Recently, Tanzania witnessed a revival of economic planning that explicitly aimed to combine rapid economic growth with accelerated structural transformation of the economy. To achieve these planning targets would require a relatively modest drop in the share of agriculture in GDP, but a dramatic fall in its share in employment by 2025. Tanzanian planners assume that labour is locked in agriculture because agricultural productivity is low, from which they conclude that, to release labour to fuel the expansion of manufacturing, it is imperative to raise agricultural productivity by appropriate land policies, leveraging private investment and developing public–private partnerships. We argue that, analytically, this planning argument leaves out the possibility that causality may run the other way – from high labour retention in agriculture to low agricultural productivity – and that, empirically, the observed patterns inherent in actual processes of economic transformation in Tanzania do not tally well with the assumptions of planners. More specifically, in so far as labour flows out of agriculture, it flows towards informal‐sector activities, both rural and urban, rather than towards formal manufacturing.