This paper addresses the issue of estimating and forecasting productivity growth trends in the US and Germany from the perspective of a business cycle researcher who wants to use the available information in time series of aggregate labor productivity to derive a model for short- and/or long-term forecasts of labour productivity. We will use stability tests and a deterministic model with structural breaks that is estimated using the methods mentioned in Hansen (2001). The methodological approach also draws on Gordon (2003) using a Kalman filter specification. We discuss the implications of unit-root assumptions for long-term forecasts and argue in favor of a near unit-root modelling. That implies a convergence of productivity growth rates in both countries within the next 15 years.