Economic Forecasting at the Fed is an article on the excellent OReilly site, about the serious uses to which the equally excellent open-source computer language Perl is now being put.
However, it also reminds me that US financial planners routinely use model simulations and forecasts “to gauge how major events, such as disruptions to foreign economies or changes in fiscal policy, may affect U.S. growth and inflation, and how monetary policy responses to such shocks might affect these outcomes.” .
The article is about Douglas Battenberg, an economist and Perl programmer at the Federal Reserve Board of the United States. His group is responsible for preparing model-based macroeconomic forecasts for the U.S. economy. Much of his work is concentrated on “FRB/US”, a large-scale econometric model with extensive simulation capabilities.
An account by Battenberg and others of FRB/US sets out in some detail the assumptions and mathematics underlying the model. And a paper on The Hidden dangers of historical simulation, by Matthew Pritsker, looks at how much historical data is required before you can confidently use it to compute Value at Risk (VAR), a measure used by large financial institutions for looking at the risk of their portfolios. Pristsker concludes that models are “under-responsive to changes in conditional risk and respond to changes in risk in an asymmetric fashion: measured risk increases when the portfolio experiences large losses, but not when it earns large gains.”