Computational Infrastructure for Economic Research

The group of economics-focused scientists at the Computation Institute is dedicated to helping the economics community by building tools and providing computational resources that advance research in major areas such as Growth and Developmental Economics and Computational Economics.

Economics is one of the sciences where the researchers have a significant amount of work to do before they can publish their results. First of all it's an imprecise science meaning that it often approximates the functions describing the actor's behavior and also many times it needs to deal with incomplete and inexact input data sets. Second, it is hard to implement and study an economic model in isolation: its components and interactions spread into and are affected by many related social systems. Third, the choices that need to be considered in combination with the complex mathematical formulation of the outcomes often make the problem size unmanageable.

 We are introducing in this column the most visible and publicly tangible product of our involvement with the economists: an integrated survey data exploration and visualization system which we have developed jointly with and dedicated to supporting research coming out of the Enterprise Initiative. This effort has been funded by the John Templeton Foundation and is led by Prof. Robert Townsend from the MIT Economics Department, with the collaboration being set up by Ian Foster, the Computation Institute director.

The main goal of the project was to create a tool that enables rapid intuition-building from the socioeconomic survey data available. We have developed a system that allows for a quick and intuitive web-based exploration of the survey data at hand. The system is used as follows. Once the variables of interest are identified, the researcher will use the set of visual tools that we have implemented to illustrate the data on the map, via the geospatial functionality of our system, and to look at significant statistical measures via plots and summaries. This new information enables the researcher to decide whether the data they are currently exploring is fit to be used in the specific economic models that they are interested in, and to extract and explore in further detail this data in more complex systems.

In order to manage large, heterogenous, multi-modal, and often messy data sets, we have chosen an architecture based on processing pipelines. We have defined and are currently improving and extending standard APIs to support future growth of the system, and we are optimizing the processing pipelines to make the research process more interactive. We are also leveraging our GIS expertise by bringing into the picture novel data exploration techniques via maps.

Our collaborations in the computational and modeling areas, as well as in applying current economic research in new settings, deserves a completely separate treatment, which we hope to provide in a future issue of this newsletter. You can get a preview of that by following The Applied General Equilibrium for Enterprise Economics reference below.

Our team, Tibi Stef-Praun, Victor Zhorin and Neil Best, is welcoming future collaborations with researchers interested in these (or related) fields, and hope to leverage the current infrastructure by encouraging other researchers to adopt and contribute to our current work.

 

 

 

 

 

 

 

 

 

 

 

 

 

The figure shows that spatial characteristics play a major role in estimation of entrepreneurship dynamics in poor northeastern provinces of Thailand compared to richer central provinces. Without the map visualization, it is very hard to determine whether massive model outputs tell a consistent story, which has significant policy-making consequences. For example, using the map, researchers can evaluate, from a model standpoint, the policy of building more roads and providing more transfers in the capital to those poor provinces.

 Further references:

The Survey Data Explorer Application: http://age3.uchicago.edu:8080/thailand
The Enterprise Initiative: http://enterpriseinitiative.org
The Applied General Equilibrium for Enterprise Economics (AGE3) http://age3.ci.uchicago.edu

Computation News; Volume 2, Issue 3.  Link to article: http://www.ci.uchicago.edu/newsletter/vol2issue3/compecon.php