Ben Bernanke, Jean Boivin and Piotr Eliasz took a fresh look at popular small-scale empirical models — vector autoregressions, or VARs — for tracking the evolution of the economy and how monetary policy affects it. These empirical models are typically based on fewer than seven or so macroeconomic indicators, because including more makes statistics unworkable. Yet in reality, many more indicators might be required to properly summarize and forecast the state of the economy. Financial markets and central banks do in fact keep track of hundreds of data series. This research solved the problem by condensing information from a large number of indicators into a few indices, an approach that requires the extra step of estimating the summary indices. The result is a factor-augmented vector autoregression, or FAVAR.
In this study, the researchers looked at data on the U.S. economy as a whole from 1959 through 2001. They compared a FAVAR model that used 120 key macroeconomic indicators to standard small-scale empirical models that used particular rather than summary variables. The results showed that the FAVAR method provides a more accurate characterization of the conduct of monetary policy and a more comprehensive picture of the effects of policy changes on various dimensions of the economy.