Aggregation in Large Dynamic Panels
M. Hashem Pesaran
University of Southern California; Cambridge University - Faculty of Economics; CESifo (Center for Economic Studies and Ifo Institute for Economic Research); Institute for the Study of Labor (IZA)
University of Cambridge
IZA Discussion Paper No. 5478
This paper considers the problem of aggregation in the case of large linear dynamic panels, where each micro unit is potentially related to all other micro units, and where micro innovations are allowed to be cross sectionally dependent. Following Pesaran (2003), an optimal aggregate function is derived, and the limiting behavior of the aggregation error is investigated as N (the number of cross section units) increases. Certain distributional features of micro parameters are also identified from the aggregate function. The paper then establishes Granger's (1980) conjecture regarding the long memory properties of aggregate variables from 'a very large scale dynamic, econometric model', and considers the time profiles of the effects of macro and micro shocks on the aggregate and disaggregate variables. Some of these findings are illustrated in Monte Carlo experiments, where we also study the estimation of the aggregate effects of micro and macro shocks. The paper concludes with an empirical application to consumer price inflation in Germany, France and Italy, and re-examines the extent to which ﾑobservedﾒ inflation persistence at the aggregate level is due to aggregation and/or common unobserved factors. Our findings suggest that dynamic heterogeneity as well as persistent common factors are needed for explaining the observed persistence of the aggregate inflation.
Number of Pages in PDF File: 56
Keywords: aggregation, large dynamic panels, long memory, weak and strong cross section dependence, VAR models, impulse responses, factor models, inflation persistence
JEL Classification: C43, E31working papers series
Date posted: February 21, 2011
© 2013 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollo4 in 0.500 seconds