How misleading is linearization? Evaluating the dynamics of the neoclassical growth model

This paper investigates the reliability of employing linearization to evaluate the dynamic adjustments to changes in productive government spending in a Ramsey growth model. If government expenditure is introduced as a flow and the dynamic adjustment is fast, linearization may yield a reasonably good approximation to the true dynamics, even for fairly large policy shocks. If government expenditure assumes the form of a stock, leading to more sluggish adjustment, linearization may yield misleading predictions. These errors occur at the beginning of the transition and weigh heavily in welfare calculations. The implications for temporary shocks and the speed of convergence are also considered.