Working papers

Yield Curve Dynamics and Monetary Policy Transmission in Bangladesh: A Dynamic Nelson–Siegel and Markov-Switching VAR Analysis

 This paper investigates the term structure of interest rates and monetary policy transmission in Bangladesh over the period April 2015 to January 2026, employing two complementary frameworks. First, a dynamic Nelson–Siegel (DNS) model decomposes the yield curve into three latent factors—level, slope, and curvature—and embeds them in a vector autoregression (VAR) augmented with macroeconomic variables (Diebold–Rudebusch–Aruoba framework). The DNS model achieves an exceptional cross-sectional fit (¯R2 = 0.991, RMSE = 24.7 bps), with Level and Slope exhibiting near-unit-root persistence ( ˆϕ = 0.905 and 0.956, respectively). Second, a two-regime Markov-switching VAR (MS-VAR) estimated via the expectation-maximization algorithm reveals pronounced regime-dependent monetary transmission: in the stable regime, a one-standard-deviation policy-rate shock raises the 10-year yield by 10.5 bps at impact, whereas in the dominant volatile regime, the same shock elicits only 0.9 bps. The volatile regime accounts for 84.9% of the sample with an expected duration of 7.8 months, indicating that conventional rate policy operates in an environment of persistently impaired transmission. These findings carry direct implications for the Bangladesh Bank’s monetary-policy framework and the development of the domestic bond market.

JEL Classification: E43, E52, C32, G12

Keywords: Yield curve, Nelson–Siegel, Markov-switching VAR, monetary transmission, Bangladesh, term structure, regime change Data