Volume 17, Issue 5

We consider the extent to which different time-series models can generate simulated data with the same business cycle features that are evident in US real GDP. We focus our analysis on whether multivariate linear models can improve on the previously documented failure of univariate linear models to replicate certain key business cycle features. We find that a particular nonlinear Markov-switching specification with an explicit “bounceback” effect continues to outperform linear models, even when the models incorporate variables such as the unemployment rate, inflation, interest rates, and the components of GDP. These results are robust to simulated data generated either using Normal disturbances or bootstrapped disturbances, as well as to allowing for a one-time structural break in the variance of shocks to real GDP growth.

A Bayesian analysis of the stochastic volatility model with regime-switching skewness in heavy-tailed errors is proposed using a generalized hyperbolic (GH) skew Student’s t -distribution. The skewness parameter is allowed to shift according to a first-order Markov switching process. We summarize Bayesian methods for model fitting and discuss analyses of exchange rate return time series. Empirical results show that interpretable regime-switching skewness can improve model fit and Value-at-Risk performance in a comparison against several other SV models with constant skewness or jump diffusions.

In this paper we model regimes and long memory in the dynamics of realized volatilities of intraday ETF and stock returns. We estimate threshold fractionally integrated (TARFIMA) models using Bayesian Markov Chain Monte Carlo (MCMC) algorithms with efficient jump. We also introduce a test based on posterior distributions of the mean squared forecast errors for model selection. Our findings are that the TARFIMA model that accounts for a different degree of long memory, persistence and variance in two regimes outperforms ARFIMA and other models using 5 day forecasts.

In this paper we estimate the single-factor Consumption Capital Asset Pricing Model (C-CAPM) over the frequency domain. We modify the standard two-step methodology (Fama and French 1992) to account for the spectral properties of consumption risk and we find that its lower frequencies explain up to 98% of the cross-sectional variation of expected returns and that the equity premium puzzle is eliminated. These results are robust to the definitions of the variables, the sample span and the set of portfolios utilized, and the maturity of interest rates.

With the adoption of a single currency and a common monetary policy, the assessment of the existence of common macroeconomic factors driving the euro area as a whole is of key importance. In this paper, we try to contribute to this debate by admitting an approximate factor model with a common component comprising global and country-specific factors. Although it is arguable to allow for common factors shared by all the euro area member countries in the context of the European economic integration, it is also reasonable to admit that there may be country-specific factors. Hence, we end up with a two-level factor model allowing for both global and country-specific factors. We discuss the consistency of the principal components estimator within such framework and propose criteria for determining the number of global and group-specific factors. The consistency of the suggested criteria is established and the small sample properties are assessed through Monte Carlo simulations. Resorting to large datasets for the four major euro area countries, we find evidence of the existence of both global and country-specific factors.

We develop hypothesis tests for binary response models estimated by maximum score. The only methods currently available for developing such tests are smoothed maximum score and subsampling. We propose a new method that uses a “discretization’’ argument to circumvent the intractable asymptotic distribution of the maximum score estimator. The resulting tests require fewer assumptions than smoothed maximum score, less computational time than subsampling and can be applied to a wide range of nested and non-nested hypotheses. The tests are based on a certain asymptotic equivalence that is obtained by discretizing the continuous covariates with the discretization becoming finer as the sample size increases. The discretization parameters (analogous to bandwidths) are specified so that discretization vanishes asymptotically which is critical for the test to be consistent. The test statistics reflect differences in the predictive accuracy of the maximum score estimates under the null and alternative hypotheses. The tests are shown to be asymptotically normal under the null, and converge to infinity under the alternative. We also investigate the size and power properties of the test through Monte Carlo simulations.