Error using regstats (RESPONSES and DATA must have the same number of rows)

Hello!

I must deliver the code for an Econometrics exam and everything is running smoothly except for the following error:

Error using regstats

RESPONSES and DATA must have the same number of rows.

Error in FinancialEconometrics_Homework (line 515)

out=regstats(PInde(:,i),F);

that refers to this part of the code:

    out=regstats(PInde(:,i),F);
    APToutBtM(i,1:6)=out.beta';
    APToutBtM(i,7:12)=out.tstat.t';
    APToutBtM(i,13)=out.adjrsquare;
    APTresBtM(:,i)=out.r;

I don't know how to solve the problem.

PInde is a 97x25 double matrix and F is a 403x5 double matrix.

Could you please help me?

Thank you

Answer 1

[Walter Roberson]

regstats

regstats(y,X,model) performs a multilinear regression of the responses in y on the predictors in X. X is an n-by-p matrix of p predictors at each of n observations. y is an n-by-1 vector of observed responses.

So the number of rows of X and y must be the same.

You are using PIndel(:,i) as the y, which will be 97 x 1

You are using F as the X, which will be 403 x 5.

The number of rows of 97 x 1 does not match the number of rows of 403 x 5, so this will fail.

[FEDERICA]

Thank you for your answer Walter,

actually I undestood the problem.

Do you think there's something wrong in creating the matrices?

I used this for PInde

% compute portfolio excess returns
PBtMe=PbyBtM-rf;
PInde=PbyInd-rf;

% CAPM estimation 
for i=1:25 
    y = PInde(:, i);
    intercept = ones(size(Me)); 
    intercept = intercept(:,1);
    X = [intercept Me];
    beta_ols = inv(X' * X) * X' * y;
    fitted_y = X * beta_ols;
    e = y - fitted_y;
    sigma2 = mean( e.^2 );
    se_beta_ols = sigma2 * inv(X' * X); 
    t_stats(i, :) = beta_ols ./ diag(se_beta_ols); 
end

% Factor model 
% CAPM estimation + Multifactor model estimation
PInde = PInde(36:end, :);
Me = Me(36:end, :);
T = 97;

and this for F

first this

% load data
PbyBtM=readmatrix('Download Data - INDEX_US_S&P US_SPX.csv');
PbyInd=readmatrix('dataset.csv');
FFF = readmatrix('North_America_5_Factors.csv');

% remove dates
PbyBtM=PbyBtM(121:end,6);
PbyInd=PbyInd(271:end,2:end);
FFF=FFF(:,2:end);

and then

% Select market excess returns and risk-free, and all factors
Me=table2array(Mkt_return);
Me=Me(121:end);
rf=0;
F=FFF(37:end,1:end-1);

[Walter Roberson]

I would need to know the size of the various matrices.

[FEDERICA]

PInde is 97x25

PbyInd is 132x25

F is 403x5

FFF is 439x6

dataset.csv to create PbyInd is 403x26 with first column dates and first row tags

Download Data - INDEX_US_S&P US_SPX.csv to create FFF is 253x6 with first column dates and fist row tags

[Walter Roberson]

    intercept = ones(size(Me)); 
    intercept = intercept(:,1);

What is the point of that? Why not just

    intercept = ones(size(Me,1),1);

[FEDERICA]

You're right, thank you for your suggestion! I'm really sorry, I'm still learning to write codes in MatLab and often I just make the same things my professor does during the lectures.

Answer 2

[FEDERICA]

Problem solved! Thank you for your help :)