for the community members, or as profits

for the community operator, or more likely as a combina-

tion of both, providing both incentives for consumers to join

a community, and incentives for the community operator to

organize a community.

Consider a population of 4 consumers, and a monopolist

supplier of a single good. Let the reservation prices of the

consumers for the good be 4, 7, 8, and 9 respectively, and the

marginal cost of producing the good is 2/unit. The optimum

price set by the monopolist will be 7, with a total profit of

15, and a total consumer surplus of 3, as shown in Fig. 1.

Now consider a pair wise grouping of these consumers

into two communities, with reservation prices 4 + 7 = 11,

and 8 + 9 = 17. Each community consists of 2 consumers

and will buy a package of 2 units of the good, if its price is

below its reservation price, or it will buy none. The marginal

cost of a two-unit package is 2 ∗ 2 = 4. The optimum price

set by the monopolist for a two-unit package is 11, the total

profits are 14, and the total consumer surplus is 6, as shown

in Fig. 2.

Clearly, the communities, by using their bargaining power,

extracted additional surplus, raising it from 3 to 6, at the

expense of the supplier, whose profits fell from 15 to 14. The

community accomplished this by pressuring the supplier to

lower its unit price from 7 to 11/2 = 5.5, as a direct result

of bargaining power gained by forming a community.

The additional consumer surplus is not necessarily

achieved at the expense of supplier profits. Communities can

Price units sold profit margin total profit consumer surplus

4 4 2 8 0+3+4+5=12

7 3 5 15(optimum) 0+0+1+2=3

8 2 6 12 0+0+0+1=1

9 1 7 7 0+0+0+0=0

Fig. 1 The total profit and theconsumer surplus for 4consumers with reservationprices 4, 7, 8, and 9, and amonopolist supplier of a goodwith marginal cost of 2

Springer

182 Inf Syst Front (2006) 8:179–194

Price units sold profit margin total profit consumer surplus

11 2 7 14(optimum) 0+6=6

17 1 13 13 0+0=0

Fig. 2 The total profit and thetotal consumer surplus for 2communities with reservationprices 11 and 17, and amonopolist supplier of a goodwith marginal cost of 4

Price units sold profit margin total profit consumer surplus

12 2 8 16(optimum) 0+4=4

16 1 12 12 0+0=0

Fig. 3 The total profit and thetotal consumer surplus for 2communities with reservationprices 12 and 16, and amonopolist supplier withmarginal cost of 4

expand markets, and generate surplus that can benefit all.

Both consumers and suppliers can be better off as a result

of communities. Consider a different pair-wise grouping of

the four consumers into two new communities, with reser-

vation prices of 4 + 8 = 12, and 7 + 9 = 16. The marginal

cost of a two-unit package is still 4. The optimum price set

by the seller will be 12, with a total profit of 16, and the total

consumer surplus will be 4, as shown in Fig. 3.

Clearly, these communities generated enough surplus to

make both parties better off. The seller’s profits went up from

15 to 16, and the total consumer surplus rose from 3 to 4. The

communities accomplished this by expanding the market,

and generating additional economic surplus for all to share.

The critical question for a community operator is how

to determine the optimum community, and to design it to

maximize the community surplus. The critical variable is the

variance of reservation prices, or market uncertainty. It is no

surprise that as the variance decreases, the supplier’s profits

go up because of better targeted pricing, and consumer sur-

plus falls. However, that is not uniformly so. At very high

variances, the trend can reverse itself, the market can shrink

considerably by dragging the consumer surplus down, and

the suppliers can appropriate some of the consumer surplus,

and can actually increase their profits. More importantly, the

reduction in transaction costs also benefits consumers as well

as suppliers as the community size increases, but the distribu-

tion of benefits depends on the community size and structure,

further complicating the results.

The community operator’s problem is to design the op-

timum community that maximizes the community surplus,

however there are some constraints to ensure that individual

community members have an incentive to join the commu-

nity. A strategy that maximizes the community surplus is not

necessarily beneficial to each and every member of the com-

munity. It may benefit some community members at the ex-

pense of others, and unless a complex surplus redistribution

scheme is devised, the community may fail. This is where

communities are fundamentally different from product bun-

dles, since community members, unlike product bundles, are

individual decision makers, and the community has to be

incentive compatible with each community member.

Consider the same four consumers with reservation prices

4, 7, 8, and 9. We have shown that a pair wise grouping with

resulting reservation prices 11 and 17 leads to an increase

in community surplus from 3 to 6, because of a reduction

in the optimum price from 7 to 5.5. However at unit price

5.5, not all community members are better off. At price 7,

the surpluses attained by consumers are 0, 0, 1, and 2, since

the first consumer is left out of the market. At price 5.5, or

11 for 2 units, all consumers are in the market, and the con-

sumer surpluses are −1.5, 1.5, 2.5, and 3.5. Clearly, the first

consumer has no incentive to join the community unless it is

subsidized by others. In other words, a community operator,

who is not able to price discriminate, would have to charge a

membership price of 5.5 to cover his costs, and the first con-

sumer with reservation price of 4 would have no incentive to

join.

There are two solutions to this incentive compatibility

problem, and they are both intrinsic to the community, auto-

matically resolving the incentive compatibility problem. The

first is the reduction in transaction costs that may create ad-

ditional incentives to join the community. When the savings

from transaction costs exceed a possible loss by any specific

consumer, the incentive compatibility problem disappears.

