PRINCETON UNIVERSITY

ECON 326 – Economics of the Internet

Fall 2020

Lectures: Tuesdays and Thursdays 3-4:20

Instructor: Swati Bhatt

[email protected] Preceptors: Mateus Dias & Jiwon Choi
JRRB 202B
Office Hours: Wed 3-5pm, Thu 12-2pm

Course Requirements:
The class is organized in a synchronous format. We will start with a 20 minute lecture
followed by a 15 minute discussion in assigned breakout rooms, and then an open
discussion for 30-40 minutes.

REVISED VERSION for online classes: I will be discussing this format in detail during the
first lecture.

TEAMS: Students will be randomly assigned to teams with 6 students per team. For example, if
there are 48 students enrolled in class, we will have 8 groups. The composition of teams will remain
constant over the entire term such that there will be a winning team with maximum points.

•   Each student’s problem set grade will be the team’s collectively submitted problem set grade.
This facilitates and encourages group work.

•   Each student’s participation grade will be the team grade. Teams points will be the sum of
all individual participation points.

Problem Sets: Six problem sets, (5 points each for a total of 30 points) submitted online. Due on
assigned Thursdays. The problems will be from the text as well an economic analysis of current
events. They will be graded on a check/check plus/check minus basis.

Class Participation: 30 points. Genuine inquiry is a process of collective discovery. Hence,

participation is an important element of the course.

We will evaluate individual participation on a scale from 1-5, along the following criteria:
If comments
•   are relevant to the class discussion or are linked to comments of others (3 points).
•   show evidence of having examined assigned reading, especially cases (4 points);
 •   move the discussion forward with new ideas and issues (5 points). Your contribution
will be enhanced by your personal experience. Note that meaningful analysis involves
embracing ambiguity with boldness.
•   0 otherwise or due to absence without prior approval

Each student will get 3 unannounced calls to participate in distinct classes. At the start of
each class, I will list names of 6 students who can accumulate participation points on this
date.

For example, class discussion runs for 40 minutes and there are 22 classes (skip the first week). Suppose there are n
= 50 students and each student comments for t = 7 minutes. Then a maximum of y = 40/7 = 5.7 or 6 students
will be called on any one date. Since we have 22*6 = 132 slots over the entire term each student will be called x =
132/n = 2.64 or roughly 3 times.

You are advised to collect your thoughts so that comments are coherent and not rambling.
If you leave class early or are absent without prior approval, your participation grade will be zero, if
your name is listed.

Case Discussion: We will be discussing cases on assigned Thursdays. I will post questions pertaining to
the case on the Sunday of that week. Feel free to bring up cases even in classes where the case is not
assigned.

Final paper: 40 points. This 20-page paper will address an important question pertaining to digitization
and the economy. I encourage you to make an appointment with me to discuss your proposed topic.

Course Description:
Using microeconomic theory, graph theory, game theory, case studies and short video clips, we will
examine the impact of digital technology on the design of markets and their economic outcomes,
both locally and globally. Significantly, this technology has connected or linked markets, so that
information is instantly, continuously and ubiquitously available to all participants.

Three major drivers have unleashed three distinct trends. Drivers: mobile connectivity; the collection
of voluminous data; and time as the new transaction unit. Trends: organizational restructuring
leading to granularity and disintermediation, powerful network effects creating organizational
behemoths and a recalibration of competition as goods and services are redefined by functionality
and evolving preferences. Human interaction generates an intricate web of connections so the
economics of networked markets will also be the study of institutions embodying human
engagements rather than mere abstractions.

We will characterize the network economy in terms of five themes: principal-agent relationships;
game theory and strategic thinking; imperfect information; network effects & externalities; copying
& behavioral economics. Two major economic problems need to be addressed, in the context of
the digital economy - massive inequality and unsustainable consumption patterns. Hence, we
will explore questions about broadband access, property rights over data, intellectual property,
privacy and innovation.

We use the term “Internet” to refer to the entire system of mobile, digital, two-way, communication
technology. The smartphone is no longer merely a communication device, but a computer with all its
technological capabilities.

