By Irving Wang
Traditional asset allocation methods for modeling the tradeo between risk and return do not fully reect empirical distributions. Thus, recent research has moved away from assumptions of normality to account for risk by looking at fat tails and asymmetric distributions. Other studies have also considered multiple period frameworks to include asset rebalancing. We investigate the use of rebalancing with fat tail distributions and optimizing with downside risk as a consideration. Our results verify the underperformance of traditional methods in the single period framework and also demonstrate the underperformance of traditional methods in a multiple period rebalancing
Advisor: Aino Levonmaa, Emma Rasiel
By Abhinay Sawant
Extreme Value Theory (EVT) is one of the most commonly applied models in financial risk management for estimating the Value at Risk of a portfolio. However, the EVT model is practical for estimation only when data is independent and identically distributed, which usually does not characterize financial returns data. This paper aims to modify this model by using high-frequency data to standardize financial returns by their realized volatility and then tests the modified model with recent equity data. The results from the paper show an improvement in the EVT model when forward volatility can be properly forecasted.
Advisor: George Tauchen
By Matthew Roqnile
We investigate the properties of several nonparametric tests for jumps in financial markets. We derive a theoretical property of these tests not observed in any of the previous literature: when they are applied to finitely sampled data, they are generally biased toward finding too many jumps. This results from bias in finite-sample estimation of several important test components. The severity of the bias corresponds to the magnitude of change in volatility over the course of a day. We use data on
an intraday volatility pattern in several US equities, which results in quantitatively significant changes in the level of volatility during the day, to undertake Monte Carlo simulations of a price process without jumps. Applying several jump tests to the simulated data, we detect one-half to two-thirds as many jumps as in the observed data, suggesting that many jumps currently detected in empirical applications of these tests are spurious. We also present several possible modifications to jump tests that limit the effect of intraday patterns in volatility, all of which produce dramatically lower estimates of the frequency and importance of jumps.
Advisor: George Tauchen
By Monitra Mohinchai
The increasing flows of internal migrants resulted from urbanization in developing countries is of great interest to policy makers. This study examines the individual-level and household-level social surveys the Nang Rong Project in 1994-1995 and 2000-2001. Individual characteristics such as gender, age, and years of schooling, and household characteristic such as family size are, significantly and consistently with the human capital model and previous empirical studies, shown to be determinants of a migration decisions. Moreover, migration selectivity differs significantly by migrant destinations. These findings indicate that policy makers should also consider different destination choice of migration, as well as the migrants’ characteristics, when they try to influence migration patterns and flows.
Advisor: Frank Sloan
By Sam Lim
This paper aims to explore how “earnings surprise”—the difference between earnings
estimates and the actual announced earnings—affects a stock’s volatility and returns using
high frequency data. The results show that earnings surprise is significantly correlated with volatility and overnight returns. Furthermore, an earnings surprise is significantly correlated with an increase in volatility in the trading period immediately following the earnings announcement, but there is no bias indicating which directions prices will go. Even with no “surprise”, the announcement tends to be followed by this increase in volatility. The findings suggest the importance of earnings on equity price valuation.
Advisor: George Tauchen, Tim Bollerslev
By Andrew Kindman
This research proposes and tests several novel strategies for enhancing the strength of conventional, signal-based currency crisis Early Warning Systems (EWS). Using country level, monthly macroeconomic time-series data, it develops an algorithmic process for identifying periods of elevated currency crisis risk and achieves robust results. The proposed changes to current EWS include: 1) an adjustment to the process by which crises are identified empirically, 2) the addition of control panels to dampen the prevalence of false positives, 3) the addition of inter-temporal interaction terms that strive to bring the forecasting model in line with contemporary theoretical models of currency crisis, and 4) the addition of an algorithm for controlling post-crisis bias in macroeconomic trends. In out-of-sample, post-estimation analysis, the system is able to identify 75% of crisis incidents while generating false positives at a rate of less than 20%. Currency crisis EWS have challenged economists for some time, and though these results are not directly comparable to current EWS based on differences in reporting strategies, they are strong enough to warrant further investigation, particularly for applicability as policy instruments.
Advisor: Charles Becker, Kent Kimbrough
By Brian Humphrey
This paper employs OLS regressions to determine whether Google search query data improves national and local existing home sales forecasts. The local dataset features metropolitan statistical area data from Texas. Initially, the national and local regressions are estimated without macroeconomic variables. Macroeconomic variables are subsequently included in order to determine if Google search queries provide information not already present in the macroeconomic variables. The impact of the Google variables is assessed using root mean squared error, p-values, and adjusted r-squared values. Finally, the top models are compared using out-of-sample testing. Both the in-sample and out-of-sample test results suggest that Google search query data improves national and local existing home sales forecasts.
Advisor: James Roberts
By Ying-te Huang
Essentially all US recessions have been preceded by oil price shocks and subsequently tighter monetary policies. (Bernanke, Gertler and Watson, 1997). Whereas some scholars, including Herrera and Hamilton (2001) claimed that such oil price shocks contributed to the recession that followed, others, including, Bernanke et al. (1997), believed that the Fed‘s endogenous reaction to the monetary policy, rather than oil price per se, led to the contraction of the economy. Which had a greater influence on gross domestic product (GDP) — oil price shocks or a change in monetary policy—has been debated for years. One of the most prominent debates is between Bernanke et al. (1997), and Herrera and Hamilton (2001). In the debate, Bernanke et al. and Herrera and Hamilton used the same model but with different lag lengths and came to different conclusions. In the current study, we contribute to the resolution of this issue by using a new methodology to examine the effects of monetary policy to the economy in response to oil price shocks. Specifically, we determine the contemporaneous causal order empirically in structural vector-autoregression (SVAR). We then examine the economic responses in counterfactual schemes where the Fed does not respond to the oil price shocks. Contrary to Bernanke et al.‘s finding, in which the economy would have done better had the Fed not held its interest rate constant during an oil price shock, we found that the Fed‘s response generates higher output but a less steady price level. This suggests that the results are dependent upon prior assumptions of the model specifications.
Advisor: Kevin Hoover