When those savings are not sufficient, the second solution

is the effect of multiple products on the distribution of the

surplus. A community procures many products for its mem-

bers. For a single product, the community surplus is likely to

be distributed unevenly, and some members may not receive

a sufficient share to keep them in the community. However,

for large product bundles, the surplus distribution evens out,

and benefits all members of the community equally. This

result holds unless the consumer reservation prices are per-

fectly correlated over many products. Because of this re-

sult, any community design that benefits the community as

a whole will also benefit every individual consumer in the

community. Intuitively, one member may have to subsidize

another for a particular product, but there is reciprocation of

Springer

Inf Syst Front (2006) 8:179–194 183

subsidies for another product, and eventually all consumers

are better off relative to no community.

Consider a bundle of four products, with reservation prices

of four consumers for each product (4, 7, 8, 9), (8, 9, 4, 7),

(7, 4, 9, 8), (9, 8, 7, 4) respectively. We know from above that

the optimum price for each product is 7, and the consumer

surpluses for each product are (0, 0, 1, 2), (1, 2, 0, 0), (0, 0,

2, 1), (2, 1, 0, 0), leading to total consumer surpluses of (3,

3, 3, 3). Now consider the pair wise groping of these con-

sumers into communities with reservation prices (11, 17),

(17, 11), (11, 17), (17, 11). The optimum price will be 11

for two units of each product, or 11/2 = 5.5, with surpluses

(−1.5, 1.5, 2.5, 3.5), (2.5, 3.5, −1.5, 1.5), (1.5, −1.5, 3.5,

2.5), (3.5, 2.5, 1.5, −1.5), and the total consumer surpluses

of (6, 6, 6, 6). Clearly, all community members have the

incentive to join the community, since their individual sur-

pluses all go up from 3 to 6 by joining the community. As the

number of products procured by the community increases,

the incentive problem for the individuals tends to diminish,

because of the redistribution effect on the surplus, unless the

reservation prices for the products are perfectly correlated.

There is a minimum number of products a community has

to procure to effectively eliminate the incentive compatibil-

ity problem. This number is dependent on the size of the

community and the correlation of reservation prices within

the community. The failure to reach that number may create

incentive problems, and this fact may explain the failure of

some early electronic communities which produced ad hoc

communities, each formed to acquire only one specific prod-

uct for the community members such as mobshop.com and

accompany.com. The second solution to the incentive com-

patibility problem is the reduction in transaction costs. The

community operator performs the search and price discovery

functions for the whole community, and has the potential to

drastically reduce transaction costs. This could improve the

incentives for individual members to join the community.

The exact impact of the transaction costs will be computed

in Section 4.

The internal structure of communities is also important,

and there are alternative structures. The community struc-

ture described above is called a “subsidizing community”,

since the reservation price for the community is the aver-

age of reservation prices for all community members, and

to achieve that price level requires community members to

subsidize each other. This need for subsidies is one of the

reasons for the incentive problem discussed above. An alter-

native structure is the “minimalist community”, where the

community reservation price is the minimum of the member

reservation prices, and hence community purchases are made

only when all community members independently agree to

a purchase without cross subsidies. Minimalist communi-

ties give rise to the same issues as subsidizing communities,

and a precise comparison is necessary to determine optimum

structures. Consider four consumers with reservation prices

(3, 5, 7, 7), and production cost of 0. Using the same anal-

ysis as before, the optimum price would be 5, with profit

15, and surplus 4. A pair wise grouping into communities

(3, 7) and (5, 7) with minimalist reservation prices 3 and 5

would lead to the optimum price 3, profit of 12, and a sur-

plus of 10, clearly benefiting the community. But again, not

all minimalist communities are beneficial. A pair of com-

munities (3, 5) and (7, 7) with reservation prices 3 and 7

would lead to the optimum price of 7, and reduce surplus to

0. Moreover, there are still incentive compatibility problems.

The optimum community with reservation prices 3 and 5,

raises individual surpluses from (0, 0, 2, 2) to (0, 2, 4, 4),

where the first consumer has no incentive to join. However,

bundling two products with reservation prices (3, 5, 7, 7)

and (7, 7, 3, 5) would change the individual surpluses from

(0, 0, 2, 2) to (4, 6, 4, 6) overcoming the incentive compati-

bility problem. The exact comparison of these two types of

communities will be provided in Section 5, and they will be

further improved by using market segmentation strategies in

Section 6.

Communities are long term structures where members re-

ceive a large bundle of goods and services over a long period

of time. Assuming that the reservation prices are not perfectly

correlated over multiple products, over the long run the in-

centive problem disappears because of the distribution effect

discussed above. As a result, the community and individual

interests merge over the long run, as long as the community

can maintain a positive overall surplus. Consequently, we

will focus on finding communities that maximize the overall

consumer surplus, since not only it is the most desirable solu-

tion for the members as a group, but also for each individual

member over the long run, hence solving the incentive

problem.

3. Optimum community size

A community operator does not necessarily organize the

whole population into a single community and become a

monopsonist. Smaller communities are also economically

feasible and may even be more desirable for the community

operator as shown informally in the previous section. Assume

a uniformly distributed reservation prices for a population

of N customers for a single bundle of goods. Community

operators organize the customers into communities of size

n, by randomly selecting n customers for each community.

The problem in this section is to determine the new distri-

bution of reservation prices faced by the supplier, and the

profits to be made by the supplier and the community oper-

ators under these conditions. Finally, we will determine the

optimum size n, to maximize the community surplus, since

the community operators have control over n, in designing

Springer

184 Inf Syst Front (2006) 8:179–194

1

0.03

c p 1

P

S

D

Fig. 4 1−F(x) when f(x) is uniform and normalized reservation prices,c = marginal cost, p = price, P = total profit of the supplier, S = totalconsumer surplus, D = deadweight (unappropriated surplus)

S

0.125

0.10

0.08

0.06

0.03

0.1 0.2 0.3 0.5 c

Fig. 5 The consumer surplus as a function of production cost c

their communities optimally. We will assume monopolistic

sellers with no price discrimination, and communities with

fixed membership fee, purchasing a single aggregate bundle

of goods from each supplier.