Readings:
•   David Easley and Jon Kleinberg. 2010. Networks, Crowds and Markets, Cambridge University
Press.
•   Bhatt, Swati. 2017. How Digital Communication Technology Shapes Markets: Redefining Competition,
Building Cooperation. Palgrave-Macmillan, Springer Publishing AG: London.
•   Bhatt, Swati. 2019. The Attention Deficit: The Unintended Consequences of Digital Connectivity.
Palgrave-Macmillan, Springer Publishing AG:
London. https://link.springer.com/book/10.1007%2F978-3-030-21848-5
•   Additional readings on the syllabus are uploaded to CANVAS.

Corequisites: ECO 300 or 310 (can be taken concurrently).

Important Guidelines
1.   You may prepare for the problem sets in groups but you are required to submit independent
answers.
2.   The text by Easley and Kleinberg is simple and does not require calculus. However, logic is
the primary tool, so spend some time thinking through some of the “proofs”.
3.   Problem Sets are due at the beginning of class on assigned Thursdays. Late Problem Set
policy: -1 point for late submission on Friday; -2 points for submission on Saturday; 0 points
thereafter.
4.   We will cover between 12 case studies in class. I will set up a class account at Harvard
Business Publishing. The cases cost approximately $4.00 each to download.
5.   Class participation will be judged on the following criteria (more details will be given in the
first class):
a.   Do your comments build on others’ ideas?
b.   Do your comments provoke discussion for the rest of the class?
c.   An excessively lengthy discussion, suggesting little attention to others’ comments,
will be judged negatively.
6.   The final paper will be judged on the following criteria:
 a.   Does it apply economic analysis to an issue or concern in markets impacted by digital
communication technology?
b.   Is the problem well defined? Is the paper clearly and logically written?
c.   Is there an innovative approach to thinking about the problem?
d.   Finally, is any attempt made toward empirical analysis in the form of a testable
proposition? You don’t need to actually run regressions, but simply propose one
along with data sources.

Course Schedule:
We will examine topics under five themes: imperfect information; game theory; principal-agent relationships; network
effects & externalities and behavioral economics.

IMPERFECT INFORMATION

Week 1:
Model 1: Paradox of connectivity: Two-person network connection, communication and behavior

Theme: Paradox of connectivity. With mobile connectivity, streaming technology and

communication via images, we are able to decouple time and content leading to
asynchronous and solo consumption of content. Digital technology has created more
connections between economic agents. Has the COVID crisis driven us even more into solo
consumption?

Reading:
•   Easley and Kleinberg, (henceforth EK), chap 1, 2 (pp 1-39); 13 (333-348)
•   Bhatt 2017, chap 1, 2, 3
•   Zweig, David, 2018. Escaping Twitter’s Self-Consciousness Machine (Demetrication).
The New Yorker, February 27, 2018. https://www.newyorker.com/tech/annals-of-
technology/escaping-twitters-self-consciousness-machine
•   Newport, Cal, 2019. Was Email a Mistake? The New Yorker, August 6, 2019.
https://www.newyorker.com/tech/annals-of-technology/was-e-mail-a-mistake

Week 2:
Model 2: Many person networks: Clustering coefficient, neighborhood overlap; Closure and affiliation networks;
Schelling Model
Model 3: Paradox of sharing: connectedness and privacy

Theme: How can we use the tools of graph theory to describe the network economy? What is
a connected graph? What is a strongly connected directed graph? What is a complete graph?
We will study closure, structural holes, homophily, bipartite graphs and distinguish between
the local closure model and global network effects model.
 Theme: Is personal information and personal space considered private property? Can we
define privacy as degree of control over personal property? Consider the privacy issues
raised by The National Security Agency’s collection of metadata in the United States,
allowing the government to follow your life. Maps can easily be drawn, tracing
communication and movement. Is this justified in the interest of national security?

