Testing Growth Ratios via Pooled Error Correction Models

The balanced growth and stochastic growth theory implies stable investment-output and consumption-output ratios. Our analysis supports cointegration between investment and output (IO-model) as well as consumption and output (CO-model) for single countries. Pooling the data we find cointegration and that output is weakly exogenous in the IO-specification. Weak exogeneity of output is not confirmed for the CO-specification on the pooled level. The growth ratio restrictions, however, are rejected. A small simulation study investigates the empirical properties of the employed panel tests in finite samples. It is found that the adopted bootstrap approach to determine critical values for common test statistics outperform the corresponding approximation.