Prevendo a volatilidade realizada de ações brasileiras : evidências empíricas

Este estudo compara previsões de volatilidade de sete ações negociadas na Bovespa usando 02 diferentes modelos de volatilidade realizada e 03 de volatilidade condicional. A intenção é encontrar evidências empíricas quanto à diferença de resultados que são alcançados quando se usa modelos de volatilidade realizada e de volatilidade condicional para prever a volatilidade de ações no Brasil. O período analisado vai de 01 de Novembro de 2007 a 30 de Março de 2011. A amostra inclui dados intradiários de 5 minutos. Os estimadores de volatilidade realizada que serão considerados neste estudo são o Bi-Power Variation (BPVar), desenvolvido por Barndorff-Nielsen e Shephard (2004b), e o Realized Outlyingness Weighted Variation (ROWVar), proposto por Boudt, Croux e Laurent (2008a). Ambos são estimadores não paramétricos, e são robustos a jumps. As previsões de volatilidade realizada foram feitas através de modelos autoregressivos estimados para cada ação sobre as séries de volatilidade estimadas. Os modelos de variância condicional considerados aqui serão o GARCH(1,1), o GJR (1,1), que tem assimetrias em sua construção, e o FIGARCH-CHUNG (1,d,1), que tem memória longa. A amostra foi divida em duas; uma para o período de estimação de 01 de Novembro de 2007 a 30 de Dezembro de 2010 (779 dias de negociação) e uma para o período de validação de 03 de Janeiro de 2011 a 31 de Março de 2011 (61 dias de negociação). As previsões fora da amostra foram feitas para 1 dia a frente, e os modelos foram reestimados a cada passo, incluindo uma variável a mais na amostra depois de cada previsão. As previsões serão comparadas através do teste Diebold-Mariano e através de regressões da variância ex-post contra uma constante e a previsão. Além disto, o estudo também apresentará algumas estatísticas descritivas sobre as séries de volatilidade estimadas e sobre os erros de previsão.

Caracterização dos retornos e das volatilidades de ativos financeiros do mercado Brasileiro : uma abordagem baseada na volatilidade realizada

The goal of this paper is twofold. First, using five of the most actively traded stocks in the Brazilian financial market, this paper shows that the normality assumption commonly used in the risk management area to describe the distributions of returns standardized by volatilities is not compatible with volatilities estimated by EWMA or GARCH models. In sharp contrast, when the information contained in high frequency data is used to construct the realized volatilies measures, we attain the normality of the standardized returns, giving promise of improvements in Value at Risk statistics. We also describe the distributions of volatilities of the Brazilian stocks, showing that the distributions of volatilities are nearly lognormal. Second, we estimate a simple linear model to the log of realized volatilities that differs from the ones in other studies. The main difference is that we do not find evidence of long memory. The estimated model is compared with commonly used alternatives in an out-of-sample experiment.

Dois ensaios em finanças

In the first chapter, we test some stochastic volatility models using options on the S&P 500 index. First, we demonstrate the presence of a short time-scale, on the order of days, and a long time-scale, on the order of months, in the S&P 500 volatility process using the empirical structure function, or variogram. This result is consistent with findings of previous studies. The main contribution of our paper is to estimate the two time-scales in the volatility process simultaneously by using nonlinear weighted least-squares technique. To test the statistical significance of the rates of mean-reversion, we bootstrap pairs of residuals using the circular block bootstrap of Politis and Romano (1992). We choose the block-length according to the automatic procedure of Politis and White (2004). After that, we calculate a first-order correction to the Black-Scholes prices using three different first-order corrections: (i) a fast time scale correction; (ii) a slow time scale correction; and (iii) a multiscale (fast and slow) correction. To test the ability of our model to price options, we simulate options prices using five different specifications for the rates or mean-reversion. We did not find any evidence that these asymptotic models perform better, in terms of RMSE, than the Black-Scholes model. In the second chapter, we use Brazilian data to compute monthly idiosyncratic moments (expected skewness, realized skewness, and realized volatility) for equity returns and assess whether they are informative for the cross-section of future stock returns. Since there is evidence that lagged skewness alone does not adequately forecast skewness, we estimate a cross-sectional model of expected skewness that uses additional predictive variables. Then, we sort stocks each month according to their idiosyncratic moments, forming quintile portfolios. We find a negative relationship between higher idiosyncratic moments and next-month stock returns. The trading strategy that sells stocks in the top quintile of expected skewness and buys stocks in the bottom quintile generates a significant monthly return of about 120 basis points. Our results are robust across sample periods, portfolio weightings, and to Fama and French (1993)’s risk adjustment factors. Finally, we identify a return reversal of stocks with high idiosyncratic skewness. Specifically, stocks with high idiosyncratic skewness have high contemporaneous returns. That tends to reverse, resulting in negative abnormal returns in the following month.

