Term premia in affine term structure models with unspanned macroeconomic factors: The case of Korea

Research output: Contribution to journalArticlepeer-review


Using the yield data for Korean government bonds, I examine several discrete-time affine term structure models with unspanned macro factors, such as output and inflation, and compares term premia implied from alternative models with different combinations of output and inflation variables. Empirical analysis shows that, except for 1-year maturity ones, there is little difference among the medium-to long-term term premia across alternative models. The model-implied term premium estimates do not show a significant pro-or counter-cyclicality in relation to output variables, but show a highly positive correlation with inflation variables. In addition, I test the traditional expectation hypothesis by fitting Campbell-Shiller long-rate regressions to the Korean bond data, the expectation hypothesis is strongly rejected as in the case of the US, due to time-varying term premia, and an additional Monte Carlo simulation study indicates that the term structure models considered in this paper show a success in matching the regression coefficients estimated from the sample.

Original languageEnglish
Pages (from-to)70-110
Number of pages41
JournalJournal of Economic Theory and Econometrics
Issue number2
StatePublished - Jun 2020

Bibliographical note

Funding Information:
∗This work received a financial support from the Bank of Korea. I am very grateful to the editor, Noh-Sun Kwark, the co-editor, Jin Seo Cho and two anonymous referees. †Corresponding author, Department of Economics, Ewha Womans University([email protected])

Publisher Copyright:
© 2020, Korean Econometric Society. All rights reserved.


  • Affine Term Structure Model with Unspanned Macro Factors
  • Expectation Hypothesis
  • Korean Government Bond
  • Term Premia


Dive into the research topics of 'Term premia in affine term structure models with unspanned macroeconomic factors: The case of Korea'. Together they form a unique fingerprint.

Cite this