1) [Difference-in-Difference] The government has designed a program to train low-skilled workers in the middle of an economic recession. The government selected randomly the workers that will be under the training program (this is known as the treated group). The non-treated workers are considered the control group. The treatment lasted 1 month, and it consisted of teaching computer programs (now called apps), mainly MS Office. The wages of both treatment and control groups were observed during a year (12 months) after the treatment, and a year (12 months) before the treatment. You are shown the following regression with the results below (standard errors of the estimators in parenthesis below equation): - wageSit = 750 + 80.60 x treatit – 20.55 x postit + 230.90 x treatit x postit (50.4) (60.80) (9.50) (40.50) Where wages is monthly wages, in dollars. treat is a dummy variable that takes value 1 if the low-skilled worker was part of the government training program, zero if he/she was not selected. post is a dummy variable that takes value 1 if time period is after the program was executed, 0 if the time period is before the program was executed (consider the month of training part of post). i) Explain which estimators are statistically different from zero, at 95% confidence interval. ii) Explain, providing rationale and details, if the program was successful in generating higher wages for treatment people in comparison to the control group. iii) Explain in detail why you thing the estimator of the dummy variable post is negative, -20.55.