Assume a uniform distribution f(x) of consumer reserva-

tion prices, normalized to the range 0–1, resulting in variance

1/12. Let c be the marginal cost of production, and let p be the

price set by the supplier. The total profit of the supplier is the

profit margin p−c times the total demand 1 − F(p) = 1 − p,

consisting of suppliers whose reservation prices are over p,

where F(x) is the cumulative frequency distribution. The sup-

plier maximizes the profits with respect to p, and will set the

optimum price p∗ = (c + 1)/2 since

P = (p − c)(1 − F(p)) = (p − c)(1 − p)

d P/dp = (1 − p) + c − p = c + 1 − 2p = 0

p∗ = (c + 1) /2

The total consumer surplus is the number of consumers pur-

chasing the product 1 − F(p) = (1 − p) times the expected

surplus per consumer (1 − p)/2, resulting in (1 − p)2/2.

At p = p∗ = (c + 1)/2, the consumer surplus is (1 − c)2/8.

Figure 4 shows the total profit and the consumer surplus

graphically. Figure 5 shows consumer surplus as a function

of cost.

Now consider a population of communities, each con-

taining n random consumers selected from the uniform

distribution above. The reservation prices of communities

approach a normal distribution with mean 1/2, and variance

1/(12n) from central limit theorem. We will approximate the

community reservation prices as a rectangular distribution

with the same mean and the same variance. This approxima-

tion is useful since the variance is the critical variable for our

results. Exact results will be obtained by numerical analysis,

and will be used to support the validity of the approximate

results. A rectangular distribution of width a, and height 1/a,

will have the variance a2/12, and setting it equal to 1/12n,

we get n = 1/a2. Communities of size 100 then would re-

sult in an approximate rectangular distribution of width 0.1

and height 10 with variance 1/1200, approximating a normal

distribution of variance 1/1200, as shown Fig. 6.

Analyzing a marketplace of communities of size n = 1/a2:

f (x) = 1/a for (1 − a)/2 < x < (1 + a)/2,

F(x) = x/a + (a − 1)/(2a) by integration and by setting

the constants appropriately,

1 − F(x) = −x/a + (1 + a)/(2a)

The supplier’s problem is to maximize profit P, by set-

ting the price p, given the communities characterized by the

parameter a:

P = (p − c)(1 − F(p)) = (p − c)((1 + a)/(2a) − (p/a))

for (1 − a)/2 < p < (1 + a)/2

Maximizing P with respect to p, we get the optimum values

p∗ and P∗:

P∗ = (p∗ − c)(1 + a − 2p∗)/(2a) and

p∗ = (2c + a + 1)/4 for (1 − a)/2 < p∗ < (1 + a)/2

Substituting for p∗:

P∗ = (a + 1 − 2c)2/(16a) and p∗ = (2c + a + 1)/4

if a > (1 − 2c)/3 and a > 2c − 1

The community organizer’s problem is to design commu-

nities, by maximizing the community surplus with respect to

parameter a, given the optimum price p∗ of the supplier:

S =∫ 1

p∗(1 − F(x))dx

S = 1/2((1 + a) / 2 − p∗)(1 − F(p∗))

Springer

Inf Syst Front (2006) 8:179–194 185

= ((1 + a)/2 − p∗)2/(2a) if (1 − a)/2 < p∗ < (1 + a)/2

Substituting for p∗,

S = (1 + a − 2c)2/ (32a)

if a > (1 − 2c)/3 and a > 2c − 1

Maximizing S with respect to a, we get S∗ by computing S at

extreme points, since S is convex and unimodal with respect

to a:

S∗ = Max(S∗1 , S∗

2 , S∗3 )S∗

1 = (1 − 2c)/6 and

a∗1 = (1 − 2c)/3 if c ≤ 0.5

S∗2 = 0 and a∗

2 = 2c − 1 if c ≥ 0.5

S∗3 = (1 − c)2/8 and a∗

3 = 1

By comparing S1 through S3, and picking the maximum:

S∗ = (1 − 2c)/6 and a∗ = (1 − 2c)/3 and

P∗ = (1 − 2c)/3 and p∗ = (1 + c) /3 if c ≤ 0.33

S∗ = (1 − c)2/8 and a∗ = 1 and

P∗ = (1 − c)2/4 and p∗ = (1 + c)/2 if c > 0.33

The community is feasible whenever a∗ < 1, since a = 1

corresponds to the original market, or community size 1.

Clearly, for marginal cost less than 0.33, a community is

economically feasible, leading to additional surplus for con-

sumers, as shown in Figs. 7 and 8.

For example, for c = 0.2, the optimum variance of reser-

vation prices is given by a∗ = 0.20, and the optimum commu-

nity size is n∗ = 1/a2 = 25. The optimum price p∗ = 0.40,

the supplier’s profit P = 0.20, and the total consumer sur-

plus S = 0.10. The consumer surplus was 0.08 without a

community, as a quick comparison of Figs. 8 to 5 shows.

The analytical results were obtained by approximating ag-

gregate distributions as rectangular distributions. The exact

results are obtained by numerical analysis using Mathemat-

ica software (Suttles, 1972), and they are similar to the an-

alytical results, demonstrating the robustness of the results.

Figure 9 summarizes the numerical results obtained by ran-

domly drawing from a uniform distribution of reservation

prices, 500 sample communities, for each size n = 1−100,

and each cost c = 0.01−1.

