Search

Using the Panel Study of Income Dynamics, I find that stockholders subject to more negative skewness in earnings growth hold a lower share of their financial wealth in stocks. Moreover, I also find that skewness in earnings growth affects the mean and skewness in consumption growth, and this effect is stronger for stockholders than nonstockholders. Using a life-cycle model incorporating business cycle variation in expected growth and skewness in earnings shocks, I investigate these relationships from an asset allocation perspective. During expansions (recessions), households consume more (less), and also invest a higher (lower) share of their wealth in the stock market, because of a higher (lower) expected future earnings growth rate. Negative skewness in the earnings process during recessions further reduces households' stock market exposure and consumption. The model shows how countercyclical skewness in earnings shocks is extremely important to match quantitatively observed portfolio choice and wealth accumulation over the life cycle, while simultaneously leading to countercyclical skewness in consumption growth. Moreover, the quantitative predictions are consistent with evidence from aggregate Flow of Funds data and match the observed degree of wealth inequality in the U.S.

I study questions related to risk premia in international equity markets and broad asset classes, and the role of financial intermediaries – market makers (dealers) and institutional investors – in shaping the price dynamics. This thesis consists of four essays. In the first essay, we study the importance of foreign institutional investors in the capital allocation process worldwide. We find that stocks that are held more by foreign institutional investors have more informative prices unconditionally and conditional on the same level of local institutional ownership. Foreign investors contribute to price informativeness about as much as local investors, especially for stocks in non-U.S. countries. The magnitude of the effect is robust to various endogeneity concerns. In the second essay, we use a unique transaction-level dataset on over-the-counter foreign exchange derivatives – forwards, swaps and cross-currency swaps – to study the failure of textbook no-arbitrage condition in FX markets, the covered interest rate parity (CIP). Empirically, we show that the newly introduced regulatory requirement on the balance sheet contributes to the violation of CIP. We solve the endogeneity issue by using an exogenous variation arising from the implementation of UK leverage ratio framework. In the third essay, we construct a strategy that buys securities with low past overnight returns and sell securities with high past overnight returns generates sizeable out-of-sample excess returns and Sharpe ratios. This strategy outperforms the conventional short-term reversal strategy for major international equity markets and futures written on equity indices, interest rates, commodities, and currencies. We find that the cross-sectional return volatility explains the returns from this strategy consistent with time-varying limits to arbitrage. In contrast, traditional risk factors fail to price these excess returns. In the last essay, we carry out the first cross-country analysis of the correlation risk premium. We examine the statistical properties of the implied and realized correlation in European equity markets and relate the resulting premium to the US equity market correlation risk and a global correlation risk factor. We find evidence of strong co-movement of correlation risk premia in European and US equity markets. The results support the hypothesis that a global correlation risk factor exists and that it is priced in international equity option markets. Finally, we document the relationship between the correlation risk premium on macroeconomic policy uncertainty and other uncertainty measures.

This thesis aims to explore the impact of media content on credit risk by using the Credit Default Swap (CDS) and to investigate the Chinese real estate market. Chapter 1 provides the motivation and a detailed summary of this thesis. Chapter 2 focuses on the impact of government bailout news on systemic risk during the European sovereign debt crisis. A market-based systemic risk measure is proposed. The empirical results suggest that crisis interventions conducted by the European Central Bank (ECB) help to stabilise the financial and sovereign sectors, and the bailout actions from the International Monetary Fund (IMF) have a dominant effect on the non-financial sector. Chapter 3 explores the impact of media content on sovereign credit risk by using a news sentiment variable from the Thomson Reuters News Analytics database. The results show that such media tone contains both noise and new information, and has a significant influence on the sovereign CDS returns. Chapter 4 studies the impact of news sentiment on a firm's credit and equity risks. A firm-specific news sentiment is constructed via performing the linguistic analysis on news articles published by the Wall Street Journal. The equity market shows consistent superior reactions to the media sentiment. Furthermore, the impact of media content concentrates on the U.S. financial crisis period. The empirical findings support explanations related to the investor inattention theory. Chapter 5 shifts the focus to the Chinese real estate market and examines the relationship between credit supply and house prices. The financial deregulation process of China in opening up its local currency business to foreign banks is used to identify the impact of foreign credit. This deregulation expands the household credit supply and increases house prices. Meanwhile, provinces with higher participation rates of foreign banks experience larger house price depreciations during the financial crisis.

