Band 14, Heft 4

This paper introduces skew-normal (SN) mixture and Markov-switching (MS) GARCH processes for capturing the skewness in the distribution of stock returns. The model class is motivated by the fact that the common way of incorporating asymmetries into Gaussian MS GARCH models, i.e., regime-dependent means, leads to autocorrelated raw returns, which may not be desirable. The appearance of the SN distribution can be explained by a pre-asymptotic behavior of daily stock returns, and can still be viewed as "generic." The dynamic properties of the process are derived, and its in- and out-of-sample performance is compared with that of several competing models in an application to three major European stock markets over a period covering the recent financial turmoil. It turns out that parsimoniously parameterized SN mixture GARCH processes perform best overall. In particular, they outperform both a skewed t GARCH specification as well as normal mixture GARCH models with skewness generated via nonzero component means.

In this paper we propose a general framework to deal with the presence of covariate measurement error (CME) in response-based (RB) samples. Using Chesher's (1991) methodology, we obtain a small error variance approximation for the contaminated sampling distributions that characterize RB samples with CME. Then, following Chesher (2000), we develop the generalized method of moments (GMM) estimators that reduce the bias of the most well known likelihood-based estimators for RB samples which ignore the existence of CME and derive a score test to detect the presence of this type of measurement error. Our approach only requires the specification of the conditional distribution of the response variable given the latent covariates and the classical additive measurement error model assumption, the availability of information on both the marginal probability of the strata in the population and the variance of the measurement error not being essential. Monte Carlo evidence is presented which suggests that, in RB samples of moderate sizes, the bias-reduced GMM estimators perform well.

The three-regime threshold autoregressive (TAR) model with a unit root process in the middle regime and the structural breaks (SB) model with level shifts are important and popular models for studying economic theories such as purchasing power parity (PPP) and economic variables with regime shifts. These nonlinear models have a significant influence on unit root tests. It is possible that the persistence of a variable becomes higher and the power of the unit root tests develops distortions if the form of the regime shifts is misspecified. This paper investigates the ability of unit root tests to detect stationarity in misspecified SB and three-regime TAR models. We show that the misspecified SB model has high persistence, similar to an autoregressive (AR) model, when the true process is a three-regime TAR process. On the other hand, unlike the AR model, the misspecified three-regime TAR model does not have persistence, when the true process is one with SB. The results of the Monte Carlo simulations demonstrate that a unit root test performed in the misspecified three-regime TAR model has more power even under processes with SB, whereas a unit root test undertaken in the misspecified SB model does not perform well under three-regime TAR processes. We also provide some instances of applications in economic variables that would have regime shifts.

The stochastic properties of volatility in spot electricity prices are only partially understood and present substantial modelling challenges. This paper develops and applies three complementary modelling approaches in order to uncover its fundamental and behavioural drivers over time and across intra-day trading periods. First, intra-day prices are related to systematic components, including economic fundamentals, strategic and market design effects. Then, residual volatility is attributed to: i) regular, non-linear agent reactions to market fundamentals (covariates of heteroscedasticity), ii) the adaptation of price formation due to substantial agent learning (time-varying effects), and iii) the transient extreme pricing in periods of scarcity (regime-switching dynamics). We find that, i) GARCH effects diminish, when each of the above sources of volatility is accounted for, and ii) allowing for the time-varying responses of prices to fundamentals can yield more precise volatility estimates than an explicit GARCH specification.

We develop Bayesian techniques for estimation and model comparison in a novel Generalized Stochastic Unit Root (GSTUR) model. This allows us to investigate the presence of a deterministic time trend in economic series, while allowing the degree of persistence to change over time. In particular the model allows for shifts from stationarity I(0) to nonstationarity I(1) or vice versa. The empirical analysis demonstrates that the GSTUR model provides new insights on the properties of some macroeconomic time series such as stock market indices, inflation and exchange rates.

We propose a new model-based, nonlinear method for seasonally adjusting time series in a multiplicative components model. The method seeks to reduce the bias inherent in linear model-based approaches, while at the same time preserving the flexibility of parametric methods. We discuss the problem of bias and the concept of recovery, and demonstrate the favorable properties of the proposed algorithm on several synthetic series.