Duncan McClements (@modelthinking)

Thanks for the reflections - I think they should be able to explain the results! Schwarze and Wilkelmann's paper tracks generational transfers after individuals have left their parents' homes: our confusion referred more to the surprisingly high level of consumption of children while they are much younger. If parental altruism fell with children's age then this would offer a way to reconcile the estimates as well: parents choose their number of children based on future utility, but choose consumption of themselves and their children in each year based on utility maximisation in that year. This hypothesis seems plausible as the Barro-Becker parameter you estimate once discounting is included is 0.15, compared to the about 0.085 (averaged over men and women) in their regression with full controls instruments etc: so assuming constant child consumption over their lifetime, 2% discount, 80 years of lifetime of which 20 are childhood this gives a coefficient of 0.25, which once some other effects are accounted for (greater concavity than log, scale economies in households, etc) should give an estimate for the current consumption of children as a fraction of their parents that is not too implausible.