Using time-stamped survey responses to measure expectations at a daily frequency

This article addresses one of the drawbacks of survey-based measures of expectations, the fact that updates are relatively infrequent, due to the monthly or quarterly frequency of survey waves. To obtain a more frequent measure, I present a state space method that uses time-stamped survey responses...

Full description

Saved in:
Bibliographic Details
Published in:International journal of forecasting 2016-04, Vol.32 (2), p.271-282
Main Author: Mokinski, Frieder
Format: Article
Language:English
Subjects:
Attitude surveys
Disaggregation
Economic crisis
Economic indicators
Expectations
Forecasting techniques
Kalman filter
Responses
State space models
Studies
Surveys
Time series
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This article addresses one of the drawbacks of survey-based measures of expectations, the fact that updates are relatively infrequent, due to the monthly or quarterly frequency of survey waves. To obtain a more frequent measure, I present a state space method that uses time-stamped survey responses as its daily measurements and the population distribution of expectations as the latent state. Augmenting the method with financial variables that measure expectations indirectly (e.g., bond yields) facilitates measurement at times when no responses are observed. An application to a survey of German financial experts shows that my method is successful for forecasting future responses. Additional analyses show that its daily estimates react in economically plausible ways to both major events, such as the September 2008 crash of Lehman Brothers, and regular releases of economic indicators.
ISSN:0169-2070
1872-8200
DOI:10.1016/j.ijforecast.2015.06.004