Micro-to-Macro Simulation: A Primer With a Labor Market Example

S imulation models, now widely used in the physical sciences, can also help economists in depicting the actions and interactions of individuals and firms through time. Such models can be built and run on any personal computer. They can illuminate a wide variety of issues that are difficult to analyze with an economist's conventional tool-kit and are especially useful in studying dynamic processes. Microsimulation models allow the analyst to produce simulated aggregate time series that are rigorously consistent with the assumptions made about behavior on the micro level. Thus, simulations provide a means of bridging the micro-macro split in economic analysis. This article provides an introduction to microsimulation: how it works, how to do it, its potential, and its drawbacks. It then allows readers, even those with no experience in computer programming, to work through the details of a simple microsimulation model. Readers can put this model on their PCs, watch it run through its paces, and experiment with their own modifications. With this model as an example, it should be fairly easy to create programs for new models on other subjects for use in theoretical exploration, empirical research, or classroom demonstrations. The demonstration model I present depicts the experience of individual workers during recession and recovery in the labor market. As the model runs, its internal "Census Bureau" performs surveys on the microlevel, and sums up to macrolevel variables. The model is simple, but it has some interesting applications. After explaining how to set it up, I use it to explore the effect of unemployment insurance on the level of unemployment, and to point up a common fallacy in current labor market literature.