Abstract
Existing cointegration tests for the savings-investment model are limited because of low testing power. In this paper the savings-investment correlation is re-examined using a panel cointegration test by which the power seems to be improved greatly. A cointegration relationship is obtained between the two variables with panel data, and the savings retention coefficient is far from zero. This seems to be consistent with results based on traditional regression methods. Thus, it is concluded that cointegration techniques do not provide the solution to the savings-investment puzzle.