A Comparison of Approaches to Select the Informativeness of Priors in BVARs

Vector autoregressions (VARs) are richly parameterized time series models that can capture complex dynamic interrelationships among macroeconomic variables. However, in small samples the rich parametrization of VAR models may come at the cost of overfitting the data, possibly leading to imprecise inference for key quantities of interest such as impulse response functions (IRFs). Bayesian VARs (BVARs) can use prior information to shrink the model parameters, potentially avoiding such overfitting. We provide a simulation study to compare, in terms of the frequentist properties of the estimates of the IRFs, useful strategies to select the informativeness of the prior. The study reveals that prior information may help to obtain more precise estimates of impulse response functions than classical OLS-estimated VARs and more accurate coverage rates of error bands in small samples. Strategies based on selecting the prior hyperparameters of the BVAR building on empirical or hierarchical modeling perform particularly well.

Modeling and Extrapolating Arctic Feedback Loops using Macroeconometric Techniques

<p>The minimum extent of arctic sea ice (SIE) in 2019 ranked second-to-lowest in history and is trending downward. Hence, there is an immediate need for flexible statistical modeling approaches that both explain endogenously the trend of SIE and permits its extrapolation to generate a long-run forecast. To that end, we propose the VARCTIC, which is a Vector Autoregression (VAR) specifically designed to capture and extrapolate feedback loops that characterize the Arctic system.  VARs are dynamic simultaneous systems of equations routinely estimated in economics to predict and understand the interactions of multiple macroeconomic time series. The VARCTIC is a compromise between fully structural/deterministic modeling and purely statistical approaches that usually offer little explanation of the underlying mechanism. Our "business as usual" completely unconditional forecast has September SIE hitting 0 around the middle of the century. By studying the impulse response functions of Bayesian VARs including different sets of variables, we single out CO2 shocks as main drivers of the long-run evolution of SIE. Additionally, we document that the corresponding responses of Sea Ice Albedo and Thickness largely amplify the long-run impact of CO2 on SIE.  Finally, we conduct conditional forecasts analysis of remedies like reducing CO2 emissions or the implementation of Albedo-enhancing Geo-Engineering technologies.</p>

Macroeconomic uncertainty and forecasting macroeconomic aggregates

Can information on macroeconomic uncertainty improve the forecast accuracy for key macroeconomic time series for the US? Since previous studies have demonstrated that the link between the real economy and uncertainty is subject to nonlinearities, I assess the predictive power of macroeconomic uncertainty in both linear and nonlinear Bayesian VARs. For the latter, I use a threshold VAR that allows for regime-dependent dynamics conditional on the level of the uncertainty measure. I find that the predictive power of macroeconomic uncertainty in the linear VAR is negligible. In contrast, using information on macroeconomic uncertainty in a threshold VAR can significantly improve the accuracy of short-term point and density forecasts, especially in the presence of high uncertainty.