4. Effect of transaction cost on community size

Communities can lead to significant reduction in transaction

costs, since they pool the demand from a large number of

consumers for a large bundle of goods, and execute a sin-

gle transaction to satisfy the pooled demand. There can be

significant economies to scale, since a community operator

f(x)

1

D1/a

P S

a (1-a)/2 (1+a)/2 (1-a)/2 c p (1+a)/2

1-F(x)

Fig. 6 Approximate frequency distribution f(x), and 1-cumulative fre-quency distribution 1 − F(x) of reservation prices for communities ofsize n. The actual distribution is also shown as a dotted line. c = marginal

cost of production, p = price, P = supplier’s profit, S = community sur-plus, D = deadweight

S for c<0.33 S for c>0.33

S* S*

a*=(1-2c)/3 1 a (1-2c)/3 a*=1 a

Fig. 7 The consumer surplus asa function of market variance a.When c ≤ 0.33, market variancecan be reduced from 1 to a∗, andthe consumer surplus increased,by forming communities. Whenc > 0.33, the optimum a∗ = 1,and it is not possible to increasesurplus by reducing marketvariance, and hence nocommunities are feasible

Springer

186 Inf Syst Front (2006) 8:179–194

a* S*

1

0.16

0.33

0.11 0.03

0.33 c 0.33 c

Fig. 8 The optimum market uncertainty a∗ and the optimum consumer surplus S∗ as a function of production cost c. For c ≤ 0.33, the marketvariance can be reduced from 1 to a∗ by forming communities, and creates an increase in consumer surplus

a*=1/√n S*

1 0.14

0.48

0.11 0.04

0.2c

0 .33 1 c

Fig. 9 The optimumcommunity and its surplus as afunction of cost, obtainedthrough numerical analysis

can do the market search for all community members, col-

lect product information and analyze it for optimum match

to its members’ needs, and negotiate on behalf of its mem-

bers. We will assume that the community operator is a pure

transaction facilitator with no involvement in the physical

handling of products, and hence the consumers’ transaction

cost is inversely proportional to the size of the community, in

effect dividing the single transaction cost among the commu-

nity members, but the seller’s transaction cost is not affected

by the community. Given a transaction cost t per transac-

tion, the effective cost for the community members is t/nwhich is equal to ta2 from the previous section. We will show

that transaction costs further justify the creation of electronic

communities, and enlarge the region in which communities

are economically feasible.

Given the same distribution of reservation prices, and the

same costs as before, the supplier’s problem is to maximize

profit P, by setting the price p, given the communities char-

acterized by the parameter a:

P = (p − c)(1 − F(p + a2t))

= (p − c)((1 + a)/(2a) − (p − a2t)/a)

for (1 − a)/2 − a2t < p < (1 + a)/2 − a2t

Substituting p for p + a2t and c for c + a2t , we get exactly

the same formulation as in Section 3, and the same results.

Similarly, the community organizer’s problem is to design

communities, by maximizing the community surplus with

respect to parameter a, given the optimum price p∗ of the

supplier:

S = 1/2((1 + a)/2 − p∗ − a2t)(1 − F(p∗ + a2t))

= ((1 + a)/2 − p∗ − a2t)2/(2a)

if (1 − a)/2 − a2t < p∗ < (1 + a)/2 − a2t

Substituting p for p + a2t and c for c + a2t , and then

substituting for p∗ from above, we get the same solutions as

in Section 3:

S = (1 + a − 2c)2/(32a)

if a > (1 − 2c)/3 and a > 2c − 1

Substituting c + a2t back for c:

S = (1 + a − 2c − 2a2t)2/(32a)

if a > (1 − 2c − 2a2t)/3 and a > 2c + 2a2t − 1

Springer

Inf Syst Front (2006) 8:179–194 187

S for c<f(t) S for c>f(t)

S*

S*

(-3+√9+8t-16ct) /4t 1 a (-3+√9+8t-16ct) /4t 1 a

Fig. 10 The consumer surplusas a function of market variancea. When c ≤ f (t), the marketvariance can be reduced from 1to a∗, and the consumer surplusincreased, by formingcommunities. When c > f (t),the optimum variance a∗ = 1,and the surplus cannot beincreased by reducing marketvariance, hence no communitiesare feasible

Maximizing S with respect to a, results in three possible

extreme points to consider:

S1 = −3 + sqrt(9 + 8t − 16ctc)/(8t)

a1 = (−3 + sqrt(9 + 8t − 16ct))/(4t)

S2 = 0 a2 = (1 − sqrt(1 + 8t − 16ct))/(4t)

S3 = (1 − c − tc)2/8 a3 = 1

And resulting in:

S∗ = S1 and a∗ = a1 if S1 > S3 iff c < f (t)

S∗ = S3 and a∗ = a3 = 1 otherwise,

where

f (t) = 1

3tc((5 − 3t)t − (22/3t(9 + 4t))/(−54t2 + 40t3

+ 3√

6 t3/2√

(1 + 4t)(27 + 2t(−9 + 4t)))1/3

+ (4t2(−27 + 20t) + 6√

6t3/2

×√

(1 + 4t)(27 + 2t(−9 + 4t)))1/3)

A community is feasible whenever a∗ <1 since then a

community can be formed to reduce a from 1 to a∗, by re-

ducing the market variance of reservation price. Figure 10

shows the community surplus as a function of market vari-

ance a, and Fig. 11 shows the feasible region for communities

as a function of production cost c and the transaction cost t.Clearly, the feasible region is considerably larger than it was

in the previous section when t was assumed 0, where the feasi-

ble region there was characterized by c < 0.33. The presence

of nonzero transaction costs considerably increases the eco-

nomic viability of communities as expected, and the benefits

from transaction costs increase with increasing transaction

costs. However at very high transaction cost levels, the bene-

fits to communities drop off, since to take advantage of very

high transaction costs, very large communities are required;

but there is a cost associated with very large communities

c

0.5

0.43 f(t)

0.33

c 1 tFig. 11 The feasible region for communities as a function of productioncost c and the transaction cost t, where the feasible region is c < f (t)is equivalent to a∗ < 1

since they decrease the market price variance too much, and

benefit the suppliers too much at the expense of consumers.

