We develop a three-sector Harris-Todaro (HT)-type model for a developing economy to analyse the failure of the employment generation program in the rural sector as a policy response to counter Covid-19-induced reverse migration of labour. First, we distinguish between two alternative modes of implementation of the program and show how they produce diverse outcomes on both unemployment and welfare. Welfare is measured in terms of Sen's (Sen, J Public Econ 4:387-403, 1974) index which is inequality (distributional) sensitive. If the authority's sole target is to lower unemployment, the policy fails miserably because it not only raises unemployment but also worsens welfare. On the contrary, if the policy aims at both rural infrastructure development and additional employment creation, it does not necessarily worsen social welfare. However, the rural employment generation program may turn out to be counterproductive in the latter case provided the urban region is relatively more labour abundant vis-a-vis the entire rural region. Numerical examples are constructed to validate the sufficient and necessary condition leading to the counterproductive outcome. Finally, we have advocated in favour of a composite policy that might succeed in minimizing further possible damage of the COVID-19 disaster.