Two Essays on Fixed Income Asset Pricing: A Generalized Framework for International Carry Trades and a PCA-Based Approach to Forecasting Correlation Dynamics

This dissertation comprises two interrelated studies that advance the understanding of fixed-income strategies and risk management by exploring novel approaches to carry trades and yield curve modeling. The overarching aim is to provide new insights into cross-currency bond strategies and the prediction of correlation structures within the U.S. Treasury market, with practical implications for portfolio allocation and risk assessment.

The first paper investigates the risk premia associated with a cross-currency carry strategy that departs from traditional formulations. Rather than focusing solely on short-term interest rate differentials or currency forwards, the strategy involves borrowing through short-term bond issuance in one currency and investing in longer-duration bonds in another. A central finding is that funding in the domestic currency systematically enhances performance by sidestepping the yield-induced depreciation typically observed in weaker currencies. Furthermore, the study highlights the importance of macroeconomic conditions in shaping bond returns. Specifically, bond valuations are positively influenced in economically robust countries—not solely due to lower interest rates, but also because of the increased investor confidence and reduced uncertainty that accompany economic strength. This interplay between bond pricing and macro fundamentals offers a nuanced perspective that extends beyond standard interest rate parity and traditional currency carry frameworks, emphasizing the multi-dimensional nature of fixed-income investment strategies.

The second paper turns to the challenge of modelling and forecasting the correlation structure of the U.S. Treasury yield curve, which is vital for fixed-income portfolio optimization and risk management. Using principal component analysis (PCA) in combination with regression-based forecasting, the study focuses on the dynamic behaviour of the first three principal components derived from changes in yields. The methodology demonstrates that even partial forecasting—limited to the eigenvalues of the correlation matrix—outperforms common benchmarks such as the rolling-window martingale and the DCC-GARCH models. More notably, full forecasting of both eigenvalues and eigenvectors yields significant improvements in predictive accuracy and temporal stability. These enhancements translate into practical benefits, as demonstrated through a simulated optimal bond portfolio that achieves a higher Sharpe Ratio when using the predicted correlation matrices. The findings underscore the utility of PCA-based approaches in modelling complex term structure dynamics, offering a scalable and interpretable alternative to more computationally intensive or opaque models.

Together, these two papers contribute to the fields of international finance and fixed-income modelling by integrating empirical insights with methodological innovation. The first paper challenges conventional wisdom about carry trades by incorporating bond maturity structures and macroeconomic indicators, while the second enhances the precision of risk estimation in bond portfolios through improved correlation forecasting. Both studies demonstrate that leveraging economic context and statistical decomposition methods can yield material improvements in strategy design and risk-adjusted performance, particularly in uncertain or volatile financial environments. The dissertation, therefore, provides a holistic view of how advanced empirical techniques can inform both tactical investment decisions and long-term risk management frameworks in the fixed-income domain.