There is an optimum market variance for the communities,

and reducing it beyond that optimum hurts the communities.

Figure 12 shows the optimum solution as a function of cost

c and transaction cost t.The analytical results were obtained by approximating

aggregate distributions as rectangular distributions. The ex-

act results are obtained by numerical analysis using Mathe-

matica software (Varian, 1996), and they are similar to the

analytical results, demonstrating the robustness of the re-

sults.Figure 13 summarizes the numerical results obtained

by Mathematica, by randomly drawing from a uniform dis-

tribution of reservation prices, 500 communities, for each

size n = 1−100, for each cost c = 0.01−1, and for each

transaction cost t = 0.01−1.

5. Community types and their demand functions

We have assumed a type of community that aggregates the

demand, and the resulting reservation price is the sum of all

n individual reservation prices for a bundle of n items. This

is a community where the community members subsidize

each other to create an expanded market, resulting in a mar-

Springer

188 Inf Syst Front (2006) 8:179–194

a* S*

1

0.160.15

0.33 tc=00.30 tc=0 tc=0.5

tc=0.5 0.060.110.10 0.05

0.33 0.5 c 0.33 0.5 c

Fig. 12 The optimum market variance a∗ and the optimum community surplus S∗ as a function of production cost c and transaction cost t

a*=1/√n S*

1 0.16

0.12

0.48 tc=0 tc=0.5 tc=0 0.45 tc=0.5

0.11 0.060.10 0.04

0.2 0.4 c 0.33 1 c

Fig. 13 The optimum community and its surplus as a function of cost and transaction cost, obtained through numerical analysis

n* Sm* and Ss*

0.18

6 0.16

0.15 Ss*

3 S m*

2 0.10

1 0.06

0.01 0.05 0.1 0.2 c 0.05 0.2 0.33 c

Fig. 14 The optimumcommunity size n∗ and theoptimum community surplusSm

∗ as a function of productioncost c. Also shown is theoptimum community surplus Ss

∗

for subsidizing communities forcomparison

Sm*, Ss* for t=0.5 Sm*, Ss* for t=0.1

0.15 0.16

0.12

0.10 0.1 Ss*

0.04 Sm* Ss* 0.09

0.03 0.06 Sm*

0.1 0.5 c 0.1 0.2 0.4 c

Fig. 15 The surplus forminimalist communities as afunction of cost, compared tothe surplus for subsidizingcommunities. The subsidizingcommunities always dominateexcept for very small t and verysmall c

Springer

Inf Syst Front (2006) 8:179–194 189

ket where the reservation price for the community converges

to the average reservation price of consumers in that com-

munity. Although this is a very intuitive type of community,

called subsidizing community, there are other types of com-

munities. A minimalist community is one where the com-

munity reservation price converges to the minimum reser-

vation price of the consumers in that community. This is a

type of community that refrains from purchasing, unless the

price is below the reservation price of all of its members.

They will be called “minimalist communities” because of

this restraint they adopt towards consumption, or “consensus

communities” because a purchase decision is made only if

all community members would have individually made the

same decision.

Given a uniform distribution of reservation prices, a ran-

dom community of n consumers would have exponentially

distributed reservation prices, when the community reserva-

tion price is defined as the minimum reservation price of the

consumers in that community.

1 − F(x) = (1 − x)n sinceallnconsumerswouldhaveto

fallbetweenxand1, fortheminimum

tofallabovex.f (x) = n(1 − x)n−1 isthedistributionofminimalist

communityreservationprices

bydifferentiation.

The profit P of a monopolist seller is given by

P = (p − c)(1 − F(p)) = (p − c)(1 − p)n

By differentiation we find the optimum price p∗ as

p∗ = (nc + 1)/(n + 1)

The consumer surplus S is given by

S =∫ 1

p∗(1 − F(x))dx +

∫ 1

p∗f (x)(1 − x)/2dx

=∫ 1

p∗(1 − x)ndx +

∫ 1

p∗n(1 − x)n−1(1 − x)/2dx

where the first term is the same as before, but the second

term is due to the minimalist nature of the community. When

a product is purchased by a community with reservation price

x, each member of the community derives an additional sur-

plus of (1 − x) / 2 on average, since x is the minimum reser-

vation price in the community, and the average reservation

price in the community is x + (1 − x)/2.

Integrating, and substituting for p∗:

S =(

n(1 − c)

n + 1

)n+1 1

2

Maximizing with respect to n to compute S∗, and compar-

ing to subsidizing communities, numerical solutions show

that minimalist communities dominate only for very small c,

approximately c < 0.05.

Similarly, when the zero transaction cost assumption is

relaxed, the subsidizing communities continue to dominate

the minimalist communities except for very small costs.

The surplus is computed by repeating the above analysis

for P = (p − c)(1 − F(p − t/n)) since the transaction cost

experienced by each consumer in a community of size n,

is t/n. The optimum price p∗ = (nc + 1)/(n + 1). Defining

consumer surplus as before, we get

S = 1

2

(n(1 − c − t

n

)(n + 1)

)n+1

Maximizing with respect to n to compute S∗, and compar-

ing to subsidizing communities, numerical solutions using

Mathematica software (Suttles, 1972) show that subsidizing

communities always dominate, except for very small costs

when c + t < 0.05. The simulation parameters are the same

as before.

