Abstract
In the talk I will discuss high-dimensional canonical correlation analysis (CCA) with an emphasis on the vectors that define canonical variables. I will show that when two dimensions of data grow to infinity jointly and proportionally, the classical CCA procedure for estimating those vectors fails to deliver a consistent estimate. This provides the first result on the impossibility of identification of canonical variables in the CCA procedure when all dimensions are large. As a countermeasure, I will derive the magnitude of the estimation error, which can be used in practice to assess the precision of CCA estimates. Finally, I will show an application of the results to the analysis of cyclical vs. non-cyclical stocks.