Current business cycle models systematically underestimate the correlation between consumption and investment. One reason for this failure is that a positive investment-specific technology shock generally induces a negative consumption response. The objective of this paper is to investigate whether positive consumption responses to investment-specific technology shocks can be obtained in a modern business cycle model. We find that the answer to this question is yes. With a combination of nominal rigidities and non-separable preferences, the consumption response is positive for general parameterisations of the model.