6. Market segmentation and communitymembership

We have assumed that communities are random collections

of individuals. In fact a very powerful tool for a community

organizer is to carefully choose its members from various

market segments. Community organizers are free to separate

individuals into multiple communities, or simply reject the

individuals that do not meet their standards. We will show

that when it is possible to segment the market, and also when

it is possible to form very large communities in each segment,

a simple 3-way market segmentation can lead to significant

benefits for communities. Moreover, we will show that, when

segmentation is possible, minimalist communities dominate

subsidizing communities everywhere. Precise segmentation

is necessary to reap the maximum possible benefits, and such

precise segmentation requires detailed knowledge about con-

sumers. There are numerous technical and organizational im-

pediments to building such consumer databases containing

detailed utility function and reservation price information.

Hence, precise segmentation may not be feasible, but when

it is, it leads to large benefits for communities at the expense

of suppliers.

Springer

190 Inf Syst Front (2006) 8:179–194

Fig. 16 1−F(x) for reservation prices when the market is segmentedinto 3 groups, (0,c), (c,b), and (b,1), shown in 3 different line types. Allsegments are structured into very large communities, leading to stepfunction for the community reservation price distribution. The shadedareas are supplier’s profit and community’s surplus

6.1. Segmented subsidizing communities

Given a uniform distribution of the reservation prices, and

assuming no transaction cost, a simple 3-way segmenta-

tion of the market can lead to significant advantages for

subsidizing communities. The first segment contains all

those whose reservation prices are below the production

cost, and they are excluded from the market. The remain-

ing population is divided into two, containing the high and

low valuation consumers. The optimum cutoff point be-

tween the two segments is critical, and will be computed

in this section. The community size is also critical, but

simulations show that the optimum solution involves very

large communities. The only remaining problem is to find

the optimum cutoff point between the two segments. Fig-

ure 16 shows the 1−F(x) for reservation prices, with 3

segments.

The problem for the community organizer is to pick

the right cutoff point b, while forcing the supplier to

keep the price at p = (b + c)/2. The constrained optimiza-

tion problem involves maximizing the surplus, the blue

shaded area, while enforcing the constraint that the sup-

plier’s profit at p is greater than his profit at (b + 1)/2, the

other possible point for the supplier’s price, hence forc-

ing the supplier to the profit shown as the green shaded

area.

MAXIMIZEb

S =(

1 + b

2− b + c

2

)(1 − b) = (1 − b)(1 − c)

2

SUBJECT TO

(b + c

2− c

)(1 − c) ≥

(b + 1

2− c

)(1 − b)

producing the solution

b∗ = 3c − 1

2+

√5c2 − 10c + 5

2= 0.38c + 0.62

S∗ =(1 − c)

(3 − 3c − √

5 − 10c + 5c2)

4= 0.19(1 − c)2

A quick comparison of these results to the results of

Section 3 and Fig. 8 shows that segmentation leads to signif-

icant increase in community surplus. The increase comes at

the expense of supplier profits. Segmentation allows commu-

nities to shape the market very precisely, reduce the suppliers’

ability to set prices by limiting their options, and expropri-

ate all excess value from the market. Segmentation is clearly

very desirable for community organizers, but requires exten-

sive technical capability to collect and process information

about the reservation prices of consumers for a variety of

goods and services, and it requires market access to build

very large communities.

6.2. Segmented minimalist communities

Assuming a uniform distribution of reservation prices, and

zero transaction costs, the analysis of the previous section can

be repeated for minimalist communities. Again, simulation

models show that the optimum segmentation is a 3-way seg-

mentation. The first segment contains all consumers whose

reservation prices are below the supplier’s price, and they

are excluded from the market. The remaining consumers are

divided into high and low valuation consumers. The cutoff

point between the two groups is critical, and will be com-

puted in this section. The optimum community size is also

critical, but simulation models show that the optimum size

for all communities is the largest community possible, hence

we will assume very large communities for all segments.

Figure 18 shows 1−F(x) of the reservation prices for

segmented minimalist communities, where F(x) is the cu-

mulative distribution function.

As in the previous subsection, the problem for the com-

munity organizer is to pick the right cutoff point b, while

forcing the supplier to keep the price at p. The constrained

optimization problem involves maximizing the surplus, while

enforcing the constraint that the supplier’s profit at p is greater

than his profit at b, the other possible point for the supplier’s

price, hence forcing the supplier to the profit shown as the

green shaded area. Utilizing the results from Section 5, where

we showed that the community surplus for very large com-

munities approaches 0.18, when the community members are

drawn from a uniform distribution of reservation prices. Sim-

ilar analysis shows that for a segment b of the population, the

community surplus for very large communities approaches

Springer

Inf Syst Front (2006) 8:179–194 191

0.18 b2 in the limit.

MAXIMIZEp,b

0.18(b − p)2 +(

(1 + b)

2− p

)(1 − b)

SUBJECT TO (p − c)(1 − p) ≤ (b − c)(1 − b)

leading to

p∗ = 0.18 + 0.82c b∗ = 0.82 + 0.18c S∗ = 0.20(1 − c)2

Once again, a comparison to these results and Fig. 19 to

the results of Section 5 and Fig. 14 shows that segmentation

leads to a significant increase in community surplus. The in-

crease comes at the expense of supplier profits, as discussed

in the previous subsection. Moreover, a comparison of the

two types of communities shows that when segmentation is

done optimally, minimalist communities dominate subsidiz-

ing communities everywhere as shown in Fig. 20. This is a

S*

0.19

0.15

0.12

0.09

0.05

0.02

0.1 0.2 0.3 0.5 0.7 1 c

Fig. 17 Community surplus S as a function of production cost c, for asegmented subsidizing community

