Embedded Files
This website reports real-time estimates of longer-run neutral interest rates – the real component of policy interest rates consistent with both economic activity and inflation at their longer-run trends. These estimates simultaneously account for a comprehensive set of determinants of neutral rates: trend productivity, demographics, global supply of safe assets, demand factors for safe assets, and global spillovers faced by each economy from the rest of the world’s developments. The methodology follows Ferreira and Shousha (2023), with pre-print draft here.
The results shown below focus on the United States (US), euro area (EA), United Kingdom (UK), Canada (CA), and Japan (JA). The underlying estimates used in the figures and tables of this website are from the October 2 vintage , as well as results for Australia, Denmark, Norway, New Zealand, Sweden, and Switzerland. For older vintages of the estimates, see end of this page.
Estimates of Real Neutral Rates in Selected Dates
(in percent)
Estimates of Real Neutral Rates in Selected Dates
(in percent)
Decomposition of Changes in Real Neutral Rates Since 2020
(in percentage points)
Decomposition of Changes in Real Neutral Rates Since 2020
(in percentage points)
Technical details on the real-time implementation
Technical details on the real-time implementation
There are some changes relative to Ferreira and Shousha (2023) that are worth clarifying:
The data on productivity growth from the original sources are only updated once a year, with many months of lag in their release. Thus, we splice these productivity series with those available from national statistic offices, whenever the later data are available.
The data on productivity growth is often subject to substantial revisions, both from national statistical offices and the original cross-country references. Thus, the recent history of neutral rates is also subject to meaningful revisions every time we update our dataset.
Because we measure trends of productivity growth and convenience yield using a Hodrick-Prescott filter, end-of-sample trend values may be sensitive to the last observations of the underlying data. To alleviate this issue, we forecast productivity growth and convenience yield up to 4 years ahead using a rolling 25-year window, then, and only then, we estimate the trends of these variables using the full sample (history and forecast).
We use data available until October 02, 2025, with model coefficients from the published paper.
Estimates will be updated monthly.
Vintages of estimates
Vintages of estimates
Google Sites
Report abuse