Question

Please I need today answer for This question and it is very important and I need solution for this issue with all the details , and help me with all the details.Please write your answer to me by typning  not by hand writing, so that I can read and understand your answer clearly.thanks in advance/H

Q.1) Consider a government that contemplates extending the mandatory school system from nine years to ten years in primary school. You are now asked to give recommendation about the implementation and are thinking about estimating the model:

?_i = ?₀ + ?₁ ?cℎ_i + ?_i,

where ?_i is the outcome (e.g. wage) of individual i, and ?cℎ_i is the years in school of the same individual. ?_i is an error term.

a) Describe in words why estimating this model can give rise to omitted variable bias.

b) Explain the research design that Meghir and Palme (2005) use in their paper to estimate ?₁

Answer 1

Following is the solution-

a) Omitted variable bias occurs when we are either excluding a relevant variable or underspecifying a model or the regressor is correlated to the omitted variable. In this model

Yi = b0 + b1 schi + ei

The independent variable is only the number of years in school. Thus, other determinants of Yi (wages of i) like abilities, experience are excluded. They end up being accounted in the error term which means that the Expected value of ei is not 0 which violates the Assumption of Zero conditional mean which means that the model is no longer unbiased. Thus, results in the omitted variable bias.

Additionally, there is a positive correlation between the error term and the independent variable leading to the problem of multicollinearity violating another Gauss Markov Assumptions for regression.

b) Meghir and Palme (2005) studied the effect of final education attainment on income of the individual. The sample is derived from Sweden as they implemented educational reforms such as – increasing compulsory schooling to 9 years, a unified curriculum at the national level and abolished selection based on grades in grade 6. The authors use the survey data of the children born in the years 1948 and 1953. The group of students from 1948 follow the old system and 1953 follow the new system in certain municipalities, in some both groups were assigned the old system and in other municipalities both groups followed the new system. This led in Meghir and Palme using what they called ‘differences-in-differences methodology’ which meant they compared results not just across groups but also across municipalities.

The model used by them is –

Yidm=b0 + b1di + b’2mi + αridm + ϒxidm + eidm

where Y is the income of individual i of group d living in municipality m. di is the dummy variable indicating the group/cohort to which the individual belongs, and mi is the dummy variable indicating the municipality in which the individual went to school. α is the average effect of reform, ridm is 1 for individual from new/reformed system, xidm is the observable characteristics of the students and eidm is the error term.

The unique element of this approach was being able to compare individuals working in the same field but products of two different education systems thus be able to compare the heterogenous effects of parental backgrounds. Meghir and Palme used this data to estimate the extent to which the educational reforms had a positive impact on those with unskilled parents and the effects of their own abilities. Their research had a key impact on policy analysis conducted by Heckman and others in 2000.