Fig. 18 1 − F(x) of the reservation prices for segmented minimalistcommunities, where F(x) is the cumulative distribution function. Thethree segments contain the consumers whose reservation prices fall in(0, p), (p, b), (b, 1), and all communities are very large, leading to acumulative distribution function approaching a step function

S*

0.20

0.16

0.13

0.10

0.05

0.02

0.1 0.2 0.3 0.5 0.7 1 c

Fig. 19 The optimum consumer surplus for segmented minimalist com-munities as a function of cost

S*

0.1 0.2 0.3 0.4 0.5 0.6 0.7

0.05

0.1

0.15

0.2

c

Fig. 20 The consumer surplus for optimum segmented and minimalistcommunities as a function of cost. The solid line shows the surplus forminimalist communities, and the dotted line for subsidizing communi-ties

complete reversal of the previous results without segmenta-

tion, where subsidizing communities dominated minimalist

communities everywhere. Segmentation is a significant tool,

since not only can it improve consumer surplus significantly,

but also it changes the structure of the optimum communities

from a subsidizing community to minimalist community.

The assumption of zero transaction cost can be relaxed,

with no changes in the conclusions. Transaction cost has no

effect on segmented communities, since the optimum com-

munities are very large, and the effective transaction cost for

very large communities approaches zero.

7. Other distributions

A uniform distribution of consumer reservation prices has

been assumed so far. Numerical analysis shows that other

distributions produce similar results, and all major conclu-

sions continue to hold. Figure 21 shows the solutions for

Normal Distribution, obtained by numerical analysis using

Springer

192 Inf Syst Front (2006) 8:179–194

S* at v0=0.1 S* at v0=0.2

0.1 0.2 0.3 0.4 0.5 0.6 0.7

0.02

0.04

0.06

0.08

0.1

0.12

0.14

c0.1 0.2 0.3 0.4 0.5 0.6 0.7

0.025

0.05

0.075

0.1

0.125

0.15

c

Fig. 21 The optimumcommunity surplus forsubsidizing communities as afunction of production cost c,transaction cost t, and initialmarket variance v0. Solid linefor t = 0, dotted line fort = 0.2, dashed line for t = 0.5,and the bottom line for nocommunity for t = 0.5

0.1 0.2 0.3 0.4 0.5 0.6 0.7

0.02

0.04

0.06

0.08

0.1

c 0.1 0.2 0.3 0.4 0.5 0.6 0.7

0.05

0.1

0.15

0.2

c

V* at v0=0.1 V* at v0=0.2Fig. 22 The optimum variancefor subsidized communities as afunction of production cost c,transaction cost t, and initialvariance v0. The solid line fort = 0, dotted line for t = 0.2,and dashed line for t = 0.5

Mathematica software (Varian, 1996). The optimum commu-

nity surplus Ss∗ the optimum market variance vs

∗ are com-

puted for subsidizing communities, as a function of produc-

tion cost c, transaction cost t, and the initial market variance

v0, in Figs. 21 and 22. Note that when the market variance is

moderate, v0 = 0.1 or 0.2, there is a great deal to be gained

from forming communities by reducing the market variance

to the optimal levels. When the market variance happens to

be smaller than the optimal to start with, there is nothing to

be gained from forming communities, since communities can

only reduce market variance. Less intuitively, when the mar-

ket variance is very large, there is also nothing to be gained

from forming communities since the surplus is already very

large, and any reduction in the market variance benefits sup-

pliers more than the community.

Similarly, for minimalist communities, the exact solutions

obtained by numerical analysis are very close to the approxi-

mate solutions obtained analytically. Figures 23 and 24 show

the solutions obtained by using Mathematica software, for

the optimum community surplus S∗ and the optimum com-

munity size n∗ as a function of cost c, transaction cost t,and market variance v0. As in the subsidizing communities,

for moderate market variance, there is a great deal that can

be gained by forming communities. However, as shown in

the analytical results, the subsidizing communities dominate

minimalist communities everywhere except for very small

costs of approximately c + t < 0.1.

The exact results obtained by numerical analysis are very

similar to approximate analytical results. As shown in the

approximate analysis, segmentation leads to significant im-

provement in the community surplus. Numerical analysis us-

ing Mathematica software shows the significance of that im-

provement in Fig. 25. As in the uniform distribution case, the

minimalist communities dominate subsidizing communities

everywhere under optimum segmentation strategy.

Simulation results show that our major conclusions remain

valid for nonsymmetric distributions such as Gamma and

Exponential, demonstrating the robustness of the results.

8. Conclusions

Electronic communities can be used to organize consumers

and to facilitate the creation of novel marketplaces. Such

commercial communities have two main effects. They in-

crease the bargaining power of consumers by allowing them

to participate in the marketplace as a single unit, and balanc-

ing the information asymmetry in favor of consumers. They

Springer

Inf Syst Front (2006) 8:179–194 193

n* at v0=0.1 n* at v0=0.2

0.1 0.2 0.3 0.4 0.5 0.6 0.7

2

3

4

5

c 0.1 0.2 0.3 0.4 0.5 0.6 0.7

1.5

2

2.5

3

3.5

4

c

Fig. 23 The optimum community size n∗ for minimalist communities as a function of cost c, transaction cost t, and the market variance v0. Thesolid line for t = 0, the dotted line for t = 0.2, and the dashed line for t = 0.5

S*

0.1 0.2 0.3 0.4 0.5 0.6 0.7

0.05

0.1

0.15

0.2

c

Fig. 24 The optimumcommunity surplus as a functionof cost c, and market variancev0. The solid line for subsidizingcommunity at v0 = 0.1, dottedline for subsidizing communityat v0 = 0.2, dash-dot line forminimalist community atv0 = 0.1, and the dashed line isfor minimalist community atv0 = 0.2. Minimalistcommunities dominatesubsidizing communitieseverywhere

also reduce transaction costs for consumers by allowing me-

diated bulk purchases by consumers. They can also benefit

suppliers by expanding markets, and reducing price variance

in the markets and allowing suppliers to better target their

pricing decisions. The optimum design of communities is

critical to derive the full economic benefits since:

1. Larger communities are not necessarily better for con-

sumers. There is an optimum size determined by a trade-

off between the bargaining power gained from size, and

the loss of focus resulting from heterogeneity in a larger

community. Larger communities can benefit consumers

by increasing their bargaining power and information, and

by reducing transaction costs, but larger communities can

also benefit suppliers since they reduce the price variance

in the market place by pooling the demand, which allows

them to better target their pricing decisions. The gains

by either party may come at the expense of the other, or

by reducing the deadweight loss and expanding markets.

