Does investment lead to greater output? A panel error-correction model analysis

In this study we attempt to determine the direction of causality between fixed investment spending and output. A literature has been developed which examines this causal structure. The results have been anything but consistent. De Long and Summers (1991) find strong evidence of the traditional causal relationship from investment to output growth. But Blomstrom et al. (1996) find just the opposite. One problem with these studies is that each fails to recognize the implied long-run relationship between fixed investment and output such that a Vector Autoregressive Representation (VAR) in these variables should be modelled as an error correction mechanism. We employ panel cointegration techniques on data from the Penn World Tables 6.1 covering 98 countries over 40 years. Our results suggest that investment and output are cointegrated and that the direction of causality generally runs in both directions. However, we find that in African- and low-income nations the direction of causality runs from output to investment, supportive of the findings by Blomstrom et al. (1996). This pattern is found in the short run as well as in the long run.