Date of Award

Spring 2021

Degree Type

Open Access Dissertation

Degree Name

Economics, PhD

Advisor/Supervisor/Committee Chair

Pierangelo DePace

Dissertation or Thesis Committee Member

Thomas Willet

Dissertation or Thesis Committee Member

Clemens Kownatzki

Dissertation or Thesis Committee Member

Hisam Sabouni


Volatility in financial markets make forecasting, or in other words estimating what will happen in the future, a difficult task. Too often forecasts are made but hardly ever revisited to see how accurate the forecast was and if not, why? The three chapters of my dissertation are focused on examining volatility in financial markets from changes in investors' trading behavior as well as studying the characteristics of forecast error of various financial securities. Often, the accuracy of these forecasts rely on the estimates made for future volatility.In my first chapter, we\footnote{This is joint work with Nasser Khalil, Clemens Kownatzki and Hisam Sabouni} analyze the predictive power of the Black-Scholes-Merton (BSM) model on a data set of options on the SPDR S&P 500 Trust ETF (SPY). We leverage the full options chain to analyze the full forecasted distribution of prices through N(d2), which we compare to the distribution of prices of SPY. Using non-parametric GOF tests, such as the Kolmogorov Smirnov and Anderson-Darling tests, we are able to analyze whether two different distributions come from the same underlying population distribution. We find that BSM tends to overestimate the tails in the implied probability distribution when further away in expiration, compared to the empirical price path of SPY. The resulting comparison gives way to visualizing and testing the ability for the BSM to predict the likelihood of options expiring in-the-money. Our findings suggest the BSM, in most cases, correctly estimates the underlying risk adjusted probabilities only a few days out from expiration, which may be attributed to the uncertainty in traders to foresee market movements until an option is close to expiration. However, this behavior is more pronounced during crisis periods, where the BSM tends to correctly estimate the likelihood of tail events occurring more often than during periods of market normalcy.In my second chapter, my co-authors and I\footnote{This is joint work with Hisam Sabouni} study the characteristics of error in economic forecasts over time. We focus on explaining the variation errors of the survey of professional economic forecasts (SPF) across three financial securities by isolating the effects of changes in fiscal and monetary policy as well as changes in various macroeconomic indicators. We examine if it is changes in government policy or changes in macroeconomic indicators (or market conditions) that is primarily responsible for increases in SPF forecast error. We use a principal component analysis to first perform orthogonal dimension reduction of our macroeconomic indicator variables and use the first two principal components as an overall measure for market conditions. We then use a linear regression to test whether market conditions or monetary and/or fiscal policy is primarily responsible for increases in SPF forecast error of three securities' yields: the three month Treasury bill, Moody's AAA corporate bond and the Ten year Treasury bill. We find increases in monetary policy via the EFFR affects the short-term security in our analysis to a large magnitude, but increases in overall market conditions affect all securities in our analysis to a smaller but significant degree. In my third chapter, I explore an anomaly that exists in the U.S. equity market that has not been documented before; investors' reactions to earnings announcements are not only asymmetric, but seasonal. Knowing which months experience larger variation than others, investors may incorporate financial derivatives such an options to hedge downside risk. Using a fixed effects linear regression, I first examine the effect an earnings beat and earnings miss have on abnormal returns; which are calculated by a CAPM-GARCH model. I find an earnings beat on average has large significant increases in firms' abnormal returns while an earnings miss, or a negative earnings surprise, has limited downside impacts. Examining this effect further at the month level, I find investors'reactions are extremely large to earnings beats announced in months June and to earnings misses announced in December compared to other months.