Explore the evidence.
Understand the reach.

This study examines whether movement out of agriculture accelerates productivity convergence in West Africa. We assemble an annual country panel for 16 West African economies over 2000–2023 from the World Development Indicators, Penn World Table 11.0 and Worldwide Governance Indicators. The estimable dynamic specification contains 237 country-year observations for 11 economies because the Penn World Table human-capital series determines the common complete-case sample. A pre-specified two-step System GMM model is rejected because second-order serial correlation is significant and the Hansen test is uninformative. Inference therefore rests on a dynamic country-and-year fixed-effects model with Driscoll–Kraay standard errors, supplemented by a first-order Nickell-bias assessment and a three-sector shift-share decomposition. Labour productivity is conditionally convergent: the lag coefficient is 0.841, implying a four-year half-life, or about 9.7 years under the conservative bias approximation. Capital deepening is positive and statistically precise, while manufacturing, services and agricultural-employment shares are not separately significant. This absence of statistical precision does not imply that structural transformation is irrelevant: employment reallocation contributes positively in all 16 economies. Regionally, within-sector gains account for 75.2% of observed productivity change, reallocation for 29.3%, and the interaction term for −4.5%. Policy should combine labour mobility with capital formation and productivity upgrading inside agriculture, manufacturing and services rather than pursue sector-share targets alone.

This study examines whether movement out of agriculture accelerates productivity convergence in West Africa. We assemble an annual country panel for 16 West African economies over 2000–2023 from the World Development Indicators, Penn World Table 11.0 and Worldwide Governance Indicators. The estimable dynamic specification contains 237 country-year observations for 11 economies because the Penn World Table human-capital series determines the common complete-case sample. A pre-specified two-step System GMM model is rejected because second-order serial correlation is significant and the Hansen test is uninformative. Inference therefore rests on a dynamic country-and-year fixed-effects model with Driscoll–Kraay standard errors, supplemented by a first-order Nickell-bias assessment and a three-sector shift-share decomposition. Labour productivity is conditionally convergent: the lag coefficient is 0.841, implying a four-year half-life, or about 9.7 years under the conservative bias approximation. Capital deepening is positive and statistically precise, while manufacturing, services and agricultural-employment shares are not separately significant. This absence of statistical precision does not imply that structural transformation is irrelevant: employment reallocation contributes positively in all 16 economies. Regionally, within-sector gains account for 75.2% of observed productivity change, reallocation for 29.3%, and the interaction term for −4.5%. Policy should combine labour mobility with capital formation and productivity upgrading inside agriculture, manufacturing and services rather than pursue sector-share targets alone.