What matters is that the productivity measure is a similarly good proxy for true productivity over time. If so, then the trend change is likely real.