Search

How can Surveys of Professional Forecasters (SPF) be used to improve inflation forecasts? By using US historical quarterly data on SPF forecasts, we provide better understanding of how we can use forecast disagreement to improve our own forecasts.

Statistical inference--both estimation and testing--for stochastic volatility (SV) models is known to be challenging and computationally demanding. We propose simple and efficient estimators for SV models with conditionally heavy-tailed error distributions, particularly the Student’s t and Generalized Exponential Distributions (GED). The estimators rely on a small set of moment conditions derived from ARMA-type representations of SV models, with an option to apply “winsorization” to improve stability and finite-sample performance. Except for the degrees of-freedom parameter, closed-form expressions are available for all other parameters, extending Ahsan and Dufour (2019, 2021), thus eliminating the need for numerical optimization or initial values. We derive the estimators’ asymptotic distribution and show that, due to their analytical tractability, they support reliable, and even exact, simulation-based inference via Monte Carlo or bootstrap methods. We assess their performance through extensive simulations and demonstrate their practical relevance in financial return data, which strongly reject the normality assumption in favor of heavy-tailed models.

We present a technical description of the Top-Down Solvency Assessment (TDSA) tool. As a solvency stress-testing tool, TDSA is used to assess the banking sector’s capital resilience to hypothetical future risk scenarios.

This paper studies how financial shocks shape the distribution of output growth by introducing a quantile-augmented vector autoregression (QAVAR), which integrates quantile regressions into a structural VAR framework. The QAVAR preserves standard shock identification while delivering flexible, nonparametric forecasts of conditional moments and tail risk measures for gross domestic product.

This paper provides an extensive evaluation of the performance of quantile vector autoregression (QVAR) to forecast macroeconomic risk. Generally, QVAR outperforms standard benchmark models. Moreover, QVAR and QVAR augmented with factors perform equally well. Both are adequate for modeling macroeconomic risks.

We use Bayesian predictive decision synthesis to formalize monetary policy decision-making. We develop a case-study of monetary policy decision-making of an inflation-targeting central bank using multiple models in a manner that considers decision goals, expectations and outcomes.

We forecast recessions in Canada using an autoregressive (AR) probit model. The results highlight the short-term predictive power of the US economic activity and suggest that financial indicators are reliable predictors of Canadian recessions. In addition, the suggested model meaningfully improves the ability to forecast Canadian recessions, relative to a variety of probit models proposed in the Canadian literature.

This paper studies non-parametric combinations of density forecasts. We introduce a regression tree-based approach that allows combination weights to vary on the features of the densities, time-trends or economic indicators. In two empirical applications, we show the benefits of this approach in terms of improved forecast accuracy and interpretability.

The fourth generation of Bank of Canada projection and policy analysis models seeks to improve our understanding of inflation dynamics, the supply side of the economy and the underlying risks faced by policy-makers coming from uncertainty about how the economy functions.

We apply two machine learning algorithms to forecast monthly growth of house prices and existing homes sales in Canada. Although the algorithms can sometimes outperform a linear model, the improvement in forecast accuracy is not always statistically significant.