Econ Papers Daily / 期刊论文
预测加拿大经济衰退：一种自回归概率模型方法
Forecasting recessions in Canada: An autoregressive probit model approach
摘要
We forecast recessions in Canada using an autoregressive probit model. Its autoregressive structure captures the persistence of the binary recession indicator, smooths forecasted probabilities and reduces false‐positive signals. Leveraging a monthly dataset from 1973 to 2022, we estimate the model using composite likelihood methods. We also perform a variable‐selection procedure on numerous macrofinancial variables. Our findings suggest that the autoregressive model improves forecasting performance relative to a range of probit models proposed in the Canadian literature. Our results also highlight the reliability of US economic activity measures, commodity prices and financial indicators in forecasting Canadian recessions.
中文摘要
我们使用自回归概率模型预测加拿大的经济衰退。该模型的自回归结构捕捉了二元衰退指标的持续性，平滑了预测概率并减少了假阳性信号。利用1973年至2022年的月度数据集，我们采用复合似然方法估计模型。我们还对众多宏观金融变量进行了变量选择程序。我们的研究结果表明，相对于加拿大文献中提出的一系列概率模型，自回归模型提高了预测性能。我们的结果还强调了美国经济活动指标、商品价格和金融指标在预测加拿大经济衰退中的可靠性。