Realized Matrix-Exponential Stochastic Volatility with Asymmetry, Long Memory and Spillovers

The paper develops a novel realized matrix-exponential stochastic volatility model of multivariate returns and realized covariances that incorporates asymmetry and long memory (hereafter the RMESV-ALM model). The matrix exponential transformation guarantees the positivedefiniteness of the dynamic covariance matrix. The contribution of the paper ties in with Robert Basmann’s seminal work in terms of the estimation of highly non-linear model specifications (“Causality tests and observationally equivalent representations of econometric models”, Journal of Econometrics, 1988, 39(1-2), 69–104), especially for developing tests for leverage and spillover effects in the covariance dynamics. Efficient importance sampling is used to maximize the likelihood function of RMESV-ALM, and the finite sample properties of the quasi-maximum likelihood estimator of the parameters are analysed. Using high frequency data for three US financial assets, the new model is estimated and evaluated. The forecasting performance of the new model is compared with a novel dynamic realized matrix-exponential conditional covariance model. The volatility and co-volatility spillovers are examined via the news impact curves and the impulse response functions from returns to volatility and co-volatility.

Volatility Smile Dynamics in Scenario Analysis

The objective of this paper is to suggest a method that accounts for the impact of the volatility smile dynamics when performing scenario analysis for a portfolio consisting of vanilla options. As the volatility smile is documented to change at least with the level of implied at-the-money volatility, a suitable model is here included in the calculation process of the simulated market scenarios. By constructing simple portfolios of index options and comparing the ex ante risk exposure measured using different pricing methods to realized market values, ex post, the improvements of the incorporation of the model are monitored. The analyzed examples in the study generate results that statistically support that the most accurate scenarios are those calculated using the model accounting for the dynamics of the smile. Thus, we show that the differences emanating from the volatility smile are apparent and should be accounted for and that the methodology presented herein is one suitable alternative for doing so.

Idiosyncratic volatility around information release date

Several papers document idiosyncratic volatility is time-varying and many attempts have been made to reveal whether idiosyncratic risk is priced. This research studies behavior of idiosyncratic volatility around information release dates and also its relation with return after public announcement. The results indicate that when a company discloses specific information to the market, firm’s specific volatility level shifts and short-horizon event-induced volatility vary significantly however, the category to which the announcement belongs is not important in magnitude of change. This event-induced volatility is not small in size and should not be downplayed in event studies. Moreover, this study shows stocks with higher contemporaneous realized idiosyncratic volatility earn lower return after public announcement consistent with “divergence of opinion hypothesis”. While no significant relation is found between EGARCH estimated idiosyncratic volatility and return and also between one-month lagged idiosyncratic volatility and return presumably due to significant jump around public announcement both may provide some signals regarding future idiosyncratic volatility through their correlations with contemporaneous realized idiosyncratic volatility. Finally, the study show that positive relation between return and idiosyncratic volatility based on under-diversification is inadequate to explain all different scenarios and this negative relation after public announcement may provide a useful trading rule.

