哈佛大学计量经济学考试题-期中试题.pdf

Harvard University Economics 1123 Department of Economics Spring 2006 Midterm – Suggested Solutions 1. (24 pts) Table 2 reports the regression coefficients and standard error estimates (in brackets) 2 for the specifi cation of the regression with and without the variable IQ. Also included is the R . Answer the following questions based on these results. (a) (4 pts) Using regression (1), what does the value 0.072 in the education row mean in practical terms. The coefficient estimate of 0.072 tells us that an extra year of education (since education is measured in years) is associated with an increase in wages of 7.2%. This means for a four year degree the effect is estimated to be 28.8% higher wages than for a male that ends their education at the end of high school. (b) (3 pts) Construct a 95% confi dence interval for this estimate. The generic confidence interval formula is estimate± cv∗se(estimate).Here, the critical value is 1.96 (since we have a 95%) confidence interval. So the result is CI = 0.072 ± 1.96 ∗0.006 = [0.060,0.084]. (c) (8 pts) Suppose a college claimed that obtaining a four year college degree increased income on average by 30% over not studying further than high school. Would this be supported by the results of regression (1)? Yes, the claim is supported by the results. If we believe internal validity, then the 95% confidence interval for the effect of four years of college on earnings is four times the confidence interval we have, i..e from 24% to 33.6%. Alternatively, notice that a 30% effect over four years is