Case Study: Federal Reserve Bank of New York achieves 10x faster DSGE model estimation with Julia Computing

The Federal Reserve Bank of New York publishes its trademark Dynamic Stochastic General Equilibrium models in Julia

The Federal Reserve Bank of New York worked with Julia Computing to modernize its complex macroeconomic modeling workflow. The bank needed a faster, more accessible way to run and maintain its Dynamic Stochastic General Equilibrium (DSGE) models, which support forecasting and policy analysis for the economy.

Using Julia and the DSGE.jl package, Julia Computing helped FRBNY rewrite its model code in a more concise and maintainable format while significantly improving performance. The results included model estimation running about 10x faster, the “solve” test running 11x faster, and some algorithms running about 6x faster than prior C++ versions, while cutting the codebase roughly in half and reducing errors.

Federal Reserve Bank of New York