You are on page 1of 58

Teaching Goals

Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Behavioral Finance

Lecture 5 - Herd Behavior:


Epidemics and Network-Externalities

Dr. Markus Demary

Ulm University

Summer Term 2021

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Payoff- and network-externalities

Teaching Goals: Students should learn

∙ how changes in behavior cascade through a network


∙ how cascading behavior can stop

Epidemiological model

∙ what the spread of infecticious deseases and the spread of information can have in common
∙ how the dynamics of epidemics work

Models with direct neighbour effects

∙ the network effects are of a local nature


∙ important for your decisions are the decisions of the agents you are connected with

Models with global network effects

∙ the network effects are of a global nature


∙ important for your decisions are the decisions of the majority of the agents

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Epidemics

Basic idea
∙ epidemics (pathogens, ideas, computer viruses) spread through a
population
∙ properties of the pathogen: contagiousness, length of infection period and
its severity
∙ properties of the social network (who knows whom or travel patterns)

Social network structure


∙ pathogens: network structure depends on the deseases mode of
transmission (e.g. HIV or Corona-virus)
∙ computer viruses: internet has a broader network compared to short-range
wireless communication
∙ ideas: different transmission for seminars, television, newspapers or social
networks

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Epidemics

Biological versus social contagion

∙ biological contagion: lack of decision making (if we abstract from


preventative behavior), modelled probabilistically
∙ social contagion: people are making decisions to adopt a new idea or
innovation, modelled by utility maximization

Simple model: branching process

∙ a persion carrying a new desease enters a population and transmits it to


other persons with probability p
∙ infectious person meets k people while contagious (these k people are the
first wave of the epidemic)

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Epidemics: A Simple Model

First wave
∙ k people are at risk of being infected, each with probability p

Second wave
∙ each person in the first wave meets k different people
∙ second wave: k × k = k 2 people
∙ each infected people in the first wave passes the desease to each of the k
second-wave people with probability p

Subsequent waves

∙ each person in the current wave meets k people and passes the desease
with probability p
∙ the contact network has the shape of a tree: single node at the top, every
node is connected to a set of nodes in the level below

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Epidemics: the branching process

3-1

2-1 3-2

3-3

3-4
1-0 2-2 3-5

3-6

3-7
2-3
3-8

3-9

Source: Similar figures can be found in Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 21
Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Epidemics: A Simple Model

The Basic Reproductive Number

∙ k people are at risk of being infected, each with probability p


∙ the basic reproductive number is the expected number of new cases caused
by a single individual
∙ in the branching process the basic reproductive number is R0 = p × k
∙ the outcome is determined, whether the basic reproductive number is
smaller or larger than one

Interpretation

∙ R0 < 1: each infected person produces less than one new case in
expectation =⇒ the size of the outbreak is trending downward
∙ R0 > 1: each infected person produces more than one new infection in
expectation =⇒ the size of the outbreak is trending upward
∙ even when R0 > 1, whenever p < 1 there is a chance that none of the first
infected people will succeed in infecting anyone else

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Assumption: three states

∙ Susceptible S(t) ≥ 0: before a node has caught the desease, it is


susceptible to infection from its neighbors
∙ Infectious I(t) ≥ 0: once the node has caught the desease, it has some
probability of infecting its susceptible neighbors
∙ Removed or Recovered R(t) ≥ 0: after an infectitous period the node is
removed from consideration (e.g. recovered because of immunity)
∙ Population N (t) = S(t) + I(t) + R(t) ≥ 0: can be stationary, growing or
shrinking by assumption
S(t) I(t) R(t)
∙ Normalization: N (t)
+ N (t)
+ N (t)
= 1 or often N (t) = 1

Initial conditions
∙ S(t0 ) = N (t0 ) = 1: before a patient zero appears, the whole population is
susceptible to infection
∙ I(t0 ) = R(t0 ) = 0: no infectious and recovered individuals

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Dynamics of the susceptible

dS S·I
=v·N −β − µS (1)
dt N

∙ v: birth rate of the population, new born vN are susceptible


∙ β = κ · q: infection rate, κ: contact rate and q: infection rate per contact
I
∙ λ = βN : probability of infection
∙ λS: newly infected individuals
∙ µ: death rate of the population (independent of infection)