Reading:
•   EK chap 3 (pp 43-62); 4; 21 (567-584)
•   Bhatt 2017, chap 7
•   Bhatt 2019, chap 1, 2, 3
•   George A. Akerlof, Rachel E. Kranton, “Economics and Identity”, The Quarterly Journal of
Economics, Volume 115, Issue 3, August 2000, Pages 715–
753, https://doi.org/10.1162/003355300554881
•   What do we mean by privacy (uploaded to CANVAS)
•   The Law and Privacy (CANVAS)
•   Wearable Devices and Privacy (CANVAS)
•   Data Privacy in a Pandemic - the old view (CANVAS)

GAME THEORY

Week 3:
Model 4: Game Theory & Auctions

Theme: Connectedness leads to interdependence, where the actions taken by each person
affects both their own outcome and other individuals’ outcome. Awareness of this
interdependence incentivizes people to act strategically. Game theory provides the tools for
modeling strategic interactions.

Reading:
•   EK 6 (pp 139-179); 8 (pp207-211); 9 (pp225-234)
•   Herskovic, Bernard, and João Ramos. July 2020. "Acquiring Information through
Peers." American Economic Review, 110 (7): 2128-52.
https://pubs.aeaweb.org/doi/pdfplus/10.1257/aer.20181798
•   Hart, Kim, 2019. The age of winner-take-all cities: How the U.S. $19.5t GDP is distributed
https://www.axios.com/era-of-winner-take-all-cities-16495b38-3df4-45fe-825e-
5913482a0250.html

Week 4 :
Model 5: Stability, Bargaining and Power in Network - Structural Balance (Balance Theorem) and Balanced (Nash
Bargaining) Outcome; labor markets

Theme: Interactions between economic agents can result in mutual gains as well as conflicts
over how these gains are distributed. How do we model power, profits and bargaining in
networks? We examine how position in a network is associated with power and how
partnerships, mergers and acquisitions change network position and power. Bargaining
 incorporates social norms and preferences that include self-interest, altruism, reciprocity and
inequality aversion.

Reading:
•   EK chap 5 (pp 107-118); 12 (pp301-320); 20 (pp537-554)
•   Bhatt 2017, chap 4
•   Greif, Avner and Guido Tabellini, 2017. “The Clan and the Corporation: Sustaining
Cooperation in China and Europe”. Journal of Comparative Economics, Vol 45, Issue 1
https://doi.org/10.1016/j.jce.2016.12.003

PRINCIPAL AGENT

Weeks 5 and 6:
Model 6: Strategy in Networks: Matching Markets & Matching Markets with Traders

Theme: Situations in which one node relies upon other nodes for effective functionality but
cannot observe their actions are defined as principal-agent problems. The appropriate design of
incentives could align the interests of all parties by creating shared responsibility and shared
risks. Platforms, where multiple parties interact in the provision of multiple goods and
services, could be either a principal or an agent in the transaction. How can network
structure be strategically designed to benefit particular nodes? Access across buyers and
sellers can be critical for the distribution of goods and services. How do we design markets
when prices cannot be used or when buyer valuation is private information and hence
unknown to other nodes?

Reading:
•   EK 10 (pp249-262); 11
•   Milgrom, Paul and John Roberts, 1988. “An Economic Approach to Influence Activities
in Organizations”, American Journal of Sociology, Vol 94.
https://web.stanford.edu/~milgrom/publishedarticles/An%20Economic%20Approach
.pdf
•   Roth, Alvin, July 2018. “Marketplaces, Markets, and Market Design”. American Economic
Review, 108(7). https://pubs.aeaweb.org/doi/pdfplus/10.1257/aer.108.7.1609

NETWORK EFFECTS & EXTERNALITIES

Weeks 7 and 8:
Model 7: Paradox of Privacy: Dynamic many-person networks: Cascades and Bayes Rule; Network effects and
externalities; Power laws

Theme: Even with perfect information, demand side externalities arise due to strategic
complementarities arising within a market and in complementarities between adjacent
markets. There are also supply side externalities due to complementarities and increasing
returns to scale in production and distribution of certain goods and services, especially
knowledge.
 What about innovation? Is it about adding value to old problems in new ways? Is it about
creating something new (that is invention)? Is it about doing the same thing better (that is
improvement)? Does the current intellectual property law (governing copyright, patents and
trade secrets) advance or retard innovation?

