Development economics

The project has to be done in STATA. The result (i.e all graphs and tables) should be copy from STATA and be presented in a word document together with a written explanation of the problems. Need a answer on all questions a to j. The data set file is appended. Part 1: Wages, race, gender and education in South Africa Data set: [login to view URL] Problems: a. Calculate the sample mean, median, and standard deviation of hourly wages. Do you think the numbers obtained are sensible (i.e. not obviously completely wrong)? Why is the mean much higher than the median? b. Produce a graph showing the frequency distribution of wages for the whole sample. Comment on the shape of the distribution. c. How do the wages of Black and White South Africans differ? Are the estimated means statistically significantly different? d. How do the wages of females and males differ? Are the estimated means statistically significantly different? e. Calculate average years of schooling and work experience, by gender and race. Do you think education is an important reason why white South Africans have higher wages than black South Africans? f. You will now estimate returns to education using the Mincerian earnings function, in which the wage variable is expressed as a log-linear function of individual human capital variables. Consider the following simple specification: In wagei = a0 + a1 x educi + εi where ln is the (natural) logarithm operator, educi denotes years of schooling, εi is a residual, and a0 and a1 are unknown parameters to be estimated. The coefficient a1 is interpreted as measuring the casual effect of another year of schooling on expected wages, in percentage terms. That is, if a1 = 0.05, for example, then we would say that one additional year of schooling increases expected wages by about 5%. Obtain OLS estimates of a0 and a1 using the South African data. Is schooling a statistically and economically significant determinant of wages? How much higher is the expected wage for someone with 10 years of education, compared to someone with no education? g. Compute predicted low wages, based on the regression in (f), and produce a scatter plot with predicted and actual wages on the vertical axis, and years of schooling on the horizontal axis. h. It is often argued in the literature that one should not assign a casual interpretation to OLS estimates of a1, based on regressions of the form in (f). One reason is that there may be lots of factors – in addition to schooling – that drive wage differences, and that may be correlated with schooling. To go some way towards addressing this concern, add to the previous specification the following explanatory variables: experience, race, gender; and re-estimated the earnings function. Based on the results, would you say there is any evidence that black South Africans earn less than while South Africans, conditional on education, experience and gender? Is there any evidence that females earn less than males, conditional on education, experience and race? i. How would you investigate whether returns to education differs by gender and race? Investigate if they do differ. j. Weil in his textbook adopts the following numbers for the returns to education: 13.4% per year for the first 4 years of schooling; then 10.1% per year for the next 4 years; then 6.8% per year beyond 8 years of schooling. For these numbers to be reasonable we must have diminishing returns to education. Is this the case for South Africa?