Instructors: Prof. J. Doyne Farmer and Dr. David R. Pugh (and co-conspirators!)
From one of Gabby’s previous emails…
On average an SFI MOOC is 10 weeks long, with 1 unit per week. Each unit is made up of roughly six-ten subunits. Most of the subunits consists of 1 to 3 video segments (ideally less than 10 minutes long), exercises (if appropriate) a quiz, and quiz solutions (in text or video form). The last subunits may not have video, instead having a homework assignment and homework solution, and an end of unit test. As an example of how a course is structured see the outline for the Introduction to Complexity MOOC.
We want to highlight the frontier research (hopefully in a way the will make it digestible to second year undergraduates).
Might make sense to have the following sub-units: Introductory video providing clear explanation of what complexity economics is (and what it isn’t). We will want to define key themes that will run throughout the course as well as terms and definitions (i.e., what is equilibrium, etc). A constructive critique of the mainstream approach to economics.
Also we will need to leave time to cover usual course logistics.
Key ideas: Illustrate the important role of expectations in economic models. Distinguish between extreme forms of expectations (i.e., naive vs rational) and stress that real expectations formation rules fall somewhere in between.
Summary: Majority of economic interactions typically involves a very small minority of the population. The tendency to focus our attention on a few individuals or activities is an attribute of the real-world that is typically omitted from the standard characterization of markets in economics. In standard models of markets, agents interact impersonally and efficiently with countless other faceless agents. This lecture looks into the consequences of including explicit connections between agents in economic decision making. Agents are assumed to occupy the nodes of a network and to interact exclusively with agents to whom they are directly linked. As motivating examples we will look at both the evolution of game strategies and the effectiveness of exchange as the topology of the underlying network is altered. Conclusion: networks matter, in particular, changes in a network’s structure can alter both the steady-state attributes of an economy as well as its dynamics.
Lecture draws heavily from Allen Wilhite's Economic Activity on Fixed Networks (Chapter 20 of the Handbook of Computational Economics, Vol 2).
Key idea: business cycles are fundamentally endogenous phenomena and are not driven by exogenous shocks as is typically (but not always!) assumed in mainstream approaches. Models of "Predator-prey" dynamics.
Will use the Solow growth model as point of departure for lecture 4 Will need to explain the basic idea behind the model to non-economics audience. Solow model can be used to motivate the importance of explicitly modeling the process of technological innovation.
Evolutionary view of technological progress. See W. Brian Arthur’s Nature of Technology. We should ask Brian if he would be interested in giving some parts of lectures 4 and 5.
This lecture will motivate the use of ABMs by covering two different ABMs of financial markets: SFI stock market (old) and Farmer, et al Leverage Causes Fat Tails and Clustered Volatility model (new).
Obvious ties with the mainstream literature. Motivated by recent (and ongoing) work of Farmer & Galla. Key idea: learning and convergence to “equilibrium”. When and under what circumstances do learning rules lead to a convergence to Nash-like equilibrium in games.
Iterated Prisoner's dilemma tournament using code from Alan Isaac's [Simulating Evolutionary Games: A Python-Based Introduction(http://jasss.soc.surrey.ac.uk/11/3/8.html). Perhaps could make use of Luzius's code to run the tournament. Each student would submit a simple Python agent with a particular strategy and then we would run face them off against one another.
Sugarscape-esque simulations to demonstrate conditions for skewed distributions of wealth. Can be tied to Edgeworth Box diagrams. Contrast the robustness of the first welfare theorem, with the fragility of the second welfare theorem.
I expect that this lecturers will fill out once we start filling in lecturers 1-8…
Topics that don’t seem to fit anywhere yet:
Rob’s model of firms… Brian’s work on increasing returns to scale and path dependence. Need to figure out a way to incorporate ScalABM to some extent. Possibly have some web application running interesting model on AWS that students can interact. Fabulous opportunity to build user community… However, ScalABM is cutting edge research tool. Pedagogically better to teach simulation and computation using Python a la Software Carpentry.
Leigh Tesfatsion’s excellent website. We need to find a way to include more of her work into the course!!!!