Problems - Ch10

Create the following simulated data.

1. n = 100 (100 cross sectional units)

2. T = 10 (10 time periods)

3. a_i ~ N(0, 3), draw one a for each 100 units

4. b_t ~ U[0,1], draw one b for each 10 time periods

5. e ~ N(0,1), draw one for each 100 units and 10 time periods (1000 all together)

6. g = 0.5

7. y_it = a_i + b_t + g*x_it + e_it, x_it = 1 if T = 10 and i = 1, 0 if T < 10, i > 1.

8. Run OLS and first differences.

9. Use Difference in Difference and fixed effect estimation to estimate the equation in (7). Explain why these estimates are or are not equal to the true value.

Create a bootstrap version of did estimator. Use it to estimate the simulated data and the Card and Krueger data.

Read Card and Krueger. Do you think it is reasonable to assume that PA and NJ have the same time effect? Explain.

Create a bootstrap version of the fixed effect estimator. How do the standard errors vary across the different estimators on the simulated data? How does the bootstrap standard errors differ from the original estimator? Explain.

Do you think workers were made better off or worse off under George W. Bush’s minimum wage increase? Explain.