Is Fixed Investment the Key to Economic Growth?
Is Fixed Investment the Key to Economic Growth? was written by Magnus Blomström, Robert E. Lipsey and Mario Zejan. It first emerged as a working paper in the National Bureau of Economic Research. It was later published in the Quarterly Journal of Economics volume 111 (1996) (DOI: https://doi.org/10.2307/2946665).
The authors group annual estimates into periods of 5 year. The measurements are average productivity growth (in GDP per capita) and investment ("fixed capital formation") in each period for each country.
The authors regress productivity growth on capital formation in previous, current, and successive periods; first split by reference period, then pooled altogether. The authors also re-run the pooled model with normalized variables; every variable is divided by the average value across the periods being used. This is meant to capture country effects, without incorporating 100+ dummy indicators. In each model, they find the greatest correlations between productivity growth and investment in the successive period.
The authors regress productivity growth in period t on productivity growth from periods t-1 and t-2; they then regress productivity growth in period t on productivity growth from periods t-1 and t-2 and investment from period t-1. The investment term does not add explanatory power to the model and is not significant.
Lastly, the authors reverse the causality test. They regress investment from period t on investment from periods t-1 and t-2; they then regress investment from period t on investment from periods t-1 and t-2 and productivity growth from period t-1. Both models feature a very high R2. The productivity growth term does add explanatory power to the model and is significant.
Altogether, investment does not cause productivity growth. Instead productivity growth causes more investment.