G'day hello_beginner,

I think you have an overly complex data structure: a Hash of Hashes of Arrays (HoHoA). I've used a Hash of Arrays (HoA): the keys are col1 values (e.g. A1, A2, etc.); the values are arrayrefs containing all the elements returned by split.

[In the code below: UNIQ refers to your A1, A2 & A3; CODE refers to your NY, SG, AUS, etc. codes.]

#!/usr/bin/env perl -l

use strict;
use warnings;
use autodie;

use constant {
    UNIQ   => 0,
    CODE   => 2,
    JOINER => ':',
};

my %data;

while (<DATA>) {
    my @cols = split;

    if (exists $data{$cols[UNIQ]}) {
        $data{$cols[UNIQ]}->[CODE] .= JOINER . $cols[CODE];
    }   
    else {
        $data{$cols[UNIQ]} = [ @cols ];
    }   
}   

print "@{$data{$_}}" for sort keys %data;

__DATA__
A1 text1 NY Jan 01
A2 text2 LN Feb 02
A3 text3 SG Mar 03
A2 text2 NY Feb 02
A1 text1 SG Jan 01
A1 text1 AUS Jan 01
[download]

Which outputs:

A1 text1 NY:SG:AUS Jan 01
A2 text2 LN:NY Feb 02
A3 text3 SG Mar 03
[download]

See also: "perldsc -- Perl Data Structures Cookbook".