skip navigation

Sunday, February 26, 2017

[ – ] Text Size [ + ]  |  Print Page

Philadelphia Research Intertemporal Stochastic Model (PRISM)

The Philadelphia Research Intertemporal Stochastic Model (PRISM) is a medium-scale economic model being developed by the Philadelphia Fed’s Research Department. PRISM is a dynamic stochastic general equilibrium (DSGE) model that is estimated using Bayesian methods. The model is under continued development and is being maintained by the Research Department’s Real-Time Data Research Center (RTDRC). While the model is used in some of the department’s forecasting and policy projects, it is important to note that the output of the PRISM model is not an official forecast of the Philadelphia Fed, the Philadelphia Fed president, the Federal Reserve System, or the FOMC.

About the PRISM Model

The PRISM model comprises about 40 equations and is estimated/simulated using the MATLAB programming language. A formal description of the model and its equations can be found in the Technical Appendix. PDF (213 KB, 10 pages) Additional documentation External Link in the form of a published paper that describes key features of the model and some extensions that allow the forecasting of nonmodeled variables or an earlier version of the published paper PDF is also available. We’ve also provided a recent version of the MATLAB source code ZIP archive (217 KB) for the model and the data used to estimate it.

The RTDRC will regularly post on its website forecasts from the model for key macroeconomic variables. These forecasts will be posted quarterly, around the time of the first release GDP data.

PRISM Forecasts

The PRISM charts show forecasts of key variables from the latest run of PRISM. Each figure shows the forecasted growth rate of the macroeconomic variable as well as the upper and lower bounds of the 68 percent probability coverage interval for the forecast (i.e., there is a 68 percent chance that the forecast falls within the bands). The vertical dotted line shows the forecast kickoff point.

The forecasted variables shown are real GDP growth, real consumption growth (measured as consumption of services plus nondurables), real investment growth (measured as gross private domestic investment plus consumption of durable goods), the logarithm of aggregate hours, core PCE inflation, and the short-term interest rate.

For details on data sources and how the series are constructed, see the Technical Appendix. PDF (213 KB, 10 pages).

Return to the main page for the Real-Time Data Research Center.

  • Last update: November 3, 2016