The first chapter provides a brief overview of the theoretical grounding of the so-called realized-measures and summarizes their statistical properties. Motivated by the superior statistical accuracy of high-frequency measures I test if they add economic value in a low-frequency investment context. I compare the profitability of the same investment strategy against two implementations of its trading signals: one that conventionally uses daily returns (LF) and the other that takes advantage of high-frequency (HF) returns. I use different lengths of the formation period to verify if the HF implementation can leverage the superior amount of observations. Although economic differences favour the HF implementation, the evidence is not statistically significant. Nonetheless, the HF implementation is more robust to the choice of parameters and provides, for the most illiquid stocks, strong economic benefits that are inversely increasing in the length of the formation period. The second chapter moves the focus onto the intraday level and seeks to establish the existence of an intraday momentum effect. Motivated by limited evidence of intraday pre- dictability both in the cross-section of US stock returns (see Heston, Korajczyk, and Sadka, 2010) and in the time-series of the aggregate stock market (see Gao et al., 2015), I reconsider the time-series dimension using all common US stocks from 1993 until 2010 and, building on this, I present the cross-sectional dimension with new and complimentary evidence. I find that statistical time-series predictability does not imply economic profitability, whereas cross-sectional sorts on past performance see stocks, which lost or won the most in the morning, earn (positive returns) above the rest of the cross-section in the afternoon, and especially during the last half-hour of trading. The effect is robust to stock characteristics, the day-of-week effect, variations in the formation and holding peri- ods, but exhibits some dependence on the sample period, suggesting that specific market mechanisms or frictions play a relevant role on intraday price formation. The third chapter looks at some claims about price acceleration constituting an informative trading signal. We build several empirical measures of acceleration and compare them to other traditional equity signals from the academic literature. We find that buying stocks whose returns are decelerating and shorting stocks whose returns are accelerating, produces a wide spread in returns. Moreover, while these profits are not explained by the state-of-the-art equity factor models, the cross-sectional variation of the average returns of portfolios sorted on the acceleration signals, are reconciled once we add our la5 factor – a simple reversal strategy with a lookback of one week. Taken together, our results cast doubt on acceleration being a separate phenomenon and suggest that the lookback period in trending strategies has been shrinking over time.

In this thesis we look to advance our understanding of how asset prices fluctuate and why. In the first chapter we provide an extensive survey on the time-series and cross-sectional variation in average returns on currency carry trade strategies. We then identify a research niche that examines the relation between uncertainty, as measured by survey forecasts' dispersion, and the exchange rates. Subsequently, in the second chapter we find that forecast disagreements on key macro variables -- economic activity growth, short-term interest rates and current account positions -- help explain a considerable part of the time variation in FX volatility, and even allow FX volatility forecasts. We continue exploring the FX market in the third chapter, which studies empirically the relation between macro uncertainty shocks and the cross-section of currency excess returns. We measure uncertainty over macro variables such as current account, inflation rate, short-term interest rate, real economic growth and foreign exchange rate using the cross-sectional dispersion of market participants' expectations from two international surveys of economic forecasts. We find evidence that typical investment currencies deliver low returns whereas funding currencies offer a hedge when current account uncertainty is unexpectedly high. In contrast, uncertainty over other macro indicators displays no significant relation with the cross-section of currency excess returns. We also find that current account uncertainty subsumes the information content of foreign exchange rate volatility risk. The forth and final chapter looks at some claims made about the price acceleration giving an informative trading signal. We built several empirical measures of acceleration and compared them to the price signals from the previous academic literature on the equity market. We do find that acceleration signal leads to the construction of the long-short (LMH) trading strategies that deliver high profits. Moreover, those profits are not explained by the state-of-the-art equity factor models. However, all of the LMH profitability and cross-sectional variation in portfolios' mean returns are reconciled once we add our la5 factor - a simple reversal strategy with a one trading week lookback horizon and one month holding period - to the off-the-shelf models. Taken together, our results cast doubt on acceleration being a separate phenomenon.

This thesis presents a model to price convertible bonds. It is the first model to my knowledge that combines a stock price tree calibrated to the implied volatility surface with an interest rate model of the users choice and a probability of default as a proxy for credit risk. The aim was to develop a pricing model which enables security pricing for hybrid derivatives with equity, interest and default risk, using observable market inputs from the equity and bond markets. The model gives the user the flexibility to choose any interest rate model they desire. Normally convertible bond models implemented on a finite difference grid or a 2 factor 3-D tree are restricted to Markovian interest rate models which can be implemented via a recombining lattice. The latest advances in interest rate modelling in the form of multi-factor HJM and Libor Market Models, that are now becoming increasing popular by practitioners, however tend to be non Markovian. The implementation of these models is restricted to inefficient non-recombining lattices/trees or Monte Carlo simulations.