In China, due to decades of the 'one-child policy' and continuous rural-urban labour migration, real population aging in rural areas is increasing more quickly than in urban areas, and the labour inputs in agricultural production are becoming ever more dependent on the elderly. Using CHARLS data, we examine the effect of health on the labour supply of rural elderly people. We construct a latent health stock index (LHSI) to eliminate measurement bias and then use this one-period lagged LHSI and the Heckman two-stage and the Bourguignon-Fournier-Gurand two-stage method to deal with the simultaneous causality of health and labour decisions and sample selectivity in model estimation. The results show that, in the overall level, the labour force participation and work time of rural elderly people increase significantly with the improvement of health. These effects on the males are sharply greater than on the females and are enhanced with age. In the subdivided agricultural and non-agricultural labour supply, health improvement is positively related with labour force participation of rural elderly and brings an employment allocation from agricultural section to non-agricultural section, especially on the males. However, as the work time, these relations are insignificant and invariant with gender and age.
Keywords: employment allocation; health; labour supply; rural elderly.