FPKM vs gene length

[Autotroph]

Hi,

I used the program RSEM to calculate the expression level of genes in my RNA-seq data (Illumina paired-end) from human cell lines.

Now i see that the FPKM and gene length are positively correlated ("kendall" rank) with a r of 0.5. Is this expected?

Can you please explain why the FPKM is correlated with gene length?

Thanks,