search

erlris / intergen_ml

Project on intergenerational mobility as a prediction problem
1 stars 0 forks source link

Becker Tomes model #1

Open jackblun opened 6 years ago

jackblun commented 6 years ago

The goal of this issue is to track work building an extension of the Becker Tomes model to suit our setting

jackblun commented 6 years ago

There is actually quite a lot in the Becker-Tomes 1979 paper that we can build on. They write several times on 'endowments' being important. These are:

'endowments of capital that are determined by the reputation and "connections" of their families, the contribution to the ability, race, and other characteristics of children from the genetic constitutions of their families, and the learning, skills, goals, and other "family commodities" acquired through belonging to a particular family culture. Obviously, endowments depend on many characteristics of parents, grandparents, and other family members...'

Further on in the paper, they say:

'The concept of the endowment may not seem to be useful because endowments cannot readily be measured and quantified'

I think we should be able to find a way to motivate a measure using the Becker-Tomes model, and write something convincing arguing that becker and tomes themselves originally did think much broader than income. However, increasingly I think the problem is going to be justifying the use of ML in estimation, particularly in the admin data setting where the added value on just looking at R2 from OLS isn't that obvious.