Crandi

Pdf High Dimensionalforecastinginthepresenceof Unitrootsandcointegrati

Leo Migdal
-
pdf high dimensionalforecastinginthepresenceof unitrootsandcointegrati

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic Research output: Working paper / Preprint › Preprint arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community?

Learn more about arXivLabs. We investigate how the possible presence of unit roots and cointegration affects forecasting with Big Data. As most macroeoconomic time series are very persistent and may contain unit roots, a proper handling of unit roots and cointegration is of paramount importance for macroeconomic forecasting. The high-dimensional nature of Big Data complicates the analysis of unit roots and cointegration in two ways. First, transformations to stationarity require performing many unit root tests, increasing room for errors in the classification. Second, modelling unit roots and cointegration directly is more difficult, as standard high-dimensional techniques such as factor models and penalized regression are not directly applicable to (co)integrated data and need to be adapted.

We provide an overview of both issues and review methods proposed to address these issues. These methods are also illustrated with two empirical applications. Keywords: high-dimensional time series, forecasting, unit roots, cointegration, factor models, penalized regression. We investigate forecasting with Big Data when the series in the dataset may contain unit roots and be cointegrated. As most macroeoconomic time series are at least very persistent, and may contain unit roots, a proper handling of unit roots and cointegration is of paramount importance in macroeconomic forecasting. The theory of unit roots and cointegration in small systems is well-developed and numerous reference works exist to guide the practitioner, see for example Enders (2008) or Hamilton (1994) for comprehensive treatments.

We discuss the problems that arise when extending the analysis to high-dimensional data and consider solutions that have been proposed in the literature. In particular, we discuss the applicability of the proposed methods for macroeconomic forecasting, reviewing relevant theoretical properties and practical issues. Moreover, by considering two big data applications —that are very different in spirit— we illustrate the issues and analyze the performance of the various methods in practically relevant situations. The empirical literature dealing with unit roots and cointegration can essentially be split into two different philosophies. The first approach is to apply an appropriate transformation to each series such that one can work with stationary time series, with the most common transformation taking first differences of a series with a... This is the most common approach in high-dimensional forecasting, as it only involves “straightforward” unit root or stationarity testing on each series.

Indeed, commonly used Big Data such as the FRED-MD and -QD datasets (McCracken and Ng, 2016) already come with pre-determined transformation codes to achieve stationarity. While this approach appears to be conceptually simple, we will argue that there are apparently minor issues that are often ignored in practice, but which can have a big impact on the performance of...

People Also Search

Research Output: Chapter In Book/Report/Conference Proceeding › Chapter › Academic

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic Research output: Working paper / Preprint › Preprint arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data ...

Learn More About ArXivLabs. We Investigate How The Possible Presence

Learn more about arXivLabs. We investigate how the possible presence of unit roots and cointegration affects forecasting with Big Data. As most macroeoconomic time series are very persistent and may contain unit roots, a proper handling of unit roots and cointegration is of paramount importance for macroeconomic forecasting. The high-dimensional nature of Big Data complicates the analysis of unit ...

We Provide An Overview Of Both Issues And Review Methods

We provide an overview of both issues and review methods proposed to address these issues. These methods are also illustrated with two empirical applications. Keywords: high-dimensional time series, forecasting, unit roots, cointegration, factor models, penalized regression. We investigate forecasting with Big Data when the series in the dataset may contain unit roots and be cointegrated. As most ...

We Discuss The Problems That Arise When Extending The Analysis

We discuss the problems that arise when extending the analysis to high-dimensional data and consider solutions that have been proposed in the literature. In particular, we discuss the applicability of the proposed methods for macroeconomic forecasting, reviewing relevant theoretical properties and practical issues. Moreover, by considering two big data applications —that are very different in spir...

Indeed, Commonly Used Big Data Such As The FRED-MD And

Indeed, commonly used Big Data such as the FRED-MD and -QD datasets (McCracken and Ng, 2016) already come with pre-determined transformation codes to achieve stationarity. While this approach appears to be conceptually simple, we will argue that there are apparently minor issues that are often ignored in practice, but which can have a big impact on the performance of...