Output in two columns

Background: I am conducting some research on equity return characteristics in relation to firms' ESG scores. Specifically, I am trying to test if investing in firms that increased their ESG scores during the past 12 months yields a higher return. Question: Why do I get two columns in “effort(i,k)”? Code is below. Thanks in advance! esg=rand(100,100); meq=rand(100,100); ret=rand(100,100); T=length(esg); nAss=size(esg,2); numPort=3; k=1:(numPort-1); prctlVals=100*(k*(1/numPort)); breakeffort=NaN(T,numPort-1); % loop for ESG effort for i=12:T; effort(i,k)=esg(i,k)-esg(i-11,k); end

% loop for breakpoints for i=1:T; breakeffort(i,:)= prctile(effort(i,:),prctlVals); end;

bestret=NaN(T,nAss); bestmeq=NaN(T,nAss); worstret=NaN(T,nAss); worstmeq=NaN(T,nAss); besg=NaN(T,nAss); wesg=NaN(T,nAss);

% allocating according to best/worst-in-class 12m holding period for i=1:11:T-11; for k=1:nAss; if esg(i,k)>=breakeffort(i,numPort-1); bestret(i:i+11,k)=ret(i:i+11,k); bestmeq(i:i+11,k)=meq(i:i+11,k); besg(i:i+11,k)=esg(i:i+11,k); end if esg(i,k)<=breakeffort(i,1); worstret(i:i+11,k)=ret(i:i+11,k); worstmeq(i:i+11,k)=meq(i:i+11,k); wesg(i:i+11,k)=esg(i:i+11,k); end end end

% Weights wbest=NaN(T,nAss); wworst=NaN(T,nAss); for i=1:T; for k=1:nAss; wbest(i,k)=bestmeq(i,k)/nansum(bestmeq(i,:)); wworst(i,k)=worstmeq(i,k)/nansum(worstmeq(i,:)); end end

% Calculating weighted and equal portfolio return bestwret=NaN(T,nAss); worstwret=NaN(T,nAss); bestVW=NaN(T,1); bestEW=NaN(T,1); worstVW=NaN(T,1); worstEW=NaN(T,1); for i=1:T; for k=1:nAss; bestwret(i,k)=bestret(i,k)*wbest(i,k); worstwret(i,k)=worstret(i,k)*wworst(i,k); end bestVW(i)=nansum(bestwret(i,:)); worstVW(i)=nansum(worstwret(i,:)); bestEW(i)=nansum(bestret(i,:)/nnz(~isnan(bestret(i,:)))); worstEW(i)=nansum(worstret(i,:)/nnz(~isnan(worstret(i,:)))); end

[Rik]

@Oliver If the answer solves your question, why not hit the 'accept' button?

Flags should be used to attract the attention of admins.

Also, can you try to fix the formatting of your code? Something like what I did below:

esg=rand(100,100); meq=rand(100,100); ret=rand(100,100); T=length(esg); nAss=size(esg,2);
numPort=3; k=1:(numPort-1);
prctlVals=100*(k*(1/numPort));
breakeffort=NaN(T,numPort-1);
% loop for ESG effort
for i=12:T; effort(i,k)=esg(i,k)-esg(i-11,k); end
% loop for breakpoints
for i=1:T; breakeffort(i,:)= prctile(effort(i,:),prctlVals); end;
bestret=NaN(T,nAss); bestmeq=NaN(T,nAss); worstret=NaN(T,nAss); worstmeq=NaN(T,nAss); besg=NaN(T,nAss); wesg=NaN(T,nAss);
% allocating according to best/worst-in-class 12m holding period
for i=1:11:T-11;
    for k=1:nAss;
        if esg(i,k)>=breakeffort(i,numPort-1);
            bestret(i:i+11,k)=ret(i:i+11,k);
            bestmeq(i:i+11,k)=meq(i:i+11,k);
            besg(i:i+11,k)=esg(i:i+11,k);
        end
        if esg(i,k)<=breakeffort(i,1);
            worstret(i:i+11,k)=ret(i:i+11,k);
            worstmeq(i:i+11,k)=meq(i:i+11,k);
            wesg(i:i+11,k)=esg(i:i+11,k);
        end
    end
end
% Weights
wbest=NaN(T,nAss); wworst=NaN(T,nAss);
for i=1:T;
    for k=1:nAss;
        wbest(i,k)=bestmeq(i,k)/nansum(bestmeq(i,:));
        wworst(i,k)=worstmeq(i,k)/nansum(worstmeq(i,:));
    end
end
% Calculating weighted and equal portfolio return
bestwret=NaN(T,nAss); worstwret=NaN(T,nAss);
bestVW=NaN(T,1); bestEW=NaN(T,1);
worstVW=NaN(T,1); worstEW=NaN(T,1);
for i=1:T;
    for k=1:nAss;
        bestwret(i,k)=bestret(i,k)*wbest(i,k);
        worstwret(i,k)=worstret(i,k)*wworst(i,k);
    end
    bestVW(i)=nansum(bestwret(i,:));
    worstVW(i)=nansum(worstwret(i,:));
    bestEW(i)=nansum(bestret(i,:)/nnz(~isnan(bestret(i,:))));
    worstEW(i)=nansum(worstret(i,:)/nnz(~isnan(worstret(i,:))));
end

Answer 1

[Cris LaPierre]

Because k=1:(numPort-1) and numPort=3