ARMA Representation of Integrated and Realized Variances

This paper derives the ARMA representation of integrated and realized variances when the spot variance depends linearly on two autoregressive factors, i.e., SR SARV(2) models. This class of processes includes affine, GARCH diffusion, CEV models, as well as the eigenfunction stochastic volatility and the positive Ornstein-Uhlenbeck models. We also study the leverage effect case, the relationship between weak GARCH representation of returns and the ARMA representation of realized variances. Finally, various empirical implications of these ARMA representations are considered. We find that it is possible that some parameters of the ARMA representation are negative. Hence, the positiveness of the expected values of integrated or realized variances is not guaranteed. We also find that for some frequencies of observations, the continuous time model parameters may be weakly or not identified through the ARMA representation of realized variances.

Do jumps matter for volatility forecasting? Evidence from energy markets

This paper characterizes the dynamics of jumps and analyzes their importance for volatility forecasting. Using high-frequency data on four prominent energy markets, we perform a model-free decomposition of realized variance into its continuous and discontinuous components. We find strong evidence of jumps in energy markets between 2007 and 2012. We then investigate the importance of jumps for volatility forecasting. To this end, we estimate and analyze the predictive ability of several Heterogenous Autoregressive (HAR) models that explicitly capture the dynamics of jumps. Conducting extensive in-sample and out-of-sample analyses, we establish that explicitly modeling jumps does not significantly improve forecast accuracy. Our results are broadly consistent across our four energy markets, forecasting horizons, and loss functions

Modeling and predicting the CBOE market volatility index

This paper performs a thorough statistical examination of the time-series properties of the daily market volatility index (VIX) from the Chicago Board Options Exchange (CBOE). The motivation lies not only on the widespread consensus that the VIX is a barometer of the overall market sentiment as to what concerns investors' risk appetite, but also on the fact that there are many trading strategies that rely on the VIX index for hedging and speculative purposes. Preliminary analysis suggests that the VIX index displays long-range dependence. This is well in line with the strong empirical evidence in the literature supporting long memory in both options-implied and realized variances. We thus resort to both parametric and semiparametric heterogeneous autoregressive (HAR) processes for modeling and forecasting purposes. Our main ndings are as follows. First, we con rm the evidence in the literature that there is a negative relationship between the VIX index and the S&P 500 index return as well as a positive contemporaneous link with the volume of the S&P 500 index. Second, the term spread has a slightly negative long-run impact in the VIX index, when possible multicollinearity and endogeneity are controlled for. Finally, we cannot reject the linearity of the above relationships, neither in sample nor out of sample. As for the latter, we actually show that it is pretty hard to beat the pure HAR process because of the very persistent nature of the VIX index.

Measuring, Modeling, and Forecasting Volatility and Correlations from High-Frequency Data

This dissertation contains four essays that all share a common purpose: developing new methodologies to exploit the potential of high-frequency data for the measurement, modeling and forecasting of financial assets volatility and correlations. The first two chapters provide useful tools for univariate applications while the last two chapters develop multivariate methodologies. In chapter 1, we introduce a new class of univariate volatility models named FloGARCH models. FloGARCH models provide a parsimonious joint model for low frequency returns and realized measures, and are sufficiently flexible to capture long memory as well as asymmetries related to leverage effects. We analyze the performances of the models in a realistic numerical study and on the basis of a data set composed of 65 equities. Using more than 10 years of high-frequency transactions, we document significant statistical gains related to the FloGARCH models in terms of in-sample fit, out-of-sample fit and forecasting accuracy compared to classical and Realized GARCH models. In chapter 2, using 12 years of high-frequency transactions for 55 U.S. stocks, we argue that combining low-frequency exogenous economic indicators with high-frequency financial data improves the ability of conditionally heteroskedastic models to forecast the volatility of returns, their full multi-step ahead conditional distribution and the multi-period Value-at-Risk. Using a refined version of the Realized LGARCH model allowing for time-varying intercept and implemented with realized kernels, we document that nominal corporate profits and term spreads have strong long-run predictive ability and generate accurate risk measures forecasts over long-horizon. The results are based on several loss functions and tests, including the Model Confidence Set. Chapter 3 is a joint work with David Veredas. We study the class of disentangled realized estimators for the integrated covariance matrix of Brownian semimartingales with finite activity jumps. These estimators separate correlations and volatilities. We analyze different combinations of quantile- and median-based realized volatilities, and four estimators of realized correlations with three synchronization schemes. Their finite sample properties are studied under four data generating processes, in presence, or not, of microstructure noise, and under synchronous and asynchronous trading. The main finding is that the pre-averaged version of disentangled estimators based on Gaussian ranks (for the correlations) and median deviations (for the volatilities) provide a precise, computationally efficient, and easy alternative to measure integrated covariances on the basis of noisy and asynchronous prices. Along these lines, a minimum variance portfolio application shows the superiority of this disentangled realized estimator in terms of numerous performance metrics. Chapter 4 is co-authored with Niels S. Hansen, Asger Lunde and Kasper V. Olesen, all affiliated with CREATES at Aarhus University. We propose to use the Realized Beta GARCH model to exploit the potential of high-frequency data in commodity markets. The model produces high quality forecasts of pairwise correlations between commodities which can be used to construct a composite covariance matrix. We evaluate the quality of this matrix in a portfolio context and compare it to models used in the industry. We demonstrate significant economic gains in a realistic setting including short selling constraints and transaction costs.

