Abstract This study explores the long‐run impact of idiosyncratic and common shocks on industry output in Ghana while controlling for the effects of investment. In order to deal with the second‐order bias problem, this study employed canonical cointegration and fully modified ordinary least‐squares (OLS) regressions, which are more robust to second‐order bias problems. Different models are, therefore, specified and estimated. Fully modifiedOLSand canonical cointegration are extended in successive steps in order to verify if the inclusion of idiosyncratic and common shocks improves the statistical properties of the model. Secondly, a backward approach, in which idiosyncratic and common shocks are excluded successively, is also adopted. Preliminary findings showed signs of long‐run equilibrium. The a priori expectation and the statistical importance of investment are established in both fully modified OLS and canonical cointegration models. This result is robust using both theBartlett andParzen kernels. However, while the elasticity value for investment is invariant to the model and kernel type used for fully modifiedOLS, the opposite result is found for canonical cointegration. Importantly, the absolute value of the investment elasticity is kept within the limits of 0 and 1. The impacts of idiosyncratic and common shocks are negative and statistically significant in the long run for both fully‐modifiedOLSand canonical cointegration. This result is robust to theBartlett andParzen kernels. Result based on the fully modifiedOLSalso showed that the sizes of the elasticity values for both idiosyncratic and common shocks are sensitive to the model type and kernel type used. Despite the differences in the elasticity values, result for both models are qualitatively similar.
Stored only when both papers are in the corpus — 27 citations exist upstream.