An increase in energy-cost can induce energy effciency improvement - a reduction in energy-output ratio. There are well-established theoretical conjectures of how this can take place. As the relative energy-cost increases, it induces firms to reallocate and selectively utilize the most energy-effcient vintages. In the long-run firms could also achieve energy effciency through investments in energy-effcient capital. This study uses the Canadian KLEMS panel data set to investigate these relationships. We employ panel vector auto regressions as well as co-integration and error correction techniques to test whether the conjectures hold in the data. Our findings support the theoretical conjectures. The channels we empirically identify suggest that the effect of increased energy-cost can be an increase in energy effciency: by decreasing energy-capital ratio and increasing output-capital ratio. The latter effect is observed only in the long-run through induced investments in new capital.