Particle emissions, volatility, and toxicity from an ethanol fumigated compression ignition engine

Particle emissions, volatility, and the concentration of reactive oxygen species (ROS) were investigated for a pre-Euro I compression ignition engine to study the potential health impacts of employing ethanol fumigation technology. Engine testing was performed in two separate experimental campaigns with most testing performed at intermediate speed with four different load settings and various ethanol substitutions. A scanning mobility particle sizer (SMPS) was used to determine particle size distributions, a volatilization tandem differential mobility analyzer (V-TDMA) was used to explore particle volatility, and a new profluorescent nitroxide probe, BPEAnit, was used to investigate the potential toxicity of particles. The greatest particulate mass reduction was achieved with ethanol fumigation at full load, which contributed to the formation of a nucleation mode. Ethanol fumigation increased the volatility of particles by coating the particles with organic material or by making extra organic material available as an external mixture. In addition, the particle-related ROS concentrations increased with ethanol fumigation and were associated with the formation of a nucleation mode. The smaller particles, the increased volatility, and the increase in potential particle toxicity with ethanol fumigation may provide a substantial barrier for the uptake of fumigation technology using ethanol as a supplementary fuel.

Are combination forecasts of S&P 500 volatility statistically superior?

Forecasting volatility has received a great deal of research attention, with the relative performances of econometric model based and option implied volatility forecasts often being considered. While many studies find that implied volatility is the pre-ferred approach, a number of issues remain unresolved, including the relative merit of combining forecasts and whether the relative performances of various forecasts are statistically different. By utilising recent econometric advances, this paper considers whether combination forecasts of S&P 500 volatility are statistically superior to a wide range of model based forecasts and implied volatility. It is found that a combination of model based forecasts is the dominant approach, indicating that the implied volatility cannot simply be viewed as a combination of various model based forecasts. Therefore, while often viewed as a superior volatility forecast, the implied volatility is in fact an inferior forecast of S&P 500 volatility relative to model-based forecasts.

Yield-factor volatility models

The term structure of interest rates is often summarized using a handful of yield factors that capture shifts in the shape of the yield curve. In this paper, we develop a comprehensive model for volatility dynamics in the level, slope, and curvature of the yield curve that simultaneously includes level and GARCH effects along with regime shifts. We show that the level of the short rate is useful in modeling the volatility of the three yield factors and that there are significant GARCH effects present even after including a level effect. Further, we find that allowing for regime shifts in the factor volatilities dramatically improves the model’s fit and strengthens the level effect. We also show that a regime-switching model with level and GARCH effects provides the best out-of-sample forecasting performance of yield volatility. We argue that the auxiliary models often used to estimate term structure models with simulation-based estimation techniques should be consistent with the main features of the yield curve that are identified by our model.