Discussion of ``A Benchmark for Models of Growth and Inflation
41 Pages
English
Downloading requires you to have access to the YouScribe library
Learn all about the services we offer

Discussion of ``A Benchmark for Models of Growth and Inflation'' by M. Marcellino

-

Downloading requires you to have access to the YouScribe library
Learn all about the services we offer
41 Pages
English

Description

New developments in economic forecasting8th Bundesbank Spring ConferenceEltville, 5-6 May 2006Todd Clark: Discussion of „A benchmark for models ofgrowth and inflation“By Massimiliano Marcellinohttp://www.bundesbank.deDiscussion of “A Benchmark for Models ofGrowth and Inflation” by M. Marcellino1Todd Clark1Federal Reserve Bank of Kansas CityMay 4, 2006Paper summaryMotivation: Models with economic content are often judged bytheir forecasting performance relative to simple univariatebenchmark models.example: Smets–Wouters modelgood GDP benchmark: AR inΔlny, estimated with allavailable datagood inflation benchmark: AR inπ orΔπ, estimated withall available datarandom walk in inflationMA in the change in inflation, estimated with a rollingsample (Stock and Watson (2005))Paper summaryMotivation: Models with economic content are often judged bytheir forecasting performance relative to simple univariatebenchmark models.example: Smets–Wouters modelgood GDP benchmark: AR inΔlny, estimated with allavailable datagood inflation benchmark: AR inπ orΔπ, estimated withall available datarandom walk in inflationMA in the change in inflation, estimated with a rollingsample (Stock and Watson (2005))Paper summaryMotivation: Models with economic content are often judged bytheir forecasting performance relative to simple univariatebenchmark models.example: Smets–Wouters modelgood GDP benchmark: AR inΔlny, estimated with allavailable datagood inflation ...

Subjects

Informations

Published by
Reads 15
Language English

Exrait

New developments in economic forecasting 8th Bundesbank Spring Conference
Eltville, 5-6 May 2006
Todd Clark: Discussion of „A benchmarkfor models of growth and inflation“
By Massimiliano Marcellino
http://www.bundesbank.de
Discussion of “A Benchmark for Models of Growth and Inflation” by M. Marcellino
1
Todd Clark
1
Federal Reserve Bank of Kansas City
May 4, 2006
aPpersumamry
Motivation: Models with economic content are often judged by their forecasting performance relative to simple univariate benchmark models. example: Smets–Wouters model good GDP benchmark: AR inΔlny, estimated with all available data good inflation benchmark: AR inπorΔπ, estimated with all available data random walk in inflation MA in the change in inflation, estimated with a rolling sample (Stock and Watson (2005))
Paperusmamry
Motivation: Models with economic content are often judged by their forecasting performance relative to simple univariate benchmark models. example: Smets–Wouters model good GDP benchmark: AR inΔlny, estimated with all available data good inflation benchmark: AR inπorΔπ, estimated with all available data random walk in inflation MA in the change in inflation, estimated with a rolling sample (Stock and Watson (2005))
aPeprsumamry
Motivation: Models with economic content are often judged by their forecasting performance relative to simple univariate benchmark models. example: Smets–Wouters model good GDP benchmark: AR inΔlny, estimated with all available data good inflation benchmark: AR inπorΔπ, estimated with all available data random walk in inflation MA in the change in inflation, estimated with a rolling sample (Stock and Watson (2005))
aPeprsumamry
Motivation: Models with economic content are often judged by their forecasting performance relative to simple univariate benchmark models. example: Smets–Wouters model good GDP benchmark: AR inΔlny, estimated with all available data good inflation benchmark: AR inπorΔπ, estimated with all available data random walk in inflation MA in the change in inflation, estimated with a rolling sample (Stock and Watson (2005))
aPeprusmmary
Motivation: Models with economic content are often judged by their forecasting performance relative to simple univariate benchmark models. example: Smets–Wouters model good GDP benchmark: AR inΔlny, estimated with all available data good inflation benchmark: AR inπorΔπ, estimated with all available data random walk in inflation MA in the change in inflation, estimated with a rolling sample (Stock and Watson (2005))
aPeprusmmary,continued
First question of this paper: Within the realm of univariate models, are these indeed good benchmark models? Is it best to difference the data or use levels? Does it help to include linear trend terms? Are the answers sensitive to the use of real time data?
Second question: Are previous results on the predictive content of economic variables robust to using stronger benchmarks? predictive content of the yield curve for output (Ang, et al. (2006)) predictive content of Phillips curves (Stock and Watson (1999))
aPperusmamr,yocntinued
First question of this paper: Within the realm of univariate models, are these indeed good benchmark models? Is it best to difference the data or use levels? Does it help to include linear trend terms? Are the answers sensitive to the use of real time data?
Second question: Are previous results on the predictive content of economic variables robust to using stronger benchmarks? predictive content of the yield curve for output (Ang, et al. (2006)) predictive content of Phillips curves (Stock and Watson (1999))
Paperusmamr,yoctnnieud
Key findings, based on U.S. data:
Some evidence suggests alternative linear models to be more accurate than the conventional benchmarks. GDP: An AR in levels with a linear trend beats an AR in growth rates. But few differences in forecast performance are statistically significant. In some prominent examples, using stronger benchmarks overturns evidence of predictive content. predictive content of the yield curve for GDP growth (Ang, et al. (2006))
Papersummary,continued
Key findings, based on U.S. data:
Some evidence suggests alternative linear models to be more accurate than the conventional benchmarks. GDP: An AR in levels with a linear trend beats an AR in growth rates. But few differences in forecast performance are statistically significant. In some prominent examples, using stronger benchmarks overturns evidence of predictive content. predictive content of the yield curve for GDP growth (Ang, et al. (2006))