Economic Modeling

views updated

Economic Modeling

As computing technologies have become more and more sophisticated, economists have come to rely increasingly heavily on computers for a variety of tasks. Today, economists use computers to analyze large quantities of statistical information, to create simulations or scenarios of what might happen to an economy if a particular economic variable, such as the price of oil, changes, and to solve complex mathematical problems that advance our knowledge of economic processes.

Perhaps the most common use of computers in economics is for the analysis of statistical information. Economists call this type of analysis "econometrics." Using the large amounts of statistical economic information that are now easily available on the Internet and elsewhere, economists can test simple but important questions about economic behavior. For example, based on data for the past two decades for the fifty states in the United States, what can one say about how an increase in the price of a pack of cigarettes will affect the number of packs of cigarettes that people buy? How will this reduce the incidence of such health problems as lung cancer and tooth loss? In order to analyze such questions, economists must have access to information on how many packs of cigarettes are sold at any time, and at what price. They must also know the relationship between the number of people who smoke cigarettes and the number of people who end up getting sick. In addition, they must have powerful computers in order to analyze such large quantities of statistical information. And finally, they must have the methods and software that enable them to answer such questions.

The most widely used method in economics for analyzing relationships between two or more variables (such as the price of a pack of cigarettes and the number of packs bought) is called regression analysis. In a regression, the economist specifies the relationship he or she is trying to analyze. In the preceding example, the regression analysis indicates the effect of the price of cigarettes on their consumption, and the effect of the consumption of cigarettes on the health of populations. Using a regression, the economist is able not only to tell whether or not cigarette consumption falls due to price increases, but also predict the amount by which consumption is decreased due to the price increase. Similarly, the economist may be able to estimate the effect of an increase in the price of a pack of cigarettes on the number of people who develop lung cancer or who lose their teeth as a consequence. Such information is of enormous value to policymakers. In the cigarette example, this information could enable people in the federal or state governments to determine a level of taxation that would reduce the consumption of cigarettes due to an increase in price, without being so large as to be burdensome to cigarette smokers.

Because computers are so integral to the analysis of economic data, a number of software packages are now available that enable economists to run regressions and to answer questions about economic behavior. Microsoft Excel, for example, has a regression tool that allows the researcher to do very basic kinds of economic analysis. Professional economists, however, use more sophisticated statistical software packages for their analyses. Among the most popular packages are SAS and SPSS, which are used by businesses, government agencies, and academic researchers alike for the analysis of economic statistics. Depending on the particular kind of analysis to be done, other packages such as RATS, STATA, TSP, Limdep, Matlab, and Gauss are also very useful.

Computers also enable economists to run simulations, in which a possible scenario for the future is created, allowing economists to answer important economic questions about this scenario. For example, an economist might ask what would happen to smoking in the United States if the price of a pack of cigarettes were to triple over the next year. Because this may never have happened in the past, the economist may not be able to use any single example from previous years to estimate the effect of this change in price on the change in consumption. The economist can, however, make an informed guess as to what might happen with the help of software.

Economists use a variety of software packages to simulate scenarios about which there is no prior statistical information. These include the General Algebraic Modeling System (GAMS) and Mathematica, both of which enable the economist to write down explicit relationships between variables that will play a role in the final outcome of the simulation. To extend the cigarette example, we know that a tripling in the price of cigarettes will cause many smokers to cut back on smoking because they have limited income to spend on cigarettes. But perhaps this will only affect casual smokers, for whom it is, presumably, easier to stop smoking. Addicts, on the other hand, may continue to smoke the quantities they did prior to the increase in the price. Depending on how we think addicts and casual users behave, we may want to model their behavior differently, making the simulation more complex. Fortunately, as computers have become increasingly powerful, economists have been able to analyze increasingly complex economic problems using computers. It is common, for example, for economists at universities and government and policy institutions to simulate the workings of an entire economy using hundreds of mathematical equations written in the GAMS software.

Given the rapid development of economic modeling and the increasing availability of statistical economic information that can be used to understand the way the economy works, it is difficult not be optimistic about the future of economic modeling. While economists have a long way to go before they can be confident about the predictions generated by their simulations, it is encouraging to recall that as recently as the 1970s, weather forecasters were unable to predict the weather with any accuracy. With the development of supercomputers and complex models of weather systems, it has now become possible to predict, with a fair amount of accuracy, the weather in the coming week. Perhaps economic modeling today is where weather forecasting was in the 1970s. If so, economists and those who stand to benefit from economic modeling will see many technological advancements in the future!

see also Decision Support Systems; Image Analysis: Medicine; Mathematics; Pattern Recognition; Simulation.

Siddharth Chandra