Reading:
•   EK 16 (pp 425-444); 17 (449-469); 18 (pp479-489)
•   Bhatt 2017, chapters 5, 6
•   Rysman, Marc, Summer 2009. “The Economics of Two-Sided Markets”, Journal of
Economic Perspectives, 23(3). https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.23.3.125
•   The 10 Ways Tik Tok Will Change Social Product Design
•   The Key to Selling Music in the Streaming Age
•   The Original Renegade – Credit and Copyright in the Streaming Age

BEHAVIORAL ECONOMICS

Week 9:
Model 8: Copying and Behavioral Economics; Advertising

Theme: Connectivity generates familiarity and copying, or strategic complementarities in

actions [FOMO (- fear of missing out]. The advertising based business model fuels copying
by supporting content provision when firms pay for the advertisement on the platforms. In
this model, social media platforms perform a “money laundering service” by selling
individual content in exchange for a bundle of time and data, which is then unpacked and
the data sold to advertisers for real dollars.

Reading:
•   EK 15 (pp385-409)
•   Bhatt 2019, chapters 4, 5
•   Akerlof, George and Rachel Kranton, 2005. “Identity and the Economics of Organizations”,
Journal of Economic Perspectives, Vol 19 (1)
https://pubs.aeaweb.org/doi/pdfplus/10.1257/0895330053147930

Week 10:
Model 8, cont: Collective action and diffusion in networks

Theme: Commons refers to a resource shared by a group of individuals. Self-governing such

resource allocation requires collective action, or voluntary action by members of this group,
together with appropriate information and institutional structures. We are in a moment of
increasing interdependence because of our connectivity. Competition in terms of a zero-sum
game may not be an option. Forming coalitions on a cooperative basis may the individually
rational way. Individuals may be influenced by the fraction of their neighbors adopting a
product or behavior rather than the fraction of the entire network adopting a product. New
payment technologies and political uprisings spread through such diffusion processes. What
about collective action?
 Reading:
•   EK chapter 19 (pp 497-517)
•   Bhatt 2019, Chapter 6, 7
•   Waze Connected City Data Now Hosted on Google Cloud
•   A Splinternet Prevention Plan

Week 11:
Model 9: Consumer Sovereignty and Artificial Intelligence: Hub-Authority, Reviews, Ratings and Recommendations

Theme: Using affiliation networks to study recommendation algorithms we examine whether

user-reviews create a winner-take-all model for digital content? Do recommendations
provide perfect filtering so that echo chambers of content are created?

Reading:
•   EK chap 14 (pp351-368)
•   Bhatt 2019, chapter 8, 9
•   Internet Shutdowns Become a Favorite Tool of Governments
•   Why People Are Sharing Their Family Secrets in Public

Week 12:
Model 10: Free Will and Attention Deficit: Cognitive Apathy and Diminished Risk-Taking

Theme: Free will is agency in the initiation and management of decisions to achieve a desired
outcome. How is it that mental effort can shrink under the weight of devices and
algorithms? The tsunami of digital content creates an attention deficit. The need to filter this
content generates further demands on cognitive bandwidth. Being focused on this scarcity,
we have less mental resources to give to other issues, such as risk evaluation and
entrepreneurship. Do recommendation algorithms further remove agency from individual
decision making?

Reading:
•   Bhatt 2017, chapter 8, 9
•   After the Pandemic - Entrepreneurship & Risk Taking
•   Business as Usual
•   After the Shutdown – Telemedicine
 Problem Sets and Cases - accessed from Harvard Business School
website
https://hbsp.harvard.edu/import/741831

Models 1 & 2: Connectivity, Clustering Coefficient and Neighborhood Overlap

Cases 1 & 2: 9/3
•   Optional Cases: The Coronavirus Pandemic and the Global Economy (Overview) and COVID-19: The
Global Shutdown (More details); Keeping it Clean During COVID-19: The Sanitizer Shortage (How to
address supply chain breakdowns? When supply and demand don’t match, should prices be raised, sales of
limited items be banned, rationing be instituted?)

