% hw3.m - suggested solutions for hw3

% load data - this file has the header information removed
load ruud/wage_nohead.dat

% isolate variables 
wage = wage_nohead(:,1);
educ = wage_nohead(:,5);
exper = wage_nohead(:,6);
age = wage_nohead(:,7);
fe = wage_nohead(:,2);
nw = wage_nohead(:,3);
un = wage_nohead(:,4);

% Table 1.1
Z = [wage educ exper age fe nw un];
tab11 = [mean(Z)' std(Z)' min(Z)' max(Z)']

% Table 1.2
selectmen = fe==0;
tab12men = [mean(wage(selectmen)) std(wage(selectmen)) sum(selectmen)];
tab12women = [mean(wage(~selectmen)) std(wage(~selectmen)) sum(~selectmen)];
diffmen = tab12men(1) - tab12women(1);

selectwhite = nw==0;
tab12white = [mean(wage(selectwhite)) std(wage(selectwhite)) sum(selectwhite)];
tab12nonwhite = [mean(wage(~selectwhite)) std(wage(~selectwhite)) sum(~selectwhite)];
diffwhite = tab12white(1) - tab12nonwhite(1);

selectun = un==1;
tab12un = [mean(wage(selectun)) std(wage(selectun)) sum(selectun)];
tab12nonun = [mean(wage(~selectun)) std(wage(~selectun)) sum(~selectun)];
diffun = tab12un(1) - tab12nonun(1);

% note i put differences in means at end for a little simplicity
tab12 = [tab12men 0 
	 tab12women diffmen
	 tab12white 0 
	 tab12nonwhite diffwhite
	 tab12un 0 
	 tab12nonun diffun]
	
% Table 1.3
menwhite = (fe==0)&(nw==0);
mennw = (fe==0)&(nw==1);
womenwhite = (fe==1)&(nw==0);
womennw = (fe==1)&(nw==1);

tab13 = [sum(menwhite(un==0)) sum(menwhite(un==1))
	 sum(mennw(un==0)) sum(mennw(un==1))
	 sum(womenwhite(un==0)) sum(womenwhite(un==1))
	 sum(womennw(un==0)) sum(womennw(un==1))]

% Table 1.4
edexp = [educ exper];

tab14 = [ mean(edexp(fe==0,:))
	  mean(edexp(fe==1,:))
	  mean(edexp(nw==0,:))
	  mean(edexp(nw==1,:))
	  mean(edexp(un==1,:))
	  mean(edexp(un==0,:))]

% Figure 1.1
plot(educ,wage,'.')

% Figure 1.2
plot(exper,wage,'.')

% Table 1.8
y = log(wage);
n=length(y);

X1 = ones(n,1);
X2 = [X1 fe];
X3 = [X1 fe nw un];
X4 = [X1 fe nw un educ exper];
X5 = [X1 fe nw un educ exper exper.^2];

b1 = inv(X1'*X1)*X1'*y;
b2 = inv(X2'*X2)*X2'*y;
b3 = inv(X3'*X3)*X3'*y;
b4 = inv(X4'*X4)*X4'*y;
b5 = inv(X5'*X5)*X5'*y;

tab18 = [ [b1; zeros(6,1)] [b2; zeros(5,1)] [b3; zeros(3,1)] [b4; zeros(1,1)] b5]


% end hw3.m