Course: Model Thinking @ Coursera
0001 Jun 1

# Why model ?

• Intelligent Citizen of the World
• Clearer Thinker
• understand and use data
• decide, strategize, design

• equilibrium
• cycle
• random
• complex

# Using and Understanding Data

• Understanding Patterns
• Predict Points
• Produce Bounds
• Retrospective analysis
• Predict Other
• Informing data collections
• Estimate hidden parameters
• Calibrate

# Using Models to Decide, Strategize, and Design

• Decision Aids
• Comparative Statics
• Counterfacts
• Identify and Rank Levers
• Experimental Design
• Institutional Design
• Helping to Choose Among Policies and Institutions

# Segregation and Peer Effects

## Sorting and Peer Effects Introduction

Approaches
Schelling's Segregation Model
Granovetter
The model consists of N individuals, each of whom has a threshold for a certain behavior. If there are enough 'extremists' in this model with very low thresholds, collective action may occur despite the presence of other individuals with very high thresholds.
Standing Ovation
Identification
Models
Equation Based Model
Agent Based Model (In agent-based modeling, we model a system that is a collection of autonomous, decision-making individuals called agents. These agents make decisions on the basis of a particular set of rules. We then look at these decisions in the aggregate to see what types of macro-level behaviors or patterns emerge.)
Individuals
Behaviors
Outcomes


## Schelling’s Segregation Model

        Threshold for Decision
Micromotives ≠ Macrobehaviour
Tipping
Exodus
Genesis


## Measuring Segregation

Index of Dissimilarity
|b/B - y/Y| / 2
dividing by 2 is a tool to change the scale of Index of Dissimilarity so that our upper limit will be 1 rather than 2.


## Peer Effects

N Persons
Tj — a Threshold for Joining


## The Standing Ovation Model

Threshold to Stand: T
Quality: Q
Signal: S = Q + E
E = Error, Diversity
Initial Rule
Stand, if S > T
Subsequent Rule
Stand, if more than X % stand
Claims
Higher Q — More People Stand
Lower T — More People Stand
Lower X — More Ovations
If Q < T, More Variation in E — More Stand


## Identification Problem

Static Data Couldn't Give a Clue, Why Segregation Appears
We Need the Dynamic Data to Know, Why Objects Are Segregated
Move = Sorting
The basic idea here is that when agents in the model - people - are choosing to surround themselves with others who are similar, we consider it sorting.
Change = Peer Effect
peer effects, in which agents are influenced by others who are around them.


Back to posts