Accelerated solution of a moral hazard problem with Swift

The study of human interactions and the modeling of social structures often require large-scale computational experiments involving complex models and substantial computational resources. The formulation of such models and computations can be taxing for researchers. We introduce the Swift parallel programming system and the SwiftScript scripting language, which together enable the rapid specification and execution of such experiments. The researcher specifies computation in SwiftScript, via a set of procedures with input-output (file) dependencies; the Swift system can then execute various sub-computations on parallel and/or distributed computing systems, ensuring reliable execution and respecting data dependencies. We present Swift features by describing their use in a computational economics application, a moral hazard problem. The benefits obtained suggest that Swift may also have utility in other social sciences.