If you use Text::CSV_XS' csv function with the filter attribute, you don't also keep the CSV data and stream the required content right into the arrays:

use 5.14.2;
use warnings;

use Text::CSV_XS qw(csv);

my (@a, @b);
my $aoa = csv (
    in      => *DATA,
    headers => "skip",
    filter  => {
        1 => sub { push @a, $_; 1 },
        2 => sub { push @b, $_; 0 },
        });

{   local $" = ", ";
    say "var secid = [@a]";
    say "var Initial_shares = [@b]";
    }

__END__
secid,Initial_shares
002826,3777
0028262,3777
0028262,3777
0028262,3777
0028262,3777
0028262,3777
[download]

→

var secid = [002826 0028262 0028262 0028262 0028262 0028262]
var Initial_shares = [3777 3777 3777 3777 3777 3777]
[download]

The entries in the filter will be used in ascending order. All but the last should return a true value. The last should return a false value to reject the record: You already stored all required data in the two arrays.

update: Alternatively, you can put everything in a hash:

my %h;
my $aoa = csv (
    in      => *DATA,
    headers => "auto",
    filter  => { 1 => sub { for my $f (keys %_) { push @{$h{$f}}, $_{$
+f} }; 0 }},
    );
[download]

→

{   Initial_shares   => [
        3777,
        3777,
        3777,
        3777,
        3777,
        3777
        ],
    secid            => [
        '002826',
        '0028262',
        '0028262',
        '0028262',
        '0028262',
        '0028262'
        ]
    }
[download]