-
首页
-
- 首页
- WORKSHOP预告
宏观workshop:Comparing Pruning Methods in Perturbation DSGE Models
发布日期：2016-03-25 03:26 来源：北京大学国家发展研究院
时间：2015年3月29日（周二）13：30－15：00
地点：北大国家发展研究院 朗润园 万众楼小教室
主持人：霍德明 王敏 赵波
主讲人：HONG LAN（兰弘）题目： Comparing Pruning Methods in Perturbation DSGE Models
摘要： We study the rationale and performance of DSGE perturbations that are pruned to guarantee stable simulations. We show that the moving average representation of the policy function is naturally pruned and express the nonlinear moving average recursively. This recursive algorithm differs from pruning algorithms and the rationale provide by series expansions in that it evaluates risk at the stochastic instead of the deterministic steady state. We compare seven different pruning algorithms at second and third order, documenting the differences between these algorithms and standard (non pruned) state space perturbations at first, second, and third order in a unified notation. The nonlinear moving average is the most accurate and the series expansion the second most accurate; yet the two algorithms perform comparably, suggesting that this choice is unlikely to be a potential source of error. Alternative ad hoc algorithms from the literature suffer a loss of accuracy to varying degree as they include terms inconsistent with or neglect terms consistent with the order of approximation.
关键词： Perturbation; DSGE; nonlinear; pruning
主讲人简介：兰弘，对外经贸大学金融学院金工系。研究方向为动态随机模型的数值方法、搜寻理论及其在劳动力市场中的应用与宏观经济学。（简历见附件 cv_honglan ）
国家发展研究院官方微信
Copyright© 1994-2012 北京大学 国家发展研究院 版权所有, 京ICP备05065075号-1
保留所有权利，不经允许请勿挪用
-
首页
-
宏观workshop:Comparing Pruning Methods in Perturbation DSGE Models
发布日期：2016-03-25 03:26 来源：北京大学国家发展研究院
时间：2015年3月29日（周二）13：30－15：00
地点：北大国家发展研究院 朗润园 万众楼小教室
主持人：霍德明 王敏 赵波
主讲人：HONG LAN（兰弘）题目： Comparing Pruning Methods in Perturbation DSGE Models
摘要： We study the rationale and performance of DSGE perturbations that are pruned to guarantee stable simulations. We show that the moving average representation of the policy function is naturally pruned and express the nonlinear moving average recursively. This recursive algorithm differs from pruning algorithms and the rationale provide by series expansions in that it evaluates risk at the stochastic instead of the deterministic steady state. We compare seven different pruning algorithms at second and third order, documenting the differences between these algorithms and standard (non pruned) state space perturbations at first, second, and third order in a unified notation. The nonlinear moving average is the most accurate and the series expansion the second most accurate; yet the two algorithms perform comparably, suggesting that this choice is unlikely to be a potential source of error. Alternative ad hoc algorithms from the literature suffer a loss of accuracy to varying degree as they include terms inconsistent with or neglect terms consistent with the order of approximation.
关键词： Perturbation; DSGE; nonlinear; pruning
主讲人简介：兰弘，对外经贸大学金融学院金工系。研究方向为动态随机模型的数值方法、搜寻理论及其在劳动力市场中的应用与宏观经济学。（简历见附件 cv_honglan ）
访问学者申请
EMAIL
北京大学
ENGLISH
