Central banks aim for a policy of shrinking their balance sheets (QT) that is as neutral as possible to the setting of their policy rate. But the identification of a neutral path for QT hinges upon the choice of an operating framework for implementing monetary policy in steady state. The framework determines whether banks’ liquid reserves are ample or scarce, and this uniquely pins down the size of the central bank balance sheet in the long run. We focus on the euro area and identify a baseline path that embodies current expectations ？ our neutrality yardstick ？ and simulate paths that converge to two polar forms of implementation frameworks: a floor system and a ceiling system. In the former, the central bank deliberately creates a large amount of non-borrowed reserves. In the latter, banks borrow as much as they need from the central bank. Using bank-level and loan-by-loan data, we find an empirically robust positive connection between non-borrowed reserves and loans, whereas no such connection exists between loans and borrowed reserves. By our neutrality score, the floor option outperforms the ceiling option as it generates a more moderate loan contraction relative to baseline. But the baseline convergence to a quantity of non-borrowed reserves consistent with the Friedman rule ？ just saturating the system ？ is preferable. We conclude that a neutral QT should run at the expected pace until mid-2026. Then, the balance sheet should start growing again to finance the secular growth in the demand for central bank liabilities.