Analysing Econometric Bias and Non-linearity in Returns toEducation of Pakistan

This study estimates the returns to education whilecontrolling endogeneity and sample selection biases in Pakistan, over atime period using Ordinary Least Square (OLS), simultaneous approachusing both Heckman Sample Selection and Instrumental Variable, and FixedEffect techniques. Household Integrated Economic Survey (HIES) data for2004-05 and 2011-12 time periods have been used in this study. Thereturns to education have been found downward biased in OLS estimatesfor both time periods. The unbiased real returns to education haveincreased on average for wage workers over time period. Landholding andNon-earned income have been used as exclusion restrictions to controlfor sample selection bias in the Heckman Sample Selection technique. Theendogeneity bias has been controlled for with the help of parentaleducation as instrument in Instrumental Variable technique. Bothtechniques have also been used collectively or simultaneously to getmore efficient estimate in simultaneous approach. Household Fixed Effecttechnique has also been used with the assumption that ability and familycharacteristics largely remain same within family or household. Theincrease in the unbiased and real returns to education shows thatprofitability still exists in investing in education whereas experiencevia skill enhancement reinforces this rise in wage. Sadly, the historicgender and regional discriminations persist or aggravate in wage market.Married persons are getting more in returns relative to the unmarriedindividuals. Having negative implications for income inequality,Convexity in education-earning relationship in Pakistan has beenconfirmed by Indicator Function technique for both time periods. Loweducation prompt low-earning workers who would be unable to bear theschooling cost of their children. This seriously inhibits earningpotential making income inequality worse. JEL Classification: I26, I24,J24 Keywords: Returns to Education, Human Capital