Risky business

released 11.12.09

By Vince Dixon

University of Illinois professors use NCSA computers to study how the economic environment and financial institutions influence entrepreneur behavior.

Before starting a software company in Texas, Pat Sullivan discussed with his wife the possibility of opening the firm. His wife feared the risks of starting a new business.

Sullivan reassured her by telling her that in Texas, they can't take your house away, they can't take your last car, and they can't take your kids.

Sullivan was later named Entrepreneur of the Year by Inc. magazine and made millions off the business. Stefan Krasa and Anne Villamil, professors of economics at the University of Illinois at Urbana-Champaign, use Sullivan's story to explain their ongoing study of the effects of economic environments and financial institutions on small firms.

The project pinpoints the risks budding entrepreneurs face when starting their businesses and why some entrepreneurs are willing to bear these risks. The professors also wanted to understand the roles financial institutions and credit play in such decisions.

"One question is, why do so many of these entrepreneurs who don't choose to be incorporated not try to protect themselves more?" Krasa says.

Modeling the question

Using NCSA resources, Krasa and Villamil developed a model in 2005 to help understand the answer to this question. With Jamsheed Shorish of the Institute for Advanced Studies in Vienna and the University of Illinois, the researchers have recently extended the model to include the effects of the banking sector.

Understanding how the environment—which includes bankruptcy policies, protections, and rules—and access to credit affects small firms is important in recognizing the overall significance small firms have in an economy, the researchers say.

In the United States, these firms are responsible for more than 50 percent of Gross Domestic Product (GDP) and 45 percent of total private payroll. Small businesses also create 60 to 80 percent of net new jobs annually, including high tech "good jobs."

"Small firms are a very dynamic part of the macro-economy because some people start out small and they eventually become big," Villamil says. "If these firms are not healthy and hiring, that's going to affect [the economy]."

The first half of the project centered on how the economic environment influences entrepreneur behavior. Countries and states have different policies that affect small businesses. Depending on the risks involved with the policies, entrepreneurs in one area may find it easier to start businesses than entrepreneurs in other areas.

U.S. institutions favor risk taking, Krasa says. New business owners like Sullivan take advantage of it and find the risks of starting a new business in the U.S. manageable.

Krasa and Villamil use the contracting model to quantify such risks. The model was "computationally intensive," taking 10,000 hours to complete on the NCSA Copper supercomputer. Information on how a firm raises money, the size of the firm's loans or interest rates, and the amount of protection granted by legal codes were included in the model's parameters.

They presented their work at two European conferences in 2007—SAET and the XVI European Workshop on General Equilibrium—and also at the Institute for Advanced Studies in Austria. Their work was published in Economic Theory in 2008.

The credit factor

The second part of the project involves expanding the first model to include the role of credit and banking institutions.

"It's tough to get credit in the first place," Krasa says. "But without credit, you can't run a business; it's like without oil you can't drive a car."

Because of the influence small firms have on the economy and job market, the professors wanted to know how access to credit impacts these businesses. Villamil used the current increase in U.S. default rates to demonstrate that if default rates on business loans rise because of poor economic conditions, a chain reaction could form. Banks would fail and workers would lose jobs, perhaps defaulting on their mortgages and other debt.

"So these things are not isolated to small firms," Villamil says. "They affect the economy as a whole."

New models

The project's new model illustrates supply and demand for firms and banks. It is more complex and will take more than 20 times the amount of hours used to complete the former model. The expanded model studies the interaction of many firms rather than focusing on the individual firm and permits the analysis of real time policy experiments.

The research involves three important steps. First, to understand how firms and banks operate and interact, the researchers had to create an economic model.

Next, Krasa, Villamil, and Shorish transform the model into computer code and develop algorithms to compute solutions. The 2005 model used MATLAB programming. The new model needed a much more complex environment because of the added parameters. The professors also wanted to make the computation easily accessible to others to use and test. Python programming seemed to be the best code language to do the job. They are currently in the process of testing this code.

The final stage in the project is to simulate the effects of alternative policy proposals in the model. Economists call this type of research a counterfactual experiment, Villamil says.

In this method, researchers ask a "what if" question and use the model to evaluate the effects. This approach allows one to compare alternative policies and better design regulations without testing the effects in the actual economy.

"That's a real virtue of having a computable model that you can go through and look at different policies in a controlled way," Villamil says.

She used banking crises as an example. While the U.S. has recently experienced its own banking crisis, such problems happen around the world each year. The team hopes their model can help governments and organizations develop better policies to use in times of crisis and create better regulations to prevent them.

The current state of the U.S. economy has raised the stakes of this research. "It certainly made it a much more interesting topic for us," says Krasa. The current economic downturn is an example of what the researchers say they hope their model would prevent.

Other researchers have used supercomputers to study economic policies, but few have used the systematic analyses of this project. Villamil says the opportunity is unique to NCSA and Illinois researchers.

"We're very lucky at the University of Illinois to have access to both the supercomputer itself and also the people at NCSA to discuss these things with," she says.

The supercomputing power at NCSA speeds the entire process and allows the researchers to develop more complex computations. With the stakes set high from the current economic crisis, this is important, she emphasizes.

The two professors say the ultimate goal of the project is to help economies around the world develop policies that could possibly avoid future crises. In order for this to happen, they need the speed of supercomputers to handle complex model computations.

"Before you do a new policy you'd really like to understand how it works in the context of a model rather than trying it out on an entire macro economy," Villamil says. "That's really the exciting part of this; that the computational environment provides a laboratory in which we can do these controlled experiments."

This project is funded by the National Science Foundation.

Team members
Stefan Krasa
Anne Villamil
Jamsheed Shorish