The Forcasting and High-Dimensional Data Analysis Group (FHDDA)
New issues related to forecasting in macroeconomics and finance.
have emerged as larger and richer data sets have become available. The FHDDA Group at CIRANO has taken up two lines of research:
We address the gains in forecast accuracy attainable in "data-rich" environments where the number of series available
for use in a forecast may number in
the dozens or hundreds. Research has focused
on useful ways to summarize their information in dimensionally-reduced
models such as factor models, or by combining models by techniques such as Bayesian model averaging.
We have investigated the potential
effects on our models, estimation procedures, and evaluations caused by
the use of preliminary data which are later revised. "Real-time" data analysis
studies the revisions to measured data across different "vintages" of
series, and the impact on our economic analyses.
The past three years CIRANO has hosted a "Real-time" conference. We will continue to do so.
This WEB page summarizes the work undertaken at CIRANO in these area and provides links to conferences and relevant data sources. Please contact us with additional information that may be of interest to the research community.