Your code is incredibly long for something that looks pretty simple, if I understand it correctly.

You haven't described your files, but I assume that the three first columns, put together, define a unique identifier for the tasks you want to process.

You first need to read the first file and store in a hash the 3-day gratuity left-overs. Something like this (untested):

my %free_days_left;
while (<$file1>) {
     chomp;
     my @fields = split /\s+/, $_;
     my $key = join "-", @fields[0..2];
     $free_days_left{$key} = $fields[5];
}
[download]

This could be made more compact, but I prefer just to explain the process as clearly as possible.

Now, the %free_days_left hash contains the free days left. You just need to read through the second file, compute the number of days, subtract the free days and output the result (again, a general idea, not tested).

while (<$file2>) {
     chomp;
     my @fields = split /\s+/, $_;
     my $key = join "-", @fields[0..2];
     my ($start_date, $end_date) = @fields[5,6];
     my $start_day = (split /,/, $start_date)[2];
     my $end_day = (split /,/, $end_date)[2];
     my $duration = $end_day - $start_day + 1:
     my $corrected_duration = exists $free_days_left{$key}? $duration 
+- $free_days_left{$key} : $duration;
     print $file3 join " ", $_, $duration, $corrected_duration, $corre
+cted_duration * $unit_price, "\n";
}
[download]

OK, the formating of the output file may need a bit more work, but I guess you've got the basic idea.