Skill formation under incomplete information

We model child development as a multi-stage investment problem, where the child's specific ability is unknown at the early stage of childhood and learnt at the late stage. We show that this form of incomplete information weakens the importance of early investments in children if inter-stage investments are easily substitutable, but strengthens their importance if substitution is difficult. The latter case is empirically relevant. We also derive interesting comparative statics and discuss the policy implications.