미국 국채 수익률곡선 및한․미 거시경제변수의 변화가한국 국채 수익률곡선에 미치는 영향:거시경제변수를 포함한2국가 Dynamic Nelson and Siegel 모형을 이용하여 (The Impact of the US Yield Curve and Korea-US Macroeconomic Variables on the Korean Yield Curve: Two-country Dynamic Nelson and Siegel Model with Macroeconomics Variables)

본고는 거시변수를 포함한 Dynamic Nelson and Siegel 모형을 2국가로 확장한 후, 이를 2단계 접근법을 사용한 구조적 벡터자기회귀 모형을 통해 분석하였다. 본 연구는 양국의 수익률곡선과 거시경제변수를 전환방정식과 상태방정식으로 구성된 하나의 모형에서 분석함으로써 국가 간, 금융 및 거시변수 간 상관관계를 확인하였다는 점에서 기존 연구왿� 차별된다. 실증분석 결과의 주요 내용을 살펴보면 우선, 양국의 수익률곡선 변수 간, 거시경제변수 간 유효한 충격반응이 나타나는 것으로 분석되었을 뿐만 아니라 수익률곡선 변수왿� 거시경제변수 사이에도 유효한 충격반응이 관찰되었다. 또한, 한․미 양국의 수익률곡선 변수 모두에 대해 단기적으로는 수익률곡선 변수의 충격이 장기적으로는 거시경제변수의 충격이 큰 영향을 미치는 것으로 분석되었다. 국채수익률에 대한 분산분해에서는 한국 국채수익률이 미국에 비해 거시경제변수의 영향을 상대적으로 크게 받는 것으로 나타났으며, 한국 단기 국채수익률의 경우 한․미 양국 거시경제지표의 영향력이 50% 내외의 큰 비중을 차지하였다. 마지막으로, 예측력 비교에 있어서 거시경제지표를 포함한 모형이 수익률곡선 변수만을 포함한 모형에 비해 한국 국채수익률 예측 시 대부분의 예측시계에 걸쳐 우수한 것으로 확인되었다.

This paper extends the Dynamic Nelson and Siegel model with macroeconomic variables (Diebold and Li, 2006; Diebold et al., 2006) to two countries, U.S. and Korea, with macro variables and analyzes it using the Structural VAR. This study contributes to the literature in that it analyzes the financial and the macroeconomic relationship between two economies, with yield curve factors and macro variables, in a single model composed of transition equations and state equations. In addition, it is confirmed that the predictive power of this model, two-country yield-macro model, is superior to that of the existing model, two-country yield-only model, even when the structural change after the Global Financial Crisis is reflected by separating the period after the GFC.
The results of the study can be summarized as follows. First, there are valid impulse responses between the yield curve factors and the macroeconomic variables, as well as valid responses in the yield curve factors and in the macroeconomic variables of the two countries. We can find that the U.S. Curvature factor has a significant effect on U.S. macroeconomic variables such as the manufacturing capacity utilization index. Korea’s Slope factor is shown to be affected by most macroeconomic variables of U.S.
and Korea, which means that Slope factor is not only sensitive to Korea’s macroeconomic cycle but also to the U.S. and Korean financial and real market conditions, and that the economic conditions of the two countries are reflected in the Korean government bonds’ yield curve through the slope variables.
Second, for all of the yield curve factors, shocks of the yield curves are influential in the short run and shocks of the macroeconomic variables are influential in the long run. In particular, the two countries’ Slope factors are responsive to the macroeconomic shocks. The variance decomposition of the yields shows that Korean governmen bond yields are affected more by macroeconomic variables than the US yields are, and two countries’ macroeconomic variables have strong effects on Korea short-term interest rates.
Lastly, in the predictability of Korean government bond yields, it is analyzed that the Yield-Macro model is superior, across most forecast horizons, to the Yield-only model. This result is valid for analysis focus on the period after the financial crises, and the same result is confirmed regardless of the bonds’ maturity.
In light of the results, we believe that the macroeconomic variables of Korea and of the U.S. should be considered as well as the U.S. government bond yields in order to more precisely predict the Korean yield curve, especially the change of the Slope factor and short-term yields. It is also necessary to consider the possibility that the influence of Korean monetary policy can be extended or curtailed by other external factors, as it has been confirmed that Korean Slope factor and short-term yields are affected by U.S. government bond yields and macroeconomic variables as well as domestic monetary policies. As this SVAR model analyzes the impact of shocks in economic variables on the Korean government bond yield curve, the model is expected to be used as a tool to predict changes happening in the Korean government bond market. In addition, the model is expected to be expanded in a useful way to improve predictability and to determine policy validity by adding additional macroeconomic variables of the two countries.