Consequently, the communities may benefit either, both,

or neither of the parties, making optimum design a critical

issue. Size of the community is the most critical and the

most easily controlled design variable.

2. The structure of the community is also an important design

variable. Two general types of communities have been

identified, in terms of community behavior, i.e. how the

community demand function is derived from the individ-

ual demand functions. The two types of communities are

shown to behave differently, and the exact conditions have

been derived under which one or the other may become

the optimum. The subsidizing communities are shown

to dominate minimalist communities everywhere except

when the costs are very small, c + t < 0.05.

3. The membership of the community is also an important

design variable. A community does not have to be a ran-

dom sample of consumers from the population in general,

Springer

194 Inf Syst Front (2006) 8:179–194

S* at v0=0.1 S* at v0=0.2

0.1 0.2 0.3 0.4 0.5 0.6 0.70.7 0.1 0.2 0.3 0.4 0.5 0.6

0.025

0.05

0.075

0.1

0.125

0.15

0.025

0.05

0.075

0.1

0.125

0.15

c c

Fig. 25 The optimumcommunity surplus S∗ forminimalist communities as afunction of cost c, transactioncost t, and the market variancev0. The solid line for t = 0, thedotted line for t = 0.2, thedashed line for t = 0.5, and thebottom line for no community att = 0.5

but it can be carefully designed, and the members can be

selected from various segments of the population. Such

segmentation was shown to lead to significant improve-

ment in the economic performance of the community. A

three-way segmentation of the marketplace was shown to

be desirable, and the minimalist communities were shown

to dominate subsidizing communities everywhere under

this segmentation.

Future research is suggested into the dynamics of

community creation, maintenance and management. The

external environment of the community is also important,

especially how communities compete with each other for

members, and what market mechanisms can be designed

for consumers to enter and leave communities and shop for

communities.

References

Anand KS, Aron R. Buying on the web: A comparison of price discoverymechanisms. Management Science 2003;49(11):1546–1562.

Armstrong A, Hagel J. The real value of online communities. HarvardBusiness Review 1996;134–141.

Bakos Y, Brynjolfsson E. Bundling and competition on the internet:Aggregation strategies for information goods. Marketing Science2000.

Bakos Y, Brynjolfsson E. Bundling information goods: Pricing, profitsand efficiency. Management Science December 1999.

Balasubramanian S, Mahajan V. Economic leverage of virtual commu-nity. International Journal of Electronic Commerce Spring 2001.

Burnett G, Buerkle H. Information exchange in virtual communities:A comparative study. Journal of Computer-Mediated Communi-cation 2004;9(2).

Burns D. Family as a consumer. Psychology and Marketing 1993;10:79–80.

Hagel J, Brown JS. Out of the Box: Strategies for Achieving ProfitsToday and Growth Tomorrow Through Web Services, HBS Press,2002.

Hanson W, Martin K. Optimal bundle pricing. Management Science1990;32(2).

Ireland N, Law P. A Cournot-Nash model of consumer cooperative.Southern Economic Journal 1983;49:706–716.

Lechner U, Hummel J. Business models and system architecturesof communities. International Journal of Electronic Commerce2002;6(3):41–82.

Mougayar W. Opening Digital Markets, McGraw Hill, 1998.Orman LV. Electronic markets, Hierarchies, Hubs, and Intermediaries.

Information Systems Frontiers 2002;4(2):207–222.Porter P, Scully G. Economic efficiency of cooperatives. Journal of Law

and Economics 1987;30:489–512.Preece J. Online Communities: Designing Usability and Supporting

Sociability, Wiley, 2000.Rheingold H. The Virtual Community: Homesteading the Electronic

Frontier. Addison Wesley, Reading, MA, 1993.Sandler T, Tschirhart JT. The Economic theory of clubs: An evaluative

survey. Journal of Economic Literature 1980;18(4):1481–1521.Schmid B. Requirement for electronic market architectures. Interna-

tional Journal of Electronic Commerce 1997;7(1):3–6.Schneider GP, Perry JT. Electronic Commerce, Thomson Learning,

2001.Suttles GD. The Social Construction of Communities, University of

Chicago Press, 1972.Varian H (ed.). Computational Economics: Modeling and Analysis with

Mathematical, Springer/Verlag Publisher, 1996.Williams RL, Cothrel J. Four smart ways to run online communites.

Sloan Management Review, Summer 2000;81–91.

Levent V. Orman is a professor of Infor-mation Systems at Cornell University, Grad-uate School of Management. He received aPh.D. degree from Northwestern University.He has taught courses and written articles onelectronic commerce, database management,decision support systems, and expert sys-tems. His recent articles appeared in Journalof Information Technology and Management,Journal of MIS, Acta Informatica, and IEEE

Transactions on Knowledge and Data Engineering. He is the asso-ciate editor of the Journal of Database Management, and serves onthe editorial board of the Journal of Information Technology andManagement.

Springer