How does the Nelson-Siegel-Svensson model parameterize the entire yield curve with just six parameters, and why do central banks prefer it?

The Nelson-Siegel-Svensson (NSS) model represents the zero-coupon yield curve as a closed-form function of maturity, using six parameters that map to economically interpretable curve features. Central banks worldwide (ECB, Bundesbank, Bank of England) use it for daily yield curve publication.\n\nThe NSS Formula:\n\nr(t) = beta_0 + beta_1 x [(1 - e^(-t/tau_1)) / (t/tau_1)] + beta_2 x [(1 - e^(-t/tau_1)) / (t/tau_1) - e^(-t/tau_1)] + beta_3 x [(1 - e^(-t/tau_2)) / (t/tau_2) - e^(-t/tau_2)]\n\n`mermaid\ngraph TD\n A[\"NSS Parameters\"] --> B[\"beta_0: Long-run level
Asymptotic yield as t goes to infinity\"]\n A --> C[\"beta_1: Slope factor
Short vs long rate spread\"]\n A --> D[\"beta_2: First curvature
Medium-term hump/trough\"]\n A --> E[\"beta_3: Second curvature
Additional hump (Svensson)\"]\n A --> F[\"tau_1: First decay rate
Controls where beta_2 peaks\"]\n A --> G[\"tau_2: Second decay rate
Controls where beta_3 peaks\"]\n B --> H[\"Determines where curve
flattens at long maturities\"]\n C --> I[\"Positive = upward slope
Negative = inverted curve\"]\n`\n\nParameter Interpretation:\n\n| Parameter | Economic Meaning | Effect on Shape |\n|---|---|---|\n| beta_0 | Long-term rate (level) | Shifts entire curve up/down |\n| beta_1 | Short-long spread (slope) | Positive = normal; negative = inverted |\n| beta_2 | Medium-term curvature | Creates hump or trough around tau_1 |\n| beta_3 | Secondary curvature | Adds flexibility (Svensson addition) |\n| tau_1 | Decay speed 1 | Sets maturity where first hump peaks |\n| tau_2 | Decay speed 2 | Sets maturity where second hump peaks |\n\nWorked Example:\nEchelon Research fits NSS to the UK gilt curve on a given date and obtains:\n\nbeta_0 = 4.25%, beta_1 = -1.60%, beta_2 = -2.10%, beta_3 = 1.35%, tau_1 = 1.8, tau_2 = 8.5\n\nInterpretation:\n- Long-run yield: 4.25%\n- Short rate approximately equals beta_0 + beta_1 = 4.25% - 1.60% = 2.65%\n- The negative beta_2 creates a trough around the 1.8-year point\n- The positive beta_3 adds a hump near the 8.5-year region\n- Overall shape: starts at 2.65%, dips slightly around 2Y, rises through a hump near 8-9Y, then converges to 4.25%\n\nAdvantages Over Splines:\n- Parsimonious: only 6 parameters vs. potentially dozens for splines\n- Always produces smooth forward curves (infinitely differentiable)\n- Parameters evolve smoothly over time, enabling time-series analysis\n- Cannot produce negative forwards (with appropriate parameter constraints)\n- Extrapolation is well-behaved (converges to beta_0)\n\nLimitations:\n- Cannot perfectly fit all market quotes simultaneously (trades fit for flexibility)\n- Calibration requires nonlinear optimization (multiple local minima possible)\n- May struggle with very unusual curve shapes (double inversions)\n\nMaster yield curve modeling in our FRM Part I course.