Noise in the German Government Bond Market

Carl Joel Hartman & Simen Spjut

Student thesis: Master thesis

Abstract

For each day from January 2000 through December 2015, we calibrate the Nelson- Siegel-Svensson model using the Differential Evolution algorithm and calculate a noise measure defined as the cross-sectional root mean squared error (RMSE) between quoted market and model implied yields for German government bonds.
The noise measure serves as a proxy for liquidity, as in times with high noise arbitrageurs do not have the same possibilities to exploit trading opportunities due to constraints in access to capital. Relating the noise measure to other popular liquidity variables, it becomes evident that the estimated noise measure adds extra value in explaining liquidity in the market.
Consistent with previous research on slow-moving capital, our results indicate that returns, a proxy for capital available to arbitrageurs, of certain hedge fund categories have a significant impact on the noise in the following month. The results thus provide further evidence that limits in funding prevents market forces to correct asset mispricings.
To test the robustness of the results, the noise measure is calculated using alternative settings in the Differential Evolution algorithm in calibrating the parameters of the Nelson-Siegel-Svensson model. The robustness check stresses the importance of careful calibration, as the series of daily curves end up of having notable worse fits, and the following hedge fund regression analysis overall show weaker relationships.

EducationsMSc in Advanced Economics and Finance, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2016
Number of pages81