Bootstrapping matrixes jointly

I have a question about bootstrapping. I have to generate bootstrapped samples in Matlab. I have found one way to do it. For example I have matrix A.

A = [ 1 2 3; 4 5 6; 7 8 9]

To obtain one bootstrapped sample I do :

s=size(A); % I calculate the dimensions of A

out=A(ceil(s(1)*s(2)*rand(s(1),s(2),1))); % then I randomly draw the observations from A to construct the bootstrapped sample.

It seems to work pretty well. But to solve my problem, I have to draw the samples from several matrixes jointly.

What I mean is: For example, I have two matrixes.

A = [1 2 ; 3 4] and B = [5 6; 7 8]

The dimensions of the matrixes are firms and time periods ( 2 firms, 2 time periods here). I have the data on characteristics ( A and B) for these firms, which are presented by matrixes above.

Here the columns are time periods and rows are firms.

So 2 (element (1,2)) is the characteristic A of the first firm in second period and 6 (element(1,2)) is the corresponding characteristic B.

When I do bootstrapping, and draw the observations from these matrixes to construct other two samples, I need that they are drawn jointly. So if I randomly draw 2 from the A matrix to form bootstrapped A matrix, I should draw 6 from the B matrix to form bootstrapped B matrix at the same time. So that each element of a bootstrapped matrix of A and bootstrapped matrix of B belonged to the same firm and time period. Is there any possibility to do it? Thanks a lot for any help in advance!

Answer 1

[Andrei Bobrov]

indata = [3 4 1 2 2 3];
[ii ii] = ismember([3 4 1 2 2 3],A);
outdata = B(ii);

[Lara]

Andrei, thanks a lot!!! It is a really easy solution to the question, that seemed so complicated to me! Thanks a lot again!!!