2004 Graduate Workshop in Computational Economics
Each student began a research project during the two-week
workshop. Below are brief descriptions of these various projects.
These projects will form the basis for dissertation chapters and/or
Simon Angus, Economics, U. of New South Wales (firstname.lastname@example.org).
Simon is modeling the emergence of networks in worlds composed of strategic
agents. Each agent in the model evolves a strategy, captured by a finite automata,
that not only dictates how the agent plays the game (here a repeated Prisoner's
Dilemma), but also whether to strengthen or weaken the tie to the particular
opponent. His initial analysis indicates that the system leads to the emergence
of interesting network dynamics as the agents learn to reciprocate ties with
other cooperative agents.
Jeremy Dalletezze, Economics, Brandeis (email@example.com).
Jeremy considers a model of competition and alliance-formation
dynamics in innovation-rich industries.
Firms must strategize about partnerships and allocate resources to develop core
expertise via research and development.
More successful firms continue in the industry and
can appropriate parts of innovations from other firms.
He calibrates this agent-based model to econometric estimates
of market share, research and development
expenditures, patents, and observed alliances, and finds that his
model is able to capture some of the key patterns in the data.
Matt Grossmann, Political Science, Berkeley (firstname.lastname@example.org).
Matt is analyzing the mobilization of social groups, with a focus on
how the individual-level attributes of social groups
aggregate into broader mobilization patterns.
In the model, innate agent characteristics and social networks
initiate a desire to mobilize that feeds into various dynamics
relating to the overall level of interest group activity and the amount of
policy-maker attentiveness. He uses survey data on five American ethnic
groups to calibrate the model and finds important differences in
across the various groups.
Kristen Hassmiller, Public Health, Michigan (email@example.com).
Kristen has implemented both differential-equations-based and agent-based
approaches to modeling the spread of tuberculosis. She then "docks" the
two approaches in an attempt to identify key similarities and
differences. She finds that modeling choices concerning granularity
and contact networks lead to
important differences in predictions and policy prescriptions.
Robert Letzler, Policy, Berkeley (firstname.lastname@example.org).
Rob is looking at heterogeneous preferences and learning
in a public goods game. He models agents with preferences
for both altruism and spite, and then explores the behavioral
dynamics implied by the collection of preferences under various
public goods institutions. He then introduces learning by allowing
agents to incorporate predictions from past trends into their choice
calculations, and finds that the system can be very sensitive to
Dan Li, Finance, Carnegie Mellon University (email@example.com).
Dan is exploring cooperation in problem solving, in particular,
market behavior. The work is based on some ideas from machine learning
the efficacy of discrimination and diversity in problem solving.
Agents attempt to make predictions in a market by using algorithms chosen
from a pool of candidate learning rules. She finds that the ability
of the system to effectively aggregate the predictions is closely tied
to the underlying market mechanism and problem difficulty.
Rodolfo Sousa, Policy, Manchester Metropolitan U. (firstname.lastname@example.org).
Rodolfo's work focuses on the the political economy of redistribution.
In the model agents of various incomes vote, using majority rule,
on redistribution policies proposed by adaptive parties.
He finds that the basic system results in an alternation of tax rates between
the parties as they exchange the incumbent position, resulting in
a greater than optimal tax rate. He also considers the linkages between
the model and the median voter theorem, as well as the impact of
simple parameter changes on the underlying dynamics.
Horacio Trujillo, Policy, RAND (email@example.com).
Horacio employs an agent-based model to analyze gentrification dynamics
and segregation. In the work, agents attempt to acquire the best locations
on a landscape.
An agent's preference for a particular location is tied to the location's inherent
quality, rent, and various externalities induced by the locations and types of
neighboring agents. He finds that the patterns of gentrification are tightly
coupled to the externalities induced by other agents.
Leanne Ussher, Economics, New School U. (firstname.lastname@example.org).
Leanne is studying the dynamics of a speculative futures market.
The focus of the model is on the impact of various regulatory
regimes on key market outcomes such as price volatility. The model is composed
of two representative hedgers (which largely influence the spot prices)
and groups of speculative
agents (that rely on three different mechanisms to predict future prices).
She finds that various regulatory requirements across
transaction taxes and margin requirements directly impact price
volatility and other key outcomes in the market.
Xing Zhong, Sociology, U. of Chicago (email@example.com)
Xing is investigating social structure and innovation dynamics.
In her model collaborations lead to either positive or negative
externalities, and agents must learn how to search for appropriate
partners across an exogenously-adjustable innovation landscape.
The model will be tested using empirical data on U.S. patent
activity across major metropolitan statistical areas.
Miller , firstname.lastname@example.org.