Micro versus macro cointegration in heterogeneous panels

• Published in: Econometrics 2010-03, Vol.155 (1), p.1-18
• Main Authors: Trapani, Lorenzo, Urga, Giovanni
• Format: Article
• Language: English
• Subjects: Aggregate analysis; Aggregation; Applications; Approximation; Asymptotic methods; Bootstrap; Bootstrap mechanism; Bootstrap method; Cointegration; Cointegration analysis; Distribution theory; Econometrics; Estimation; Exact sciences and technology; Heterogeneous panels; Heterogeneous panels Aggregation Cointegration Spurious regression Bootstrap; Insurance, economics, finance; Mathematics; Monte Carlo simulation; Nonparametric inference; Parametric inference; Probability and statistics; Regression analysis; Sciences and techniques of general use; Spurious regression; Statistics; Studies
• ISSN: 0304-4076; 2225-1146; 1872-6895
• DOI: 10.1016/j.jeconom.2009.07.005

We consider the issue of cross-sectional aggregation in nonstationary and heterogeneous panels where each unit cointegrates. We derive asymptotic properties of the aggregate estimate, and necessary and sufficient conditions for cointegration to hold in the aggregate relationship. We then analyze the case when cointegration does not carry through the aggregation process, and we investigate whether the violation of the formal conditions for perfect aggregation can still lead to an aggregate equation that is observationally equivalent to a cointegrated relationship. We derive a measure of the degree of noncointegration of the aggregate relationship and we explore its asymptotic properties. We propose a valid bootstrap approximation of the test. A Monte Carlo exercise evaluates size and power properties of the bootstrap test.