•   ByteDance Beyond China: Leveraging AI from Toutiao to Musical.ly and TikTok

•   Tesla, Inc (Is Tesla a car company or a technology company?

Problem Set 1: 9/10

`````````````````````````
Models 3 & 4: Closure and Cascades
Cases 3 & 4: 9/17
•   Majid Al Futtaim in 2019: Driving Digital Transformation in a Physical World (End of retail?)
•   Paris Saint-Germain: Building One of the World’s Top Sports Brands (Sports without fans?)

Problem Set 2: 9/24

```````````````````````````
Model 5: Network Effects
Cases 5 & 6: 10/1
•   The Walt Disney (Disruption of media and entertainment by streaming?)
•   Walmart Flipkart: A Deal Worth its Price? (M&A and valuing a tech startup)

Problem Set 3: 10/8

``````````````````````````````
Model 6: Structural Balance, Stability and Power in Networks
Case 7 & 8: 10/15
•   Fake News at Der Spiegel (A) and (B)
•   Forgery scandal at Der Spiegel at end of 2018. Was this systemic or the result of one bad apple?

Problem Set 4: 10/22

``````````````````````````````
Model 7 & 8: Matching Markets
Cases 9 & 10: 10/29
•   Tyson Foods and Alternative Proteins: Where to Invest for Sustainable Growth
•   Breakfast at the Paramount: Dine-in or Take out

Problem Set 5: 11/5

``````````````````````````````
Model 9: Hub Authority Model and AI (Recommendation Algorithms)
Case 11 & 12: 11/12
•   A Quest for Achieving Financial Inclusion: M-Pesa Versus UPI
•   Ant Financial: The Road to Financial Inclusion in China through QR Codes and Technology as a Service

Problem Set 6: 11/19

````````````````````````````````
Model 10: Attention Deficit

Some definitions
An Algorithm is a sequence of precise rules that executes some function and results in a succinct
answer. Artificial intelligence consists of a set of algorithms that find patterns in a vast, labelled
data-set such that “information flows backwards as well as forwards, so that by iterating its processing, the network
improves its recognition at each layer”, and therefore provides a range of answers. Kernighan, p217.

Bandwidth is a measurement of bit rate available or data communication resources available along
particular pathways, expressed in bits per second (bps) or multiples of it (Kbps, Mbps, Gbps, etc).
Expressed in terms of waves or frequency along the electromagnetic spectrum, we have frequencies
from 0 to 300 gigahertz. For example, home wireless systems provide up to 600 Mbps, 4G cell
phones can provide 1Gbps while fiber optics can provide 100 Gbps or more.

Broadband refers to the wide bandwidth characteristics of a transmission pathway and its ability to
transport multiple signals and traffic types simultaneously. Broadband is a communication
bandwidth of at least 256Kbps. The wider or broader the bandwidth, the larger the communication
capacity.

Digital communication is the physical transfer of data (a digital bit stream or a sequence of 0s and
1s) over a point-to-point pathway. Examples of such pathways are copper wire, coaxial cable, DSL
and optical fiber. The data are represented as electromagnetic signals and can be transmitted via a
sequence of electrical or light pulses (baseband transmission) or a set of continuously varying waves
(passband transmission), via a modulated sine wave signal – used in wireless communication.

Technically, the Internet is a system of computer networks coordinated by standards and protocols
that specify how the various elements in the system communicate with each other. We will use the
term Internet more broadly to refer to the four components of digital technology: universal digital
representation of information (bit streams of 0 or 1); universal digital processors (computers);
universal digital networks and massive amounts of digital data.

Brian Kernighan, “Understanding the Digital World”, Princeton University Press, 2017 is a useful
reference for understanding the basics of digital communication technology.