Polish sovereign term-premium decomposition
Adrian-Crump-Moench (ACM) affine term-structure model and the Bauer-Rudebusch-Wu small-sample bias correction (BRW), estimated weekly on the LW-NSS zero-coupon panel. Cross-country comparison runs the same ACM specification on the GSW US Treasury panel and the ECB euro-area AAA panel.
Read the short — "Are survey-based rate expectations informative?"PL term-premium statistics by horizon — click a row to drive the chart below
Cross-sectional summary statistics of the PL ACM and BRW term premia at standard horizons (1y, 2y, 3y, 5y, 7y, 10y) over the selected window. The "% rank" column shows where the latest observation sits in the windowed distribution: 0% = at the historical minimum, 100% = at the historical maximum, 50% = at the median. Click any row to load that horizon into the time-series chart below.
PL term premia by horizon — ACM versus BRW
Time-series of PL ACM and BRW term premia at the horizon selected above. Both lines are weekly-Friday closes from the LW-NSS curve. The window pills at the top of the page restrict the visible range.
Cross-country term-premia comparison
Polish (BRW) vs US (ACM) vs euro-area AAA (ACM) term premia at the selected horizon. US and EA term premia are estimated on the GSW Treasury and ECB AAA panels respectively, using the same Adrian-Crump-Moench five-factor specification. Monthly resampling (last observation of month).
Correlation heatmap — PL BRW × US ACM
Pearson correlations between PL BRW term premia and US ACM term premia at all horizon pairs. Off-diagonal cells tell you how Polish term premia at one tenor co-move with US term premia at another tenor.
Correlation heatmap — PL BRW × EA ACM
Same construction with euro-area AAA term premia. Higher PL-EA correlations than PL-US would be the prior, given closer business cycle and monetary policy linkages.
Methodology: ACM five-factor affine term-structure model, OLS regression of monthly excess returns on lagged Nelson-Siegel principal components, no cross-sectional restrictions imposed. BRW correction follows Bauer, Rudebusch and Wu (2012, 2014), iterated bootstrap with common-random-numbers and Kilian shrinkage on the risky-recursion VAR. US and EA panels are interpolated to a 120-month TTM grid via cubic spline before fitting the same specification. See about page for full references.