In contrast to the Balassa–Samuelson hypothesis, many fast‐growing Asian countries have experienced little trend real exchange rate appreciation, or even depreciation. Moreover, their long‐run real exchange rate trend seems to be dominated by movements in traded goods prices. A model is developed which is consistent with these observations. As in the Balassa–Samuelson model, productivity growth is concentrated in the traded goods sector. Nevertheless the real exchange rate may exhibit trend depreciation, driven by persistent deviations in the price of traded goods from those in the reference country. The key feature of the model is the presence of endogenous productivity growth in the distribution services sector.