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Dynamics of the infected

dI S·I
=β − γI − µI (2)
dt N

∙ β = κ · q: infection rate, κ: contact rate and q: infection rate per contact


I
∙ λ = βN : probability of infection
∙ λS: newly infected individuals
∙ γ: recovery rate, γI: individuals who have recovered, who are no more
susceptible to infection
∙ µI: infected individuals who have died

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Dynamics of the recovered

dR
= γI − µR (3)
dt

∙ γI: recovered individuals


∙ µR: recovered individuals who have died

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Assumption: stationary population: µ = v = 0 and N = 1

dS
= −βS · I (4)
dt
dI
= βS · I − γI (5)
dt
dR
= γI (6)
dt

The epidemic starts

dI
 
= βS · I − γI = I βS − γ (7)
dt

∙ I(t) > 0
∙ βS > γ or R0 · S > 1 with R0 = β/γ as the basic reproductive number

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Assumption: growing population: µ > 0 and v > 0 and N = 1

dS
= v − βS · I − µS (8)
dt
dI
= βS · I − γI − µI (9)
dt
dR
= γI − µR (10)
dt

The epidemic starts

dI
 
= βS · I − γI = I βS − γ − µ (11)
dt

∙ I(t) > 0
∙ βS > γ + µ or R0 · S > 1 with R0 = β/(γ + µ) as the basic reproductive
number
Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model


Equilibrium points

Equilibrium condition:

dI S·I
=β − γI = 0 (12)
dt N

Desease-free equilibrium
I ∗ = 0 leads to
∗ ∗ ∗ ∗
E1 = (S1 , I1 , R1 ) = (1, 0, 0) (13)

Endemic equilibrium
βS/N = γ, which means S ∗ = N/R0 which leads to
∗ ∗ ∗ ∗
E2 = (S2 , I2 , R2 ) = (1/R0 , µ/(R0 − 1)β, 1 − 1/R0 − µ(R0 − 1)/β) (14)

∙ case R0 = 1 leads to E2∗ = (1, 0, 0) = E1∗


∙ case R0 < 1 each infected individual produces on average less than one infected individual,
which leads to a convergence to E1∗ = (1, 0, 0)
∙ case R0 > 1 the infection is able to invade the susceptible population and the infections
becomes endemic

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model


Examples for the endemic equilibrium

R0 = 8.3

∙ 2 percent of the population die naturally: µ = 0.02


∙ 10 contacts and 10 percent chance of infection leads to a mean
transmission rate of β = 10 · 0.1 = 1
∙ 10 percent recover per unit time γ = 0.1
1
∙ infectious period γ
= 10 days
β 1
∙ this leads to a basic reproductive number of R0 = γ+µ
= 0.1+0.02
= 8.3
∙ Infections stabilize at S ∗ = 1/R0 = 1/8, i.e. one eight of the population
are still suceptible

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model


Examples for the endemic equilibrium

R0 = 4.8

∙ 2 percent of the population die naturally: µ = 0.02


∙ 50 contacts and 0.5 percent chance of infection leads to a mean
transmission rate of β = 100 · 0.005 = 2.5
∙ 50 percent recover per unit time γ = 0.5
1
∙ infectious period γ
= 5 days
β 2.5
∙ this leads to a basic reproductive number of R0 = γ+µ
= 0.5+0.02
= 4.8
∙ Infections stabilize at S ∗ = 1/R0 = 1/5, i.e. 20 percent of the population
are still suceptible

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Application: influenza in a British boarding school

∙ estimated active infectious period: 1/γ = 2.2 days


∙ mean transmission rate: β = 1.66 per day
∙ basic reproductive number: R0 = 3.652
∙ estimated curve of the infected reaches peak of 5 weeks

Application: online social networks

∙ Cannarella and Sprechler: information diffusion on Twitter


∙ Wang and Wang: rumor spreading on social networks

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Application: viral marketing

∙ encourages people on social networks to share product information with


their friends
∙ more impact because information was recommended by friends and peer
networks
∙ spread of information is comparable to epidemic

Application: audience applause

∙ Crowd psychology: individuals change their behavior in response to others


∙ Mann et al.: individuals’ probability of starting clapping increased in
proportion to the number of other audience members already affected by
social contagion

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Application: diffusion of ideas

∙ the diffusion of ideas holds many similarities to the spread of infections


∙ Bettencourt et al.: spreading of Feynaman diagrams in theoretic physics
communities
∙ Funk: diffusion of public campains

