There appears to be a disconnect between the importance of the zero bound on nominal interest rates in the real-world and predictions from quantitative DSGE models. Recent economic events have reinforced the relevance of the zero bound for monetary policy whereas quantitative models suggest that the zero bound does not constrain (optimal) monetary policy. This paper attempts to shed some light on this disconnect by studying a broader range of shocks within a standard DSGE model. Without denying the possibility of other factors, we find that risk premium shocks are key to building quantitative models where the zero bound is relevant for monetary policy design. The risk premium mechanism operates by increasing the spread between the rates of return on private capital and risk-free government bonds. Other common shocks, such as aggregate productivity, investment-specific productivity, government spending and money demand shocks, are unable to push nominal bond rates close to zero as the same risk premium spread mechanism is not at play.