In our discussion of table 14.3, we note that there are provocative results regarding race and ethnicity. Table 14.6 presents another stratification of the pooled regression in table 14.3. In this case, the sample for the first regression consists of only nonminorities, or whites. The sample for the second regression consists of all minorities, regardless of racial or ethnic identity.

(a) Why does the first regression omit all the dummy variables for racial or ethnic identity?

(b) How many parameters are estimated by the stratified regressions of table€14.6? How many by the pooled regression of table 14.3?

(c) The pooled regression of table 14.3 is a restricted version of the model comprising the stratified regressions of table 14.6. How many restrictions does the pooled regression embody? List them.

(d) Use the test of equation (14.21) to compare the pooled regression of table€14.3 to the stratified regressions of table 14.6. What is the

critical value for this test at 5% significance? Which model is superior, the pooled or the stratified?

(e) How do the returns to human capital compare for whites and minorities?

Transcribed Image Text:

Explanatory variable TABLE 14.6 Earnings regressions stratified by minority status, sample from section 7.6 Regression for nonminorities Regression for minorities Intercept -40,589 -8,220.7 (38.79) (13.97) Sex -25,003 -11,910 (74.29) (58.95) Years of schooling 6,076.4 2,172.1 (98.81) (84.03) Age 194.28 252.65 (13.55) (29.08) Hispanic 1,273.6 (3.83) American Indian or Alaskan Native -1,116.6 (1.64) Asian 6,709.7 (18.50) Native Hawaiian, other Pacific Islander 2,944.6 (2.59) R Adjusted R F-statistic p-value for F-statistic Sum of squared errors Number of observations .1530 .1378 .1529 .1377 5,405.5 2,049.2