Non-Gaussian score-driven conditionally heteroskedastic models with a macroeconomic application

We contribute to the literature on empirical macroeconomic models with time-varying conditional moments, by introducing a heteroskedastic score-driven model with Student's t-distributed innovations, named the heteroskedastic score-driven -QVAR (quasi-vector autoregressive) model. The -QVAR model is a robust nonlinear extension of the VARMA (VAR moving average) model. As an illustration, we apply the heteroskedastic -QVAR model to a dynamic stochastic general equilibrium model, for which we estimate Gaussian-ABCD and -ABCD representations. We use data on economic output, inflation, interest rate, government spending, aggregate productivity, and consumption of the USA for the period of 1954 Q3 to 2022 Q1. Due to the robustness of the heteroskedastic -QVAR model, even including the period of the coronavirus disease of 2019 (COVID-19) pandemic and the start of the Russian invasion of Ukraine, we find a superior statistical performance, lower policy-relevant dynamic effects, and a higher estimation precision of the impulse response function for US gross domestic product growth and US inflation rate, for the heteroskedastic score-driven -ABCD representation rather than for the homoskedastic Gaussian-ABCD representation. © The Author(s), 2023. Published by Cambridge University Press.

Score-driven cryptocurrency and equity portfolios

This paper discusses whether the Bitcoin exchange-traded fund (ETF), which tracks the value of Bitcoin, improves equity portfolios, by using a robust portfolio performance analysis. The equity portfolio is represented by an ETF that tracks the Standard & Poor's 500. We use data from a turbulent investment period within the coronavirus pandemic, to study the diversification benefits of Bitcoin. We compare the performances of diverse portfolios composed of both ETFs, which include 40 classical dynamic volatility model-based portfolios and 900 score-driven portfolios. For the score-driven portfolios, the dynamic association is modelled by score-driven Clayton, rotated Clayton, Gumbel, rotated Gumbel and Student's t copulas. We compare portfolio strategies using the model confidence set test. We find that score-driven portfolios outperform classical volatility model-based portfolios and the equity portfolio. Our results may provide suggestions for cryptocurrency investors on portfolio optimization and may also have policy implications for regulators and policymakers. © 2023 Informa UK Limited, trading as Taylor & Francis Group.