Main files -- empirical analysis

sample_period_1.dta, sample_period_2.dta

Stata data files from which all the analysis is made
Appendix A in the text explains how I built this matched
sample from its original CPS files step by step

paper.do

Stata execution file that delivers five set of
results: (i) the empirical facts about hours and wages (table 2.1); 
(ii) Heckman two step estimation (shown in appendix B); 
(iii) statistics about several variables that are needed to get the
empirical transition probabilities (presented in section 2, and also used to
solve the model in section 4) -- average group age, education, occupation,
shares and median wages for the discretized grid used in the model, and
finally the percentage of individuals that enter the market in each of the
wage groups (newborns in the model); (iv) robustness of the
empirical facts by age, skills, occupation (tables A.3, A.4, A.5, A.6, A.7
in the paper); (v) composition effects raw statistics (individual
sub-group averages, shares, group averages), which following what is
described in the text leads to table 2.2.

run line 3 or line 4 depending on which time period you want
to see the results of (1st or the 2nd sample period); then run the code
throughout and follow the comments written on the file. That is: 

- If sample period 1: run line 3, all part 1, part 2 selecting
the corresponding line for this sample period as indicated in the comments,
all part 3, all part 4, part 6 selecting the corresponding line for this
sample period as indicated in the comments, all part 7, and all part 8.

- If sample period 2: run line 4, all part 1, part 2 selecting
the corresponding line for this sample period as indicated in the comments,
all part 3, all part 5, part 6 selecting the corresponding line for this
sample period as indicated in the comments, all part 7, and all part 8.

prob_values_ME.m

Matlab code that gives the probability values and
marginal effect of hours in these probabilities as presented in tables 2.3
and 2.4

For each sample period (as indicated in the file) follow two
steps. First, run the code start to finish; that yields the probability
values presented in table 2.3. Second, run line 88 (if 1st sample period) or
line 438 (if second), and from then on to the end to obtain the new
probability values when individuals work 5 more hours per week. The
difference between the new set of results and the first provide the marginal
effects presented in table 2.4. 


Main files -- model

SS2.m, SS1.m

Matlab program files that give solutions in table 4.2 (model
results in the paper)
Just run the code as it is; results show up at the end

SS2_taxes1.m, SS2_transitions1, SS2_taxes1.m

3 Matlab files that give solutions for each of the
counterfactual exercises presented in table 4.3
Just run the code as it is; results show up at the end


Additional files

Tables_2_5_and_2_6.m

Matlab code that gives the probability values and marginal
effect of hours in the transition probabilities using the PSID (presented in
tables 2.5 and 2.6)

Where indicates 77-79 (first sample period): run the code
start to finish; that yields the probability values presented in table 2.5.
To obtain the values in table 2.6 follow the same two steps described above
for the CPS in each sample period. First, run code start to finish; second,
run line 447 (if 1st sample period) or line 92 (if second), and from then on
to the end to obtain the new probability values when individuals work 5 more
hours per week. The difference between the new set of results and the first
provide the marginal effects for each sample period. The difference in the
marginal effects between the two sample periods is presented in table 2.6.

Table_2_7.m

Matlab code that gives the probability values and
marginal effect of hours in the transition probabilities using the PSID
using only the hourly paid workers (presented in table 2.7)

For each sample period as indicated in the file follow two
steps. First, run code start to finish; that yields the probability values.
Second, run line 444 (if 1st sample period) or line 93 (if second), and from
then on to the end to obtain the new probability values when individuals
work 5 more hours per week. The difference between the new set of results
and the first provide the marginal effects presented in table 2.7. 

Figure2_CPS.m and figure2_PSID.m

Matlab code that gives the yearly dollar increase
with one extra weekly hours over time in the CPS and the PSID (results
summarized in figure 2)
Simply run each of the codes for the corresponding dataset;
to get figure 2, take the difference between the result for the 5th and 1st
quintiles in each sample period.