HANDOUT 3

Question 1

Why do we use dummy variables?

Answer 1

For categorical variables

Previously have assumed all X variables continuous

Question 2

2 types of categorical variables & egs

Answer 2

1. Ordered e.g. qualifications

2. Unordered e.g. gender

Question 3

If we have K categories, how many variables include?

Answer 3

K - 1

The left out one is the default

Question 4

Additive dummy variables allow

Answer 4

An intercept shift

Question 5

The coefficient on female if

ln(wages) = alpha + B1school + B2female + Ei

Answer 5

B2 = the expected proportionate rise in wages for a female relative to a male at a given level of school
Or 100(e^B2 - 1) exact % change in wages

Question 6

Multiplicative dummy variables allow

Answer 6

Rotation of the line

e.g. gender effect on return (in terms of wages) to schooling

Question 7

ln(w) = alpha + B1school + B2female + B3(female x school) + Ei. Interpret B2.

Answer 7

B2 = the expected proportionate increase in wages for a female c.f. male when schooling=0

Question 8

ln(w) = alpha + B1school + B2female + B3(female x school) + Ei. Interpret B3.

Answer 8

The additional expected proportionate rise in wages from an extra year of school for a female c.f. male.

Question 9

Interactive dummy variables allow

Answer 9

multiply 2 categorical variables together

difference in difference

Question 10

ln(w) = B1 + B2female + B3Nwh + B4(female x Nwh) + Ei. Interpret B2.

Answer 10

B2 = the gender effect on wages for whites

Question 11

ln(w) = B1 + B2female + B3Nwh + B4(female x Nwh) + Ei. Interpret B1.

Answer 11

B1 = expected wage for a white male

Question 12

ln(w) = B1 + B2female + B3Nwh + B4(female x Nwh) + Ei. Interpret B3.

Answer 12

B3 = the ethnicity effect for males

Question 13

ln(w) = B1 + B2female + B3Nwh + B4(female x Nwh) + Ei. Interpret B4.

Answer 13

Difference in difference
B4 = the additional female vs male effect on wages for Nwh vs whites
OR B4 = the additional Nwh vs white effect on wages for females vs males