Value at Risk (vaR) is a financial widely used risk measure to manage and control risk market. Th... more Value at Risk (vaR) is a financial widely used risk measure to manage and control risk market. The measure risk estimation can be done by methods such as econometric approach and Theory of Extreme Values (TVE). However, these methods have some weaknesses such as inappropriate assumption of erros distribution, rsulting in inaccurate measures. Therefore, in this article we discuss and compare Auregressive Conditional Value at Risk (CAViaR) and Expectil Autoregressive Condiotional (CARE) models. We apply the proposed modeling approach to calculate the Value at Risk (VaR) and Expected Shortfall (ES) to the stock IBOVESOPA market index
O objetivo desse artigo é introduzir estimadores suavizados de cópulas via ondaletas para o caso ... more O objetivo desse artigo é introduzir estimadores suavizados de cópulas via ondaletas para o caso de séries temporais. As propriedades dos estimadores são avaliadas por meio de simulações e seu desempenho comparado com outros estimadores. Aplicações a dados reais também são feitas
In this paper, we consider estimating copulas for time series, under mixing conditions, using wav... more In this paper, we consider estimating copulas for time series, under mixing conditions, using wavelet expansions. The proposed estimators are based on estimators of densities and distribution functions. Some statistical properties of the estimators are derived and their performance assessed via simulations. Empirical applications to real data are also given.
ABSTRACT Stationarity has always played a major role in time series analysis. The statistical tec... more ABSTRACT Stationarity has always played a major role in time series analysis. The statistical techniques for stationary processes, based on the spectral analysis or parametric models, are well developed and are often employed. But in many applications the assumption of stationarity fails to be true and there exists no natural generalization from stationary to nonstationary processes. This paper reviewes some definitions of the spectrum for non-stationary processes: evolutionary spectra, time-varying spectral density, Wigner-Ville spectrum and their estimators with properties. We also present an interesting application of M. B. Priestley’s [J. R. Stat. Soc., Ser. B 27, 204–237 (1965; Zbl 0144.41001)] evolutionary spectrum: a test of non-stationarity. We perform some simulations and apply the techniques to actual series of daily mortality, air pollution and temperature in São Paulo. The results show that Priestley’s evolutionary spectra and the short-time periodogram preserve the local energy distributions over frequency and the pseudo-Wigner estimator preserves the classical frequency concept, i.e., the serie’s periodicity can be identified.
Value at Risk (vaR) is a financial widely used risk measure to manage and control risk market. Th... more Value at Risk (vaR) is a financial widely used risk measure to manage and control risk market. The measure risk estimation can be done by methods such as econometric approach and Theory of Extreme Values (TVE). However, these methods have some weaknesses such as inappropriate assumption of erros distribution, rsulting in inaccurate measures. Therefore, in this article we discuss and compare Auregressive Conditional Value at Risk (CAViaR) and Expectil Autoregressive Condiotional (CARE) models. We apply the proposed modeling approach to calculate the Value at Risk (VaR) and Expected Shortfall (ES) to the stock IBOVESOPA market index
O objetivo desse artigo é introduzir estimadores suavizados de cópulas via ondaletas para o caso ... more O objetivo desse artigo é introduzir estimadores suavizados de cópulas via ondaletas para o caso de séries temporais. As propriedades dos estimadores são avaliadas por meio de simulações e seu desempenho comparado com outros estimadores. Aplicações a dados reais também são feitas
In this paper, we consider estimating copulas for time series, under mixing conditions, using wav... more In this paper, we consider estimating copulas for time series, under mixing conditions, using wavelet expansions. The proposed estimators are based on estimators of densities and distribution functions. Some statistical properties of the estimators are derived and their performance assessed via simulations. Empirical applications to real data are also given.
ABSTRACT Stationarity has always played a major role in time series analysis. The statistical tec... more ABSTRACT Stationarity has always played a major role in time series analysis. The statistical techniques for stationary processes, based on the spectral analysis or parametric models, are well developed and are often employed. But in many applications the assumption of stationarity fails to be true and there exists no natural generalization from stationary to nonstationary processes. This paper reviewes some definitions of the spectrum for non-stationary processes: evolutionary spectra, time-varying spectral density, Wigner-Ville spectrum and their estimators with properties. We also present an interesting application of M. B. Priestley’s [J. R. Stat. Soc., Ser. B 27, 204–237 (1965; Zbl 0144.41001)] evolutionary spectrum: a test of non-stationarity. We perform some simulations and apply the techniques to actual series of daily mortality, air pollution and temperature in São Paulo. The results show that Priestley’s evolutionary spectra and the short-time periodogram preserve the local energy distributions over frequency and the pseudo-Wigner estimator preserves the classical frequency concept, i.e., the serie’s periodicity can be identified.
Uploads
Papers by Clelia Toloi