The Long Run Shift-Share: Modeling the Sources of Metropolitan Sectoral Fluctuations.

ABSTRACT: Attempts to analyze the sources of regional-sectoral fluctuations have been reductions of a VAR that have simplified either the VAR’s orthogonalization or long-run multiplier matrix. These shift-share decompositions are shown to be largely rejected by the data, mainly because they fail to properly account for local supply shocks. However the violence done to the data is more severe in short run than in the long, making the long run shift share a viable structure. ACKNOWLEDGMENTS: This paper was written for inclusion in a volume celebrating the career of Robert Engle. Comments welcome. My thanks go to Mark Watson for the invitation, and to participants at presentations at the 2006 Regional Science Association and the Federal Reserve Banks of New York and St. Louis for helpful comments. 11. In tro d u c tio n In this tribute to the career of Robert Engle, attention should be given to an aspect of his early career that is not universally recognized, that of urban and regional economist. As related in his interview with Diebold (2003), upon arriving at MIT Engle was asked by Franklin Fisher and Jerome Rothenberg to collaborate on the construction of a multi-equation structural model of the Massachusetts economy, and this led to a number of publications at the outset of his career. Hi