讲座一：
后疫情时代的通货膨胀预测：三种机器学习方法的比较
Lecture 1: Forecasting inflation in the aftermath of Covid19: A comparison of three machine learning methods
时间：2025年6月6日 14：00-15：30
地点：大学城校区验楼C203
报告人：Paul McNelis教授
讲座一内容
本研究衡量三种机器学习的方法在进行covid19前中后时期的通货膨胀率的预测方面相对于线性模型的表现有何种优势，以及哪些变量对于通货膨胀的变化影响相对而言更加重要。我们使用了一个相当广泛的变量集合，可以将这些变量划分为本国宏观经济指标，外部宏观经济指标，以及危机指标，这些危机指标中包括了与covid19有关的感染率以及死亡率。本研究针对菲律宾展开，并正在扩展至其他东南亚经济体。
Lecture 1:
This paper evaluates the performance of three Machine Learning methods, relative to the benchmark linear model, for both forecasting accuracy as well as for assessing the relative importance of key variables driving inflation before, during and in the aftermath of the COVID19. We make use of a wide set of variables, which can be grouped into domestic macroeconomic indicators, external macroeconomic indicators, and crisis variables, which include COVID19 infections and mortality rates for the Philippines.
讲座二：
利用大型开放经济模型衡量日本的宏观经济政策表现
Lecture 2: A large open economy model to assess Macroeconomic policy performance in Japan
时间：2025年6月13日 14:00-15:30
地点：大学城校区实验楼C203
报告人：Paul McNelis教授
讲座二内容
本文开发并估计了一个动态随机一般均衡（DSGE）模型，该模型反映了日本经济的几个关键特征，即：一个大型开放经济体、财政赤字巨大且国内居民持有的债务不断增加，以及贸易商品市场中的垄断竞争定价。该经济体系受到一系列反复出现的冲击驱动，包括生产率冲击、政府支出冲击、资本质量冲击、外国利率冲击、国内消费冲击、全球需求冲击以及出口和进口加价定价冲击。本文比较了三种简单规则在降低银行系统持有的政府债务方面的效果：消费税率的最优简单规则、带有负利率的泰勒规则（因此不受零利率下限约束）和量化宽松（QE）规则。这三种规则在减轻负面冲击的不利影响方面表现良好。量化宽松规则在减少债务方面具有相对优势，而税率规则在不利时期稳定消费方面更具优势。这两种政策规则的效果与在一个无零利率下限、价格粘性的泰勒规则世界中获得的结果非常接近。
Lecture 2:
This paper develops and estimates a dynamic stochastic general equilibrium (DSGE) model representing several key characteristics of Japan, namely, a large open economy, large fiscal deficits with increasing amounts of debt held by domestic residents, as well as monopolistically-competitive pricing in traded goods. The economy is driven by recurring sets of shocks to productivity, government spending, quality of capital, foreign interest rates, domestic consumption, global demand and export and import markup price setting. The paper compares optimal simple rules for consumption tax rates, a Taylor rule with negative interest rates (and thus not subject to the zero lower bound) and a quantitative easing rule, for reducing government debt held by the banking system. All three rules do very well in mitigating the adverse effects of negative shocks. QE rules have a comparative advantage for reducing debt, while tax-rate rules have the advantage for stabilizing consumption in adverse periods. Both of these policy rules are close to the results prevailing in a Taylor-rule world with no zero bound, with sticky prices.
报告人介绍
Paul McNelis教授，现任Boston College 经济系访问教授，在Journal of Monetary Economics, Review of Economics and Statistics, Journal of Applied Econometrics, Journal of Economics Dynamics and Contro, Journal of International Money and Finance, Journal of Financial Stability等期刊发表论文多篇。研究领域为计算宏观经济学。
图文：金融学院
编辑：屈凡恺
初审（校）：来特
复审（校）：邓超 张璇
终审（校）：徐昶斌
