Abstract We show that when real-wage flexibility is defined in terms of cointegration of the product real-wage and productivity, the wage formation system in Finnish manufacturing maintains considerable real-wage flexibility. We develop an empirical error correction model of wages and show that the responsiveness of wages to changes in unemployment is much less important in explaining real-wage flexibility than existing studies suggest. For example, we find hysteresis effects, which are usually associated with real-wage rigidity rather than flexibility. This puzzle is resolved by pointing out that theoretically hysteresis is only a necessary condition for rigidity. Empirically, strong equilibriating mechanisms, including labor market policies, have induced considerable wage flexibility even in the presence of hysteresis. Our conclusions are substantiated by tests of parameter invariance and encompassing.