Here is an awk script, add_reference.awk that I wrote for this task. It adds a new sample called "Reference" that will be homozygous 0/0 at each locus.

#! /usr/bin/awk -f 

BEGIN {
    FS = "\t"
    OFS = FS
}

#print the header info as-is
/^##/ {
    print
    next
}

#add sample named "Reference" to the list of samples
/^#CHROM/ {
    print $0"\tReference"
    next
}

#add homozygous reference allele to every locus.
{
    print $0"\t0/0:0,0,255:100:0:100,0:1,0,0:150"
}

For gzipped VCF files, I run it like this:

zcat my_file.vcf.gz | add_reference.awk | bgzip > my_file_with_reference.vcf.gz 

You will probably have to change it to match the format of your VCF. My fields are in the format "GT:PL:DP:SP:AD:GP:GQ" so I went with an entry of "0/0:0,0,255:100:0:100,0:1,0,0:150" so that it would pass my downstream filters, but you can change it to match whatever fields you have in your VCF.

It assumes that all loci are unphased diploid loci and that the reference sample is not heterozygous at any site. Caveat: ignoring heterozygous sites in the ref sample may cause inaccuracies in certain downstream analyses depending on what it is used for.