This question is about Infant health as measured by LBW:

Suppose there is a policy that would specifically target infant birth weight and you wish to assess whether this policy would substantially impact infant health. To answer this question, you set up the following regression: h; = at bw;B+ Mij (1) where u;; = Xly +a; +E;; Here hij is the underlying health of newborn j for mother i; bwij is birth weight, X ; is a vector of mother specific observable determinants of health; a; reflects mother specific unobservable determinants of health; and & are newborn specific idiosyncratic factors. Assume that birth weight and Eij are uncorrelated (note: the same assumption cannot necessarily be made of other factors in Mij). 1. Write down the relationship between your estimate (Bois) and the 'true' B?" 2. Suppose the policy you are considering targets birth weight primarily. What concerns might you have with the conclusions suggested by results from model (1)? Provide specific examples of where the analysis based on (1) would be incorrect. 3. Within the context outlined above, describe an identification strategy that would account for your concerns outlined above. In your description include the model you would run, discussion of the problem it solves, and interpretation of the estimate of B in your new model. 4. What does the literature in this area tell us of the difference between results from (1) and the model described in (c)? Does this hold for the longer-term implications of birth weight?<