We investigate the dynamics of a closed-loop make-to-stock system consisting of a remanufacturer and a manufacturer with a first order vector auto-regressive demand and return process. Remanufactured products are considered as-good-as-new and stocked to partially satisfy market demand; newly manufactured products make up the remainder. The manufacturer and the remanufacturer cooperate to minimize the sum of the capacity costs at the remanufacturing and manufacturing processes, a finished goods inventory holding and backlog cost, and the cost to dispose returned items that are not re-manufactured. The remanufacturer, after receiving the returned products, inspects them in a triage process. Only a predetermined proportion (the triage yield) of the returns is re-manufactured. The triage yield is a decision variable. The impact of the triage yield on the system-wide cost is investigated. From the sustainability viewpoint, the ideal triage yield rate is obviously 100%. In addition, if the system is cost-conscious and the unit cost of remanufacturing is lower than that for manufacturing, it seems reasonable to conjecture that higher triage yield rates lead to lower system-wide costs. Surprisingly, it is shown that the system-wide cost is always convex in the triage yield rate, suggesting that system-wide costs could increase in the triage yield rate, even when the unit remanufacturing cost is lower than the manufacturing cost. This yield paradox is caused by higher yields causing the variability of the remanufacturing to increase. Boundary conditions are investigated. To overcome this yield paradox, we reveal that larger expected returns, lower unit remanufacturing costs, and higher disposal costs mitigate the paradox.