* Examples of dummy independent variables 
Data " cps_msa2013 : 2013 Current population survey for 3 MSA
3 MSA : NY, LA, CHICAGO
*/  

use https://bigblue.depaul.edu/jlee141/econdata/cps_data/cps_msa2013

gen lhrwage = log(hrwage) 
label var lhrwage "Log of Nominal Hourly Wage"
/* 1. Creating continuous education variable 
Number of Years of Education */

// tablulate the educ92 
tab educ92
// tablulate the educ92 without label 
tab educ92, nolabel 

gen educyr = 0 
replace educyr =  1 if educ92 == 1 
replace educyr =  4 if educ92 == 2 
replace educyr =  6 if educ92 == 3 
replace educyr =  8 if educ92 == 4 
replace educyr =  9 if educ92 == 5 
replace educyr =  10 if educ92 == 6 
replace educyr =  11 if educ92 == 7 
replace educyr =  12 if educ92 == 8  |  educ92 == 9  
replace educyr =  14 if educ92 == 10 | educ92 == 11 | educ92 == 12 
replace educyr =  16 if educ92 == 13 
replace educyr =  18 if educ92 == 14  
replace educyr =  20 if educ92 == 15 
replace educyr =  21 if educ92 == 16 

/* 2. Creating simple dummy variable 
Here are some examples to create simple dummy variables */

gen young = 0 
replace young = 1 if age < 25 

gen retired_age = age > 65 
gen retired_man = (age > 65 & female == 0 )
			\end{lstlisting}
		\end{column}
		
		\begin{column}{0.5\textwidth}
			\begin{lstlisting}[basicstyle=\tiny]
/*  3. Creating multiple dummy variables  
Here are some examples to create multiple dummy variables 
wbho (white, black, hispanic, others) 
-> generate four dummy variables (race1, race2, race3, race4) 
*/

tab wbho 
tab wbho, gen(race)
sum race1-race4

/* 4. Regression using dummy variables aid factor variables*/

regr lhrwage educyr female
regr lhrwage educyr race2-race4  
regr lhrwage educyr i.wbho  
predict fit1 
tw (line fit1 educyr if race1 == 1 ) ///
(line fit1 educyr if race2 == 1 ) ///
(line fit1 educyr if race3 == 1 ) ///
(line fit1 educyr if race4 == 1 ) ///
legend(lab(1 "Fitted for White") lab(2 "Fitted for Black")) ///
legend(lab(3 "Fitted for Hispanic") lab(4 "Fitted for Others")) ///
ytitle("Fitted log of hourly wage")  ///
xtitle("years of education")

gen female_educyr =  female*educyr 
regr lhrwage female  educyr female_educyr
regr lhrwage i.female##c.educyr

predict fit2
tw (line fit2 educyr if female == 0)   ///
(line fit2 educyr if female == 1),  ///
legend(lab(1 "Fitted for Male") lab(2 "Fitted for Female")) ///
ytitle("Fitted log of hourly wage")  ///
xtitle("years of education")


/* Functional Forms in Regression Model */
/* House Price in Chicago (zip == 60629) */
clear all
use https://bigblue.depaul.edu/jlee141/econdata/hp2013chi.dta
keep if zip == 60629
gen log_hprice = log(hprice*1000)
gen log_sqft = log(sqft)
gen price = hprice*1000

/* Linear Model */
regr price sqft

/* Log-Linear Model */
regr log_hprice sqft

/* Linear-Log Model */
regr price log_sqft

/* Double Log Model */
regr log_hprice log_sqft


/* Nonlinear Relationship between wage and eductions */
clear all
use https://bigblue.depaul.edu/jlee141/econdata/wage1
tw ( scatter wage educ) (lfit wage educ) (qfit wage educ)
regr wage educ 
regr wage educ educ_sq