Problems - Ch6

  1. Create the simulated data presented at the start of Chapter 6.

    1. Create a bootstrap OLS estimator where the data is restricted (x < 0.6).

    2. Create a bootstrap probit estimator.

    3. Create a bootstrap Tobit estimator.

    4. Compare the standard errors around the parameter estimates for each case. Which is the most efficient? That is, which has the smallest variation around the parameter estimates?

  2. Download data from NLSY97. You need to create an account here: https://www.nlsinfo.org/investigator/pages/register.jsp. After creating an account, you can select “NLSY 97” from “Select the study you want to work with”.

    1. Download variables similar to the ones used by David Card in his 1993 paper using the NLSYM 66 (See Chapter 2).

    2. Create a summary table of the variables.

    3. Replicate Table 2 (column 1) of Card’s paper, to the extent possible.

    4. Using number of siblings as an instrument, estimate an IV model.

    5. Using the bounds approach estimate returns to schooling.

    6. Estimate a Heckman selection model.

    7. Given the different estimates, which do you find the most persuasive and why?