Now, I realise that my post is not going to make any ground-breaking contribution anymore this late in the process ... and after the neat bit-wise operation solution suggested by almut there is little to be added. Still, I would like to assert that this is a neat problem and offer my humble solution which may at least win a prize for conciseness (and if not - might at least get considered in the Perl obfuscation section ;-). The noteworthy feature about this implementation is that it accomplishes the entire analysis process in a single regex!

Here it goes:

#! /usr/bin/perl/ -w

use strict;

my @strings = qw ( cooling rooting hooting looking doormat );

our @common_letters;

my $reference = shift @strings;

() = $reference =~ /(.)(?{    
                my $letter = $1; 
                my $position = $-[0]; 
                   my $bolean = 1; 
                  
                for ( @strings ) 
                { 
                    if ( substr( $_, $position, 1 ) ne $letter )
                    { 
                        $bolean = 0; 
                        last
                    } 
                }
                
                $common_letters[ $position ] = $letter if ( $bolean );
+ 
            })/gx;

print ( $common_letters[ $_ ] || "-" ) for ( 0 .. length $reference );
[download]

P. S.: It certainly will challenge on the regex-side as you had intended originally ;-) Lemme know what you think!