Should the recent decline in inflation raise doubts about the reliability of inflation forecasts? No, say Cleveland Fed researchers
In early 2012, Federal Reserve policymakers and many private sector economists were projecting that inflation, as measured by the personal consumption expenditures (PCE) index, would be about 1.7 percent in 2013. Today, this measure stands at about 1.2 percent. Should the unanticipated decline in the inflation rate since early 2012 raise doubts about the reliability of inflation forecasts and common forecasting models?
The answer, according to Federal Reserve Bank of Cleveland researchers Todd Clark and Saeed Zaman, depends on the magnitude of the surprise relative to historical norms and the ability of the models to explain (after the fact) the slowing of inflation.
Using a common forecasting model, Clark and Zaman find that the gap between actual inflation and forecasts made in early 2012 hasn't been especially large and falls well within the normal range of uncertainty. In addition, the researchers say their model explains most of the falloff in inflation as a response to other economic developments.
Magnitude matters, say the researchers, because the outlook for inflation and other macroeconomic variables is always uncertain; what's important is whether the deviation of actual inflation from the model forecast falls within the normal range of uncertainty around the path the model projected.
The ability of models to explain the deviation matters because it reflects how well the model is constructed. Inflation may have followed a path different from the one expected because the economy experienced surprise movements in some of the determinants of inflation that are built into the model, such as GDP growth or unemployment.
Using their model to project the most likely path of core PCE inflation from the second half of 2013 through 2015, the researchers say it appears that core PCE inflation has bottomed out and will gradually rise over time toward the Federal Reserve long-term inflation goal of 2 percent. However, they caution that the recent unanticipated falloff in inflation should serve as a useful reminder of the uncertainty that always surrounds forecasts.
A second study from the Cleveland Fed shows how accounting for a slow-moving underlying inflation trend helps to improve the accuracy of inflation forecasts. Saeed Zaman, a senior economic analyst at the Bank, finds that incorporating the trend (by modeling inflation as the gap from an estimated underlying trend) leads to substantial gains in forecast accuracy of about 20 to 30 percent, two to three years out.