Application: spread of computer viruses

∙ spread of computer viruses through the interconnected computers is similar


to the transmission of an infectious desease

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The SIR Epidemic Model

Application: rational expectations epidemiological models

∙ connection between epidemiological models and rational decision making


∙ fighting pandemics is not only in the hands of policymakers but also in the
hands of rational decision makers who change their behavior in response to
a pandemic
∙ there are behavioral implications of public health policies

Application: financial network contagion

∙ macroprudential risks: bank-specific risks contribute to financial stability


∙ losses can spread through the interbank market, e.g. after the failure of
Lehman Brothers
∙ losses in mortgage-backed securities were transmitted to banks which hold
similar assets after the burst of the US housing bubble

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Payoff- and network-externalities

Basic idea
∙ an individual’s choices depend on what other people do
∙ new ideas and innovations are adopted by a population

Economic structure
∙ interactions at a local level: social network structure is important
∙ individuals often do not care as much about the full population’s decisions
as about the decisions made by friends and collegues
∙ in a work setting: choice of technology such that it is compatible with
collegues’ technology

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Assumptions

Decisions
∙ individuals make decisions based on the choices of their neighbors
∙ the benefits of adipting the new behavior increases as more and more of
the neighbors adopt it

Coordination game

∙ each network node has a choice between two alternatives: A or B


∙ If nodes v amd w are linked, then there is an incentive to have their
behaviors match

Payoffs

∙ if v and w both adopt A: they each get a payoff of a > 0


∙ if v and w both adopt B: they each get a payoff of b > 0
∙ if they adopt oppositve behaviors, they each get a payoff of 0

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Assumptions

Multiple games

∙ individuals are playing this game with all of their neighbors


∙ v’s choice of strategy will be based on the choices made by all of its
neighbors, taken together

Basic question

∙ initially, v and all neightbors are using B


∙ suppose that some of the neighbors adopt A
∙ what should v do in order to maximize its payoff?

Payoff depends on

∙ the relative number of neighbors using A or B


∙ the relation between the payoff values a and b

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Decision rule

Decision rule
∙ Suppose that a p fraction of v’s neightbors are using A
∙ and a 1 − p fraction of v’s neighbors are using B
∙ if v has d neighbors: pd adopt A and (1 − p)d adopt B

Payoffs

∙ Payoff of choosing A: pda


∙ Payoff of choosing B: (1 − p)db

A is better than B if

pda ≥ (1 − p)db (15)


b
p ≥ (16)
a+b

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Decision rule

Interpretation

∙ if a fraction of at least q = b/(a + b) of your neighbors use A, then you


should, too
∙ when q is small: A is the more enticing behavior, and it only takes a small
fraction of your neighbors engaging in A for you to do so as well
∙ if q is large, then B is the attractive behavior, and you need a lot of your
friends to engage in A before you switch to A

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Decision rule

Example

∙ Payoff of using A: a = 3 and payoff of using B: b = 2


∙ 3 neigbours are using A and 4 neigbours are using B

Payoffs
3
∙ Payoff of choosing A: pda = 7
·7·3=9
4
∙ Payoff of choosing B: (1 − p)db = 7
·7·2=8

A is better than B because

pda ≥ (1 − p)db (17)


b
p ≥ (18)
a+b
3 2 2
p= ≥ = (19)
7 3+2 5
Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Cascading behavior

Multiple equilibria

∙ One equilibrium, in which everybody uses A


∙ One equilibrium, in which everybody uses B
∙ How easy is it, to tip the network from one equilibrium to the other?
∙ What are other intermediate equilibria, in which one group uses A and
another group uses B?

Assumption

∙ Initial behavior: everybody is using B


∙ Then: a small set of inital adopters / fashion leaders decide to use A
∙ Then: all other agents re-evaluate their payoffs

Questions

∙ Do all the other agents choose to follow the fashion leaders?


