2000 Graduate Workshop in Computational Economics
Student Projects
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
journal articles.
Josh Anderson, UCSB (josh@econ.ucsb.edu).
Josh is developing a more cognitive-based theory of finance. In his
model, two types of agents attempt to buy and sell a security. The
first type of agent bases its behavior on reinforcement (associative)
learning concepts arising from cognitive psychology. In the initial
model, these agents base their behavior on price and volume
observations from the previous k time periods (in future models,
agents will adaptively identify key features to focus upon). The second type
of agent uses the more traditional decision approach of attempting to
predict and discount the future value of a security. The model will
explore the dynamics of such a world.
Yann Bramoulle, U of Maryland
(bramoulle@arec.umd.edu).
Yann has created a simple model of crime and social interactions. In
his model, agents must decide whether or not to participate in a life
of crime versus joining the legitimate labor
market. This choice is influenced by social networks that form among
the agents---agents who are connected to other agents pursuing lives of
crime tend towards that choice as well (influenced either by norms or lower
information costs). By manipulating the level of property rights (that
is, the amount of legitimate activity that is immune from crime) and
network connections, Yann finds that very different social outcomes
result, ranging from stable, low-crime worlds to highly unstable
scenarios in which wave after of wave of crime hits society.
Elizabeth Bruch, UCLA (bruch@ucla.edu).
Elizabeth has extend Schelling's basic tipping model to multiple
minority groups so that she can investigate key demographic issues
surrounding multi-ethnic segregation patterns. The model suggests that
additional minority groups can buffer segregation patterns. She is
currently refining the model and will eventually link its basic
elements (for example, preferences for living with other groups) and
outcomes to empirical data.
Stephanie Chow, Cal Tech (steph@hss.caltech.edu).
Stephanie is exploring the dynamics of political party formation. In
her model, citizens who cannot find a suitable candidate from among the
current slate, form new parties that are tested during the primary
elections. Final party positions depend on both the candidate's
inherent preferences and a willingness to compromise in order to
win. She is currently investigating the party formation dynamics
implied by this model. Early results indicate that such a system tends
toward good policies and the domination of elections by a small number
of parties.
Eldar Nigmatullin, U of Wisconsin
(enigmatu@ssc.wisc.edu).
Eldar is focusing his work on how neighborhood interactions can
influence socioeconomic dynamics. Specifically, he is developing a
theoretical and empirical model of pre-marital births. In the model,
the likelihood of having a pre-marital birth is tied to the occurrence of
such events in an agent's social network. Early results indicate that
such neighborhood interactions can dramatically alter the hazard rate
and that the efficacy of policy is closely linked to
understanding key interactions.
Paolo Patelli, Pisa (paolo@black.gelso.unitn.it).
Paolo is analyzing the decentralized information processing inherent in
large, hierarchical organizations. In his work, organizations are
modeled as hierarchical trees with each node being capable of limited
information processing. The nodes employ adaptive behavior to try and
refine their individual information processing. The model will allow
Paolo to investigate issues ranging from how to provide effective
feedback to the nodes to
ways of duplicating activity that will improve performance.
Daniel Reeves, U of Michigan, (dreeves@umich.edu).
Dan is exploring both optimal and heuristic bidding in the synchronous
k-double auction. He has created a "strategy-zoo" of heuristic rules
for bidding (for example, shave your true value by a fixed percent),
and is currently studying the performance of these strategies in
mixed populations. He is also deriving
"optimal" strategies for this auction environment. Eventually he hopes
to refine current strategic theories about these auctions as well as
implement these results in actual web-based auction markets.
Darren Schreiber, UCLA (dschreib@ucla.edu).
Darren is analyzing the dynamics underlying political party formation.
In his model, agent behavior is guided by a small set of nested
heuristics, for example, form an alliance with agents that have similar
preferences, change your position towards winning platforms, etc. The
nested heuristics allow various types of parties to form, ranging from
pure, preference-based coalitions to ambitious, office-seeking
parties. The current analysis of the model is focused on understanding
the dynamics of party formation and heuristic interaction, as well as
testing some long-standing theoretical ideas (for example, Duverger's
law).
Troy Tassier, U of Iowa (troy-tassier@uiowa.edu).
Troy has created a model of referral networks in labor markets. In the
model, firms need workers for both low and high-skilled jobs. Workers,
who also have different skill levels, learn about job openings either
through general advertising or via social networks of knowledgeable
friends. The current focus of the analysis is on the dynamics of job
acquisition in a market where there is a minority group with an
identical distribution of skills as the majority group. By exploring
the impact of social networks and segregation, key policy questions
such as the speed at which minorities can attain equal job access,
etc., can be investigated.
Charles Williams, U of Michigan (fcw@umich.edu).
Charlie is exploring a simple model of market-driven labor
specialization. In his model, agents begin to specialize their
production activities by forming long-term trading networks with other
agents who also specialize. The model can be used to investigate the
effectiveness of decentralized versus centralized organizations, the
dynamics of adaptive trade network and supply chain formation, the impact of
technological shocks, etc.
John H.
Miller , miller@zia.hss.cmu.edu.