Academic Commons

Theses Doctoral

Essays in Macro-Finance and International Finance

Liu, Yu

This dissertation contains three essays on macro-finance and international finance.
In Chapter 1, Richard Clarida and I study the term structure of US interest rates using observable macro factors as inputs to a Taylor-type rule that can account for the time path of the short term interest rate. Using a standard essentially affine model, we build directly on the pioneering work of Ang and Piazzesi, Rudebusch and Wu, and others but extend their analysis to a framework in which all macro factors are observable. We focus on the period since 1997 when US inflation expectations have been well anchored and inflation indexed bonds - which provide useful information on expected inflation and expected future real interest rates - have been issued by the US government. In contrast to many previous studies that - of necessity - focus on earlier periods when low frequency movements in expected inflation appeared to dominate, in our sample variation in expected inflation at longer horizons is modest and the yield curve is importantly driven by the evolution of the `neutral' real policy rate as estimated by Laubach and Williams. Deviations of the policy rate from the Taylor rule path are found to have a marked impact at the front end of the yield curve. In any factor model yields are linear combinations of factors, and principal components, are linear combinations of yields. In our model, we can solve explicitly for the mapping from macro factors to traditional `level' `slope' and `curvature' factors. Our model exhibits surprising robustness in a post-crisis out-sample study. We also propose a novel, but simple regression based approach to generate initial values - required to implement the non-linear GMM estimation technique we use - for the affine model's deep structural parameters.
In both Chapters 2 and 3, I study the portfolio problem associated with currency carry trade. In Chapter 2 specifically, I analyze the carry trade threshold portfolios. I prove that under general assumptions, the optimal mean-variance portfolio gives a higher weight to carry trade having larger forward premium. I then proposes a more robust version of the mean-variance optimal portfolio: the threshold portfolio, where I construct carry trade threshold portfolios using thresholds that depend upon forward premium. And I show that empirically, up to the optimal threshold value, higher-threshold portfolios outperform lower-threshold portfolios. The financial performance then decreases, as the threshold goes higher. I model the threshold effect in a random-walk model of exchange rates. The model predicts the optimal threshold value and the relative gain of an optimal threshold portfolio. The model is calibrated, and the predictions are tested. I also discuss the threshold effect in a model which features global risk factor. Following Jurek (2014) and using crash-hedged portfolios, I test the crash risk explanation for outperformance of threshold portfolio, I show that the crash risk premium can explain around 25 percent of the excess performance of higher threshold portfolios.
In Chapter 3, I study the hedging problem associated with currency carry trade. I propose theoretical frameworks and divide hedging instruments into three categories: insurance, technical rule, and the market neutral strategy. I then propose and empirically test four hedging strategies: FX options strategy, VIX future strategy, "Stop-loss" rule and CTA strategy. Based upon empirical evidence from 2000 to 2012, I find that CTA is the preferred hedging strategy because it upgrades both return and volatility. The stop-loss strategy reduces risk but fails to affect return. Both the currency options strategy and the VIX future strategy offer good hedges against tail risk, while also reducing volatility. Unfortunately they are costly to implement. I also compare the VIX strategy to various currency option strategies, to determine if VIX is a cheaper form of systematic insurance as compared to the currency options. With respect to CTA, I study its risk-return aspect, I also provide a new methodology for replicating the returns of the benchmark CTA index.



  • thumnail for Liu_columbia_0054D_12583.pdf Liu_columbia_0054D_12583.pdf binary/octet-stream 2.23 MB Download File

More About This Work

Academic Units
Thesis Advisors
Clarida, Richard H.
Ph.D., Columbia University
Published Here
May 7, 2015