I would script this sort of thing. I'm sure there is an R way to do it, but I tend to use python for this sort of thing. There might be a clever way to do it in awk, but I don't know it.

If you are happy with installing things, cgatapps has code for handling GTF files. This sort of thing could be achieved with:

from cgat import GTF
from cgatcore import iotools

input_file = iotools.open_file("my_transcripts.gtf.gz")

gtf_entries = gtf.iterator(input_file)

transcripts = gtf.transcript_iterator(gtf_enteries)

for transcript in transcripts:
    exons = [e for e in transcript if e.feature == "exon"]

    if len(exons) == 0:
        continue

    if transcript[0].strand == "-":
        exons = sorted(exons, key=lambda x: -1*x.end)
    else:
        exons = sorted(exons, key=lambda x: x.start

    last_exon = exons[-1]

    print (str(last_exon) + "\n")

The same thing can be achieved with pysam with a little more effort if cgat is too big an installation.