Carrier aggregation (CA) is an important feature of next-generation cellular networks where multiple component carriers from contiguous or non-contiguous frequency bands are combined to increase the effective transmission bandwidth. In this work, we consider a system which has 2 non-contiguous bands separated by nearly 1 GHz so that the path loss exponent in each band is significantly different. We then derive the coverage probability and average rate in frequency reuse δ by considering two Schemes for CA. Essentially, Scheme 1 uses the carrier with higher path loss exponent for the cell-centre users, and the carrier with the lower path loss exponent for the cell-edge users, and Scheme 2 is the other way. We show that at low SNR for the cell-edge users corresponding to carrier having higher path loss exponent, Scheme 1 may be preferred over Scheme 2. However, at high SNR for the cell-edge users corresponding to carrier having higher path loss exponent, Scheme 2 may be preferred over Scheme 1.