The -scale takes an integer or a float, so you need to do the calculation of an appropriate factor yourself. It is nothing more than a factor that all bedGraph values are multiplied by. A simple workaround would be to use a scaling factor that is directly derived by the bam file you used:

TmpScale=$(bc <<< "scale=6;1000000/$(samtools view -f 0 -c mybamfile.bam)")
bedtools genomecov -ibam mybamfile.bam -bg -scale $TmpScale -g mm10.chrom.sizes > myBedGraphfile.BedGraph

The TmpScale counts the number of reads in the BAM and multiplies it by 1mio, rounded to 6 digits. This is then the factor that all values in the bedGraph are multiplied by. Depending on your needs, you can change the flag in SAMtools view, lets say only count the forward read in case of paired-end data (that would then be -f 1).

You may also try bamCoverage (http://deeptools.readthedocs.io/en/latest/content/tools/bamCoverage.html#usage-examples-for-rna-seq), which produces RPKM values from a BAM File in bedGraph format. It is designed for both ChIP- and RNA-seq. I have used it to good effect for ChIP-seq, but not yet tried it for RNA-seq.