∙ What could cause the spread of A to stop?
Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Initial situation: all agents are using option B

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Agents v and w are early adopters of option A

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Agents r and t switch to option A

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Agents s and u switch to option A

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Initial situation: all agents are using option B

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Agents v and w are early adopters of option A

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Agents r and t switch to option A

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Agents s and u do not switch to option A

r s

v w

t u

Source: Easley/Kleinberg (2010), Networks, Crowds and Markets, Chapter 19


Dr. Markus Demary Behavioral Finance
Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The economy without network effects

Assumptions

∙ market with huge number of potential purchasers


∙ each is small enough relative to the entire market, total mass is equal to 1
∙ individuals are arranged in [0, 1], ordered in decreasing reservation prices
∙ r(x) is the reservation price of consumer x, a strictly decreasing function

Market equilibrium

∙ market price p
∙ consider r(x) = p: all consumers between 0 and x buy the product, all
others don’t
∙ equilibrium quantity: r(x∗ ) = p∗
∙ result is based on the absence of network effects

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The economy with network effects

Assumptions

∙ potential purchasers take their reservation prices as well as the total


number of users into account
∙ a z fraction of the population is using the good
∙ reservation price of consumer x: r(x)f (z), where f (z) measures the
benefit from having others also using the good
∙ f (z) is increasing in z

Decision rules
∙ each consumer has to predict z in order to evaluate whether to purchase
∙ at market price p∗ : consumer x expects a z fraction to use the good
∙ x wants to purchase provided that r(x)f (z) ≥ p∗

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The economy with network effects

Self-fulfilling expectation

∙ we assume that consumers predictions are perfect


∙ then all form a shared expectation of z
∙ if everyone expects that a z fraction of the population will purchase the
product, then this expectation is in turn fulfilled by people’s behavior

Market equilibrium

∙ if z = 0 then reservation price = r(x)f (0) = 0, which is below p∗


∙ no one wants to purchase and the shared expecation z = 0 is fulfilled
∙ 0 < z < 1: if a z fraction of the population purchases the good, to which
set of individuals does this correspond to?
∙ if x0 purchases the good, x < x0 will also purchase the good
∙ therefore 0 < x < z will purchase the good

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The economy with network effects

Market equilibrium

∙ lowest reservation price has consumer z, which is r(z)f (z)


∙ for these consumers to purchase the good, we must have p∗ = r(z)f (z)
∙ if the price p∗ > 0 with the quantity 0 < z < 1 form a self-fulfilling
expectations equilibrium, then p∗ = r(z)f (z)

Example

∙ r(x) = 1 − x, decreasing in x
∙ f (z) = z, increasing in z
∙ r(z)f (z) = z(1 − z), parabolic shape

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

The economy with network effects

Market equilibrium for p∗ > 1/4

∙ if p∗ > 1/4 then there is no solution to p∗ = r(z)f (z) = z(1 − z)


∙ since the right-hand side has a maximum value of 1/4 at z = 1/2
∙ the only equilibrium is when z = 0

Market equilibrium for 0 < p∗ < 1/4

∙ there are two solutions z1 and z2 to p∗ = z(1 − z) where the horizontal


line y = p∗ slices through z(1 − z)
∙ if people believe a z fraction to buy the good, a z fraction between 0 and
z will buy it

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

1 Teaching Goals

1 Epidemiological Models
The Basic Idea
The SIR Epidemic Model
Applications of the SIR Model

1 Network-Models with Direct Neighbor Effects


Assumptions
Decision rule
Cascading behavior
Stopping cascades

1 Network-Models with Global Network Effects


The economy without network effects
The economy with network effects
Stability, instability and tipping points

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Stability, instability and tipping points

What happens at other values of z

∙ if z is between 0 and z1 : there is downward pressure on consumption,


since r(z)f (z) < p∗
∙ if z is between z1 and z2 : there is upward pressure on consumption, since
r(z)f (z) > p∗
∙ if z is larger than z2 : there is downward pressure on consumption, since
r(z)f (z) < p∗

Dr. Markus Demary Behavioral Finance


Teaching Goals
Epidemiological Models
Network-Models with Direct Neighbor Effects
Network-Models with Global Network Effects

Stability, instability and tipping points

Market equilibria

∙ z = 0 and z = z2 are stable equilibria, demand drives z to either 0 or z2


∙ z = z1 is an unstable equilibrium, demand drives z away from z1
∙ z1 is a critical point / tipping point in the success of the good: if the
producer can get the consumers’ expectations for the number of
purchasers above z1 , they can use the upward pressure of demand to get
their market share to the stable equilibrium z2
∙ if consumers’ expectations are below z1 , then the downward pressure will
tend to drive the market share to 0

Dr. Markus Demary Behavioral Finance

You might also like